基于改进型遗传算法的混沌神经网络在电力负荷预测的应用

通过对陕西省省网历史负荷数据进行混沌特性分析,重构系统相空间,并计算最大Lyapunov指数,指出该时间序列具有混沌特性,从而采用混沌神经网络对该时间序列进行短期负荷预测。神经网络模型采用改进型遗传算法对权值和阈值进行学习和训练,优化神经网络权重,充分发挥遗传算法的全局寻优能力和神经网络的局部搜索性能。然后采用该网络进行预测,预测结果表明:该模型预测算法优于纯BP网络方法的预测结果,较大地提高了预测精度。

Application of Chaotic Neural Network Based on Improved Genetic Algorithm in Load Prediction

The analysis is carried out on the chaotic characteristics of the historic load data of power grid of shaanxi province to reconstruct the phase-space of the system and to calculate the largest Lyapunov index. It points out that the time series have chaotic characteristics. Thus the time series can be used on the chaotic neural network in short-term load forecasting. Improved genetic algorithm is used in neural network model to train and study the weights and thresholds,to optimize the weights of neural network,and to maximize the capacity of global optimization of genetic algorithms and the local search performance of the neural network. Then this neural network model is used on load prediction. The prediction results show that the method by predictive algorithm is superior than by pure BP network method. It greatly improves the prediction accuracy.