A Model Optimization Method for Accelerated Degradation Test Data

Abstract

Aiming at the problem of fitting error in accelerated degradation data processing, a model optimization method was proposed. Firstly, the best fitting model was selected according to the minimum Residual Sum of Squares, and pseudo-failure lifetime was calculated for each degradation data group. Secondly, the optimization method of the lifetime distribution model was proposed and the best lifetime distribution was obtained according to the minimum fitting error. Finally, a case example was given, which indicated that the deviation may occurred without the model optimization.