基于特征工程的XGboost风速短期预测

Short-term XGboost wind speed prediction based on feature engineering

为了解决现有风速预测模型精度不足以及数据利用度不高的问题,实现更高精度更快速度的风速短期预测,提出了一种基于特征工程的极限梯度提升算法(eXtreme Gradient boosting, XGboost)风速预测模型。XGboost算法是一种boosting集成学习算法,具有精度高速度快的特点。所提方法通过提取风速序列的5个统计特征,并与原始风速序列进行结合获得新的模型训练输入集,从而实现数据的充分利用,并采用XGboost算法对新输入集进行风速预测,提高了模型的预测精度。以江阴市某风电场实测风速数据进行预测,预测结果表明:基于时间-特征序列的XGboost风速预测模型具有精确的预测结果与快速的训练速度,与长短期记忆网络(long short-term memory,LSTM)、门控循环单元(gated recurrent unit,GRU)、卷积神经网络-长短期记忆网络(convolutional neural networkslong short-term memory,CNN-LSTM)及轻梯度提升机(light gradient boosting machine,LightGBM)等4种算法预测的结果进行对比,结果表明所提方法更具有效性。

In order to solve the problems of insufficient accuracy and low data utilization of existing wind speed prediction models and realize the short-term wind speed prediction with higher accuracy and faster speed, an eXtreme Gradient boosting(XGboost) algorithm wind speed prediction model based on feature engineering is proposed. XGboost algorithm is a boosting integration learning algorithm, which has the characteristics of high precision and fast speed. By extracting five statistical features of the wind speed sequence and combining with the original wind speed sequence, the proposed method obtains a new model training input set, so as to make full use of the data. The XGboost algorithm is used to predict the wind speed of the new input set, which improves the prediction accuracy of the model. Based on the measured wind speed data of a wind farm in Jiangyin City, the prediction results show that the XGboost wind speed prediction model based on timecharacteristic sequence has accurate prediction results and fast training speed. Compared with the prediction results of long short-term memory(LSTM), gated recurrent unit(GRU), convolutional neural networks-long short-term memory(CNN-LSTM) and light gradient boosting machine(LightGBM), the results show that the proposed method is more effective.