基于模糊聚类神经网络的电站锅炉故障诊断研究  

Faulty Diagnosis Research for Power Plant Boiler Based on Fuzzy Clustering and Neural Networks

作　　者：牛培峰[1] 张泽[1] 王怀宝[1] 

机构地区：[1]燕山大学电气工程学院计算机控制工程河北省重点实验室,秦皇岛066004

出　　处：《微计算机信息》2010年第7期40-42,共3页Control & Automation

基　　金：基金申请人:牛培峰;项目名称:循环流化床锅炉燃烧过程聚类融合控制研究;基金颁发部门:国家自然科学基金委(60774028)

摘　　要：建立了电站锅炉故障样本及其故障诊断的模块化模糊网络模型,并用可能性聚类算法对所建模型进行了样本聚类,有效地避免噪声对聚类结果的影响,保证了聚类正确性。实验结果表明其网络训练速度和精度明显提高,解决了BP网络应用于复杂系统故障诊断时容易陷入全局最小问题。且该网络采用多输出结构不仅能对故障是否发生进行故障诊断,而且还能判断故障发生的严重程度。A fuzzy modular networks model and the samples presented for the fault diagnosis of power plant＇s boiler.Due to possibilistic sample clustering algorithm,the noise was eliminated effectively,which ensure the correctness of result.The test result indicates that speed and precision of sample training are increased because of sample clustering for fuzzy modular networks,the problem of slow training speed and minimum point are avoided when BP networks are applied in the fault diagnosis of complex boiler.Multi output of the fuzzy modular networks can decide whether the fault of boiler happens,and the severity degree of boiler fault.

关 键 词：锅炉 模块化模糊网络 模糊聚类 故障诊断 

分 类 号：TP[自动化与计算机技术]