基于元胞负荷特性分析的RBF神经网络空间负荷预测方法  

A Spatial Load Forecasting Method Based on RBF Neural Network and Cellular Load Characteristics Analysis

作　　者：肖白[1] 刘庆永[1] 牛强 綦雪松 王皓[1] 

机构地区：[1]东北电力大学电气工程学院,吉林省吉林市132012 [2]国网吉林省电力有限公司吉林市供电公司,吉林省吉林市132001

出　　处：《电网技术》2018年第1期301-307,共7页Power System Technology

基　　金：国家自然科学基金项目(51177009)~~

摘　　要：针对元胞历史负荷数据的特点,结合RBF神经网络在非线性拟合方面的优势,提出了一种基于元胞负荷特性分析的RBF神经网络空间负荷预测方法。该方法以元胞的历史负荷数据为基础,横向从元胞的年历史负荷特性入手,描述各元胞年负荷峰值段的发展趋势;纵向从元胞日最大负荷出现的时刻入手,刻画峰值时段的最大负荷数据群。在充分地考虑元胞负荷的季节特性、双峰特性和转移特性对空间负荷预测精度影响的基础上,构建出一种合理的峰值时段最大负荷数据群的双向选取模型。然后采用RBF神经网络对所选取的最大负荷数据群进行训练和预测,进而得到目标年各元胞的年最大负荷预测值。工程实例验证了该方法的实用性和有效性。Aiming at characteristics of historical load data about lattice cell and combining advantages of RBF neural networks in terms of nonlinear fitting, a spatial load forecasting method based on RBF neural networks and cellular load characteristic analysis is proposed. Based on historical load data of the cell, development trend of peak load of each cell is described with annual load characteristics of the cell, and maximum load data group of peak period is described with moment of daily maximum load of the cell. Considering the influence of seasonal characteristics, bimodal characteristics and transfer characteristics of the cell load on accuracy of spatial load forecasting, a reasonable bidirectional selection model of maximum load data group is established. Then RBF neural network is used to train and forecast the selected maximum load data group, and annual maximum load forecast value of each cell in target year is obtained. Practicality and validity of the proposed method is verified with engineering example.

关 键 词：电力负荷特性 RBF神经网络 空间负荷预测 负荷最大值 元胞 

分 类 号：TM715[电气工程—电力系统及自动化]