Abstract: In view of the complexity?of running status?and?the uncertainty of health?diagnosis, the fault prediction method of the large complex?equipment is studied. A fault prediction model based on extension clustering algorithm is put forward. In these models, the running status and the diagnostic data are described formalized?and modeled by matter element?model. Using the correlation functions of extension theory, the qualitative?and?quantitative?prediction analysis method of the equipments state is studied. Based on these, a rapid?prediction?of fault states?of complex equipment is achieved, and it provides?a support to the?computer aided design of?the equipment fault prediction. An example is provided to prove its feasibility and validity.

Key words: fault prediction, extension clustering algorithm, matter element, extension theory, aritificial intellegence

摘要： 针对大型复杂设备运行状态的复杂性和健康状态诊断的不确定性，研究了复杂设备故障预测问题，给出了一种基于可拓聚类方法的智能化复杂设备故障预测分析模型。该模型利用可拓理论进行被诊断设备诊断状态的物元建模，并基于物元模型进行诊断数据的形式化和模型化描述，利用可拓理论关联函数对复杂设备故障预测进行定性和定量相融合的分析，从而达到对复杂设备故障状态的快速预测，为设备维修的计算机辅助设计顺利实施提供支持。将模型与方法应用于某动力系统装备的实例中，验证了模型的有效性。

关键词: 故障预测, 可拓聚类, 物元, 可拓理论, 人工智能