基于XGBoost和自适应阈值的电厂风机故障预警  

Power Plant Fan Wind Turbine Fault Early Warning in Power Plant Based on XGBoost and Adaptive Threshold

作　　者：夏文苗 黄伟 XIA Wen-miao;HUANG Wei(School of Automation Engineering,Shanghai University of Electric Power,Shanghai 200090,China)

机构地区：[1]上海电力大学自动化工程学院,上海200090

出　　处：《计算机仿真》2023年第2期108-112,共5页Computer Simulation

基　　金：上海市“科技创新行动计划”地方院校能力建设专项项目(19020500700);中国华能集团有限公司2019年度科技项目(K-522019007)。

摘　　要：为了对电厂风机实现故障预警,提出了基于极端梯度提升(XGBoost)算法的数据驱动的故障预警方法。首先,通过对电厂原始数据进行数据特征提取和Box-Cox变换,建立基于XGBoost算法的风机轴承温度预测模型;其次,将模型预测值和真实值的偏差用相似度函数表示,并设计了基于区间估计思想的自适应阈值方法;最后利用某电厂送风机数据进行仿真,并将XGBoost算法与支持向量机(SVM)算法、梯度提升树(GBDT)算法进行对比。结果表明该方法能实现风机早期故障预警,验证了该故障预警模型的有效性。In order to realize fault early warning for wind turbine in power plant fan,a data-driven fault early warning method based on extreme gradient boost(XGBoost)algorithm is proposed.Firstly,through the data feature extraction and box Cox transformation of the original data of the power plant,the fan bearing temperature model based on XGBoost algorithm iwas established;secondlySecondly,the deviation between the predicted value and the real value of the model is was expressed by through similarity function,and the adaptive threshold method based on interval estimation idea is was designed;finallyFinally,the data of a power plant blower is were used for simulation,and the XGBoost algorithm is was compared with the support vector machine(SVM)algorithm and the gradient lifting tree(GBDT)algorithm.The results show that the method can realize the early fault warning,and also verify the effectiveness of the fault warning model.

关 键 词：电厂风机 极端梯度提升算法 自适应阈值 相似性 故障预警 

分 类 号：TM743[电气工程—电力系统及自动化]