Multi-Stage Wiener Process Based Remaining Useful Life Prediction for Load Cells

Abstract Nowadays, sensor technology is used in a wide range of fields, and the demand for its reliability continues to increase. Accurate estimation of the remaining useful life of sensors is critical. Since the performance degradation of sensors involves multiple stages, the single-stage Wiener degradation model is difficult to accurately characterize. To resolve this issue, a multi-stage Wiener process degradation model is established. The model utilizes the cumulative sum algorithm to identify change points among stages. Then, the performance of the multi-stage Wiener model for characterizing degradation trends is validated with load cell accelerated aging test data. Furthermore, the results of experiment demonstrate that the proposed method provides a more accurate remaining useful life estimation compared to the single-stage Wiener model, and the relative error is reduced by 18.82%.