基于最优动态功率补偿的电力负荷预测方法

Power Load Forecasting Method Based on Optimal Dynamic Power Compensation

电力系统负荷主要由全社会综合经济状况和气象条件这2个关键影响因素决定,为了提高短期和中远期负荷预测的精度,提出一种基于最优动态功率补偿的负荷预测方法。首先,按照时间顺序将负荷数据划分为分段1和分段2,通过相似性矩阵计算,分别提取分段1和分段2最优的训练体感温差曲线和目标体感温差曲线;其次,利用构建的径向基神经网络分别实现分段1和分段2最优动态补偿功率的预测;在此基础上,利用推导的基于最优动态功率补偿的负荷预测通用公式进行修正,实现短期和中远期负荷曲线的有效预测。实例分析结果表明,所提方法具有可行性、有效性和通用性,预测精度较高。

The load of power system is mainly determined by two key influence factors of comprehensive economic index and meteorology.To improve the precision of short-term and mid-long-term load forecasting,a load forecasting method based on optimal dynamic power compensation is proposed in this paper.Firstly,the load data is divided into segment 1 and segment 2 according to time sequence,and through calculation of the similarity matrix,the optimal training and target curves of somatosensory temperature differences of these two segments are extracted respectively.Secondly,the radial basis function(RBF)neural network is used to predict the optimal dynamic compensation power of segment 1 and segment 2 respectively.Finally,the general formula of load forecasting based on optimal dynamic power compensation derived is used for modification,which can realize the effective forecast of short-term and med-long-term load curves.The results of the case analysis indicate that the proposed method is feasible and effective with high prediction accuracy and good versatility.