The electric power load is the basis of development of electric power system,so the improvement of predicting accuracy of the former is beneficial to the latter. Elman neural network forecasting method is easy to fall into local solution and its convergence rate is slow. The artificial fish swarm algorithm has better global convergence ability and higher optimization speed. To improve the accuracy of short-term power load forecasting,the artificial fish swarm algorithm is adopted to optimize the initial weights and thresholds of Elman neural network. Then a new forecast model of short-term power load is built and applied to forecast the short-term power load. With the historical load data of a city as training samples,the artificial fish swarm neural network prediction model and the traditional Elman neural network prediction model are employed in contrast experiments. The results show that it has smaller computational error,higher accuracy and faster convergence rate compared with the traditional Elman neural network prediction model and has better application prospects in short term power load forecasting.%由于电力负荷是电力系统发展的基础,提高电力负荷预测的准确性有利于电力系统的快速发展.Elman神经网络预测方法容易陷入局部解,且收敛速度慢,而人工鱼群算法具有较优的全局收敛能力及较快的寻优速度.为了提高短期电力负荷预测的精度,利用人工鱼群算法对Elman神经网络的初始权值和阈值进行了优化,提出并建立了一种新的人工鱼群神经网络短期负荷预测模型.以某市的历史负荷数据作为训练样本,将人工鱼群神经网络预测模型与传统Elman神经网络预测模型进行对比实验.实验结果表明,相对于传统Elman神经网络预测模型,人工鱼群神经网络模型的计算误差更小,预测精度更高,收敛速度更快,具有较好的短期电力负荷预测应用前景.
展开▼
