Statistic Analysis of Accelerated Reliability Growth Testing Based on Proportional Intensity Model

Abstract:

To solve the question of reliability evaluation for accelerated reliability growth testing (ARGT), firstly the test procedure of constant stress ARGT is described. Secondly,by combining proportional intensity model which is a nonparametric statistical regression model for stochastic point process with AMSAA model for reliability growth, assumptions of data analysis are proposed. Then the statistical analysis model of CSARGT is established and the maximum likelihood estimation method of unknown parameters is given. Furthermore, according to characteristic of nonhomogeneous poisson process, a simulation sampling formula that can simulate failure data of CSALT is derived, and simulation examples under different stress levels are discussed. Finally the availability and reasonableness of the proposed statistical method are verified.

Key words:accelerated reliability growth testing; proportional intensity model; Monte Carlo simulation; parameter estimation