Performance reliability assessment for products based on time series analysis

Time series model has the advantage of strong self-adjustment for stochastic process and high precision for prediction. For the high-reliability and long-life products in the field of aeronautics and astronautics, two methods based on time series model were proposed to evaluate reliability and predict lifetime using performance degradation data. Firstly, assuming that the degradation measure follows the same distribution family but its parameters may change with time. Non-stationary time series were used to fit distribution parameters. According to the relation between reliability and degradation measure distribution, the corresponding reliability functions were developed. Then, degradation paths of all samples were described by time series model. False failure interval and the false lifetime value could be predicted and reliability was obtained by statistical method from complete lifetime test data. Finally, the alloy fatigue crack data was used to evaluate reliability and predict lifetime, and the reasonable results show that the proposed method has better robustness.