Accelerated Life Data Processing Method Based on Quantile Regression and Its Application to Maintenance Strategy

Traditionally, accelerated life data processing methods depend on the distribution assumption, which might result in the precision uncertainty of parameter estimation. Therefore, in this paper we propose the accelerated life data processing method based on the quantile regression. By minimizing the absolute value deviation estimate, the estimators of the accelerated model under different quantiles are obtained. Then the corresponding quantile life can be extrapolated under the normal stress. Based on the quantile life, we can derive the reliability function and the corresponding life distribution function of the product. Furthermore, we apply the proposed method to the maintenance strategy of the repairable system. We calculate the optimal preventive maintenance period by minimizing the long-run average cost of the system using the interior-point method. Finally, taking the double-stress accelerated life test of the seal ring as an example, we illustrate the implementation and effectiveness of the proposed method.