Study on hybrid optimization method of accelerated life test to hydraulic pump

Hydraulic pump is the key component in hydraulic system, its performance and life affects airplane's safety and passenger's life directly[1]. So how to determine the life of hydraulic pump is very important in application. Life testing is a normal way to assess the reliability and lifetime of components, whereas it is not suitable to products with long life and high reliability. Especially to some expensive products, normal life test needs a large of test samples and long test time, so Accelerated Life Testing (ALT) emerges as the times requires, which can promote the failure process and determine the life with short experimental time and low testing cost [2]. Aiming at the ALT of hydraulic pump, this paper investigates its failure mechanism and obtains the multiple accelerated stresses. Through analyzing the failure accumulative process and regressing the prior test data, this paper establishes the accelerated model based on hybrid Weibull distribution under variable synthetic stresses. With multiple estimated parameters, complex load spectrum and nonlinear cycle execution, the maximum likelihood function based on cumulative exposure theory in ALT of hydraulic pump becomes complex and its parameter estimation is difficult to achieve global optimization using traditional optimum method^[1]. After investigating the variable rule of estimated parameters, this paper determines the initial value bound of estimated parameters. Directing to the contradiction between the global optimization and high searching efficiency in traditional optimal methods, this paper designs a Hybrid Optimization Method (HOM) based on Genetic Algorithm (GA) and Powell method, which can take the advantages of both method and overcome their shortcoming through suitable switch point. The application of hydraulic pump indicates that HOM not only can search the global optimum value but also can increase the searching efficiency greatly.