基于主成份分析法的神经网络电力负荷预测
|
摘要
影响电力负荷的变量具多,其具有非线性程度高、冗余程度高等特点,传统方法预测结果精度不高。为了提高结果精度,利用主成份分析方法对人工神经网络进行优化,达到提高预测精度的目的。首先,利用粒子群算法优化、改进径向基函数神经网络。然后,对输入量进行主成份分析、筛选,把经分析、筛选的输入量重新输入神经网络。最后,进行训练、预测,得出结果。利用经过优化、改进的模型对某地级市2016年的电力负荷进行验证。结果表明,径向基函数神经网络经过粒子群算法的优化以及主成份的分析,负荷预测精度得到了提高。
There are many variables influencing the power load, which have the characteristics of high degree of nonlinearity, high degree of redundancy, etc. The accuracy of traditional methods is not high. In order to improve the accuracy of results,principal component analysis methods are used to optimize the artificial neural network, to improve the prediction accuracy. First, the particle swarm optimization algorithm is used to optimize and improve the radial basis function neural network. Then, the principal component analysis and screening are performed on the input quantity, and the analyzed and filtered input quantity is re-inputted into the neural network. Finally, conduct training, forecasting, and get results. Use an optimized and improved model to verify the electricity load of a prefecture-level city in 2016. The results show that the radial basis function neural network is optimized by the particle swarm optimization and the analysis of the principal components, and the load forecasting accuracy is improved.
There are many variables influencing the power load, which have the characteristics of high degree of nonlinearity, high degree of redundancy, etc. The accuracy of traditional methods is not high. In order to improve the accuracy of results,principal component analysis methods are used to optimize the artificial neural network, to improve the prediction accuracy. First, the particle swarm optimization algorithm is used to optimize and improve the radial basis function neural network. Then, the principal component analysis and screening are performed on the input quantity, and the analyzed and filtered input quantity is re-inputted into the neural network. Finally, conduct training, forecasting, and get results. Use an optimized and improved model to verify the electricity load of a prefecture-level city in 2016. The results show that the radial basis function neural network is optimized by the particle swarm optimization and the analysis of the principal components, and the load forecasting accuracy is improved.
引文
[1]晨辉.电力系统负荷预报理论与方法[M].哈尔滨:哈尔滨工业大学出版社,1987.
[2]康重庆,夏清,刘梅.电力系统负荷预测[M].北京:中国电力出版社,2007.
[3]赵登福,张涛,杨增辉,谷庆利,夏道止.基于GN.BFGS算法的RBF神经网络短期负荷预测[J].电力系统自动化,2003,27(4):33-36.
[4]俞俊霞,赵波.基于改进粒子群优化算法的最优潮流计算[J].电力系统及其自动化学报,2005,17(4):83-88.
[5]邓聚龙.灰理论基础[M].武汉:华中科技大学出版社,2002.
[6] CONEJO A J,PLAZAS M A,ESPINOLA R,et al.Day-ahead electricity price forecasting using the wavelet transform and A RIM A models[J].IEEE Trans on Power Systems,2005,20(2):1035-1042.
[7] DE SOUZA A C Z,DE SOUZA J C S,DA SILVA A M L.On-line voltage stability monitoring[J].IEEE T rans on Power Systems,2000,15(4):1300-1305.
[8]王志征,余岳峰.姚国平.主成分分析法在电力负荷预测中的应用[J].电力需求侧管理,2003,5(3):21-23.
[9]姚祥光,周永权,李咏梅.人工鱼群与微粒群混合优化算法[J].计算机应用研究,2010,27(6):2084-2087.
[2]康重庆,夏清,刘梅.电力系统负荷预测[M].北京:中国电力出版社,2007.
[3]赵登福,张涛,杨增辉,谷庆利,夏道止.基于GN.BFGS算法的RBF神经网络短期负荷预测[J].电力系统自动化,2003,27(4):33-36.
[4]俞俊霞,赵波.基于改进粒子群优化算法的最优潮流计算[J].电力系统及其自动化学报,2005,17(4):83-88.
[5]邓聚龙.灰理论基础[M].武汉:华中科技大学出版社,2002.
[6] CONEJO A J,PLAZAS M A,ESPINOLA R,et al.Day-ahead electricity price forecasting using the wavelet transform and A RIM A models[J].IEEE Trans on Power Systems,2005,20(2):1035-1042.
[7] DE SOUZA A C Z,DE SOUZA J C S,DA SILVA A M L.On-line voltage stability monitoring[J].IEEE T rans on Power Systems,2000,15(4):1300-1305.
[8]王志征,余岳峰.姚国平.主成分分析法在电力负荷预测中的应用[J].电力需求侧管理,2003,5(3):21-23.
[9]姚祥光,周永权,李咏梅.人工鱼群与微粒群混合优化算法[J].计算机应用研究,2010,27(6):2084-2087.