核电站混合式故障诊断系统的开发和评价

Development and Evaluation of Hybrid Fault Diagnosis System for Nuclear Power Plant

提出了一种基于规则和人工神经网络的混合式故障诊断方法,用于在线诊断核事故。基于规则的诊断方法用于事故类型的识别,人工神经网络用于准确判断事故的具体位置以及严重程度等基于规则难以诊断的信息。正常工况下的仿真试验验证了该方法的有效性。

A hybrid approach based on rule-based method and Artificial Neural Network （ANN） is proposed for on-line fault diagnosis of nuclear power plant. Rule-based method is used for identifying accident type and ANN is used for accurately estimating some information that is difficult to be diagnosed by rule-based method, such as accident location and severity level. Validity of the proposed method was verified by a simulation experiment in normal operation condition.