| 1,074 | 17 | 263 |
| 下载次数 | 被引频次 | 阅读次数 |
针对光伏功率预测中特征因素太多、关键特征与功率间映射关系难以有效挖掘和预测精度不高的问题，提出一种基于随机森林RF(random forest)算法特征选择和灰狼优化算法GWO(grey wolf optimizer)优化高斯过程回归GPR(Gaussian process regression)模型相结合的组合预测模型。首先，采用皮尔逊和斯皮尔曼相关系数对特征进行相关性分析，并进行初步筛选；接着，基于随机森林算法对特征进行重要性评价，并选取最优特征子集；然后，采用灰狼优化算法对高斯过程回归模型进行优化；最后，将最优特征子集输入到组合预测模型RFGWO-GPR中进行短期光伏功率预测。应用某光伏电站实测数据的仿真实验结果表明，提出的模型在不同天气条件下可以对特征进行有效选择，与未进行特征选择的单一模型相比，预测精度显著提高，并且明显优于其他优化算法与GPR模型组成的组合预测模型。
Abstract：Aimed at the problems in photovoltaic power prediction such as too many characteristic factors,difficulty in effectively mining the mapping relationship between key features and power,and low prediction accuracy,a combined prediction model is proposed,which integrates the feature selection using the random forest(RF)algorithm and the Gaussian process regression(GPR)model optimized by the grey wolf optimizer(GWO)algorithm. First,Pearson and Spearman correlation coefficients are used to conduct a correlation analysis of features and further perform preliminary screening. Second,based on the RF algorithm,feature importance evaluation is conducted to select the optimal subset of features. Third,the GWO algorithm is employed to optimize the GPR model. Finally,the optimal feature subset is inputted into the combined prediction model of RF-GWO-GPR for short-term photovoltaic power prediction. The results of simulation experiments based on the measurement data from one photovoltaic power station show that the proposed model can effectively select features under different weather conditions. Compared with that of the single model without feature selection,the prediction accuracy of the new model is significantly improved,and it is significantly better than the combined prediction model composed of other optimization algorithms and GPR model.
[1]王蓓蓓，亢丽君，苗曦云，等（Wang Beibei,Kang Lijun,Miao Xiyun,et al）.考虑可信度的新能源及需求响应参与英美容量市场分析及思考（Analysis and enlightenment of renewable energy and demand response participating in UK and US capacity markets considering capacity credibility）[J].电网技术（Power System Technology）,2022,46(4):1233-1247.
[2]刘洁，林舜江，梁炜焜，等（Liu Jie,Lin Shunjiang,Liang Weikun,et al）.基于高阶马尔可夫链和高斯混合模型的光伏出力短期概率预测（Short-term probabilistic forecast for power output of photovoltaic station based on high order Markovchain and Gaussian mixture model）[J].电网技术（Power System Technology）,2023,47(1):266-274.
[3] Wang Kejun,Qi Xiaoxia,Liu Hongda. A comparison of day-ahead photovoltaic power forecasting models based on deep learning neural network[J]. Applied Energy,2019,251:13315.
[4] Zhang Chu,Tian Peng,Nazir M S. A novel integrated photovoltaic power forecasting model based on variational mode decomposition and CNN-BiGRU considering meteorological variables[J]. Electric Power Systems Research,2022,213:1-14.
[5] Scott C,Ahsan M,Albarbar A. Machine learning for forecasting a photovoltaic(PV)generation system[J]. Energy,2023,278:127807.
[6]朱旭坤，姚李孝，杨国清（Zhu Xukun,Yao Lixiao,Yang Guoqing）.基于PSOEM和神经网络的光伏电站功率预测（Power prediction of photovoltaic power plants based on PSOEM algorithm and neural network）[J].电网与清洁能源（Power System and Clean Energy）,2021,37(7):115-120,135.
[7]成燕，庄飞鸯，徐万万，等（Cheng Yan,Zhuang Feiyang,Xu Wanwan,et al）.基于改进的极限学习机光伏出力短期预测（Short-term prediction of photovoltaic output based on improved extreme learning machines）[J].现代电力（Modern Electric Power）,2023,40(5):679-686.
[8]宋绍剑，李博涵（Song Shaojian,Li Bohan）.基于LSTM网络的光伏发电功率短期预测方法的研究（Short-term forecasting method of photovoltaic power based on LSTM）[J].可再生能源（Renewable Energy Resources）,2021,39(5):594-602.
[9]孟安波，陈嘉铭，黎湛联，等（Meng Anbo,Chen Jiaming,Li Zhanlian,et al）.基于相似日理论和CSO-WGPR的短期光伏发电功率预测（Short-term photovoltaic power generation prediction based on similar day theory and CSOWGPR）[J].高电压技术（High Voltage Engineering）,2021,47(4):1176-1184.
[10]王灿，张雪菲，凌凯，等（Wang Can,Zhang Xuefei,Ling Kai,et al）.基于区间概率不确定集的微电网两阶段自适应鲁棒优化调度（Two-stage adaptive robust optimal scheduling based on the interval probability uncertainty set for microgrids）[J].中国电机工程学报（Proceedings of the CSEE）,2024,44(5):1750-1763.
[11]刘嘉诚，刘俊，赵宏炎，等（Liu Jiacheng,Liu Jun,Zhao Hongyan,et al）.基于DKDE与改进mRMR特征选择的短期光伏出力预测（Short-term photovoltaic output forecasting based on diffusion kernel density estimation and improved max-relevance and min-redundancy feature selection）[J].电力系统自动化（Automation of Electric Power Systems）,2021,45(14):13-21.
[12]高博，茆超，张冲标，等（Gao Bo,Mao Chao,Zhang Chongbiao,et al）.基于时空图网络的分布式光伏发电出力预测（Output prediction of distributed photovoltaic power generation based on spatial-temporal graph neural network）[J].电力系统及其自动化学报（Proceedings of the CSU-EPSA）,2023,35(3):125-133.
[13]吴珺玥，赵二刚，郭增良，等（Wu Junyue,Zhao Ergang,Guo Zengliang,et al）.基于Spearman系数和TCN的光伏出力超短期多步预测（Ultra-short-term photovoltaic power multi-step prediction based on Spearman coefficient and TCN）[J].太阳能学报（Acta Energiae Solaris Sinica）,2023,44(9):180-186.
[14]王鹏翔，沈娟，王菁旸，等（Wang Pengxiang,Shen Juan,Wang Jingyang,et al）.基于PCA-LMD-WOA-ELM的短期光伏功率预测（Short term photovoltaic power prediction based on PCA-LMD-WOA-ELM）[J].智慧电力（Smart Power）,2022,50(6):72-78.
[15]孙顺远，刘康康（Sun Shunyuan,Liu Kangkang）.基于变量选择与组合协方差的GPR建模方法及其应用（GPR modelling method and its application based on variable selection and combined covariance）[J].控制工程（Control Engineering of China）,2021,28(12):2336-2342.
[16] Pires S,Mascarenhas C. Cyber threat analysis using Pearson and Spearman correlation via exploratory data analysis[C]//International Conference on Secure Cyber Computing and Communications. Jalandhar,India:IEEE,2023:257-262.
[17] Breiman L. Random forests[J]. Machine Learning,2001,45:5-35.
[18]李争，罗晓瑞，张杰，等（Li Zheng,Luo Xiaorui,Zhang Jie,et al）.基于改进麻雀搜索算法的光伏功率短期预测（Short term prediction of photovoltaic power based on improved sparrow search algorithm）[J].太阳能学报（Acta Energiae Solaris Sinica）,2023,44(6):284-289.
[19]黄南天，齐斌，刘座铭，等（Huang Nantian,Qi Bin,Liu Zuoming,et al）.采用面积灰关联决策的高斯过程回归概率短期负荷预测（Probabilistic short-term load forecasting using Gaussian process regression with area grey incidence decision making）[J].电力系统自动化（Automation of Electric Power Systems）,2018,42(23):64-71.
[20] Chen Minghao,Xu Zuhua,Zhao Jun,et al. Nonparametric identification based on multi-inherited Gaussian process regression for batch process[J]. Industrial and Engineering Chemistry Research,2020,59(47):20757-20766.
[21] Mirjalili S,Mirjalili S M,Lewis A. Grey wolf optimizer[J].Advances in Engineering Software,2014,69:46-61.
[22] Qu Jiaqi,Qian Zheng,Pei Yan. Day-ahead hourly photovoltaic power forecasting using attention-based CNNLSTM neural network embedded with multiple relevant and target variables prediction pattern[J]. Energy,2021,232:120996.
[23] Gao Mingming,Li Jianjing,Hong Feng,et al. Day-ahead power forecasting in a large-scale photovoltaic plant based on weather classification using LSTM[J]. Energy,2019,187:115838.
基本信息:
DOI：10.19635/j.cnki.csu-epsa.001474
中图分类号:TM615
引用信息:
[1]张赟宁,魏广军.考虑特征选择的短期光伏功率组合预测模型[J].电力系统及其自动化学报,2024,36(08):122-132.DOI:10.19635/j.cnki.csu-epsa.001474.
基金信息:
国家自然科学基金资助项目（62233006）