Abstract: Remaining life evaluation of products based on accelerated life testing (ALT) is becoming a hot spot of research, and how to improve the precision of evaluation is one of the most important issues. Related tasks on remaining life evaluation based on ALT, however, exist some problems, such as regarding the samples as new products, taking the extrapolated condition from ALT as the real operated condition, and neglecting field data. In view of these problems, a more accurate life model is built by taking the impact of initial failure and environment difference into account. The model is statistically analyzed by fusing ALT testing data and field data, so as to increase much more multisource life information within evaluated data. Simulation shows that, the result of remaining life evaluation based on ALT becomes more precise due to improving life model and increasing information.

Key words: Accelerated life testing, Data fusion, Initial failure, Monte Carlo simulation, Remaining life evaluation