Reliability Modeling And Misspecification Analysis Base On Inverse Gaussian Process

Nowadays,the quality and reliability of products has been gaining more and more attention.It is has become the key issue on the research on how to assess the reliability of products accurately by using related information.Compared with traditional methods which are based on lifetime data,the reliability assessment methods based on the performance degradation of products are more suitable to those products with high quality and long lifetime.For high reliability and expensive-test-cost products,it is difficult to get invalid data in a short period of the test even under the conditions of accelerated test.However,performance degradation data can be obtained in the operating period of the product via real-time measurement.In recent years,a number of studies about the reliability assessment methods based on the performance degradation have been conducted by experts.Most of the studies focused on products with single failure mode.The problem of competing failure hasn't gained enough attention yet.one of the hot spots and important development directions of the reliability study.Inverse Gaussian process is a random process with many excellent probability and statistics properties,befitting for modeling products with monotonous degradation trajectory.Unlike the Gamma process which has intuitive physical interpretation,IG process hasn't been widely studied and discussed in the course of performance degradation assessment to product reliability for a long time.This paper studies the reliability assessment and the MTTF modeling issue based on inverse Gaussian degradation process,and analyzes the influence on the estimated accuracy of product's mean time to failure when misspecification of the IG and Wiener process occurs.The main research works include the following aspects: 1.When a product only occurs degradation failure and the degradation process obeys the Inverse Gaussian process,the reliability assessment model and mean time to failure(MTTF)model of the product are established.And then parameters are estimated by using maximum likelihood estimation method.2.When the product occurs degradation failure as well as catastrophic failure,the reliability assessment model of competing failure and the method of parameter estimation are put forward which taking the relationship between the two kinds of failure into consideration.3.The influence on the estimated accuracy and precision of product's mean time to failure is studied when the IG process and Wiener process are misspecified.The models of evaluation indicators are built base on the models of MTTF and then the influence is analyzed via the indicators.The sensitivity analysis of parameters,sample size and terminal time is made at last.