Comprehensive Reliability Analysis And Assessment Of System Under Multi-source Uncertainty And Complex Failure Characteristics

Reliability analysis and assessment of large-scale electromechanical systems such as spacecraft,naval vessels and computer numerical control machine tools is an important measure to ensure the safe and efficient operation of the system.However,with the continuous improvement of modern large-scale electromechanical systems’intelligence,digitization and integration,not only the number and types of components in the system have increased sharply,but also the functions and composition of components,modules and subsystems have become increasingly complex.In the process of design,production and service,due to factors such as incomplete experiments,design defects,processing errors,cognitive limitations and working environment,the uncertain information affecting the reliability of complex system is more diversified;at the same time,the interaction relationship among components,modules and subsystems is highly coupled,which makes the failure characteristics of complex systems more complicated.However,the current traditional reliability analysis and assessment methods are mainly aimed at random-parametric uncertainty and common cause failure of redundant systems,which do not meet the reliability assessment requirements of modern large-scale electromechanical systems characterized by multi-source uncertainties and secondary failure.Therefore,in order to effectively and accuratly analyze and assess the reliability of modern large-scale electromechanical systems,it is of great significance and value to carry out system reliability assessment which integrates multi-source uncertainties and complex failure characteristics.In order to solve the above-mentioned problems,this dissertation is funded by the project of National Natural Science Foundation of China.It is based on reliability assessment of multi-source uncertainties and common cause failure,and focuses on reliability assessment of multi-source uncertainties and secondary failure,and conducts research from the aspect of quantification and unification of multi-source uncertainties,reliability assessment under secondary failure and comprehensive reliability assessment methods.Finally,this dissertation establishes a comprehensive system reliability assessment framework which combines multi-source uncertainties and complex failure characteristics,and verify its feasibility in engineering cases.The main research work of this dissertation is as follows:(1)Conduct an Interval Bayesian Network(IBN)based on explicit analysis method andβ-factor model to realize the reliability analysis and assessment of complex systems under random-parametric uncertainty.IBN is an extended model of the traditional Bayesian Network(BN),which can effectively represent random uncertainty and parametric uncertainty.At the same time,to comprehensively evaluate the impact of common cause failure on system reliability,theβ-factor model is introduced into the IBN by adding independent nodes to the network,thereby establishing an IBN based on the explicit analysis method andβ-factor model.Aiming at the problem that the IBN constructed above cannot analyze multi-order common cause failure,with the help of Markov method to model dependent-failure systems,an inaccurate continuous-time Markov chain is constructed.Through simulation analysis and case verification,it is proved that the proposed method can effectively comprehensively evaluate the reliability of complex systems under multi-source uncertainties and common cause failure.(2)Construct a nonhomogeneous continuous-time Markov chain(NHCTMC)based on copula theory to realize the reliability analysis and assessment of system under secondary failure.The value of the state transition rate that characterizes the secondary failure in Markov model mostly relies on expert experience and subjective assumptions,so that the dependability of reliability assessment is low.To solve the problem,copula theory is introduced into continuous-time Markov chain,and the method of calculating Markov state transition rate through copula function is explained in detail.Then a NHCTMC based on copula function is proposed to realize the reliability analysis and assessment of the system under secondary failure.At the same time,in order to comprehensively evaluate the influence of random-parametric uncertainty on system reliability,interval values are used to characterize the uncertain parameters of component life distributions.Aiming at the state explosion problem faced by Markov chain modeling and solving,the scale of Markov chain is reduced with the help of hierarchical model.The simulation analysis and case verification prove that the method can effectively realize the reliability analysis and assessment of the system under secondary failure.(3)Construct a probability-box BN to solve the unified quantification problem of multi-source uncertainties in system reliability analysis.In view of the coexistence of multiple uncertainties in practical system reliability modeling,probability-box is used to uniformly quantify the uncertain parameters of multiple representation forms such as Dempster-Shafer Structure,probability distribution,interval distribution,and interval information.Combing the advantages of BN for uncertainty modeling and reasoning,a probability-box BN is proposed,and the reasoning mechanism of the network is clearly defined.Through simulation analysis and case verification,it is proved that the method can effectively realize the reliability analysis and assessment of the system under multi-source uncertainties.(4)Construct a probability-box BN based on copula theory to realize the comprehensive reliability assessment of system under multi-source uncertainties and deterministic secondary failure.In order to comprehensively consider the impact of various uncertain parameters on the system reliability,a probability-box BN based on copula function is established.This model converts the m-dimensional integration operation into 2~m difference operation,which has high computational efficiency.The simulation analysis and case verification prove that the above method can effectively realize the comprehensive assessment of system reliability under multi-source uncertainties and deterministic secondary failure.(5)Construct a probability-box BN based on affine algorithm to realize system reliability analysis and assessment that integrates multi-source uncertainties and non-deterministic secondary failure.Facing the more common non-deterministic secondary failure problem in practical engineering,the correlation analytical method based on copula theory is no longer applicable.In order to solve the above problems,a probability-box BN based on affine algorithm is proposed.By comparing with Frechet’s inequality,the calculation result of this method has less uncertainty and better effect.The case analysis proves that the method can effectively realize the comprehensive assessment of system reliability under multi-source uncertainty and non-deterministic secondary failure.From theoretical model,mathematical reasoning,simulation and case analysis,it has proved that the comprehensive system reliability assessment method constructed in the dissertation is effective and has high practical value and guiding significance for the reliability assessment of modern electromechanical systems.