Reliability Analysis And Maintenance Strategy Research On Multi-state Complex System

In recent years, with system towards larger size, more complex, higher precision, as well as the further understanding of system failure characteristics and mechanism, it has been observed that multiple states occur during the failure processes of systems and components. These systems and components cannot easily be divided into states of success and failure, and the traditional two states theory is unable to resolve issues of reliability and maintenance. The multi-state system (MSS) is an important theory to solve the problems existing in the complex system reliability analysis. Multi-state reliability theory can not only reveal the essential characteristics of complex system state evolution, but can also provide qualitative and quantitative performance analysis for MSS.The goal of this paper is to discuss some problems in the development process of the MSS reliability theory. Research has been conducted with the following aspects:system reliability modeling, quantitative calculation, dynamic reliability analysis, importance evaluation for multi-state components, and maintenance strategy optimization for MSS with multiple redundant components. The main research contents and results are as follows:(1) Probability Matrix Algorithm (PMA) is introduced to simplify reliability modeling and quantitative analysis for complex MSS on the basis of the improved GO Methodology. Two concepts, the state probability matrix and the joint state probability matrix, are introduced. Through the multiplication of these probability matrices, ali possible state combinations between operator and signal flow are easily determined. Details on the new formula derivation of the frequently-used GO operators are discussed based on operators' inherent rule. The new algorithm easily analyzes the state coupling relation, and also has the advantages while using intuitive formulas that are easily programmed. A new algorithm is presented through the establishment of equations using probability matrixes to avoid the shortage of the existing correction algorithm for shared signals. The new correction algorithm not only avoids deriving the complex mathematical formula, but also breaks the number limitation of shared signals and state dimension. Finally, system reliability models of some three-axis Inertial Navigation System (INS) and system static reliability analysis through the simulation program are provided. The results show that the proposed measures, including PMA and the shared signals correction algorithm are not only feasible, but effective.(2) A new GO operator called the multi-state operator is put forth in order to improve the accuracy and flexibility of GO Methodology in building reliability analysis model for complex MSS, and the quantitative formula of this operator based on PMA is given. The proposed multi-state operator can realize a variety of complex coupling relationships between input and output signals. In principle, it can fully instead of the other GO operators, and is not sensitive to the state dimension and algorithm complexity of components. The multi-state operator makes up for the shortage of the existing GO operators, and improves GO Methodology theory for MSS.(3) Combining with GO Methodology and Markov process theory, state transition equation about reliability parameters are built to achieve system dynamic reliability while changing over time. The dynamic reliability analysis model for components and system are subsequently presented. As long as the instantaneous state probability distribution of components can be obtained combining the periodic covariant sampling, the state probability matrix of each operator and the dynamic reliability of system can be determined. Finally, the proposed measure is applied to the Strap-down Inertial System (SINS). Two proving simulations are processed when components instantaneous state transition probability are set as constants and variables. The results show that the proposed dynamic reliability analysis algorithm is feasible, and the evaluation results can be used as a credible support for real-time performance judgment to system and components.(4) Aiming at the deficiency of the existing importance evaluation for multi-state components and the emphasis points about components in maintenance activities, the concept of Maintenance Importance (MI) for multi-state components is introduced. MI arises from the probability important degree in fault tree analysis, risk evaluation of failure mode, and Birnbaum importance for multi-state components. It balances three factors:risk degree, components importance to system reliability, and detection difficulty. When combined with real-time operating conditions, MI can provide quantitative maintenance priority results for all components at any given point. The evaluation results of MI has more comprehensive and higher degree of confidence, even though it only uses a single evaluation index, when compared with the other important evaluation methods. This is due to the consideration of the structure relations and the economic relationship between components. The proposed importance evaluation method is applied to some SINS, and maintenance priorities of all components are easy to obtain during each maintenance interval. The evaluation results can provide maintenance sequence data for the following activities.(5) A comprehensive platform is constructed to manage system health through combining the GO Methodology and the concept of Reliability Centered Maintenance (RCM). Reliability assessments, residual life predictions, and components of MI estimation are fully integrated in this unified platform. The platform keeps pace with the actual system through perpetual covariant sampling in order to obtain dynamic reliability analysis of components and system. Accurate performance assessment of system and components are achieved through the prediction of residual life based on dynamic reliability. Since redundant components are taken into account, maintenance actions are triggered only when system reliability fails to meet the set threshold, thereby unnecessary (?)entive maint(?) is reduced, while system availability is increased. A health management platform is established for SINS based on GO Methodology. The simulation results prove that the proposed management platform is feasible, and its evaluation results can provide support for the follow up maintenance strategy-making.(6) A dynamic rolling horizon grouping maintenance strategy based on components MI and GO Methodology is introduced for complex structure systems with multiple redundant components and multiple states. Maintenance time is defined when system reliability fails to meet the set threshold, and maintenance sequence is determined by components maintenance importance. The maintenance scope and maintenance task are optimized by an established maintenance unit time cost model. The proposed dynamic grouping maintenance strategy can be divided into two stages. In the first stage, quantitative components MI are calculated to order the maintenance sequence of components. In the second stage, an optimization model is applied to select the optimal group of maintenance components and the corresponding maintenance task. In addition, grouping strategy for maintenance with many ways has been discussed, and the corresponding maintenance strategy optimization model has been also established. A maintenance unit time cost model is built for SINS and the corresponding maintenance scope and maintenance task are made with goals of optimizing the unit time cost. Compared with the traditional grouping maintenance strategy, the dynamic maintenance strategy in this paper greatly simplifies the online optimization procedure, thus improving the real-time processing performance.A series of measures, including a new quantitative assessment algorithm for system reliability, a correction algorithm for shared signals based on PMA, multi-state operator for the application of multi-state components modeling, and dynamic reliability analysis model for MSS, are put forward to focus on the reliability modeling and analysis for complex MSS. In view of predictive maintenance of complex MSS, a concept of maintenance importance for multi-state components is introduced; health management platform based on GO Methodology and a dynamic rolling horizon grouping maintenance strategy are put forward. The proposed theories and algorithms are applied to complex INS, and the achievements of theoretical innovation and practical application are obtained. The research results resolve some existing problems of reliability theory about complex MSS, while enriching and improving the theory of GO Methodology. Basic theory and technical support for system reliability analysis and maintenance strategy optimization of MSS are adequately discussed.