基于SSA-VMD-INGO-RF的短期风电功率预测  

Short⁃term wind power prediction based on SSA⁃VMD⁃INGO⁃RF

作　　者：汪繁荣[1,2] 梅涛 张旭东 汪筠涵 肖悦 WANG Fanrong;MEI Tao;ZHANG Xudong;WANG Yunhan;XIAO Yue(College of Electrical and Electronic Engineering,Hubei University of Technology,Wuhan 430068,China;Xiangyang Industrial Institute of Hubei University of Technology,Xiangyang 441100,China)

机构地区：[1]湖北工业大学电气与电子工程学院,湖北武汉430068 [2]襄阳湖北工业大学产业研究院,湖北襄阳441100

出　　处：《现代电子技术》2024年第24期88-96,共9页Modern Electronics Technique

基　　金：国家自然科学基金项目(52307239)。

摘　　要：为解决风电功率输出的不确定性、弱化电网波动以及电网的提质增效等问题,提出一种基于变分模态分解(VMD)、Piecewise混沌映射、北方苍鹰优化(NGO)算法和随机森林(RF)的组合模型。该模型采用麻雀搜索算法(SSA)对VMD核心参数(K值和惩罚系数α)进行寻优,通过SSA-VMD将原始功率序列分解为多个有限带宽的特征模态分量,以降低原始数据的复杂度和非平稳性对预测精度的影响;然后,构建模态分量并在改进的北方苍鹰算法优化随机森林中进行预测;最后,将各分量预测结果叠加,得到最终预测值。以内蒙古某风电场的实测数据为研究对象,将所提组合模型与另外6种模型进行比较。结果表明,所设计模型预测结果平均绝对百分比误差(MAPE)为1.734%,均方根误差为0.068 MW,R^(2)为0.992,证明了该模型的有效性。In order to solve the uncertainty of wind power output,weaken the grid fluctuation as well as improve the quality and efficiency of the grid,a combined model based on variational mode decomposition(VMD),Piecewise chaotic mapping,northern goshawk optimization(NGO)algorithm and random forest(RF)is proposed.In this model,the sparrow search algorithm(SSA)is used to optimize the core parameters of the VMD(K values and penalty coefficientsα),and the original power sequence is composed into multiple finite bandwidth characteristic modal components by means of SSA-VMD to reduce the complexity of the original data and the impact of non-stationarity on prediction accuracy.The modal components are constructed and predicted in an optimized RF using an improved NGO.The predicted results of each component are overlayed to obtain the final predicted value.With the measured data of a wind farm in Inner Mongolia,the proposed combined model is compared with the other 6 models.The results show that the designed mean absolute percentage error(MAPE)of the designed model prediction results is 1.734%,the root-mean-square error is 0.068 MW,and R^(2) is 0.992,which can prove the validity of this model.

关 键 词：短期风电功率预测 北方苍鹰算法 Piecewise混沌映射 随机森林 变分模态分解 麻雀搜索算法 

分 类 号：TN919-34[电子电信—通信与信息系统] TM614[电子电信—信息与通信工程]