Reserch On Product Reliability Evaluation Based On Multi-source Heterogeneous Information Fusion

In reliability engineering,most new products or materials have the characteristics of long life and high reliability.Therefore,it is very important for the development of science and technology and the renewal of products to adopt appropriate methods to evaluate the reliability of products based on less on-site data.Therefore,in order to comprehensively utilize all feasible life information,three information fusion models are established and gradually improved in this paper to realize the variable weight fusion of heterogeneous life information from multiple sources,which can be applied to the reliability assessment of small samples.The main research content of this paper includes the following three aspects.Firstly,under the condition that each group of heterogeneous life information obtained by natural conditions is independent from each other,the multi-source heterogeneous life information is fused with equal weight by combining Bayesian estimation and maximum likelihood estimation,and a convenient reliability evaluation model is proposed,and the validity of the model is verified by simulation.Secondly,because the sample information obtained from each information source has different degree of reference,the weight of each group of life information can be predicted by the tail gradient feature of likelihood function.The field data is integrated into prior information,considering the influence of weight on multi-source fusion,combined with Bayes method,the multi-source life information is fused with variable weight,and the parameter fusion results and product reliability evaluation analysis are obtained.Simulation examples show that the weight assignment method is reasonable,and the evaluation effect of the variable weight fusion model is better than that of the equal weight fusion model,which can effectively improve the evaluation accuracy.Finally,considering the applicability of the model,the accelerated test data under constant addition and step addition were incorporated into the pseudo-life data under normal stress level by the least square method and acceleration factor method.In these and other rights and variable weight fusion model,on the basis of the pseudo data service life and the life of the natural environment information respectively as two parts as a whole,put forward in three different fusion in order to evaluate the reliability of the variable weight fusion model,can be used to accelerate the test data and hardware-in-the-loop simulation,field data and other information service life of variable weight fusion.Simulation further verifies the rationality of the weighting method,and the three different fusion models are effective,which can avoid the analyst’s subjectivity.