基于NGO-VMD-SSA-ESN的短期电价预测  

NGO-VMD-SSA-ESN-based Short-term Electricity Price Prediction

作　　者：郭庆辉 林浩哲 李媛[1] 谢露露 刘桁宇 GUO Qinghui;LIN Haozhe;LI Yuan;XIE Lulu;LIU Hengyu(School of Science,Shenyang University of Technology,Shenyang 110870,China;Electric Power Research Institute of State Grid Liaoning Electric Power Co.,Ltd.,Shenyang 110055,China)

机构地区：[1]沈阳工业大学理学院,辽宁沈阳110870 [2]国网辽宁省电力有限公司电力科学研究院,辽宁沈阳110055

出　　处：《电工技术》2024年第2期130-136,共7页Electric Engineering

基　　金：辽宁省兴辽英才计划项目(编号XLYC2008005)。

摘　　要：针对电价波动性和非线性的特点,为提高电价预测的精度,提出了一种基于回声状态网络的短期电价混合预测模型。首先,基于北方苍鹰优化算法(NGO)优化后的变分模态分解(VMD)对原始电价进行分解,降低电价的波动性;然后,利用麻雀搜索算法(SSA)对回声状态网络(ESN)的参数进行优化,使其能针对NGO-VMD分解后的不同子序列自适应地调整参数进行预测,降低参数经验设置的随机性;最后,根据分解子序列与原始数据的皮尔逊相关系数,选择合适子序列的预测结果重构合成最终预测结果,消除了噪声的影响。以美国PJM电力市场为例,与其他电价预测模型对比验证所提出的混合模型具有更好的预测精度。In view of volatility and nonlinearity characteristics of electricity price,aiming at improving prediction accuracy of electricity price,a short-term hybrid prediction model based on echo state network is proposed.First the variational modal decomposition(VMD)optimized by northern goshawk optimization algorithm(NGO)is used to decompose original electricity price and reduce electricity price volatility.Then sparrow search algorithm is used to optimize the parameters of echo state network so that it can adaptively adjust the parameters for different sub-sequences after NGO-VMD decomposition,and to reduce the randomness of parameter experience settings.Finally according to Pearson correlation coefficient between the decomposition sub-sequence and the original data,the prediction result suitable with sub-sequence is selected and reconstructed to form final prediction result,eliminating the influence of noise.Taking the PJM electricity market in the United States as an example,compared with other electricity price prediction models,the proposed hybrid model has better prediction accuracy.

关 键 词：电价预测 回声状态网络 变分模态分解 

分 类 号：TM715[电气工程—电力系统及自动化]