基于卡尔曼滤波的日前电价混合预测方法  

Hybrid Forecasting Method of Day-ahead Electricity Price Based on Kalman Filter

作　　者：伏玥橦 张政[1,2] 袁凯 李天然[1,2] Fu Yuetong;Zhang Zheng;Yuan Kai;Li Tianran(School of NARI Electric and Automation,Nanjing Normal University,Nanjing Jiangsu 210023,China;NARI Group Corporation,Nanjing Jiangsu 210006,China)

机构地区：[1]南京师范大学南瑞电气与自动化学院,江苏南京210023 [2]南瑞集团有限公司,江苏南京211106

出　　处：《电气自动化》2022年第4期57-60,共4页Electrical Automation

基　　金：智能电网保护和运行控制国家重点实验室开放课题“基于混合仿真的电力市场与碳市场互动仿真分析与调控机制研究”(SGNR0000KJJS1907532)。

摘　　要：为提高电价预测准确性,提出利用卡尔曼滤波、遗传算法和BP神经网络模型的电力市场日前电价预测方法。利用卡尔曼滤波的最优估计能力对历史数据进行预处理,清除噪声等干扰,修正异常点。将预处理后的历史数据作为模型输入,建立基于遗传算法的BP神经网络日前电价预测模型,有效地提高了预测准确性。最后利用基于美国PJM现货市场公开数据,验证了所提方法的可行性和有效性。In order to improve the accuracy of electricity price prediction,a day-ahead electricity price prediction method in the electricity market using Kalman filtering,genetic algorithm and BP neural network model was proposed.The optimal estimation ability of Kalman filter was used to preprocess the historical data,remove the noise and other interference,and correct the abnormal points.Taking the preprocessed historical data as the model input,the BP neural network day-ahead electricity price forecasting model based on genetic algorithm was established,which effectively improved the accuracy of forecasting.Finally,the feasibility and effectiveness of the proposed method are verified by using the public data of PJM spot market in the United States.

关 键 词：电力市场 预测 卡尔曼滤波 神经网络 机器学习 

分 类 号：TM715[电气工程—电力系统及自动化]