基于XGBoost算法的汽轮机转子故障原因定位方法  

Fault Location of Steam Turbine Rotor Based on XGBoost Algorithm

作　　者：王新伟 钱虹 冷述文 杨宝清 WANG Xinwei;QIAN Hong;LENG Shu-wen;YANG Baoqing(School of Automation Engineering,Shanghai University of Electric Power,Shanghai 200090,China;Shanghai Key Laboratory of Power Station Automation Technology,Shanghai 200090,China;Huaneng Shandong Power Generation Co.,Ltd.,Jinan 250014,China;Huaneng Linyi Power Generation Company,Linyi 276016,Shandong Province,China)

机构地区：[1]上海电力大学自动化工程学院,上海200090 [2]上海市电站自动化技术重点实验室,上海200090 [3]华能山东发电有限公司,济南250014 [4]华能临沂发电公司,山东临沂276016

出　　处：《动力工程学报》2021年第6期460-467,共8页Journal of Chinese Society of Power Engineering

基　　金：上海市2019年度“科技创新行动计划”高新技术领域资助项目(19511103700);上海市科委地方能力建设资助项目(18020500900)。

摘　　要：提出了基于XGBoost算法的汽轮机转子故障原因定位方法,首先对由故障类型和相关参数组成的原始样本集进行特征分析,评估各特征的重要度,然后利用XGBoost算法构建汽轮机转子故障原因定位模型,利用转子故障数据对模型进行训练和测试,最后将具体的故障原因链接到故障知识库,采取相应的故障修复措施。结果表明:相比随机森林(RF)和梯度提升决策树(GBDT)模型,XGBoost模型可有效识别汽轮机转子3种故障类型下的9种故障原因,其分类准确率更高。A fault location method of steam turbine rotor was proposed based on XGBoost algorithm.Firstly,the characteristics of the original sample set composed of fault types and related parameters were analyzed to evaluate the importance of each feature.Then,the XGBoost algorithm was used to build fault location model of steam turbine rotor,so as to use rotor fault data to train and test the model.Finally,specific fault causes were linked to the fault knowledge base,based on which,corresponding fault repair measures were taken.Results show that compared with random forest(RF)and gradient boosting decision tree(GBDT)model,XGBoost model can identify 9 fault causes of turbine rotor under three types of faults effectively,which shows higher classification accuracy.

关 键 词：汽轮机转子 XGBoost算法 故障类型 运行参数 故障原因定位 故障知识库 

分 类 号：TK267[动力工程及工程热物理—动力机械及工程]