基于BP神经网络与马尔可夫链的短期电价预测  

Short-term Electricity Price Forcasting Based on BP Neural Network and Markov Chain

作　　者：黄羹墙 杨俊杰[1] 

机构地区：[1]上海电力学院电子与信息工程学院,上海200090

出　　处：《上海电力学院学报》2017年第1期1-3,10,共4页Journal of Shanghai University of Electric Power

摘　　要：提出了一种基于BP神经网络与马尔可夫链的短期电价预测方法.在采用BP神经网络模型进行短期电价初步预测的基础上,按照模糊C-均值聚类法划分预测误差的马尔可夫链状态区域,再根据状态转移概率矩阵对预测误差进行修正,得到最终预测结果.算例仿真结果表明,所提出的方法比单纯采用神经网络的预测精确度更高.A new method based on the BP neural network and Markov chain is established to predict short-term electricity price. BP network is firstly used to make a preliminary forecast, then the Markov chain state region is divided based on the fuzzy c-means algorithm. Finally the predic- tion error is corrected according to the state transition probability matrix and the final forecasting result is obtained. The results show that the proposed method is more accurate than those by tradi- tional BP network.

关 键 词：电价预测 BP神经网络 马尔可夫 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] F426.61[自动化与计算机技术—控制科学与工程]