基于负荷内部特性和外部随机因素的短期负荷预测模型

根据电力系统负荷序列的混沌特性,提出将其划分为基本混沌负荷分量和外部随机因素负荷分量,依据不同的理论分别构造预测模型。前者通过混沌动力学机理和动态递归时延神经网络融合来构造模型;后者在依据日类型和气象特征进行数据挖掘聚类的基础上利用统计分析与智能识别融合来构造模型。大量的仿真计算证明了所提出的短期负荷预测模型能有效保证全年的预测精度及其稳定性,对夏季高温区和特殊类型日的预测精度有明显提高。

A New Short-Term Load Forecasting Model Based on Internal Characteristic of Power Load and External Stochastic Factor

According to chaotic features of power load time series, it is proposed that at first the load time series are divided into basic chaotic load component (BCLC) and external stochastic factor load component (ESLC), and the forecasting models for the two components are built separately by different theories. The forecasting model of BCLC is constructed based on the fusion of chaotic dynamical mechanism and recursive time-delay neural network; the forecasting model of ESLC is constructed by the fusion of statistical analysis and intelligent recognition on the basis of the clustering of data mining considering the date type and the external meteorological factors. A lot of simulation proves that the proposed short-time load forecasting model can effectively ensure year-round forecasting precision and stability, especially, the forecasting precisions for high temperature periods in summer and special holiday periods are evidently improved.