基于卷积神经网络与LightGBM的短期风电功率预测方法

考虑到风力发电存在的波动性和不确定性,提出一种基于卷积神经网络(CNN)和LightGBM相结合的风力发电机功率预测模型。先对相邻风电机组原始数据进行时序特征相关性分析,构建新的特征集;其次,应用CNN从输入数据中提取信息,并通过比较实际结果调整网络参数;再次,考虑到单一卷积模型在预测风电时的局限性,将LightGBM分类算法集成到模型中,从而提高预测的准确性和鲁棒性;最后,将提出的算法与已有的支持向量机、LightGBM、CNN进行仿真对比,结果表明所提出的融合模型具有更好的精度和效率。

Short-term Wind Power Prediction Based on Convolution Neural Network and LightGBM Algorithm

Considering the fluctuation and uncertainty of wind power generation.a wind power forecasting model based on convolution neural network and lightgbm was proposed.First of all,the time series feature correlation analysis of the original data of adjacent wind turbines was carried out to build a new feature set.Secondly,the convolution neural network(CNN)was used to extract information from the input data and adjust the network parameters by comparing the actual results.Then,considering the limitations of a single volume model in the prediction of wind power.the lightgbm classification algorithm was integrated into the model to improve the pre prediction The accuracy and robustness of the measurement.Finally.the proposed algorithm is compared with the existing support vector machine.lightgbm and CNN,and the results show that the proposed fusion model has better accuracy and efficiency.