Product reliability prediction with failure information fusion

A Bayesian statistical approach is proposed to improve product reliability prediction by fusing product failure information from both field performance data and accelerated life testing data. Through this approach a calibration factor is developed, which compensates the difference of failure time distribution when the product is under the operational condition comparing to the lab testing condition. An example, based on the Arrhenius lifetime-stress function of temperature, is used to illustrate how to estimate the calibration, factor as well as other important parameters of the failure time distribution.