Remaining useful life prediction of high voltage relays based on performance degradation information

Considering the challenges associated with limited performance degradation data and low prediction accuracy for high-voltage relays in new energy vehicles, this paper proposes a residual life prediction method that incorporates data cleaning and the Wiener process. In this approach, the contact resistance measured during the on-off test of a high-voltage relay serves as the input variable. Standard deviation and mean filtering techniques are employed to identify and remove anomalous data. The independence and normality of the degradation increments are assessed using statistical tests, and the parameters of the Wiener process model are estimated through the maximum likelihood method. The point at which the contact resistance of the relay first reaches a predetermined threshold is considered the end of product life, and the remaining life is predicted based on this failure threshold. Utilizing the performance degradation data of a high-voltage relay and employing the mean absolute percentage error (MAPE) as a metric, the practicability and effectiveness of the proposed method are demonstrated.