摘要
主泵是核电厂非常重要的设备,它直接关系到整个核动力装置能否安全运行,对其进行有效的故障诊断十分必要。支持向量机(SVM)具有使用较少的训练样本达到较好分类效果、不需要故障分类的先验知识的特点,可以应用于主泵的故障诊断。为此,首先使用小波变换提取某主泵的转子质量不平衡、转子不对中、碰摩等三种典型故障的故障信息,然后使用最小二乘支持向量机(LS-SVM)方法对故障模型进行训练,最后对训练得到的模型进行故障诊断。诊断结果较好,从而验证了该方法的有效性。
The main coolant pump is an important device which relates to the safety of the whole nu- clear power plant, so its fault diagnosis is very important. Because of the better ability to classify faults in using fewer training samples and no need for prior knowledge, SVM can be used for the fault diagnosis of main coolant pump. Thus, firstly the information about three typical kinds of faults of the main coolant pump was extracted by means of wavelet transformation. Secondly, the fault diagno- sis model was trained by use of LS-SVM. Finally, the trained model was used to diagnose the faults data. The diagnosis results were good which proves the validity of this method.
出处
《海军工程大学学报》
CAS
北大核心
2012年第5期82-85,共4页
Journal of Naval University of Engineering
