故障机理的不确定传播及系统故障自动推演建模方法研究

With the urgent requirement of large scale engineering system reliability design, analysis and evaluation, this project aims at deepen the understanding of system failure and evolution, explore failure mechanism uncertainty propagation rules and failure scenario automatic reasoning system modeling method. Based on the academic theories of system science, Physics-of-Failure and uncertainty qualification, interpret the laws during the occurrence, development of failure mechanisms, illuminate and qualify the uncertainty of failure mechanism propagation under the synthetic contribution of environment, load, structure, material, manufacture defection and human factors. Combine history data and domain knowledge to excavate the association rules of failure mechanism uncertainty propagation. Construct failure mechanism evaluation model using multi-level asynchronous and heterogeneous cellular automata method and structure modeling method to explore the emergence feature of failure from microcosmic mechanism to macroscopic system. Failure route identification method, system failure route sensitivity analysis and route blocking or impeding strategy is proposed based on route searching theories. Case study and modeling verification will be carried out with FADEC system. Results can not only provide new theory and method for reliability assurance for complicated engineering system, but also offer technology support for formulating maintenance and supportability strategy.

本项目以大型工程系统的可靠性设计分析和预测评估需求为研究背景，以提高系统故障发生和演化规律认识为目的，探索故障机理不确定传播规律及故障场景演化自动推理建模新方法。基于系统科学、故障物理以及不确定性理论，诠释故障机理发生、发展过程中可能的机制，阐明在环境、载荷、结构材料、工艺缺陷、人因等的综合作用下，故障机理传播的不确定性并加以量化。将历史数据与领域知识相结合，挖掘故障机理不确定传播的关联规则。利用多层异步异质元胞自动机及结构化建模方法，建立故障机理演变的传播模型，探索故障从微观机理到系统涌现过程的特征。基于路径搜索理论提出故障传播路径识别、系统故障路径敏感性分析及路径阻断/阻滞决策设计方法。针对全权限发动机数字电子控制系统进行案例应用与验证。研究成果不仅可为复杂工程系统的可靠性保证提供新的理论方法，也可为制定维修、保障策略提供有效的支撑。

本项目以大型工程系统的可靠性设计分析和预测评估需求为研究背景，以提高系统故障发生和演化规律认识为目的，探索故障机理不确定传播规律及故障场景演化自动推理建模新方法。分析单点故障机理可能的耦合特性，提出载荷和时域多场耦合作用下故障交互理论，构建两两相互作用的故障机理演化为主线的系统故障模型，描述简单系统在复杂环境-材料-结构交联系统中故障机理逐渐发展直至引发系统故障的规律性特征。将历史数据与领域知识相结合，利用FP-growth算法和最大频繁项数，挖掘故障机理不确定传播的关联规则。研究了多层异步异质元胞自动机建模方法流程，建立了故障机理演变的传播模型。基于深度优先算法和A*优化算法实现故障场景及路径的智能识别和推理，并分析了系统故障路径的敏感性。针对发动机数字电子控制系统进行案例分析，通过自动推理的方法获得了系统的故障传播路径,同时评估系统的可靠性，为维修性和保障性提供了技术支撑。