Multiple stresses optimization design of constant-stress accelerated degradation test based on Wiener process

Abstract

Constant-stress accelerated degradation test (CSADT) is an effective means of evaluating the reliability of products, to ensure accurate assessment of reliability-related indicators under limited funds. The optimized design of CSADT has been widely applied. The advantage of Wiener process in capturing the random nature of non-monotonic degradation paths caused by inherent uncertainties in products has also been widely used in the application of accelerated degradation tests (ADT). To address the drawback of traditional Wiener process with constant diffusion coefficients leading to low accuracy in test evaluation, a multi-stress coupled constant acceleration degradation model with stress-related diffusion coefficients is proposed. This includes the construction of a D-optimization criterion that minimizes the variance of model parameter estimation as the optimization target, and a scheme optimization method based on Particle Swarm Optimization (PSO) with cost constraints. Through the analysis of a case study of ADT for LED lamps, comparison and parameter sensitivity analysis of four accelerated degradation models with or without considering stress-related diffusion coefficients, the effectiveness, and robustness of the model in the paper are validated.