Optimization design method for multi-stress accelerated degradation test based on Tweedie exponential dispersion process

Abstract

In the operational environment of electronic products, simultaneous multiple stresses contribute to product degradation, with their interactions often being intricate. To effectively analyze and optimize multi-stress accelerated degradation tests, we introduce a Multi-Stress Generalized Coupling accelerated degradation model based on the Tweedie exponential dispersion process, coupled with a multi-decision optimization approach for test design. The model integrates stress interactions elucidated through fuzzy correlation analysis, reflecting the accelerated degradation behavior consistent with the Tweedie exponential dispersion process. Addressing the complexities of multi-parameter estimation, we propose a parameter estimation method underpinned by the Sine Cosine Algorithm for multi-objective optimization within the proposed framework. Considering budgetary constraints, D-optimization and V-optimization strategies are deployed to finetune test scheme allocations, optimizing resource utilization. Validation through a three-stress accelerated degradation test on LED chips confirms the model's efficacy in reliability assessment. Our findings indicate the superior descriptive power of the Tweedie exponential dispersion model for complex degradation phenomena and the proposed model's closer approximation to stress conditions in real-world settings, with a methodological error maintained below 2 %. Furthermore, the test optimization method enhances the D-optimization and V-optimization objective values by 21.20 % and reduces them by 4.08 %, respectively, marking a substantial improvement over traditional schemes and endorsing the holistic advancement of multi-stress accelerated degradation testing methodologies.