Virtual FMEA and Its Application to Software Verification of Electric Power Steering System
2017-01-0066
03/28/2017
- Features
- Event
- Content
- This paper presents the “Virtual Failure Mode and Effects Analysis (vFMEA)” system, which is a high-fidelity electrical-failure-simulation platform, and applies it to the software verification of an electric power steering (EPS) system. The vFMEA system enables engineers to dynamically inject a drift fault into a circuit model of the electronic control unit (ECU) of an EPS system, to analyze system-level failure effects, and to verify software-implemented safety mechanisms, which consequently reduces both cost and time of development. The vFMEA system can verify test cases that cannot be verified using an actual ECU and can improve test coverage as well. It consists of a cycle-accurate microcontroller model with mass-production software implemented in binary format, analog and digital circuit models, mechanical models, and a state-triggered fault-injection mechanism. In this paper, the vFMEA method was applied to the verification of the safety mechanisms implemented on an ECU of an EPS system. It was revealed that the software-implemented safety mechanism detected the drift fault injected into a current monitor circuit in the ECU and shut down the system properly as designed. This means that the software was verified with the vFMEA method for the case of a drift fault as well as an open fault and a short fault. In addition, a simulation result was compared with the experimental ones using an actual ECU in the case of a sinusoidal steering input. It was also revealed that motor torque for driver assistance gives results within range of the experimental ones. Therefore, we confirmed the validity of the vFMEA system as a simulation platform for safety-mechanism verification.
- Pages
- 7
- Citation
- Nakao, S., Hyodo, A., Itabashi, M., Sakashita, T. et al., "Virtual FMEA and Its Application to Software Verification of Electric Power Steering System," SAE Technical Paper 2017-01-0066, 2017, https://doi.org/10.4271/2017-01-0066.