Reliability sensitivity analysis using variable environment test data

Many products always operate under various complex environment conditions. To describe the dynamic influence of environment factors on their reliability, a method of reliability sensitivity analysis using variable environment test data is proposed. In this method, a location scale model is introduced to relate the reliability to environment factors. And the location parameter is assumed as a function of relevant environment variables while the scale parameter is assumed as an unknown positive constant. Then, the location parameter function is constructed by using the method of radial basis function. Using the varied environment test data, the log-likelihood function is transformed to a generalized linear expression by describing the indicator as Poisson variable. With the generalized linear model, the maximum likelihood estimations of the model coefficients are obtained. Thus, the influences of environment factors on the reliability of the product are measured quantitatively, especially the interaction of varied environment factors. With the reliability model, the reliability sensitivity is obtained. An instance analysis shows that the method can be used to analysis the dynamic variety character of reliability along with environment factors and is straightforward for engineering application.