一种短期电力负荷预测新方法的研究与应用

通过对电力负荷变化规律和影响因素的分析,提出了一种新的短期电力负荷预测模型。首先,鉴于模糊聚类方法易陷入局部最优解及运算速度慢的缺点,采用蚁群算法中pij(t)改进模糊聚类分析;然后以每天的24点负荷数据、天气数据以及天类别数据为指标,将历史数据聚分成若干簇团,并采用动量BP神经网络针对每一簇团建立相应的预测模型。对山东地区1年的实际数据进行预测分析的结果表明,该模型不仅对普通工作日有较高的预测精度,对双休日、节假日和一些特殊情况(夏季典型日负荷)也有较好的预测精度。

Study and Application of Novel Short Term Load Forecasting Method

A novel short-term load forecasting model was proposed. Because fuzzy C-means algorithm (FCM) was unsatisfied in circulating precision and speed, to improve the performances of FCM algorithm, the fuzzy membership function and clustering centers were initialized by ant colony algorithm. The practical historical data within one year was divided into several groups by ant colony -fuzzy clustering algorithm. A separate module based on momentum BP neural networks models was used for each group. Using data from the Shandong Province, the satisfactory accurate results were obtained on the weekday, weekend and holidays. Moreover, the model is robust, and produces accurate results in some special cases (the type summer hourly loads).