PHM (Prognostics & Health Management) technology, a health monitoring technology, has been widely used in aviation, military, and railroad fields, but has not been properly used in vehicles. Since the technology was developed based on data sufficient for fault diagnosis or prediction, there are few restrictions on the application of the required sensor, and it is being used in high value-added industries. In commercial vehicles, PHM would be a essential technology because sharing platform cannot earn money during the period of breakdown. Therefore, we tried to develop a PHM methodology suitable for vehicles targeting the most expensive motor system among vehicle parts. In the first stage, the first thing to do is to review the failure modes that can occur in the electrification system and appropriate detection methods such as the type and location of sensors to detect failure modes. Therefore, several critical failure modes were selected by gathering from design, analysis, and test fields through FMEA(Failure Mode Effect Analysis). The critical failure modes were bearing failure, gear failure, permanent magnet demagnetization, and motor shaft eccentricity. In the second stage, we tried to find a suitable diagnosis method according to each failure mode. For gear defects in failure modes, an index was generated based on a rule, and the validity of the index was verified through an experiment. In the failure mode, the bearing failure was also approached based on the rule(Short Time Fourier Transform , Hilbert Transform), and the degree of determining the presence or absence of failure could be utilized. However, there were limitations in predicting the location of the failure, so we tried again based on the data by AI(Artificial Intelligence) and were able to confirm the validity of the method. In the failure mode, for eccentricity( static, dynamic, mixed) and demagnetization defects, the difference in signals from normal and failure was confirmed using an CAE analysis model, and similar signal differences were confirmed in the test results of the target product. Based on this hybrid approach(CAE & DATA), it was possible to create a robust diagnostic model with only a small number of experimental data. In the third stage, we created a device that can collect data, and a device with edge computing capabilities that can analyze and diagnose signals from actual vehicles. In the last stage, we have built a platform to periodically check the condition of individual vehicles. In conclusion, the following results can be obtained. (1) FMEA approach can be universally applied to completely different fields where motors are applied (wind power generation, UAM). (2) A hybrid method that can efficiently develop health indicators for motor systems was proposed, and indicators for other types of motor systems could be efficiently created by mainly using analytical models and minimizing the number of tests in the target test. (3) It was confirmed that the PHM method can be applied to actual vehicles by identifying the factors that were the limitations in applying PHM to vehicles and finding and applying methods to supplement them.

Mr. Hong Suk Chang, senior research engineer, 현대자동차