改进神经网络的短期负荷预测模型及仿真  

Short-Term Power Load Forecasting Model and Simulation Based on Neural Network

作　　者：杨廷志[1] 文小飞[1] 万俊[1] 李书[1] 

机构地区：[1]国网重庆市电力公司綦南供电分公司,重庆401420

出　　处：《计算机仿真》2014年第10期145-150,176,共7页Computer Simulation

摘　　要：在电力负荷准确预测问题的研究中,电力负荷具有周期性、随机性和非线性的特点,而传统方法存在负荷预测精度低的难题,为了提高负荷的预测精度,提出一种改进神经网络的短期负荷预测模型(CPSO-BPNN)。首先利用非线性预测能力强的BP神经网络对短期负荷建模;然后结合混沌粒子群优化算法挖掘短期负荷的变化规律以提高短期负荷预测精度;最后对模型性能进行仿真。仿真结果表明,CPSO-BPNN解决了传统方法存在的难题,提高了短期负荷的预测精度,为非线性负荷预测提供了依据。Power load is cyclical and random, and traditional method has low prediction accuracy. In order to ob- tain better forecasting results of short - term load, a novel short - term load forecasting model was proposed based on chaotic particle swarm optimization algorithm and neural network （ CPSO - BPNN）. Firstly, the chaos theory was in- troduced into PSO algorithm to improve the global search ability, and then the BP neural network parameters were op- timized by chaos particle swarm algorithm to establish the optimal short - term power load forecasting model. Finally, the model was simulated and tested. The simulation results show that, the CPSO - BPNN can improve the forecasting accuracy of the short - term load, which provides a new research idea for the nonlinear load forecasting.

关 键 词：短期负荷 预测精度 混沌粒子群算法 神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]