Research On Modeling And Application Of Burn-in Optimization

Burn-in is a widely used reliability screening technique in order to reduce or eliminate early failures in industrial production,especially electronics manufacturing as well as many other areas,such as military and aerospace industries.Burn-in is of particular importance,as it improves quality and reliability of products,reduces the loss due to early failures,and saves warranty and maintenance costs.Over the last half century or so,engineers,technicians and researchers conducted extensive research on burn-in,centering on how long and under what conditions should the burn-in process be conducted in order to get the most profit or the highest reliability,as referred to as the optimal burn-in problems.Efficient burn-in optimization models are imperative in deciding on the optimal burn-in,usually based on the intention of manufacturers and decision-makers,life characteristics of products as well as burn-in implementation process.Optimal burn-in settings and warranty/maintenance policies are then derived by solving these models.However,existing researches haven't payed enough attention to the burn-in optimization modeling under accelerated stress and are inefficient for highly reliable and long-lifetime products.Furthermore,research on the burn-in optimization modeling under complex failure process isn't in depth.For the above reasons,and motivated by a couple of practical burn-in optimization problems in engineering,this dissertation investigates modeling and application of burn-in optimization in combination with failure modeling,parameter estimation for accelerated models and life distributions as well as commonly used warranty/maintenance policies for a certain number of products with typical failures,including products with/without wearout period,Wiener degradation failure products and competing risks products involving traumatic and degradation failures.The main research content includes modeling and applications of failure-based accelerated burn-in optimization,degradation-based burn-in optimization as well as burn-in optimization under competing risks involving traumatic and degradation failures.The main contribution and innovation of this dissertation are listed as follows.(1)Aiming at products with/without wearout period,two failure-based accelerated burn-in optimization modeling methods are developed in combination with warranty and maintenance,by which problems in accelerated burn-in optimization for MOS capacitors and an electro-mechanical component for optical path control are solved.Failure model for products in accelerated burn-in is built using linear accelerated failure time regression model.Mathematical expressions of life distribution and failure rate are also derived.On this basis,a failure model for products without wear-out period is constructed using Weibull-exponential distribution and an accelerated burn-in optimization model involving one-dimensional mixed warranty is proposed.In addition,for the product including wear-out period,a failure model is established based on mixed Weibull distribution and a model for joint optimization of accelerated burn-in and block replacement is developed in combination with minimal repair.The proposed approaches and models are applied to solve the problems in failure modeling,parameter estimation,and accelerated burn-in optimization for MOS capacitors and an electro-mechanical component for optical path control in “SHENGUANG III” large laser facility.The results show that the accelerated burn-in has the ability to shorten testing time noticeably as compared with burn-in under normal stress.Methods and models proposed in this dissertation make up for the deficiency of traditional failure-based burn-in optimization in the aspect of accelerated failure modeling and has practical value for the application of traditional burn-in methods in engineering.(2)Two degradation-based burn-in optimization models with screening efficiency restriction are proposed in combination with warranty and maintenance,which is on the basis of in-depth study on the failure model and screening process of burn-in testing under normal and accelerated stresses for Wiener degradation failure products.In addition,the burn-in optimization problems in metallized film capacitors are solved.For degradation-based burn-in under normal stress,the life distribution of Wiener degradation failure products is analyzed in-depth.In the case of a batch of products composed of defective items and normal items being burnt-in simultaneously,a decision rule to screen out defective products is established based on the degradations after burn-in and expressions of reliability function and residual life distribution for items passed through the burn-in test are also given.On this basis,two dagradation-based burn-in optimization models with screening efficiency constraint are proposed in view of two types of misclassification risks existing in burn-in.One model chooses the maximum expected profit as optimization objective from the manufacturer's point of view,taking selling price and warranty cost into account and the other maximizes the expected net yield in terms of the whole benefit of manufacturer and user,taking gain of products in actural use and age replacement into consideration.For burn-in under accelerated stress,accelerated degradation model is studied and two degradation-based accelerated burn-in optimization model are proposed,which are similar to the case of burn-in under normal stress.At last,the optimal burn-in for metallized film capacitors in “SHENGUANG III” large laser facility is conducted using the proposed methods and models.The results show: as compared with failure-based burn-in,degradation-based burn-in,especially under accelerated stress,can not only shorten testing time greatly,but also reduce harmful effect on the normal products and increase profit or yield.This research furnishes novel burn-in optimization methods for highly reliable,long-lifetime products and has merits in both theory and application.(3)Aiming at the typical competing risks products involving traumatic and degradation failure,failure models under normal and accelerated stresses are established,a maximum likelihood estimation method for unknown parameters is given and the burn-in optimization problem in pulsed xenon lamp is solved.In the case of degradation path following Wiener process,competing risks model is built when traumatic failure is related to the degradation using proportional hazard model.A maximum likelihood estimation method is given for the unknown parameters.For burn-in testing of this typical competing risks products under normal and accelerated stresses,two types of burn-in optimization models are proposed,one of which chooses reliability criteria including mission reliability and the mean residual life(MRL)as objectives to optimize burn-in time,and the other optimizes burn-in time and age replacement interval simultaneously by choosing the average cost rate as objective.The problems in failure modeling,parameter estimation and burn-in optimization for pulsed xenon lamp in “SHENGUANG III” large laser facility are solved using the proposed methods and models.Comparative analysis is also conducted with other optimal policies in which only burn-in or preventive maintenance is considered.The results show: burn-in can notably enhance mission reliability of the pulsed xenon lamp;age replacement is more efficient than burn-in in the aspect of reducing the average cost rate of the pulsed xenon lamp;as compared with burn-in under normal stress,accelerated burn-in is more efficient.The above studies deepen the research on optimal burn-in method for products under complex failure process,particularly competing risks involving traumatic and degradation failures and have significantly theoretical and practical meaning.