Research On The Modeling Method For The State Degradation Of The Equipment Component Aiming At The Condition Based Maintenance

Condition Based Maintenance(CBM)is a preventive maintenance strategy based on the actual operating states of the equipment.The information in the operating process is recorded by the operation and maintenance data which is a foundation for accurately evaluating the states of the equipment components and effectively implementing the CBM.How to describe the complex coupling relations between the data and the states and built the degradation model is a problem in the implementation of the CBM.Based on the study of the relations among the data,the states and the stresses in the operating conditions and environment,a unified description framework of the relations among the data,the stresses and the states is established.For different stress(es)-data combinations,a modeling method for the state degradation of the component aiming at the condition based maintenance is proposed.The main contributions of this dissertation are summarized as follows.(1)Aiming at the problems of how to analyze the relations between the data and the states of the component,an analysis method for the causes of the states changes based on the stress environment model is proposed with formal and standardized ideas.The model is abstracted from the functional model of the component.By standardizing the expressions and the analysis process of the failure modes and failure mechanisms,the relations between the status and the data is established based on the model and the types of the data are also decided corresponding to the different failure modes.This method effectively avoid the arbitrariness of the traditional failure analysis methods,which may cause duplication and omissions in the analysis results.The method based on the stress environment model can form a unified modeling framework described by stresses,which is conducive to mathematical modeling under the different types of the data.(2)Based on the unified modeling framework,the modeling methods for the states degradation of the component aiming at four kinds of common stress(es)-data combinations are studied as follows.1)A modeling method for the type of the cyclic stress is proposed.Based on the characteristics of the cyclic stress,the degradation model of the component state is derived and established with the relevant data.The adaptability and the effectiveness of the method is verified through a case.2)A modeling method for the type of the multi-shocks is proposed.Based on the characteristics of the multi-shocks,two situations including single failure threshold and multiple failure thresholds are considered and the degradation model of the component state is derived and established with the relevant data.In the parameter estimation process,an approximate estimation method based on the conservative Markov state space for the hybrid phase-type distribution(PH distribution)and an approximate estimation method based on dynamic reconstruction of the PH distribution are proposed to solve the problem that the parameters cannot be decoupled in the likelihood function of the censored samples in both situations.The effectiveness of the method is verified through a case.3)A modeling method for the type of the continuous stress is proposed.Based on the characteristics of the continuous stress,a multi-stages degradation model of the component state is derived and established with the relevant data due to different degradation rates caused by the change of the shape and material in the degradation process.A fuzzy interval boundary is introduced in order to define the boundary between two adjacent degradation stages.The effectiveness of the method is verified through a case.4)A modeling method for the environment including both shock and continuous stress is proposed.Based on the characteristics of the shock and the continuous stress,the degradation model of the component state is derived and established with the relevant data.In the parameter estimation process,an estimation method based on extending Markov state space with variable transition rates is proposed in order to solve the problem that the transition rates of the phases in the Markov state-space can not satisfy the conservative properties of the infinitesimal generator.The effectiveness of the method is verified through a case.In summary,with the unified modeling framework,the relations between the data and the states of the component are established based on the stress environment model and the degradation modeling methods of the states under four common stress(es)-data combinations are introduced.The research findings provide theoretical and methodological support for expanding the capability of the existing CBM,and also provide reference for the subsequent relevant researches.