XGBoost算法在风机主轴承故障预测中的应用

针对传统机器学习算法处理海量风机数据采集与监视控制(SCADA)监测数据效率低和准确度差的问题,提出利用极端梯度提升(XGBoost)算法预测风机主轴承故障。首先,对风机主轴承SCADA数据开展特征分析,挖掘和发现特征与故障之间的关联关系,并评估各特征的重要性;然后利用XGBoost算法构建主轴承故障预测模型,进行模型评估;最后,依据SCADA系统收集的实测数据对模型进行训练和测试,并调整XGBoost模型的主要参数,提高预测准确率。通过与经典梯度提升决策树(GBDT)算法诊断结果相对比,结果表明XGBoost在风机主轴承故障预测的效率和准确度方面均优于GBDT算法,是处理SCADA大规模数据集的有效工具。

Application of XGBoost algorithm in prediction of wind motor main bearing fault

To improve efficiency and accuracy of the processing of scale wind turbine SCADA data, XGBoost algorithm is used to predict the main bearing fault of the wind turbine instead of traditional machine learning algorithm. Firstly, the feature analysis is carried out on the SCADA data of main bearing, the relationship between the feature and fault is excavated and discovered, and the importance of each feature is evaluated. Then the XGBoost algorithm is used to construct the main bearing fault prediction model, and the model is evaluated. Finally, the fault prediction model is trained and tested according to the measured data collected by SCADA system, and the main parameters of the XGBoost model are adjusted to improve the prediction accuracy. Compared with the diagnosis results of classical GBDT(Gradient Boosting Decision Tree) algorithm, the results show that XGBoost is superior to GBDT algorithm in terms of efficiency and accuracy on wind turbine main bearing fault prediction, and it is an effective tool to deal with large scale data sets of SCADA.