Reliability Assessment from Performance Degradation Data Based on Time Series Model

To evaluate reliability and predict lifetime for products from performance degradation data. a new method based oil degradation measure distribution is proposed. Assume that the degradation measure follows the same distribution family but its parameters may change with time, dynamic data of time-dependent parameters can be modeled by autoregressive integrated moving average (ARIMA), and the corresponding reliability functions are then developed. The method has the advantages of self-adjustment with time series model for stochastic process and high precision for prediction. and effectively overcomes human factors influences caused by the hypothesis of time-dependent parameter distributions, so the method has better robustness. An example of reliability assessment is given at last, which can illustrate the validity and feasibility of the presented method.