考虑风速时空相关特性的元启发式支配预测模型  

Meta-heuristic Dominance Prediction Model Considering Wind Speed Spatio-temporal Correlation Characteristics

作　　者：潘超[1] 王典 蔡国伟 杨雨晴 于凤娇 PAN Chao;WANG Dian;CAI Guowei;YANG Yuqing;YU Fengjiao(Key Laboratory of Modern Power System Simulation and Control&Renewable Energy Technology(Northeast Electric Power University),Ministry of Education,Jilin 132012,Jilin Province,China)

机构地区：[1]现代电力系统仿真控制与绿色电能新技术教育部重点实验室(东北电力大学),吉林省吉林市132012

出　　处：《电网技术》2020年第11期4105-4113,共9页Power System Technology

基　　金：国家重点研发计划重点专项(2016YFB0900100);国家自然科学基金项目(51507028)。

摘　　要：为了提高大型风场风功率预测的精度和效率,构建动物种群仿生算法分别嵌入空间提取及多步预测双模块的组合预测模型。对于风速特征提取模块,利用基于动物寻巢行为的仿生算法嵌套卷积神经网络。优选典型风机位置信息,结合典型风机数据重构空间风速矩阵,通过降低输入数据复杂度以提高提取效率。利用卷积神经网络提取重构矩阵特征信息,实现风速特征信息降维。针对多步预测模块参数设置问题,采用基于动物繁衍行为的寻优算法优化支持向量回归模型参数,结合空间特征信息对风速进行多步预测,将所得风速预测值通过等效风能利用系数法进行风功率预测。最后,将文中方法应用于实际风场的风速及风功率预测,通过对比分析验证所提方法。In order to improve the accuracy and efficiency of wind power prediction in large-scale wind farms,a combination prediction model with the animal population bionic algorithm embedded into the dual module of the space extraction and the multi-step prediction is constructed.In the wind speed feature extraction module,a bionic algorithm based on the animal nesting behavior is used to nest the convolutional neural network.The positions of the typical wind turbines optimized,the spatial wind speed matrix is reconstructed firstly in combination with the typical wind turbine data so that the extraction efficiency is improved through reducing the complexity of the input data.Then the convolutional neural network is used to extract the feature information of the reconstruction matrix in order that the wind speed feature dimension is reduced.Aiming at the parameter setting of the multi-step prediction module,an the parameter optimization of the support vector regression model is achieved initially by using the animal reproduction behavior.Secondly the multi-step prediction of the wind speed is conducted with the spatial characteristic information.The wind power is predicted through the equivalent wind power utilization coefficient method with the obtained wind speed prediction value.Finally,the method in this paper is applied to the prediction of wind speed and wind power in the actual wind field,and it is verified through comparative analysis.

关 键 词：风速预测 时空相关性 卷积神经网络 支持向量回归 

分 类 号：TM721[电气工程—电力系统及自动化]