Development of Noise Diagnosis and Prediction Technology for Column-Based Electric Power Steering Systems Using Vehicle Controller Area Network Data

The steering system is a critical component for controlling a vehicle's direction. In the context of Advanced Driver Assistance Systems (ADAS) and autonomous vehicles, where drivers may not always be actively holding the steering wheel, early detection of precursor noise signals is essential to prevent serious accidents resulting from the loss of steering system functionality. It is therefore imperative to develop a device capable of early detection and notification of steering system malfunctions. Therefore, the current study aimed to quantify the noise levels generated within the Column-based Electric Power Steering (C-EPS) system of a D-segment sedan. To this end, we measured the uniaxial acceleration in nine noise-generating areas while simultaneously collecting data from three Controller Area Network (CAN) sources that are directly related to steering operation. The results indicated that we have successfully developed a meaningful machine learning model by analyzing the correlation between steering noise acceleration and CAN data. Altogether, we have designed an algorithm that is capable of predicting steering system malfunctions when abnormal noise acceleration occurs, thereby enhancing accident prevention capabilities.