Modeling And Optimization Of Accelerated Degradation Test Based On Geometric Brownian Motion

As the key technology in reliability engineering,prediction and health management technology has been widely used in aerospace,weapon equipment and other fields.Accelerated degradation modeling and accelerated degradation test optimization are important research contents of prediction and health management technology,which provides theoretical guidance for the prediction of high reliability and long-life products,and has important academic significance.However,with the deepening of research,in the prediction research of some highly reliable products with nonlinear and random degradation,the problems of model fitting accuracy and life prediction accuracy are increasingly revealed.Aiming at the above problems,the accelerated degradation modeling and accelerated degradation test optimization are studied based on geometric Brownian motion respectively,which provides a new idea for solving the existing problems in the field of reliability engineering and makes some useful attempts.Aiming at the problems of nonlinear degradation and random fluctuation in the process of accelerated degradation of some products,an accelerated degradation model is established based on geometric Brownian motion.Combined with unconstrained optimization,a two-step maximum likelihood estimation method is proposed.Compared with Wiener process,the effectiveness of the geometric Brownian motion accelerated degradation model is verified by using Monte Carlo simulation and carbon film resistor example.Aiming at the problem that geometric Brownian motion is difficult to describe the individual differences and measurement errors of similar products,an accelerated degradation model of uncertain geometric Brownian motion considering both individual differences and measurement errors is proposed.Combined with genetic algorithm,a two-step maximum likelihood estimation method is proposed.Compared with the accelerated degradation model based on geometric Brownian motion,the effectiveness of the proposed method is verified by using Monte Carlo simulation and carbon film resistor example.In order to solve the problem that test configurations based on different optimization criteria are contradictory,a multi-objective optimization method based on accelerated degradation model of uncertain geometric Brownian motion is proposed,in which the objective functions are the maximum determinant of Fisher information matrix(D-optimization criterion)and the minimum asymptotic variance of p-quartile lifetime(V-optimization criterion).Under the constraints of cost and test conditions,multi-objective optimization of accelerated degradation test is carried out.the multi-objective optimal test configuration scheme set and Pareto front are obtained considering the accuracy of life estimation(V-optimization criterion)and model fitting(D-optimization criterion).And compared with the single objective optimization,the advantages of the multi-objective optimization method based on uncertain geometric Brownian motion model are illustrated.Finally,the effectiveness of the optimized test configuration is verified by simulation experiments combined with the objective functions of D-optimization and V-optimization.