Accelerated Life Testing Under Competing Exponential Failure Distributions

摘要

Accelerated life testing of a product is commonly used to estimate parameters of devices with extremely long lifelines. The problem of analyzing accelerated life tests when the product can fail from any one of p independent causes is considered. For each component it is assumed that the time to failure distribution is exponential with a hazard rate which follows a 'power rule.' The method of maximum likelihood is used to obtain estimators of the power rule parameters when the life test is either type I, type II, or progressively censored. Estimators of each component's mean failure time and survival function are obtained. Estimators of system or subsystem survival functions are also obtained. Results of a simulation study are presented. (Author)