多状态条件下复杂电子装备状态监测与故障诊断新方法研究

Presently, complex electronic equipment with a multi-state,multi-parameter characteristics, the current working state of equipment cannot be reflected and real-time condition monitoring and failure prediction cannot be implemented by the traditional fault diagnosis method. Hereto, the condition monitoring, fault diagnosis and failure prediction technology of integrating multi-source information is researched in this application. On the basis of reasonable analysis of key components and important parameters,combined to equipment BIT（Built-in Test，BIT）information, fuzzy sets partition based optimal sensor configuration of equipment is researched,and the effectiveness of multi-source information is improved. According to regularity and sensitivity of the information data, the status information pretreatment technology is investigated,the automatic feature selection or extraction model is builded to classify the feature samples,and the condition monitoring technology based on multi-source status information fusion is researched. For the multi-state signal charictistics of complex equipment with non-stationary and uncertain, a kind of comprehensive diagnosis method based on the state information and D-S evidence theory is researched. Through the information determine model and the fault diagnosis model, choice is made according to the degree of conflict between the state information, to correctly reflect the differences of the various information on the judgement, and to improve the accuracy of fault diagnosis. The application has important academic value and prospective applications. The application has important academic value and prospective applications.

针对复杂电子装备具有多状态、多参数的特性，传统故障诊断方法不能反映装备的当前工作状态、更不能对装备进行实时状态监测和故障预测的现状。该申请研究多状态条件下集状态监测、故障诊断与预测等功能于一体的复杂电子装备状态监测与故障诊断技术。在合理分析装备关重件和重要参数的基础上，结合装备自身机内测试（Built-in Test，BIT）信息、研究基于模糊集分割的装备最优感知器配置，提高多源信息的有效性。针对信息数据的规律性、敏感性，探讨状态信息预处理技术，构建自动的特征选择或提取模型，进行特征样本分类，研究基于多源状态信息融合的状态监测技术。针对复杂装备的多状态信号具有非平稳、不确定的特点，研究基于状态信息和基于D-S 证据理论融合的综合诊断方法，通过信息判断模型和故障诊断模型，根据状态信息间的矛盾程度进行选择，从而正确反应各信息在判断上的差异，提高故障诊断的准确性。具有重要的理论价值和应用前景。

项目以多状态复杂电子装备为对象，主要研究了综合诊断过程中的感知器优化配置、参数特征自动提取、以及利用各类多源数据信息综合诊断维修的理论与方法。并且，针对FPGA等大规模电路广泛应用于电子装备的现状，拓展了研究内容，探讨研究了数字电路故障自检测和自修复的方法。取得的主要研究成果有：①结合电子装备测试的特点，建立了节点优选问题的数学模型，提出了基于改进遗传算法的测试节点优选方法；在考虑各故障发生概率和测试代价的情况下，引入遗传算法中的变异方法，提出了分层加权的蚁群信息素更新策略，为大规模复杂系统的测试序列优化提供了一种新的解决途径。②针对故障诊断中常以测试信号参数和波形作为判断故障的依据，提出了一种改进粒子群的参数提取方法，使测试信号模型参数的提取具有更好的稳定性，提高了波形匹配的精度，为特征参数的自动提取提供一种新的方法。③综合考虑现场状态数据、备件数量、维修代价、装备的维修历史数据等因素，分别从“部件级”、“系统级”角度，提出了基于机会策略的多态系统视情维修方法和基于Semi-Markov模型的多态系统不完全维修方法，在保证系统可靠性的前提下，降低了系统的维修资源，提高了装备系统的任务执行能力。