Product Reliability Modeling Based On Performance Degradation Data

In engineering applications,most of the systems fail under the joint action of multiple failure modes.In particular,the failure modes of complex systems interact with each other and the relationship is complicated.During the operation of the system,there are not only normal and failure states,but a continuous degradation process.Degradation failure is the degradation of the performance of the system with time.The system failure after exceeding a certain threshold is a random process.The performance degradation process occurs during operation and storage until the performance parameters in the system do not reach the values??required to complete the specified function.Modern highly reliable products often have very complex system structures,and there are many types of degradation,that is,one product has multiple components,each component has its own degradation process,and even one component can exhibit multiple degraded mode,which causes a variety of degradations in the product.The typical failure rate function presents a bathtub shape,so a single degradation mode for a component will also exhibit different degradation processes during the operation: early failure periods,mainly due to design errors,process defects,and assembly problems,or due to poor quality inspections and other reasons.The period of accidental failure is the best working phase of the product.During the wear failure period,product performance degradation is gradually deteriorated due to aging,fatigue and wear.This paper first studies the degenerate product modeling of the Wiener process model and the Gamma process model.It is considered that the Wiener process and the Gamma process are generally applicable to linearly degradation modeling,and in fact the product degradation process is mostly non-linear;and the product is affected by random factors in the manufacturing,materials,and operating environments so that shows individual differences.The differences between the two need to study the impact on product reliability assessment.Two time-scale transformations—mixed-effect Wiener process model,time-scale transformation—mixed-effect Wiener process and nonlinear mixed-effect Gamma process model are discussed.The two models not only take into account the nonlinearity of the degraded system and the individual degradation between products,but also avoid the shortcomings of complex and difficult to generalize the model.Secondly,taking the productwith two correlation performance mixed degradation indicators as an example,the Copula selection is carried out by scatter plot,correlation coefficient test and AICc criterion.A hybrid correlation process reliability modeling method based on Copula function selection is proposed.Finally,considering the degradation failure of many products is not a smooth single degradation process,but a two-stage or even multi-stage situation in the degradation process.For the degradation process with change-points,the reliability analysis of two-stage degradation failure is carried out.Firstly,the BIC information criterion is used to determine the interval where the change point is located.Then the Wiener process and the time-scale Wiener process are selected to analyze the reliability of the two-stage degradation failure.Considering the individual differences between products and the influence of external factors in the experiment,it is assumed that the change-point is a random variable obeying the normal distribution,and the estimated value of the change-point is obtained by using the minimum squared residual sum criterion for each sample,and all the change-points are obtained.The estimated value is regarded as a set of samples from the distribution of the change-points,and the parameters of the change-point distribution are obtained by the maximum likelihood estimation.Then consider the 3? interval of the change-point as the transition phase,and use the probability value of the change-point as the weight to the reliability of the first and second stages to obtain the reliability of the transition stage,and then establish the real-time reliability evaluation model of the process with change-point.