Browse Topic: Fault detection

Items (514)
Given the nonlinear and nonstationary nature of fault signals in rotating machinery, this paper proposes a fault feature extraction and identification method based on refined composite multiscale dispersion entropy (RCMDE) and Gath–Geva (GG) clustering. In the proposed approach, RCMDE is first employed to extract feature sets from the vibration signals. Principal component analysis (PCA) is then applied for dimensionality reduction, and the resulting low-dimensional features are fed into the GG clustering algorithm to achieve fault diagnosis. The effectiveness of the proposed method is validated on a typical bearing fault dataset, with classification performance evaluated using the classification coefficient and partition entropy. Experimental results demonstrate that the proposed RCMDE-GG achieves superior performance in fault feature extraction and exhibits excellent classification capability.
Wang, Guohui, Wang, Haiming, Wang, Tiecheng, Liu, Yongqiang
As a core carrier of urban underground infrastructure, underground utility tunnels have an increasingly urgent demand for AAV swarm technology in operation and maintenance tasks such as internal pipeline inspection and fault diagnosis. However, the special environment of utility tunnels, characterized by GPS signal loss, narrow and enclosed space, and complex types of obstacles (including fixed pipeline supports and temporary construction components), poses severe challenges to AAV swarm operations. In this scenario, while maintaining the predetermined formation configuration, AAV swarms are required to complete three tasks simultaneously: accurately avoiding complex static obstacles in the tunnels, real-time evading other AAVs within the swarm to prevent intra-formation collisions, and strictly maintaining a safe distance from the tunnel walls. The superposition of these three constraints and the narrow operating space greatly increases the difficulty of swarm control. Traditional control methods often suffer from problems such as formation collapse, jerky obstacle avoidance, or wall scraping. To address the above pain points, this paper proposes an AAV swarm Model Predictive Control (MPC) method integrated with the Adaptive Perceptual Cauchy Mutation Pigeon-Inspired Optimization (APCMPIO) algorithm. This method optimizes the receding horizon parameters and constraint weights of model predictive control through the APCMPIO algorithm to achieve rapid response to complex dynamic scenarios, and embeds a dynamic safe distance threshold mechanism to balance formation integrity and individual safety. A simulation experiment was built for a utility tunnel inspection scenario with 6 AAVs. It significantly improves the operational adaptability and robustness of AAV swarms in constrained utility tunnel environments, providing reliable technical support for the swarm-based operation and maintenance of underground infrastructure.
Li, Tian-Qi, Ma, Lei, Liu, Yan-Jun, Li, Shu, Li, Dapeng, Liu, Lei
In high-end manufacturing and aerospace equipment, reliable identification of welding defects is the key premise for intelligent fault diagnosis and structural health management because undetected or misclassified defects may lead to serious safety risks. However, the actual welding inspection scene is usually subject to some problems, such as limited training data, serious imbalance of categories, and low contrast of defect patterns, which significantly reduce the robustness of traditional vision-based inspection methods. In order to meet these challenges, this paper proposes a tail-sensing-enhanced welding defect detection framework based on YOLOv8, which is specifically aimed at the distribution of common long tail defects in industrial inspection tasks. On the premise of not changing the original network architecture, this method systematically enhances the learning ability of underrepresented defect categories through collaborative optimization in data, training, and loss levels. The specific measures include tailperceived data enhancement, convergence-oriented training strategy, and loss re-weighting of classification priority. The experimental results on multi-class welding defect datasets show that the proposed framework significantly reduces the false alarm rate and inter-class confusion of small samples and low contrast defects, and improves the reliability and robustness of the overall diagnosis. These results show that the proposed method provides an effective and practical solution for intelligent welding fault diagnosis and has great potential to be integrated into aerospace manufacturing quality inspection and health monitoring systems.
Fan, Yuhang, Zhang, Jiyu, Huang, Xingyu
Rapid electrification of the automotive industry has significantly increased the functional complexity and safety expectations of Battery Management Systems (BMS) used in electric vehicles. While Model-in-the-Loop (MIL) testing remains a standard practice for early-stage verification of control algorithms, it is inherently limited in its ability to represent real time execution effects and physical hardware interactions. Critical BMS functions, including protection timing validation, sensor interface behavior, contactor actuation logic and fault detection mechanisms are strongly influenced by hardware characteristics such as signal conditioning, analog to digital conversion, execution latency, and nonlinear component behavior, which cannot be accurately captured in a software only environment. This paper presents the application of Hardware-in-the-Loop (HIL) testing as an essential validation methodology for safety critical BMS functions by interfacing the production intent BMS controller with a real time simulated battery environment. The proposed HIL setup enables realistic interaction between embedded software and physical I/O interfaces, allowing systematic evaluation of BMS behavior under normal operating conditions as well as controlled electrical, thermal, and communication fault scenarios. Key validation aspects include protection of response timing, robustness of fault handling logic, and system behavior under transient and abnormal conditions. The study demonstrates that HIL testing effectively bridges the validation gap between MIL simulations and vehicle level testing by enabling early identification of hardware–software integration issues. This approach enhances confidence in functional safety readiness, reduces development risk, and supports compliance with established automotive development workflows. The presented methodology aligns with established automotive development practices and provides a practical framework for robust BMS validation at bench level prior to vehicle level.
Patowary, Suman, Kumar, Aashish, Choudhary, Abhilash, Kumar, Abhishek, Bora, Abhijit
Gas path faults in aero-engines are a major factor affecting flight safety and maintenance costs. To achieve the transition from time-based maintenance to condition-based maintenance and enhance fault prediction and health management capabilities, this paper conducts research on intelligent diagnostic methods based on performance parameters for engine gas path components. Firstly, a component-level nonlinear steady-state model of a dual-spool turbofan engine is established, employing a variable specific heat capacity method to improve computational accuracy and provide a healthy baseline for diagnosis. Secondly, addressing the strong nonlinearity and measurement underdeterminacy in fault diagnosis, a diagnostic framework integrating a multi-population genetic algorithm and multi-operating-point analysis is proposed. The multi-population genetic algorithm effectively enhances global optimization capability and solution stability through parallel evolution, migration operators, and an elite population strategy; multi-operating-point data increases information content and strengthens fault isolation capability. Finally, simulation experiments under ground test conditions are conducted to validate the diagnosis of single-component faults in the low-pressure compressor, high-pressure compressor, high-pressure turbine, and low-pressure turbine. The results show that the proposed method can accurately invert component performance degradation, with a high fault identification accuracy rate, verifying the effectiveness of the proposed model and algorithm. This research provides a feasible technical solution for intelligent condition monitoring and precise diagnosis of aero-engines.
Cai, Na, Li, Ming, Li, BianJiang, Yu, GuangHui, Ma, ZhongYuan
Acoustic emission (AE) signals generated by incipient aero-engine damage are usually weak, intermittent, and highly dependent on sensor position, which makes diagnosis based on a single channel unreliable. To address this issue, this paper develops a multi-sensor AE diagnosis framework that combines entropy-guided feature fusion with a convolutional neural network (CNN). Time-domain and frequency-domain descriptors are first extracted from AE waveforms collected by multiple sensors. The consistency of each feature type across channels is then evaluated using feature entropy. Features with lower entropy are considered more informative for damage discrimination and are therefore retained for fusion. Based on the normalized channel contributions of these selected features, an adaptive weighted fusion matrix is constructed and used as the input to the CNN. The method is validated on a self-built aero-engine structural test rig equipped with 12 AE sensors. Fifteen representative initial-damage cases are designed, including signal excitation, crack-like excitation simulated by pencil-lead break (PLB), and impact-induced micro-deformation cases. Experimental results show that the proposed framework achieves a best average classification accuracy of 96.80% and provides more balanced class-wise performance than conventional machine-learning and deep-learning baselines. Additional noise-injection tests further indicate that the method maintains good robustness over a wide signal-to-noise-ratio range. These results suggest that entropy-guided multi-sensor AE fusion is a practical solution for intelligent diagnosis of aero-engine initial damage.
Jin, Yucheng, Han, Cong, Zhang, Leixuan, Yang, Guoan
Rolling bearings are key components of aero-engine rotor systems, and their degradation can trigger abnormal dynamics and safety risks. Unplanned bearing failures increase downtime and inspection cost; early detection enables condition-based maintenance and reduces conservative replacement. Early faults are weak, intermittent, and often masked by operating variability, so diagnosis based on a single sensing modality is usually incomplete. Vibration signals describe global dynamic behavior and resonance responses, whereas acoustic emission (AE) captures micro-damage processes and short impulsive events. Given the differences in bandwidth and noise sources, combining vibration and AE can provide a fuller description of bearing states. This study proposes an STFT–MFDCNN framework for bearing fault diagnosis using fused vibration–AE information. First, synchronized vibration and AE signals are converted into time– frequency images via the short-time Fourier transform (STFT), preserving spectral content and temporal non-stationarity in a unified representation. Second, a dual-branch multi-scale feature dense convolutional neural network (MFDCNN) is constructed. In each branch, parallel convolutions with different kernel sizes capture fault patterns across receptive fields, and dense concatenation connects all preceding feature maps to improve feature reuse and gradient flow. Third, global average pooling generates compact feature vectors for both branches, which are concatenated in a fusion head and classified by a Softmax layer. Experiments on a vibration–AE dataset built in this work show that the proposed method reaches 97.78% identification accuracy. Compared with dual-branch fusion baselines based on AlexNet, ResNet, and a shallow CNN, accuracy increases by 3.34%, 4.45%, and 5.00%, respectively. Compared with single- modality models (Only-AE and Only-Vib), accuracy improves by 14.71% and 6.77%.
Li, Zhichuang, Wang, Xingfeng, Zhang, Yanpu, Yang, Guoan
Civil aircraft engine thrust reversers are easy to have overspeed faults when they are being deployed, therefore this condition is very important for flight safety and maintenance work efficiency. The low sampling frequency, about 1Hz, of onboard data recording systems, for example WQAR, together with external environment interferences such as sudden strong winds, therefore causes a quite large number of wrong and missed alarms in fault data. Therefore, the labeled data that can be obtained are not only few in number but also have low quality. In this article, a fault diagnosis framework which is based on semi-supervised learning is put forward to solve the problems of high labeling expenses and fault diagnosis under the condition of small samples. This framework makes use of a small amount of labeled data and a large amount of unlabeled data. It uses Semi-Supervised Support Vector Machines, Semi-Supervised Neural Networks, and Semi-Supervised Random Forests for dynamic self- training to realize classification of three kinds of states: “Normal State,” “Low-Speed Thrust Reverser Deployment Exceedance,” and “Medium-Speed Thrust Reverser Deployment Exceedance”. The experiment outcomes show that the Semi-Supervised Neural Network model which uses self-training method has the highest classification accuracy on the test set, achieving 94.8%. Further robustness analysis, such as label noise and feature noise disturbance experiments, indicates that this model displays the smallest performance lowering under noisy environments, hence it represents the best overall performance. This research gives an effective method for the diagnosis of engine thrust reverser function problems in an effective and strong way.
Yuan, Xiaojiang, Liu, Nan
Planetary gears are key transmission components in spacecraft drive systems and are prone to failures under long-term operation in harsh environments. To address the problem that fault diagnosis and life prediction are treated separately in existing research, this paper proposes an integrated diagnosis-prediction method. First, order tracking and time synchronous averaging are used to preprocess vibration signals, eliminating speed fluctuations and noise interference. Then, fault features are extracted via VMD/EEMD decomposition, and the gear health condition is identified through a deep convolutional neural network. Finally, ensemble learning prediction models are established separately for different fault states to realize life prediction. Experimental results show that the VMD+CNN method achieves higher diagnostic accuracy than traditional methods for gears in healthy, cracked, spalled, and pitted conditions. The ensemble prediction models matched with different states can further reduce the life prediction error. This method provides effective technical support for condition-based maintenance of spacecraft planetary gears.
Liu, Yanyan, Gao, Tongxin, Zhu, Yangsong
Spacecraft operate in a highly complex in-orbit environment. Telemetry data from power systems is often plagued by multi-source disturbances such as plasma noise, electromagnetic interference, and communication packet loss, which places very strict demands on fault diagnosis model reliability. Using dynamic disturbance test manifolds and the comprehensive robustness quantification index (RQI) for evaluation, we found a critical vulnerability: baseline models such as the joint loss enhanced global sparse attention transformer (JLE-GSA-Transformer) perform well in resisting continuous signal disturbances, but their performance drops sharply when temporal structures are disrupted due to data missingness. This finding highlights that static diagnostic models have fragile robustness boundaries when operating under extreme non-stationary conditions. To solve this problem, this paper puts forward an adaptive optimization method that combines deep reinforcement learning (RL) with quantitative robustness feedback. By making use of the dynamic disturbance manifold, this method embeds the RQI into the reward function of the RL algorithm. Through continuous updates to the agent’s policy, it drives the adaptive adjustment of key parameters in the JLE-GSA-Transformer, mainly including the coding structure of global sparse attention (GSA) and the discriminative boundaries of temporal cosine-center joint loss. This closed-loop mechanism can significantly improve the stability, adaptability, and long-term reliability of the model when it works in complex environments with multi-source disturbances.
Liu, He, Wei, Cheng, Zhang, Zexu
With the continued expansion of low Earth orbit (LEO) satellite constellations, inter-satellite links (ISLs) have become a critical infrastructure for network connectivity and end-to-end forwarding capability. Existing studies often show a gap between the physical layer and the system layer: physical factors such as link budget, pointing error, and Doppler mismatch are frequently simplified, while link performance analysis is difficult to validate in a closed loop using a real protocol stack. In addition, dynamic topology and time-varying link states make fault detection and localization highly complex.To address these problems, this paper constructs a decentralized satellite network emulation platform and proposes a decentralized addressing and connectivity verification method for full inter-satellite link interfaces. At the physical layer, an optical inter-satellite link (OISL) model is developed by incorporating free-space path loss, Doppler mismatch, and BER-FEC-PLR mapping, with its dynamic evolution driven by orbital parameters. At the system layer, based on a one-satellite-one-container architecture, OpenClaw is introduced to perform link detection and log analysis, thereby generating diagnostic reports.Experimental results show that the proposed method supports interface-level link detection, identifies dynamic link faults such as transient impairments caused by Doppler mismatch in real time, and generates corresponding diagnostic reports.
Zhao, Jiangtong, Luo, Jiangtao, Zhang, Jinshuo, Ran, Yongyi
Against the performance degradation and failure risks that reusable launch vehicles (RLV) encounter over multiple mission cycles, together with the absence of a unified quantitative foundation for pre-launch health assessment, fault diagnosis validation, and remaining useful life (RUL) prediction verification, this paper presents a health management (HM) indicator system construction and allocation method tailored for RLV. Consistent with international standards for health management and test engineering, the method uses a “core capability decomposition—hierarchical metric mapping” framework to develop a two-dimensional metric system spanning functional and physical domains. Centered on fault detection, fault isolation, and fault prognosis, the approach breaks down top-level HM requirements into concrete technical indicators at the vehicle, system, and component levels, with explicit definitions, calculation formulas, and validation approaches to ensure engineering measurability and comparability. Integrating system criticality analysis and expert judgment, this work adopts the analytic hierarchy process (AHP) and a weight allocation scheme to decompose system-level indicators to critical subsystems including propulsion and flight control systems. A data-driven validation environment is established using a custom HM prototype and high-fidelity simulation platform. Multi-scenario simulation and fault injection yield labeled operational data for computing subsystem-level performance metrics. Comparing bottom-up aggregated indicators with top-down allocated targets forms a complete engineering closed loop from indicator allocation to system implementation and performance verification. Sensitivity analysis confirms that vehicle-level indicators remain within design thresholds under moderate weight perturbations, demonstrating robustness. Results show the method is feasible and consistent in simulation, offering solid support for the design and validation of RLV health management systems.
Shen, Tingzheng, Wu, Jiaxin, Li, Xiaochao, Qian, Ze, Yu, Jinsong
Bearing signals measured under different loads rarely keep the same waveform characteristics. Their amplitudes, spectral components, and statistical properties may change at the same time, which makes a classifier trained on one load condition less reliable on another. This study investigates a cross-load diagnosis method that trains a multi-scale convolutional squeeze-excitation network (MCSECNN) with multi-kernel maximum mean discrepancy (MK-MMD) and a domain-adversarial neural network (DANN). The convolutional branch captures vibration patterns over several temporal ranges, and the SE unit adjusts the importance of the channels generated by the parallel kernels. MK-MMD penalizes the distance between load-dependent feature distributions, whereas DANN weakens the load identity retained in the features. Tests using CWRU bearing data show that the integrated model gives better accuracy than the basic MCSECNN and the two single-adaptation variants. Ablation results indicate that the statistical and adversarial constraints contribute to transfer performance in different ways.
Ming, Mei, Ai, Yumei, Li, Shengnan, Wan, Haozheng, Sang, Shentian, Zhang, Haifeng
The present study proposes an absorptive coating defect detection method based on an improved YOLOv11 model, with a view to addressing the issues of low detection accuracy, false positives, and missed detections in existing absorptive coating defect detection algorithms when handling small-sized targets. Specifically, given the limited availability of relevant datasets, this study collected and constructed a dataset of visible light and infrared images of invisible coating materials containing typical defects, and processed the data using a variety of advanced data augmentation methods. Subsequently, the model was enhanced with a global attention mechanism (GAM) to improve feature extraction and detection accuracy for small objects. Regarding the loss function, this study adopted the Wise-IoU loss function, which adjusts the model’s attention to different samples. Finally, comparative and ablation experiments were conducted on the self-built dataset. The experimental findings demonstrate that the improved YOLOv11 model attained a precision of 93%, accompanied by a recall of 88%. This signifies an augmentation of 3% and 10%, respectively, in comparison to the conventional YOLOv11 model. This serves to substantiate the model’s exceptional performance in the task of detecting defects in absorptive coating materials.
Dou, Jiaxi, Xu, Yanlong, Wang, Yanling, Xu, Rui, Peng, Zhaoqin
The Electro-Mechanical Brake (EMB) system is a dry-type Brake-by-Wire technology that eliminates hydraulic components and directly controls friction braking using electrical actuators at each wheel. The EMB architecture consists of a Main Center Control Unit, a redundant Backup Center Control Unit, and four Wheel Control Units communicating via CAN FD. Due to its direct involvement in vehicle braking, compliance with ISO 26262 functional safety requirements is critical. As system complexity increases, potential risks such as hardware failures and communication faults must be systematically addressed. The proposed TSC was developed according to ISO 26262, covering the concept phase (Part 3), system-level development (Part 4), and software implementation (Part 6). Safety goals and Functional Safety Requirements derived from HARA are used to guide system architecture design and TSC development. Key design principles include modularity, redundancy, fault detection, and fail-safe operation. Verification is conducted at both system and vehicle levels using ECU-in-the-Loop Simulation (EILS), Hardware-in-the-Loop Simulation (HILS), and real-vehicle tests. Fault scenarios, including Main Center Control Unit failures and CAN communication losses, are injected using a custom LabVIEW-based fault injection tool. The study evaluates Fault Tolerant Time Interval (FTTI) settings, error handling mechanisms, and control handover strategies under fault conditions. The results show that redundancy and localized communication enable stable operation and smooth control transfer within the FTTI window without noticeable impact on braking performance or driver awareness. This study demonstrates the robustness of the proposed EMB architecture. Future work will focus on prognostics and maintenance strategies to support safe deployment in autonomous and electric vehicles. [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
Kim, Dokun
This work deals with the topic of integrating a multi-GNSS solution within existing avionics, specifically concentrating on airworthiness approval issues, thereby addressing risks associated with susceptibility in standalone GPS solutions. An older aircraft, still relying on existing GPS solutions, remains susceptible to jamming, spoofing, or single-point failures, which, in turn, pose risks associated with integrity requirements necessary for ADS-B or TAWS operation. An approach for a fault detection and isolation (FDI) scheme applicable to a BDS/GPS hybrid satellite system configuration suitable for airworthiness projects would thus be relevant. A transparent proxy architecture is proposed, which consists of a form-fit GPS/BDS dual-mode antenna replacement and an RF power splitter. This design makes it possible to provide BeiDou Navigation Satellite System (BDS) signals with neither changes in the interfaces nor structural changes being required to existing flight management systems. It includes a signal processing module that combines adaptive quality weighting on carrier-to-noise ratios, satellite elevation angles, and code-minus-carrier differences, as well as a carrier phase smoothing process via the Hatch filter. A multiple-channel approach in a parallel position system is designed, with different channels for GPS-only, BDS-only, as well as hybrid solutions, with weighted least squares estimation incorporating the Huber robust reweighting rule, along with Receiver Autonomous Integrity Monitoring (RAIM) with fault detection and exemption in every channel. A cross-constellation integrity monitoring algorithm is presented for the purpose of identifying constellation-level spoofing attacks that evade conventional single-system RAIM integrity. Simulation results, carried out for four different interference levels, show that the hybrid channel provides a 34% improvement in horizontal positioning accuracy compared with the GPS-only mode of operation. The cross-check method is able to effectively detect GPS spoofing attacks with position divergence exceeding 50m within a detection time of 25 seconds, along with a navigation availability of 99.9% under nominal operating conditions.
Ou, Chongjie, Zhang, Kezhi, Zhu, Haijie
This paper addresses the determinacy issue of multi-task execution in the Remote Data Conversion Unit of an integrated modular avionics (IMA) system in a non - operating system environment. A three - level hierarchical static scheduling table architecture for the Remote Data Conversion Unit is proposed. In this architecture, the maximum execution cycle of functional parameters is used as the device scheduling table cycle, the minimum execution cycle is used as the scheduling block cycle, and the worst - case execution time is used as the functional execution time. Through hierarchical design, the orderly connection of functions, scheduling blocks, and devices is achieved. A periodic interrupt mechanism is adopted between scheduling blocks to ensure time alignment at the scheduling block cycle level. Inside the scheduling block, a polling mechanism is used to perform static sorting according to the worst - case execution time of functions, and a wait function is introduced to achieve time alignment and fault isolation. For fault - tolerant faults, a delayed response strategy is adopted to avoid violating atomicity. For non - fault - tolerant faults, rapid detection and restart processing are achieved relying on periodic interrupts. A Simulink model is used to conduct a comparative simulation of the Remote Data Conversion Unit using a competition mechanism and the scheduling table mechanism. The results show that under normal and fault conditions, this architecture significantly reduces data transmission jitter, improves system determinacy and fault - tolerance ability, meets the requirements of the DO - 297 standard for functional independence and safety, and provides an effective solution for improving the determinacy of the civil aviation Remote Data Conversion Unit.
Hao, Yongqi, Wu, Meng, Miao, Zhiqi, Xu, Guanglei
The finite width of ultrasonic array elements results in a non-uniform angular radiation pattern of elastic waves in solids, deviating from the ideal point-source assumption commonly adopted in reverse-time migration (RTM). This angular radiation non-uniformity produces a crack tip-dominated imaging amplitude with weak crack flank representation, manifesting as a pronounced depth-dependent amplitude imbalance along vertically oriented defects. As a result, cracks may be misinterpreted as point reflectors, which compromises the reliability of characterization in ultrasonic nondestructive testing (NDT). This study proposes an ultrasonic frequency-domain reverse-time migration (FDRTM) imaging method incorporating element directivity correction. A longitudinal-wave directivity model in solids is formulated in the frequency domain and normalized at each frequency to ensure consistent scaling during multi-frequency stacking. During backward wavefield reconstruction using full matrix capture (FMC) data, the angular-dependent energy distribution associated with the receiving direction is explicitly corrected, rebalancing the angular energy distribution in the reconstructed wavefield. Defect imaging is then performed using a frequency-domain cross-correlation imaging condition, followed by stacking over frequency and normalization. Validation experiments were conducted on an artificially manufactured vertical crack in a 7075 aluminum alloy specimen. The results indicate that, relative to conventional RTM, the proposed method reduces crack tip dominance and enhances the relative visibility and continuity of crack flanks. Compared with conventional RTM, the peak imaging amplitude increases by 22.44%, and the amplitude at a depth of 6.8 mm is enhanced by 24.39%. In addition, the proposed method outperforms the total focusing method (TFM) in crack profile continuity and the relative visibility of crack flanks. The results confirm that incorporating element directivity correction into frequency-domain RTM mitigates depth-dependent amplitude imbalance and restores crack flank visibility, thereby improving the reliability of crack defect characterization in ultrasonic NDT.
Chen, Si, Zhang, Yifeng, Ma, Tengfei, Xu, Zheng, Jiang, Jiansheng, Gao, Jiaqi
Driven by the growing demand for higher efficiency and load-bearing capacity in fields such as new energy vehicles and heavy-duty engineering machinery, planetary gear sets are increasingly operating at elevated rotational speeds, coupled with a corresponding expansion of their revolution radii. This dual trend directly induces a substantial surge in centrifugal acceleration acting on the internal needle roller bearings. Under the cyclic stress inherent to transmission operations, such enhanced acceleration not only accelerates the initiation of spalling faults on the inner bores of planet gears but also exacerbates the propagation and deterioration of these faults throughout the service life. To elucidate the influence mechanism of inner bore spalling on the dynamic response of planetary gear bearings, this study develops a specialized dynamic model. This model explicitly incorporates the compound kinematic effects of simultaneous rotation and revolution, thereby ensuring a high-fidelity reconstruction of actual operating scenarios. The research systematically investigates how different spalling types and dimensional parameters affect the system’s dynamic behavior. Numerical results demonstrate a positive correlation between the severity of the spalling defect and the dynamic response intensity. Specifically, the expansion of defect dimensions under harsh operating regimes markedly exacerbates both the contact impulses at the needle-roller interface and the overall vibration acceleration amplitudes. Notably, the amplitude increment of the needle rollers is far more pronounced than that of other components. These findings enrich the theoretical understanding of fault-induced dynamic responses in planetary gear systems and provide a solid theoretical and model-based foundation for optimizing the fault diagnosis, condition monitoring, and maintenance strategies of the associated needle roller bearings.
Zou, Desheng, Lai, Junbin, Guo, Wei, Dong, Peng, Xu, Xiangyang, Sun, Qiang
Due to the interference of oscillatory components and noise, the periodic impulses associated with localized bearing faults become difficult to extract, leading to unreliable diagnostic performance. To solve this problem, the study proposes a simultaneous impulse and oscillatory component decomposition method (SIOCD). The method designs and solves a novel optimization model to decompose oscillatory components and fault impulse components from noisy vibration signals. To achieve component separation in the optimization model, distinct penalty functions are introduced for oscillatory and impulse components. For oscillatory components, a regularization term is applied to achieve their extraction by minimizing the component bandwidth in the frequency domain. For impulse components, a penalty function is designed to achieve their decomposition by enhancing both sparsity within groups (SWG) and sparsity across groups (SAG) in the time domain. Then, an iterative solver is derived using an alternating minimization framework and the majorization-minimization (MM) algorithm. Finally, the proposed method’s effectiveness is demonstrated through comprehensive simulation and experimental analyses, and the results demonstrate that it achieves superior performance over existing approaches in fault feature extraction and enhancement.
Sun, Haoran, Zhang, Jinduo, Han, Tianyu, Shi, Xi
This article focuses on the research and development of a remote cab controller for pure electric loaders, aiming to address the threats posed by traditional loaders operating in harsh and hazardous environments to drivers’ health and safety. First, the functional requirements of the controller were analyzed, based on which the hardware design with a multicore microprocessor as the core was completed, featuring functions such as signal acquisition, controller area network (CAN) communication, and H-bridge driving. On this basis, a control algorithm framework for remote driving was developed, including modules for signal input, analysis and processing, and signal output. Detailed control strategies were formulated for key components: For the pedal sensor, algorithms for opening degree calculation, automatic zero-position calibration, and dual-signal redundant fault diagnosis were proposed; for the steering module, precise angle calculation and force feedback feel simulation were achieved; and for the electric control handle, a hysteresis control algorithm was developed to suppress shocks caused by overly fast operations. In addition, a hierarchical fault diagnosis mechanism was established to ensure system safety. To verify the controller performance, a complete remote driving system was built. Field test results show that the system exhibits good signal following and control responsiveness in terms of traveling and working functions. Efficiency tests indicate that the remote driving efficiency can reach 80% of that of in-person operation under short-term test conditions, demonstrating the technical feasibility and control effectiveness of the developed controller. While the prototype exhibits promising performance for pilot deployment, long-term reliability metrics such as mean time between failures (MTBF) remain to be validated through extended field operation.
Lu, Yueqi, Ji, Shaobo, Yu, Qiuye, Li, Meng, Xu, Haozhi, An, Meng
Inertial Friction Welding (IFW) equipment is essential for the welding process of aircraft engine shaft components. However, the absence of comprehensive fault-handling standards for domestically produced inertial friction welding equipment has hindered its further development. This study focuses on the connecting rod and motor of the 30T-IFW equipment, employing a model-based fault detection method. Through simulation, the deformation of the connecting rod and the frequency response of motor vibration acceleration under different working conditions are obtained. Additionally, a monitoring platform is proposed to collect real-time data on connecting rod deformation and motor vibration from actual welding equipment. By establishing a quantitative correlation model of connecting rod deformation-force and revealing the coupling mechanism between motor eccentricity faults and modal frequency vibrations, a hybrid diagnostic framework that combines simulation of primitive warning and measurement of calibration is proposed. At last, the simulation and experimental results verify the effectiveness of the fault diagnosis method proposed in this paper.
Yang, Haifeng, Yuan, Mingqiang, Sun, Tao, Liang, Wu, Gong, Maolin, An, Xingyi, Wang, Qisong, Liu, Dan
Conventional aero-engine fault detection techniques tend to have problems simultaneously extracting local anomalies in sensor data and long-term temporal dependencies. To solve this problem, we propose a new fault detection scheme that only uses a Dual-Path Temporal Convolutional Network (Dual-TCN) and a Gated Recurrent Unit (GRU) module. The model, in turn, takes advantage of dual parallel branches of TCNs to extract local and global features and integrates these features with the GRU to model the progression of faults in time. Validated on the dataset of the National Aeronautics and Space Administration, called C-MAPSS, the proposed technique achieves a detection accuracy of 91.39%, which is better than CNN and LSTM baselines, demonstrating interesting improvements in the precision, recall, and F1-score. Experimental results further demonstrate the effectiveness of the dual-path feature extraction and GRU fusion strategy; this method is potentially useful to realize the real-time and accurate detection of faults in complex aero-engine systems.
Yan, Shaokai, Zhang, Yongjian
The inconsistency in bearing data distributions under diverse conditions often affects the representations of the faulty data and leads to indistinct decision boundaries and even negative transfer resulted from overlapping class distributions, greatly limiting the accuracy of the diagnosis model. To cope with the challenge, a pseudo-label-guided dual-supervised alignment (PDSA) method is developed for bearing fault diagnosis across diverse operating scenarios in this paper. To address the fixed alignment strategy issue, an adaptive distribution alignment layer is incorporated to ResNet18 to achieve dynamic data distribution alignment under varying condition, To enhance classification performances, a dual-supervised mechanism, comprising shallow-layer supervised contrastive learning is introduced through target domain pseudo-labels in target domain and deep-layer regularization class consistency. Experiments on two publicly available bearing datasets demonstrated this model realizes refined class-level alignment, strengthens fault states representation, and shows notable superiority in both accuracy and robustness.
Sun, Hao, Ren, Shijin, Gu, Zhangqing
The rotary storage mechanism is a critical component responsible for transferring cylindrical units. To accurately simulate the nonlinear dynamics characteristics of the rotary storage mechanism, a dynamics model incorporating uncertain parameters is established based on the Lagrange method. Utilizing an optimization approach, uncertain parameters of the rotary storage mechanism are identified based on test data. The Stellar Oscillation Optimization (SOO) algorithm is employed, which balances exploration and exploitation by simulating the periodic expansion and contraction of stars to achieve optimal solutions. The results show that the output of the identified dynamics model under two operating conditions closely matches the test data, validating the accuracy of the model and the effectiveness of the identification process. This provides strong support for subsequent reliability analysis and fault diagnosis studies of the rotary storage mechanism.
Li, Ang, Chen, Guangsong, Huang, Peng, Li, Hanning
Conveyor belt fault detection is critical for ensuring the safety and efficiency of industrial material transportation. In this study, a screen-printed flexible strain sensor based on a thermoplastic polyurethane (TPU) substrate and graphene conductive ink was fabricated. The sensor exhibited excellent flexibility, mechanical robustness, and stable electromechanical performance. Comprehensive evaluations were conducted, including microstructural analysis, strain sensitivity, hysteresis, dynamic response, and long-term cycling stability. The results demonstrated that a two-layer graphene configuration achieved an optimal balance between sensitivity and structural stability, showing high gauge factor, fast response, and reliable cyclic performance. Furthermore, the sensor was applied to conveyor belt fault monitoring. Experiments validated its ability to detect both halting faults and foreign object intrusions, with distinctive resistance signal features enabling not only fault occurrence detection but also identification of fault location, type, and severity. These findings highlight the potential of the proposed flexible sensor system as a promising solution for intelligent conveyor belt monitoring in harsh industrial environments.
Zhang, Bo, Ai, Shigeng, Zhang, Xiaobo, Sun, Wanting, Li, Pengfei
During the high-speed operation of packaging machines, if the abnormal components evolve into faults, the packaging machines often stop for inspection or even damage, causing production stagnation and huge economic losses. If key variables are predicted and faults are identified before the evolution of packaging machine failures, it is of great significance to ensure equipment safety and reduce maintenance costs and losses for enterprises. The purpose of fault prediction is to use the information modeling of equipment historical data to output the changes in key features before component failures in the future. Firstly, for the redundant data of multiple measurement points of the same variable in the packaging machine process variables, Pearson correlation analysis is used to obtain more accurate variable data. We reuse adaptive empirical mode decomposition (EEMD) for signal processing and feature extraction, reduce redundant information, use convolutional neural network (CNN) models for spatial feature learning, and then use bidirectional long short-term memory models to capture temporal dependencies of CNN information for capturing time series data. A model is established on the normal training set to fit the normal state of the packaging machine, identify different types and degrees of equipment fault characteristics through normal test set data, and send the predicted results of the equipment state to the fault classifier for judgment to determine whether to issue a fault warning. The results indicate that this article has validated the effectiveness of the model in fault feature extraction and high-precision fault classification through training on equipment status data.
Wu, Aimin, Liu, Shixian, Zhao, Lihui, Liu, Zhao, Wu, Tao, Li, Lianbing
In recent years, drone technology has seen widespread application in both civilian and military fields. By 2025, China will introduce supportive policies from multiple dimensions, including industrial development, technological innovation, and application promotion, to significantly increase the number of UAVs in use and their frequency. However, drones are prone to malfunctions due to factors such as bad weather and electromagnetic interference, which may result in serious consequences, including property damage and casualties. Therefore, improving the accuracy of fault detection and the response time of drones is of great significance. Although current research has made progress, there are still deficiencies: First, most of them rely on a single or limited data source, resulting in incomplete information and vulnerability to interference, which leads to low detection accuracy and reliability; Second, traditional methods are mostly based on fixed thresholds or simple rules, lacking real-time dynamic monitoring and adaptive analysis capabilities, making it difficult to issue timely warnings of potential faults. To this end, this study proposes a multi-scale time series prediction model based on multimodal and multi-branch, integrating multimodal data, constructing a dual-branch architecture, and combining deep learning and attention mechanisms to enhance the anomaly detection effect of unmanned aerial vehicles. A dual-branch anomaly detection model based on 1DCNN-BiLSTM and continuous wavelet transform is proposed, including a trajectory prediction difference branch and a full time series data branch. In the dual-branch output stage, the attention gating mechanism is utilized to fuse features and improve the detection performance. The experimental results show that this model performs excellently in both normal trajectory prediction and anomaly detection, providing an effective solution for drone anomaly detection.
Pu, Zhenglin, Zhang, Lin
Aiming at problems such as low efficiency and poor accuracy in fault identification for traditional small satellites, this paper proposes a multi-model fusion method based on machine learning. By constructing a telemetry data preprocessing module based on the Data Generation Adversarial Network, it effectively deals with outliers and fills missing values. Combining single model methods such as polynomial curve fitting, the grey model, and the ARMA model, and introducing the Long Short-Term Memory network and Gated Recurrent Unit to fuse with these models enhance the ability to process complex data features. The prediction results of each model are fused using machine learning methods, and finally, the fused value is taken as the final prediction result. The numerical simulation results show that this prediction method can predict the anomalies of different types of satellite telemetry parameters and has achieved good results.
Liu, Biyan, Chen, Ye, Guo, Qi
As a key component of unmanned aerial vehicles (UAVs), the stable operation of motor bearings is of vital importance to the stability of UAVs. In view of the incomplete data set in the actual diagnosis process, samples not encountered during model training are highly likely to appear. This paper proposes an Adaptive Class-Incremental Learning(ACIL) intelligent fault diagnosis method. This method construct a ResNet framework embedded with Coordinate Attention as the base architecture for class-incremental learning. Furthermore, the Information Preservation Example Selection(IPES) method is utilized to alleviate catastrophic forgetting and update the model from the previous phase using knowledge distillation under coordinate attention. The effectiveness of this method is verified through experiments on the bearing test dataset. The results show that, both average incremental accuracy and average incremental forgetting rate achieve state-of-the-art performance, which means that the performance of the proposed method outperforms than those of other methods.
Song, Ziyang, Lu, Jiantao, Wu, Wei, Li, Shunming
Nowadays, the majority of intelligent fault diagnosis approaches are still centered on individual faulty components, while only a limited number of models are capable of performing integrated diagnosis for rotating systems that consist of shafts, bearings, and gears. Under variable-speed operating conditions, the large scale of vibration data further complicates the process of effective feature extraction. To improve these challenges, this study develops a comprehensive diagnostic framework for rotating components, termed WGAN-SAFC. The proposed architecture integrates a Wasserstein Generative Adversarial Network (WGAN) with a hybrid structure of stacked autoencoders and sparse filtering (SAFC). SAFC integrates the feature-learning capability of SAE and the sparsity-driven representation of SF, while incorporating adversarial data generation to address sample imbalance and enhance fault diagnosis performance. Experimental verification on collected vibration datasets demonstrates that WGAN-SAFC achieves superior diagnostic accuracy and robustness compared with existing methods.
Li, Shunming, Feng, Mengqi
The gearbox is a key component of the mechanical transmission system, and its fault diagnosis is essential to the reliability of the equipment. However, obtaining fault samples under actual working conditions for gearbox fault diagnosis is challenging. In this paper, the rigid-flexible coupling dynamic simulation model of the gearbox is established, and the co-simulation of gear normal, crack, and breakage is carried out in the ADAMS and MATLAB environments. The comparison between the simulated and measured signals shows that the simulation method can accurately reflect the key characteristics, such as rotation frequency and meshing frequency, and verify its reliability and accuracy. The research results can provide effective data support for gearbox fault diagnosis and improve the operational safety of mechanical systems.
Li, Dongxiao, Zhang, Qianqi, Zhang, Zhongzheng, Li, Yongbo
The Active Wheel-Corner (AWC) integrates driving, braking, steering, and suspension systems into the wheel end, forming a fully drive-by-wire, four-wheel independent steering and four-wheel independent driving (4WIS&4WID) vehicle platform. While improving vehicle control performance, the full by-wire architecture also places higher demands on system reliability and fault tolerance. The steer-by-wire system has electrical, communication, and software failure risks, which may cause the vehicle to lose steering capability and trigger severe traffic accidents. This article proposes a hierarchical active fault-tolerant control strategy based on fault information reconstruction (FAST-FTC), enabling fault diagnosis and active fault-tolerant control when the steer-by-wire system fails, effectively ensuring the steering maneuverability and lateral stability of the vehicle under fault conditions. First, the strategy designs an adaptive observer combined with the Dugoff tire model to estimate nonlinear tire forces, while introducing a fault factor to achieve quantitative grading of steering system faults. Second, a hierarchical controller is designed for steering system faults. The upper-level controller, based on Adaptive Super-Twisting Sliding Mode Control (AST-SMC), determines the generalized forces required to track the desired trajectory under different fault conditions. The lower-level controller, based on the Fault-Aware Model Predictive Control (FA-MPC) strategy, dynamically adjusts weight matrices according to the fault factor and tire reconstructed stiffness, coordinating the allocation of four-wheel driving and the steering of healthy wheels to ensure lateral stability. Finally, the effectiveness of the proposed active fault-tolerant control strategy is validated through Hardware-in-the-Loop (HIL) simulation and real-vehicle tests.
Xiao, Feng, Jiang, Yueyong, Cheng, Rui, Xu, Changhe, Tang, Xiangjiao, Gao, Fengling, Li, Jianhua
This paper investigates the integration of Artificial Intelligence (AI) within radar-based perception for Advanced Driver Assistance Systems (ADAS) under safety considerations aligned with ISO 26262 [1] for functional safety and ISO 21448 (SOTIF) [2] for performance-related safety of the intended functionality. The study evaluates a hybrid architecture in which AI-based perception modules are combined with deterministic supervisory mechanisms to maintain safety compliance. A simulation-based case study using CARLA with radar sensor modeling is presented to compare a deterministic radar perception pipeline with an AI-enhanced approach under nominal and degraded environmental conditions. Performance is evaluated using precision, recall, and F1 score metrics. Results indicate improved recall and F1 score under adverse scenarios for the AI-based perception module, accompanied by a moderate increase in false positives. The paper discusses architectural constraints required to limit non-deterministic behavior, including confidence gating, deterministic supervision, and scenario-based validation. The findings are limited to simulation and are intended to provide preliminary insights into the technical and safety implications of incorporating AI-based radar perception within ISO 26262-compliant ADAS architectures.
Jain, Yesha
In recent years, the automotive industry has actively explored the application of various AI-based models such as Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, Autoencoders, and Transformers to improve defect detection rates at the End-of-Line (EOL) stage. However, implementing these approaches in the Noise, Vibration, and Harshness (NVH) area face several practical challenges: ① extended evaluation times compared to other data types, which limit the quantity of training data and lead to overfitting; ② label imbalance caused by the relatively small amount of defect data; ③ reduced labeling accuracy due to human error; ④ decreased robustness under domain shifts such as changes in jig fixtures, test environments, and signal-to-noise ratio (SNR); ⑤ diminished model reliability when new defect arise during development; and ⑥ constraints imposed by compatibility requirements with existing test equipment. This study proposes a Convolutional Autoencoder (CAE) based framework trained on NVH datasets collected from normal and defective Column-type Electric Power Steering (C-EPS) systems. Latent variables at the bottleneck layer are used for dimension reduction, enabling visualization and unsupervised classification using a clustering algorithm. A classification model derived from the encoder is fine-tuned with clustered data, and Gradient-weighted Class Activation Mapping (Grad-CAM), an eXplainable AI (XAI) technique, is applied to extract Feature Frequency Maps (FFM) highlighting defect-related noise and vibration characteristics. The proposed approach does not rely on the deep learning model to directly classify defect. Instead, it utilizes extracted FFM as weights(mask) to detect defect. This method enables quantitative data representation and ensures high applicability with existing EOL equipment. Post-processing within the FFM enables root cause analysis, reducing issue resolution time and supporting integration with conventional signal analysis techniques.
Park, Jun-Seo, Jo, Hyeon-Choel, Cho, In-Je, Seo, Jae-Yong, Yoo, Seong-Sik
Modern aircraft depend on extensive electrical wiring networks for power distribution, avionics, and control systems; however, these wiring systems are vulnerable to wear, insulation degradation, and arcing over time, leading to safety risks and costly unscheduled maintenance. This paper introduces an advanced Electric Health-Monitoring Wiring (E-Wiring) system that integrates temperature, current, insulation, vibration, and environmental sensors directly into aircraft wiring harnesses to enable continuous monitoring and intelligent fault detection. Data from these embedded sensors are processed through a distributed edge AI network, forming an Electrical Health Monitoring System (EHMS) capable of real-time diagnostics, predictive maintenance, and fault localization. The architecture comprises smart cable segments with sensor nodes, local harness gateways for edge processing, aircraft-level EHMS integration via AFDX/Ethernet, and cockpit or maintenance displays linked to ground-based cloud analytics for fleet-wide insights. We have an existing method to detect by using acoustic sensing method which can detect ongoing insulation chafing or a cut, they are limited in identifying pre-existing damages and by adding multiple acoustics in the existing wire harnesses it’ll add extra load to the aircraft. To overcome this, the system incorporates Time Domain Reflectometry (TDR) technology to detect both existing and potential wiring faults. The TDR circuitry interfaces with onboard devices, injecting test signals into wiring to pinpoint insulation anomalies or conductor breaks without adding significant weight or complexity. The proposed E-Wiring and EHMS solution enhances aircraft safety, reduces maintenance costs, and improves operational availability, offering a scalable approach for both retrofit and new-generation aircraft.
Tammana, Bala Sai Sri Rohit, Murthy, Harsha, Mendu, HarikaSivaniSunandha
The monorail crane is important in mining operations, and its operation affects both safety and efficiency. Currently, fault diagnosis for monorail cranes has several challenges, such as heterogeneous mixing of multimodal data, poor use of knowledge, low real-time requirements, and high deployment costs for large-scale models. To solve these problems, we present an agent framework using a multimodal knowledge graph and a lightweight large model. In particular, we construct a fault knowledge graph for monorail cranes, organizing professional knowledge about components, failure modes, symptoms, and maintenance. By employing retrieval-augmented generation (RAG) technology, the knowledge graph is merged with the Qwen lightweight large model (low-rank adaptation) for fine-tuning to develop a diagnostic agent with task planning, tool invocation and memory. The experimental results show that the agent framework reduces “machine hallucination” and outperforms conventional diagnostic accuracy, response speed and resource efficiency, thus offering a safe and efficient solution for intelligent operation and maintenance of mining equipment.
Zhang, Yixuan, Xue, Shun, Bi, Xiang, Wei, Xing, Kang, Ranyu, Jue, Jie, Cheng, Liruiran
This study investigates the post-failure flight dynamics of a 1200 lb classical octocopter under single motor inoperative condition using nonlinear time-domain simulations with a baseline feedback controller. A physics based propulsion sizing strategy is developed using IEC duty cycle definitions where continuous requirements are derived from nominal hover with margin and short time capability is used to accommodate elevated post failure loads. The selected motor satisfies both regimes and enables transient overdrive without excessive weight penalty. Simulation results in hover and forward flight at the best range speed showing that the vehicle can recover from any single motor failure and retrim using inherent redundancy without fault identification. However, recovery involves significant transient attitude excursions and altitude loss, and requires substantial increases in motor power, with multiple motors exceeding S1 power limits. Post-failure maneuver simulations indicate retained controllability with some degradation and increased coupling. These simulations demonstrate that the proposed motor sizing enables necessary operation post-failure while avoiding unnecessary oversizing.
Lemelin, Dakoda, Gandhi, Farhan, Fong, Weston
This study presents a distinct methodology for the early detection of faulty cells in electric vehicle (EV) battery systems, leveraging temporal voltage deviation patterns under real-world charging scenarios alongside outputs from a physics-based model. A comparative longitudinal analysis was conducted on a fleet of twelve EVs—six exhibiting stable performance and the other six demonstrating early-stage anomalies characterized by intermittent transitions from drive to neutral mode. These behavioral cues were investigated as precursors to deeper battery degradation. The analysis focused on cell-level voltage dispersion in battery pack during the mid-to-high state-of-charge (SoC) range (approx. 20–30% to full charge). Vehicles in healthy condition consistently displayed minimal voltage deviation between BMS-measured cell voltages and physics-based model predictions, whereas those with latent faults showed markedly higher variance, particularly between the highest battery and model-expected cell voltages. Notably, this voltage divergence was often accompanied by a modest yet recurrent thermal rise of 2–3°C, suggesting early-stage thermal non-uniformity. All vehicles were monitored over extended distances under diverse, real-world driving and environmental conditions, enhancing the robustness and generalizability of the findings. The proposed approach underscores the diagnostic value of tracking voltage deviation trajectories as a non-intrusive, scalable means of forecasting cell-level degradation. This framework could significantly advance predictive maintenance strategies, improving both the reliability and operational lifespan of EV battery packs.
Jawle, Bharat Sanjay, Selvakumar, Ashwin, Puttoji Rao, Nagaraj Kumar
The reliability of Drive Unit (DU) oil pumps is critical to the performance and safety of electric vehicles, as these pumps provide essential lubrication and thermal management. In modern EV architectures, real-time health monitoring of these pumps typically relies on indirect signals than dedicated sensing hardware, a design choice optimized for cost, weight, and system complexity. This makes early fault detection a non-trivial challenge. To address this limitation, we present a novel, data-driven anomaly detection framework that leverages large-scale customer fleet telemetry and advanced machine learning to identify incipient pump degradation that traditional diagnostic methods often fail to capture. Specifically, we develop an XGBoost regression model trained on time-series features—including commanded pump speed, oil temperature, and historical pump current—to predict expected current behavior under nominal conditions. Deviations are quantified using the Mean Absolute Percentage Error (MAPE) between predicted and actual currents, providing a continuous and interpretable measure of anomaly severity. A fully automated pipeline ingests daily telemetry, performs session segmentation, executes predictive modeling, and records anomaly outcomes in backend databases for continuous monitoring and engineering review. The proposed framework enables continuous, fleet-wide predictive maintenance of DU oil pumps. It improves early detection of degradation, reduces vehicle downtime, enhances safety, and increases customer satisfaction. More broadly, it highlights the potential of large-scale data analytics and machine learning to advance predictive maintenance and reliability in electric vehicle (EV) systems.
Li, Jingman, Yao, Mengqi, Rahimi, Sahil, Lin, Joanne
To address the challenges of recognizing abnormal states, detecting subtle early warning signs, and quantifying fault severity in scenarios involving simultaneous multiple faults in lithium-ion batteries, this study proposes a dual-layer fault diagnosis framework that integrates One-Class Support Vector Machine (OCSVM) and Robust Local Mahalanobis Distance Quantile (RLMQD) algorithm. First, a three-dimensional multi-scale feature space, incorporating voltage, kurtosis, and voltage change rate, is constructed to detect abnormal battery states via OCSVM and dynamically filter abnormal time periods with improved adaptability. Second, a computationally efficient RLMQD-based quantization algorithm is developed, which employs a small-scale sliding window and adaptively selects healthy cells to construct reference distributions. By incorporating low-quantile thresholds, the algorithm enhances early abnormality detection and significantly reduces false positives. Subsequently, fault severity is quantified through scale-weighted fusion and normalization, enabling accurate evaluation across diverse abnormal modes. Finally, The diagnostic performance of the proposed method is comprehensively validated through three sets of simulation experiments and real-vehicle data collected under realistic operating conditions. The results demonstrate that the proposed method accurately identifies both single-point and clustered anomalies, corresponds closely with actual fault conditions and exhibiting strong generalization capability. In real vehicle validation, the method achieves 95.79% accuracy, 100% recall, and a 93.3% F1 score in abnormal detection tasks. Furthermore, It demonstrates robustness and interpretability, enabling multi-type abnormal detection and fault severity evaluation without reliance on extensive fault datasets, thereby offering high suitablility for online monitoring and early warning in actual Battery Management Systems.
Wei, Fuxing, Yang, Libing, Wang, Zonglei, Xia, Xuelei, Shen, Jiangwei, Chen, Zheng
The rapid integration of intermittent renewable energy sources (RES) poses significant operational challenges for modern power systems. Lithium-ion battery (LIB)–based battery energy storage systems (BESS) have become vital for grid stability and energy management. However, large-scale deployment of BESS has led to increasing incidents such as fires and explosions, raising serious concerns regarding their safety and reliability. To overcome the limitations of traditional reliability assessment methods—such as reliability block diagrams (RBD), fault tree analysis (FTA), and Markov models—this study proposes an integrated fault detection and reliability analysis framework that combines FTA, failure mode and effects analysis (FMEA), and a Bayesian Fault Propagation Network (BFPN). The framework systematically models fault propagation across component, subsystem, and system levels, dynamically updating the prior probabilities of basic failure events using a Gaussian Mixture Model (GMM) and Expectation–Maximization (EM) algorithm. Conditional Probability Tables (CPTs) are recalculated through Maximum Likelihood Estimation (MLE) with logical relationships to achieve accurate and adaptive fault probability estimation. A multi-feature fusion indicator, the State Severity Indicator (SSI), is further introduced to evaluate system health in real time. A qualitative comparison with representative fault modeling and detection approaches—including Bayesian Network, FTA-DBN, and various machine learning methods—shows that the proposed BFPN offers a well-balanced trade-off between interpretability and real-time performance. Simulation experiments under both single- and multiple-fault scenarios demonstrate that the proposed framework accurately detects typical fault events and provides early warnings before fault escalation. Under complex coupled fault conditions, it effectively captures fault interactions and predicts cascading failures across subsystems and the overall BESS, showing strong robustness and diagnostic capability for real-time reliability assessment in modern energy storage systems.
Yang, Zhan, Chen, Xiaobo, Zheng, Ruixiang, Li, Mian
With the rapid expansion of the electric vehicle market, the safety of lithium-ion batteries, which serve as the main power source, has become a critical concern. Current mainstream methods for battery fault detection generally face a technical bottleneck of struggling to balance high accuracy with a low false alarm rate. Furthermore, constrained by algorithmic complexity and data processing efficiency, detection speeds often fail to meet the practical demands of real-time monitoring. As a result, developing more efficient and accurate fault detection technologies has emerged as a key challenge urgently needing to be addressed in the industry. This paper proposes a hierarchical fault detection framework for lithium-ion batteries that integrates voltage change characteristics with a Local Outlier Factor (LOF) scoring mechanism. The framework aims to achieve early identification and accurate diagnosis of abnormal battery states through multi-dimensional feature extraction and algorithmic fusion. In the first layer, decentralized voltage data are standardized using the 3σ rule to identify potentially anomalous batteries. In the secondary analysis phase, a sliding time-window mechanism is adopted to dynamically capture voltage sequences. Within each window, voltage variations are calculated along both vertical and horizontal directions, and statistical metrics, including mean, standard deviation, range, and increment are derived. Principal component analysis is then applied to extract key features, and battery anomalies are evaluated and confirmed based on the maximum LOF score. Experimental validation using datasets from vehicles that experienced thermal runaway events, along with data from 1,000 normal vehicles, demonstrates that the proposed method significantly improves the accuracy of battery fault detection. It also provides early warnings up to 17 days prior to the occurrence of thermal runaway.
Gao, Zhengpeng, Gao, Pingping, Chang, Penghui, Liu, Gang, Wu, Ji
Wheel-corner brake failures can significantly deteriorate vehicle stability and safety, since unbalanced braking forces may introduce an undesired yaw moment. This work investigates a fault-tolerant control strategy for Active Wheel-Corner Systems, exploiting Four-Wheel Independent Steering (4WIS) to mitigate such effects and preserve vehicle stability when brake actuator malfunctions occur. Unlike many existing approaches, the proposed framework does not require explicit fault detection or quantification as a prerequisite for corrective action, eliminating potential delays and uncertainties associated with fault-diagnosis schemes. A reference model for yaw rate and sideslip angle, incorporating combined longitudinal and lateral dynamics, is proposed, and a Weighted Pseudo-Inverse Control Allocation (WPCA) scheme is employed to distribute corrective actions among the four steering angles according to each tire’s capability, compensating for yaw moment imbalances caused by degraded braking performance. The overall control framework is evaluated using a high-fidelity vehicle model implemented through VI-CarRealTime, with control algorithms and fault scenarios integrated via MATLAB/Simulink, providing a flexible and realistic platform for systematic analysis. The strategy is tested in cornering maneuvers with fault injection representing the worst-case scenario of a complete failure of the outer-front wheel brake. Results demonstrate that 4WIS can effectively recover the desired vehicle response, reducing deviations in yaw rate and sideslip compared to a baseline vehicle without reconfiguration. The study highlights the potential of steering redundancy as a complementary solution to braking and torque-vectoring systems for improving fault tolerance in future Active Wheel-Corner systems.
Sonnino, Samuel, Melzi, Stefano, Caresia, Pietro, Manzoni, Alessandro, Vaini, Gianluca
Software-defined vehicles are those whose functionalities and features are primarily governed by software, thus allowing continuous updates, upgrades, and the introduction of new capabilities throughout their lifecycle. This shift from hardware-centric to software-driven architectures is a major transformation that reshapes not only product development and operational strategies but also business models in the automotive industry. An SDV operating system provides the base platform to manage vehicle software and enable those advanced functionalities. Unlike traditional embedded or general-purpose operating systems, it is designed to meet the particular demands of modern automotive architectures. Reliability, safety, and security become crucial because even minor faults may have serious consequences. Key challenges to be handled by the SDV OS include how to handle software bugs, perform real-time processing, address functional safety and SOTIF compliance, adhere to regulations, minimize attack surface exposure, and protect against remote access and data breaches. This is achieved via sound architectural principles, including a CSM for fine-grained access control, a lean and minimal kernel to reduce vulnerabilities, secure and efficient inter-process communication, and user-level drivers to provide better fault isolation. The key novelty of this approach rests on the fact that it uses open-source kernels, libraries, and tools that guarantee flexibility, clarity, and community-driven innovation. It provides a flexible runtime environment and OS-level isolation using virtualization, safe hardware sharing, and adherence to safety standards to set up the SDV OS as a resounding, secure, and future-ready base for next-generation automotive systems.
Khan, Misbah Ullah, Gupta, Vishal
Due to the rapid transformation of EVs and the battery storage system, the battery management system (BMS) is essential to ensure optimal performance of the battery storage piles. A BMS monitors and controls parameters such as SOC, voltage, current, and temperature. A traditional BMS has a minimum support of analytics, and it’s limited to local processing. However, when the battery information is uploaded to the internet, it becomes easier to manage maintenance and track the battery’s performance from anywhere in the world. This Cloud-based system is easy and made earlier, thereby giving a system alarm before the issue becomes big. Managing many batteries at once saves a significant amount of money in places like EV charging stations and Energy Storage Systems (BESS). Software updates to the system can also be sent remotely. Also, a BMS connected to the cloud can be used to support weaker grids in an instant if it needs the reactive power support. Cloud integration of BMS with the grid network will help in better planning of energy management at load dispatch centers. A BMS managing a pack of batteries at a renewable energy system can help to understand power demand and decide when the best time is to charge or discharge. So, this can monitor all the batteries without being near them. Further, identifying the problems is work that focuses on an ML-RL-based battery management system connected to the cloud to control and monitor the Voltage, temperature, Cell balancing, SOC, SOH, and fault identification. This BMS system has easy scalability to thousands of batteries connected. As the demand for EVs and clean energy soars, this cloud-integrated BMS would play an important role in managing the batteries that are part of that system, making it smarter, efficient, and reliable. The proposed Q-learning–based Cloud BMS achieves 96.5% energy efficiency, 3.2% SOC RMSE, and zero safety violations across 75,000 simulated samples, using an adaptive 6,000-state Q-learning agent validated through real-time cloud integration.
R, Rajarajeswari, N, Kalaiarasi, Francis, Elgin Calister
Fault detection in autonomous VTOL aircraft is critical because even minor degradations can quickly destabilize multirotor vehicles in safety-critical environments. However, real-flight fault detection remains challenging due to sensor noise, environmental disturbances, and the nonlinear aeromechanics of multirotor platforms. This study proposes a comprehensive machine-learning framework for rotor fault detection, isolation, and severity prediction using real flight data. A convolutional neural network (CNN) architecture is developed to learn spatio-temporal patterns from multivariate flight dynamics, enabling direct inference of both the faulty rotor and its damage level. The framework is first validated using simulated data generated by our in-house flight dynamic model. Next, to verify the framework using real flight data, a hexcopter was designed, fabricated and flight tested for both nominal and faulty cases by introducing controlled blade-tip breakage. The trained model achieves rotor-wise fault classification accuracies above 99% and sample-wise severity estimation accuracy of 96% within a ±1% tolerance in experimental data, demonstrating strong generalization and supporting real-time health monitoring for autonomous VTOL systems.
Sarker, Rippon, Dabaghian, Pedram, Halder, Atanu, Goyal, Raman
Traction motors technology has, driving the EV industry forward with more efficient, lightweight, and durable solutions. However, despite these advancements, noise testing at the end of the production line remains a critical stage for identifying manufacturing defects in traction motors. Hence early fault detection in traction motors is crucial to ensure safety and reliability of EV. This research contributes a solution that predicts early-fault detection, supporting improved reliability, reduced material cost and minimizing process time in the series production line. To identify the root cause of this problem, historical quality data has been acquired from manufacturing plants to enable efficient analysis. Feature selection was then carried out using embedded and wrapper methods to identify the most important features. These selected features were subsequently used as input for ML models. The best accuracy was achieved using SVC model for early-stage motor failure prediction.
Gaikwad, Pooja, Nangare, Kapilraj, Suryawanshi, Chaitanya
As the automotive industry moves from conventional function oriented embedded ECU-based systems to Code-driven system, the core electrical and electronic (E&E) architecture is also being redesigned to support more software-driven functionality. Modern and centralized architectures promise scalability and software-driven flexibility, but they also introduce significant challenges in power distribution—an area that remains underexplored despite its critical role in overall vehicle safety and performance. Our paper aims at the adoption of the traditional power distribution approach for Next Gen vehicle architecture. It requires a fresh look at how power is distributed. In a novel E&E architecture, a single power harness supplies battery voltage to each zone. If there's a failure or voltage drop, it can affect multiple functions within that zone at once, and management of voltage regulation, thermal dissipation, and EMI/EMC compliance becomes crucial. Adding to the complexity, safety-critical systems need power redundancy and isolation to meet Functional Safety standards. Mixed-criticality designs further complicate power management, as they demand strict segregation between critical and non-critical power loads to preserve functionality under fault conditions. The integration of software-controlled power switching and dynamic power management introduces additional failure modes previously unrecognized. Consequently, real-time monitoring and power fault detection are becoming vital for maintaining the health of a vehicle’s power distribution network. Traditional diagnostics, such as On-Board Diagnostics, offer limited checks and periodic alerts, primarily for engine and transmission faults. Advanced capabilities are essential. Through an investigative lens, this paper identifies the key bottlenecks in power distribution and proposes areas for further research and innovation aimed at ensuring resilience, safety, and performance in next-generation vehicles.
Borole, Akash, Warke, Umakant, Chakra, Pipun, Jaisankar, Gokulnath
Items per page:
1 – 50 of 514