Reliability Evaluation of Complex Equipment Based on Virtual Samples and Performance Degradation

Abstract

Aiming at the difficulties in reliability evaluation of small samples and multi-performance parameter products, a reliability evaluation method based on virtual samples and performance degradation is proposed. Firstly, the multi-performance parameter distance concept is introduced, and the original parameter is virtual augmented with the performance parameter distance. Secondly, the improved Elman neural network is used to process the sample to obtain the complete degradation trajectory. Finally, this method is combined with the performance prediction method based on performance degradation to process the degradation data of a certain type of space relay and obtain a lifetime of 128 hours. The result shows the method effectively solves the processing problem of the rare sample data in the accelerated degradation test, which has certain reference significance.