基于kNN-SVR模型的风电功率预测  

Wind Power Prediction Based on kNN-SVR Model

作　　者：徐正华 刘三明[1] 王致杰[1] XU Zheng-hua;LIU San-ming;WANG Zhi-jie(Shanghai Dianji University,Shanghai 201306,China)

机构地区：[1]上海电机学院

出　　处：《电力学报》2019年第5期411-416,共6页Journal of Electric Power

基　　金：国家自然科学基金（11201267）;上海市教育委员会科研创新项目（15ZZ106）

摘　　要：风能是一种清洁的可再生能源能源,已开始在我们的日常生活中使用。由于风能是一种高度波动的资源,所以风电功率很难预测。研究提出一种基于最近邻支持向量回归(kNN-SVR)模型的多风速融合方法,提高了短期风电预测的精度。首先,设计了一种kNN算法,从历史数据集中选取最接近预测点的历史风速点。然后,在此基础上,建立基于最近历史预报点三个独立数值天气预报的融合风速SVR模型,提高风速预报的精度。最后,利用融合风速模型,以及反向传播神经网络(BPNN)对风电功率进行预测。通过算例分析验证该方法的有效性。Wind power is a kind of clean and pollution-free renewable energy,and start to be used in our daily life.However,wind speed is a highly volatile resource.So wind power is difficult to predict.A multi-wind speed fusion method based on the nearest neighbor support vector regression(kNN-SVR)model is proposed to improve the accuracy of short-term wind power prediction.Firstly,a kNN algorithm is designed to select the historical wind speed point closest to the predicted point from the historical data set.On this basis,an integrated wind speed SVR model based on three independent numerical weather forecasts of recent historical forecast points is established to improve the accuracy of wind speed forecast.Finally,the fusion wind speed model and the back propagation neural network(BPNN)are used to predict the wind power.The effectiveness of the method is verified by an example.

关 键 词：kNN-SVR模型 风电功率预测 反向传播神经网络 

分 类 号：TM715[电气工程—电力系统及自动化] TM614