NRS和PSO算法优化最小二乘支持向量机的短期电力负荷预测
|
摘要
针对依赖经验选取影响短期电力负荷相关因素的不准确性以及最小二乘支持向量机(LS-SVM)模型中参数难以确定的问题,建立结合邻域粗糙集(NRS)理论和粒子群优化(PSO)算法的最小二乘支持向量机短期电力负荷预测模型。为了从经验选择的属性中挖掘出与负荷密切相关的因素,避免选取过多属性而加长训练时间以及冗余属性对预测精度的影响,采用邻域粗糙集理论对属性进行约简,使其结果作为LS-SVM法对模型参数进行寻优,避免依赖经验选择的参数对模型的影响。最后用上述方法对某地区负荷进行预测分析,仿真结果表明上述方法能有效提高负荷预测精度。
A short-term power load forecasting model based on least square support vector machine (LS-SVM) optimized by neighborhood rough set (NRS) and particle swarm optimization (PSO) algorithms is established to solve the uncertain selection of relation factor influencing short-term power load and difficult determination of parameters in LS-SVM. In order to mine the factors closely relating to the load in experience selection attributes,and avoid prolonging the training time due to excessive attributes selection and influence of redundant attribute on prediction accuracy,the theory of NRS is used to reduce the attributes,and its reduction result is taken as the input variables of the LS-SVM model. The PSO algorithm is used to optimize the model parameters while establishing the LS-SVM model,which can avoid the influence of parameters selected by experience on the model. The method is applied to the prediction analysis of the load of a certain area. The simulation results show that the method can greatly improve the forecasting accuracy of load.
A short-term power load forecasting model based on least square support vector machine (LS-SVM) optimized by neighborhood rough set (NRS) and particle swarm optimization (PSO) algorithms is established to solve the uncertain selection of relation factor influencing short-term power load and difficult determination of parameters in LS-SVM. In order to mine the factors closely relating to the load in experience selection attributes,and avoid prolonging the training time due to excessive attributes selection and influence of redundant attribute on prediction accuracy,the theory of NRS is used to reduce the attributes,and its reduction result is taken as the input variables of the LS-SVM model. The PSO algorithm is used to optimize the model parameters while establishing the LS-SVM model,which can avoid the influence of parameters selected by experience on the model. The method is applied to the prediction analysis of the load of a certain area. The simulation results show that the method can greatly improve the forecasting accuracy of load.
引文
[1]陈冬沣,郑舟,吴永峰.基于时间序列的电力系统短期负荷预测问题分析[J].自动化应用,2017(11):99-101.CHEN Dongfeng,ZHENG Zhou,WU Yongfeng. Analysis of short-term load forecasting in power systems based on time se-ries[J]. Automation application,2017(11):99-101.
[2]张健美,周步祥,林楠,等.灰色Elman神经网络的电网中长期负荷预测[J].电力系统及其自动化学报,2013,25(4):145-149.ZHANG Jianmei,ZHOU Buxiang,LIN Nan,et al. Prediction of mid-long term load based on gray Elman neural networks[J].Proceedings of the CSU-EPSA,2013,25(4):145-149.
[3]王玉龙,崔玉,李鹏,等.基于小波分析改进的神经网络模型电力系统负荷预测[J].电网与清洁能源,2015,31(2):16-20.WANG Yulong,CUI Yu,LI Peng,et al. Load forecasting of power system based on wavelet analysis improved neural net-work model[J]. Advances of power system & hydroelectric engi-neering,2015,31(2):16-20.
[4]汤义强,毛军军,李侠,等.基于粗糙集属性简约的电力供应量SVM回归预测[J].计算机技术与发展,2012,20(9):48-52.TANG Yiqiang,MAO Junjun,LI Xia,et al. China′s power supply SVM regression forecast based on rough set attribute re-duction[J]. Computer technology and development,2012,20(9):48-52.
[5]拜润卿,何欣,赵红,等.基于粒子群优化方法的短期电力负荷组合预测[J].电子测试,2015(4):46-48.BAI Runqing,HE Xin,ZHAO Hong,et al. Short-term power load combination forecasting based on PSO algorithm[J]. Elec-tronic test,2015(4):46-48.
[6]石晓艳,刘淮霞,于水娟,等.鲶鱼粒子群算法优化支持向量机的负荷预测[J].计算机工程与应用,2013,49(11):220-223.SHI Xiaoyan,LIU Huaixia,YU Shuijuan,et al. Short-time load prediction based on support vector machine optimized by catfish particle swarm optimization algorithm[J]. Computer en-gineering and applications,2013,49(11):220-223.
[7]马晓博.基于小波变换和BP神经网络的短期风电功率预测[J].电力科学与技术学报,2015,30(2):92-97.MA Xiaobo. Short-term wind power prediction based on wavelet transform and BP neural network[J]. Journal of electric power science and technology,2015,30(2):92-97.
[8]刘晓菲,商立群.非线性主成分分析和RBF神经网络的电力系统负荷预测[J].电网与清洁能源,2016,32(1):47-52.LIU Xiaofei, SHANG Liqun. Nonlinear principal component analysis and RBF neural network for power system load fore-casting[J]. Advances of power system & hydroelectric enginee-ring,2016,32(1):47-52.
[9]贾逸伦,龚庆武,占劲松,等.基于有效度和相对熵的短期电力负荷组合预测方法[J].电力科学与技术学报,2016,31(2):3-9.JIA Yilun,GONG Qingwu,ZHAN Jinsong,et al. Combination forecast method of short-term power load based on forecasting availability and relative entropy[J]. Journal of electric power science and technology,2016,31(2):3-9.
[10]何传,舒勤,贺含峰.ICA特征提取与BP神经网络在负荷预测中的应用[J].电力系统及其自动化学报,2014,26(8):40-46.HE Chuan,SHU Qin,HE Hanfeng. Application of BP neural network and ICA feature extraction in power load forecasting[J]. Proceedings of the CSU-EPSA,2014,26(8):40-46.
[2]张健美,周步祥,林楠,等.灰色Elman神经网络的电网中长期负荷预测[J].电力系统及其自动化学报,2013,25(4):145-149.ZHANG Jianmei,ZHOU Buxiang,LIN Nan,et al. Prediction of mid-long term load based on gray Elman neural networks[J].Proceedings of the CSU-EPSA,2013,25(4):145-149.
[3]王玉龙,崔玉,李鹏,等.基于小波分析改进的神经网络模型电力系统负荷预测[J].电网与清洁能源,2015,31(2):16-20.WANG Yulong,CUI Yu,LI Peng,et al. Load forecasting of power system based on wavelet analysis improved neural net-work model[J]. Advances of power system & hydroelectric engi-neering,2015,31(2):16-20.
[4]汤义强,毛军军,李侠,等.基于粗糙集属性简约的电力供应量SVM回归预测[J].计算机技术与发展,2012,20(9):48-52.TANG Yiqiang,MAO Junjun,LI Xia,et al. China′s power supply SVM regression forecast based on rough set attribute re-duction[J]. Computer technology and development,2012,20(9):48-52.
[5]拜润卿,何欣,赵红,等.基于粒子群优化方法的短期电力负荷组合预测[J].电子测试,2015(4):46-48.BAI Runqing,HE Xin,ZHAO Hong,et al. Short-term power load combination forecasting based on PSO algorithm[J]. Elec-tronic test,2015(4):46-48.
[6]石晓艳,刘淮霞,于水娟,等.鲶鱼粒子群算法优化支持向量机的负荷预测[J].计算机工程与应用,2013,49(11):220-223.SHI Xiaoyan,LIU Huaixia,YU Shuijuan,et al. Short-time load prediction based on support vector machine optimized by catfish particle swarm optimization algorithm[J]. Computer en-gineering and applications,2013,49(11):220-223.
[7]马晓博.基于小波变换和BP神经网络的短期风电功率预测[J].电力科学与技术学报,2015,30(2):92-97.MA Xiaobo. Short-term wind power prediction based on wavelet transform and BP neural network[J]. Journal of electric power science and technology,2015,30(2):92-97.
[8]刘晓菲,商立群.非线性主成分分析和RBF神经网络的电力系统负荷预测[J].电网与清洁能源,2016,32(1):47-52.LIU Xiaofei, SHANG Liqun. Nonlinear principal component analysis and RBF neural network for power system load fore-casting[J]. Advances of power system & hydroelectric enginee-ring,2016,32(1):47-52.
[9]贾逸伦,龚庆武,占劲松,等.基于有效度和相对熵的短期电力负荷组合预测方法[J].电力科学与技术学报,2016,31(2):3-9.JIA Yilun,GONG Qingwu,ZHAN Jinsong,et al. Combination forecast method of short-term power load based on forecasting availability and relative entropy[J]. Journal of electric power science and technology,2016,31(2):3-9.
[10]何传,舒勤,贺含峰.ICA特征提取与BP神经网络在负荷预测中的应用[J].电力系统及其自动化学报,2014,26(8):40-46.HE Chuan,SHU Qin,HE Hanfeng. Application of BP neural network and ICA feature extraction in power load forecasting[J]. Proceedings of the CSU-EPSA,2014,26(8):40-46.