Reliability Modelling for Degradation Based on Type-ⅡGeneralized Logistic Distribution and Particle Swarm Optimization

作者：Zhang, Jinbao; Zhao, Yongqiang; Liu, Ming; Kong, Lingxian

来源：Journal of South China University of Technology(Natural Science Edition), 2019, 47(5): 96-102.

DOI：10.12141/j.issn.1000-565X.180091

• Degradation data
• Generalized logistic distributions
• Particle swarm optimization algorithm
• Performance degradation datum
• Probabilistic distribution
• Quantile
• Reliability Evaluation
• Reliability modelling

As the generalized probabilistic distribution can describe the characteristics of performance degradation data with high precision and reduce the impacts caused by the outliers and the fault selection of probabilistic distribution, the type-Ⅱgeneralized logistic distribution(Ⅱ-GLD)is applied to the degradation reliability evaluation in this paper. In the investigation, the parameters of location and scale are introduced in the modeling of time-dependent degradation data, the objective function is established according to the mean-square error(MSE)between the quantiles of Ⅱ-GLD and the experiment data, and particle swarm optimization(PSO)algorithm is utilized to estimate parameters simultaneously. Then, the proposed approach is applied to a practical example for the evaluation of degradation reliability, followed with a verification and comparison. The results show that the relative errors between the mean and the standard deviation are below 8%, and that the reliability evaluation results match well with the pseudo-lifetime data in different failure thresholds. Moreover, it is indicated that, as compared with the normal distribution and the Weibull distribution, Ⅱ-GLD can explore the tail characteristics of degradation data more effectively and represent the initial degradation of the product faithfully.