High reliability and long-lifetime products usually work in multi-stress environment such as temperature, humidity, electricity, and vibration. How to evaluate the reliability of the product under multi-stress condition is an urgent problem to ensure the safe and reliable operation of the product. Accelerated test provides an efficient and feasible way; however, the existing acceleration models have some shortcomings, such as less stress type, neglecting the stress coupling, and multi-parameter estimation difficulties. Therefore, in this article, first, a new universal multi-stress acceleration model is derived based on the classical Arrhenius model. Second, a multi-parameter estimation method for multi-stress model is proposed by combining particle swarm optimization and maximum likelihood estimation. Six simulation cases are used to verify the effectiveness of the proposed multi-parameter estimation method. The results of Case 1 to Case 3 show that the maximum mean square error of five parameters in the multi-stress model without considering stress coupling is 3.71%. The results of Case 4 to Case 6 show that the maximum mean square error of nine parameters in the multi-stress model considering stress coupling is 7.69%. Finally, an application example is performed to investigate the performance of the universal multi-stress acceleration model and multi-parameter estimation method.
Stiffness Parameter Estimation for Forearm Skeleton Model by Using Particle Swarm Optimization
