Optimal design of ADT based on non-parametric statistics

Abstract

Optimal design of Accelerated Degradation Testing (ADT) to obtain more useful data within the limited cost is a crucial research in ADT technology. In this paper stochastic process is used to describe the degradation process of products. For analyzing the accelerated degradation data, parametric statistical methods needs to assume the distribution function of parameter, and error will be caused if assuming a wrong distribution. To solve this problem non-parametric statistical method which is distribution free is proposed to analyze the accelerated degradation data to establish a suitable regression model by the data itself, and then obtain the mean time of products under normal condition. The optimal design of ADT is conducted with the objective that minimizing the mean square error (MSE) of the estimation of mean time of products under normal condition under the constraints of experimental cost. The optimal plan can provide variables including: stress levels, interval of performance inspection, sample size and number of inspection at each stress level. Finally a simulation example is used to illustrate the proposed ADT optimization design method.