Research On Validation Technology Of Prognostics Capability Of Equipment PHM System

The rapid development of equipment prognostics technology brings two problems,which need to be solved urgently.The first one is what the capability indexes of equipment prognostics system are.The second one is how these indexes can be validated.Improving the validation technology of the prognostics system capability,it provides technological support to the contract indexes and validate the capability indexes by two acquisition sides.Obeying the statistical rules,it is the most urgent problem to construct a complete test flow and method of equipment prognostics system.Currently,the PHM system prognostics capability validation technology is mainly for verifying the prognostics capability of a single prognostics algorithm or a single product.There are three main problems in the existing research.Firstly,it can not meet the multi-failure mode or the overall prognostics capability validation of complex products.Secondly,there are many validation indexes for the single prognostics algorithm or single product prognostics ability.Thirdly,obeying the statistical test rules,how to design the PHM system prognostics ability verification test scheme has not been fully studied.From the system perspective,this thesis focused on solving the two core problems of “what to validate” and “how to validate” on the prognostics capability validation of a PHM system.Two metrics for prognostics capability of PHM system are proposed.On this basis,the overall design flow of the prognostics capability validation scheme of the PHM system is proposed.Moreover,the failure mode-degradation process two-dimensional sample size determination,sample allocation and selection method for the prognostics capabilityof the equipment PHM system is deeply studied.The research content and research conclusions of the thesis include:1.Construction of the prognostics capability indexes of a PHM system and design of validation test planFrom the perspective of equipment system,the paper analyzes the two-dimensional characteristics(failure mode dimension and degradation process dimension)of the prognostics sample.Two indexes describing prognostics capability of the equipment PHM system is proposed,which takes the failure occurrence time as the output target.Among them,the relative accuracy is an index that measures the accuracy of the PHM system for the time of failure occurrence.The FPR(failure prediction rate)is a counting index,which measures the accurate number of times of the PHM system predicts the time of failure.According to the engineering practice requirements of the prognostics capability validation of a PHM system,the overall work flow of the equipment design of the PHM system prognostics capability validation test plan is proposed.2.Research on time-varying characteristics of prognostics indexes based on stochastic processAnalyzing the prognostics capability index of the equipment PHM system,the occurrence characteristics of each failure mode will affect the variation characteristics of the overall system prognostics index along with time.The existing sampling test theory can only solve the situation that the index does not change during the whole time.Based on the failure occurrence rate model under the three maintenance behaviours,including the perfect maintenance,minimum maintenance and imperfect maintenance,the paper studies the time-varying characteristics of the prognostics index of the multi-failure mode system.The results show that under the perfect maintenance condition(constant failure occurrence rate),the statistical characteristics of the system prognostics and prognostics index do not change along with time.Under the two maintenance conditions of minimum maintenance and imperfect maintenance,the statistical characteristics of the prognostics capability index change with time.Existing sampling test theory cannot meet the validation of such index.Therefore,the thesis assumes that the system prognostics capabilityindex does not change,which does not consider the minimum maintenance,imperfect maintenance and other complex maintenance behaviours.3.Design of failure prognostics capability validation test plan of failure mode dimensionThe core prognostics capability validation test plan of a PHM system of the failure mode dimension is sample size,decision criteria determination and sample size allocation.(1)Sample size and decision criterion determination of the failure mode dimension method based on hypothesis testing theory●Sample size and decision criterion determination method based on the double side risks and sequential theoryOn the basis of assuming that the relative accuracy of prognostics sample obeys the normal distribution,a method for determining the fault model dimension sample size and decision criteria based on normal distribution is proposed,which considers the risks of both the purchaser and the contractor.Aiming at the problem that the sample size requirement of the fixed experiment theory is large,the sample size determination method is given based on the sequential test theory under the assumption of normal distribution.●Sample size determination method based on the improved Bayes posterior riskThe sample size determination of the failure mode dimension by the classical statistical theory has a problem,which lead a too large sample size.Moreover,the traditional Bayes posterior method has a problem that the prior distribution is incompatible with the field data.In order to avoid the problem that the prior distribution is incompatible with the field test data.Based on the compatibility test method,a sample determination method of the failure sample dimension based on the improved Bayes posterior risk is proposed.Compared with the sample size determination method based on the two-party risk based on classical statistical theory,this method can reduce the sample size more effectively.(2)Sample comprehensive weighted allocation of the failure mode dimensionSolving the failure mode dimension sample allocation problem,the various factors affecting the prognostics capability validation of a PHM system must be fully considered.Based on the three factors of fault occurrence rate,severity,and failure impact degree,the failure mode dimension sample comprehensive weighted allocation method is proposed,which is to determine a set of failure mode samples of good representativeness and excellent sample structure.4.Design of prognostics capability validation test plan of the degradation process dimension(1)Sample size and decision criterion determination of the degradation process dimension based on hypothesis testing theoryDue to the delay of the prognostics samples,the degradation process samples can only be determined by the fixed sample size determination method.Considering whether the equipment system can obtain the relative accuracy and risk constraints of each single failure mode,the paper focuses the sample size determination method of the degradation process based on the risk of both parties on two cases.(2)Sample selection of the degradation process dimension based on the piecewise Wiener processAiming at the problem of sample selection in the degradation process dimension,a comprehensive weighted selection method of the prognostics degradation processes based on the degradation process feature parameters as the selection weighting factor is proposed.Based on the historical degradation data,a failure degradation process model based on the segmentation Wiener process is established to describe the initial degradation amount,the degradation amount increment,the degradation rate and the variance of each degradation stage of the failure degradation process.The parameters are used to determine the sample selection weights of each degradation stage of the failure degradation process by using the four parameters as the sample selection factors,which realizes the sample optimization selection of the degradation process dimension.