关键词: 多阶段退化, 测量误差, Wiener过程, 变点分布区间, 寿命预测

Abstract: In the existing multi-stage Wiener degradation modeling and reliability analysis, most of them ignored the measurement errors and did not consider the differences of change points in the degradation processes. Considering measurement errors， a two-stage Wiener model was established and the maximum likelihood estimation method and multidimensional search algorithm were combined to obtain the estimated values of measurement errors. Considering the differences of change points, the change point interval was estimated based on BIC minimum criterion, at the same time, the estimated values of drift coefficient and diffusion coefficient were determined according to the positions of change points, and then the estimated values of change points were obtained by using SSE minimum criterion. A reliability evaluation method was given based on the distribution interval of change points. Finally, based on the established model, an example of RV reducer transmission accuracy degradation data was analyzed. Considering the uncertainty impacts of measurement errors and change point positions on degradation modeling, the reliability evaluation accuracy of the products is further improved and the life of the products may be predicted more accurately.

Key words: multi-stage degradation, measurement error, Wiener process, change point distribution interval, lifetime prediction