基于Bi-LSTM和支持向量机的风机叶片短期覆冰状态预测模型

Short-term Icing Status Prediction Model of Wind Turbine Blades Based on Bi-LSTM and SVM Models

摘要:

风机叶片覆冰灾害严重危害风电场安全经济运行，对风机叶片覆冰状态预测是预防覆冰灾害的有效手段。针对传统覆冰状态预测方法精度较差问题，本文基于风电场SCADA监测数据，提出一种基于Bi-LSTM和SVM的风机叶片短期覆冰状态预测模型。首先采用PCA对风机叶片覆冰状态监测特征指标进行降维，筛选可以反映风机叶片覆冰状态的特征指标；其次基于大量历史数据，对Bi-LSTM和SVM模型进行训练，训练结果表明模型有较好精确度；最后将多组实际数据集输入Bi-LSTM预测模型，预测输出值输入SVM模型，对风机叶片是否会出现覆冰故障进行判别。结果表明所提方法可准确实现叶片覆冰状态预测，准确率可达95%。

Abstract:

The icing disaster of wind turbine blades severely impair the safe and economic operation of wind farms, while the prediction of icing status of wind turbine blades is an effective method to prevent icing disaster. Based on the data of supervisory control and data acquisition (SCADA), an short-term icing status prediction model based on bidirectional long short-term memory (Bi-LSTM) and support vector machine (SVM) is proposed to solve the problem of inaccuracy of the traditional prediction method. Firstly, principal component analysis (PCA) are employed to reduce the dimension of characteristic index of wind turbine icing status, and the characteristic index that can reflect the fan blade icing status are screened. Secondly, Bi-LSTM and SVM models are trained based on a large number of historical data, and the training results show that the model has good accuracy. Finally, SVM is used to predict the icing status of the Bi-LSTM forecasting output data set to judge whether the fan blade will have icing failure. The results show that the proposed method can accurately predict the icing status of blades with an accuracy of 95%