基于IGA-SVM的汽轮机故障诊断研究  

Research on Fault Diagnosis of Steam Turbine Based on Improved Genetic Algorithm Optimization-Based Support Vector Machine

作　　者：孙凯[1] 田国清[2] 田宏[3] 段文超[1] 田洋[1] 陈立军[1] 

机构地区：[1]东北电力大学自动化工程学院,吉林吉林132012 [2]华能新华发电有限责任公司,黑龙江大庆163815 [3]包头钢铁职业技术学院自动化系,内蒙古包头014010

出　　处：《东北电力大学学报》2012年第3期26-30,共5页Journal of Northeast Electric Power University

摘　　要：针对支持向量机相关参数很难预先确定合适的取值,而这些相关参数又对其分类精度有着很大影响的问题,本文利用改进的遗传算法(IGA)对支持向量机的相关参数进行了优化。将改进遗传算法优化的支持向量机(IGA-SVM)算法应用于汽轮机故障诊断中,并与标准遗传算法优化的支持向量机(GA-SVM)算法的识别结果进行比较。结果表明,IGA-SVM算法对故障数据能够得到较优的分类辨识结果,对汽轮机的故障诊断有显著的指导作用。Aiming at the shortcoming that the parameters of support vector machine are not easily chosen, which have a big effect on classification accuracy, this paper utilizes the improved genetic algorithm （IGA） to optimize the parameters of SVM. Finally, the improved genetic algorithm to optimize the support vector machine （IGA-SVM） algorithm is applied to the fault diagnosis of steam turbine, and compared with the standard genetic algorithm to optimize the support vector machine （GA-SVM） algorithm identification results. The resuits show that IGA-SVM algorithm can obtain better classification result than the standard GA-SVM algorithm, which has a significant guidance on the fault diagnosis of steam turbine.

关 键 词：支持向量机 遗传算法 参数优化 汽轮机 故障诊断 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]