天气数据模糊化处理对神经网络短期负荷预测模型的影响研究  

Research on Impact of Fuzzy Processing of Weather Data on Neural Network Model of Short-term Load Forecasting

作　　者：夏昌浩[1] 向学军[1] 段莉梅[1] 

机构地区：[1]三峡大学电气与新能源学院,湖北宜昌443002

出　　处：《三峡大学学报（自然科学版）》2010年第4期52-55,共4页Journal of China Three Gorges University：Natural Sciences

摘　　要：介绍了一种简洁实用的基于模糊集的神经网络电力系统短期负荷预报方法,计及了天气和日期特征量.应用模糊集理论将天气和日期特征量模糊化后作为BP神经网络的一部分输入进行训练,同时考虑实际历史负荷数据构造了短期负荷预测模型,预测未来24 h负荷.通过典型算例与普通BP方法预测结果相比,表明该方法是有效的并具有较高的预测精度.A simple and practical short-term load forecasting approach is presented by using artificial neural network based on fuzzy set and considering weather variables and date variables. Fuzzy processing to some sensitive factors such as weather variables and date variables was finished using fuzzy set technology. Processed data were used to train BP neural network as a part of network inputs. Taking into account the historical load data, the forecasting model is constructed to forecast hourly loads for the next 24 hours. The comparison between typical example and usual forecasting result by BP model indicates that this method is effective and of high accuracy.

关 键 词：短期负荷预测 人工神经网络 模糊集 BP算法 MATLAB 

分 类 号：TM715[电气工程—电力系统及自动化]