黎静华;黄乾;韦善阳;黄玉金;

• 1:广西电力系统最优化与节能技术重点实验室(广西大学)

摘要(Abstract)：

为提高光伏出力的预测精度,提出了一种改进深度学习算法的光伏出力预测方法。首先,针对传统的深度学习算法采用批量梯度下降(batch gradient descent,BGD)法训练模型参数速度慢的问题,利用随机梯度下降(stochastic gradient descent,SGD)法训练快的优点,提出了一种改进的随机-批量梯度下降(stochastic-batch gradient descent,S-BGD)搜索方法,该方法兼具SGD和BGD的优点,提高了参数训练的速度。然后,针对参数训练过程中容易陷入局部最优点和鞍点的问题,借鉴运动学理论,提出了一种基于梯度累积(gradient pile,GP)的训练方法。该方法以累积梯度作为参数的修正量,可以有效地避免训练陷入局部点和鞍点,进而提高预测精度。最后,以澳大利亚艾丽斯斯普林光伏电站的数据为样本,将所提方法应用于光伏出力预测中,验证所提方法的有效性。

关键词(KeyWords)： 光伏出力预测;深度学习算法;梯度下降法;梯度累积量;参数训练;神经网络;随机-批量梯度下降

Abstract:

Keywords:

基金项目(Foundation): 国家重点研发计划支持项目(2016YFB0900100);; 国家自然科学基金项目资助(51377027)~~

作者(Author): 黎静华;黄乾;韦善阳;黄玉金;

Email:

DOI: 10.13335/j.1000-3673.pst.2017.1393

参考文献(References)：

• [1]程泽,刘冲,刘力.基于相似时刻的光伏出力概率分布估计方法[J].电网技术,2017,41(2):448-455.Cheng Ze,Liu Chong,Liu Li.A method of probabilistic distribution estimation of PV generation based on similar time of day[J].Power System Technology,2017,41(2):448-455(in Chinese).
• [2]黎静华,左俊军,汪赛.大规模风电并网电力系统运行风险评估与分析[J].电网技术,2016,40(11):3503-3513.Li Jinghua,Zuo Junjun,Wang Sai.Analysis and assessment of operation risk for power system with large-scale wind power integration[J].Power System Technology,2016,40(11):3503-3513(in Chinese).
• [3]Alquthami T,Ravindra H,Faruque M O,et al.Study of photovoltaic integration impact on system stability using custom model of PV arrays integrated with PSS/E[C]//North American Power Symposium.Arlington,TX,USA:IEEE,2010:1-8.
• [4]龚莺飞,鲁宗相,乔颖,等.光伏功率预测技术[J].电力系统自动化,2016,40(4):140-151.Gong Yingfei,Lu Zongxiang,Qiao Ying,et al.An overview of photovoltaic energy system output forecasting technology[J].Automation of Electric Power Systems,2016,40(4):140-151(in Chinese).
• [5]Ma T,Yang H G,Lu L.Solar photovoltaic system modeling and performance prediction[J].Renewable&Sustainable Energy Reviews,2014,36(36):304-315.
• [6]Lorenz E,Hurka J,Heinemann D,et al.Irradiance forecasting for the power prediction of grid-connected photovoltaic systems[J].IEEE Journal of Selected Topics in Applied Earth Observations&Remote Sensing,2009,2(01):2-10.
• [7]丁明,徐宁舟.基于马尔可夫链的光伏发电系统输出功率短期预测方法[J].电网技术,2011,35(1):152-157.Ding Ming,Xu Ningzhou.A method to forecast short-term output power of photovoltaic generation system based on markov chain[J].Power System Technology,2011,35(1):152-157(in Chinese).
• [8]赵唯嘉,张宁,康重庆,等.光伏发电出力的条件预测误差概率分布估计方法[J].电力系统自动化,2015,39(16):8-15.Zhao Weijia,Zhang Ning,Kang Chongqing,et al.A method of probabilistic distribution of conditional forecast error for photovoltaic power generation[J].Automation of Electric Power Systems,2015,39(16):8-15(in Chinese).
• [9]兰飞,桑川川,梁浚杰,等.基于条件Copula函数的风电功率区间预测[J].中国电机工程学报,2016,36(S1):79-86.Lan Fei,Sang Chuanchuan,Liang Junjie,et al.Interval prediction for wind power based on conditional copula function[J].Proceedings of the CSEE,2016,36(S1):79-86(in Chinese).
• [10]高阳,张碧玲,毛京丽,等.基于机器学习的自适应光伏超短期出力预测模型[J].电网技术,2015,39(2):307-311.Gao Yang,Zhang Biling,Mao Jingli,et al.Machine learning-based adaptive very-short-term forecast model for photovoltaic power[J].Power System Technology,2015,39(2):307-311(in Chinese).
• [11]Bae K Y,Han S J,Dan K S.Hourly solar irradiance prediction based on support vector machine and its error analysis[J].IEEE Transactions on Power Systems,2017,32(2):935-945.
• [12]袁晓玲,施俊华,徐杰彦.计及天气类型指数的光伏发电短期出力预测[J].中国电机工程学报,2013,33(34):57-64.Yuan Xiaoling,Shi Junhua,Xu Jieyan.Short-term power forecasting for photovoltaic generation considering weather type index[J].Proceedings of the CSEE,2013,33(34):57-64(in Chinese).
• [13]Chow S K H,Lee E W M,Li D H W.Short-term prediction of photovoltaic energy generation by intelligent approach[J].Energy and Buildings,2012,55(55):660-667.
• [14]单英浩,付青,耿炫,等.基于改进BP-SVM-ELM与粒子化SOM-LSF的微电网光伏发电组合预测方法[J].中国电机工程学报,2016,36(12):3334-3342.Shan Yinghao,Fu Qing,Geng Xuan,et al.Combined forecasting of photovoltaic power generation in microgrid based on the improved BP-SVM-ELM and SOM-LSF with particlization[J].Proceedings of the CSEE,2016,36(12):3334-3342(in Chinese).
• [15]茆美琴,龚文剑,张榴晨,等.基于EEMD-SVM方法的光伏电站短期出力预测[J].中国电机工程学报,2013,33(34):17-24.Mao Meiqin,Gong Wenjian,Zhang Liuchen,et al.Short-term photovoltaic generation forecasting based on EEMD-SVM combined method[J].Proceedings of the CSEE,2013,33(34):17-24(in Chinese).
• [16]Malvoni M,Giorgi M G D,Congedo P M.Photovoltaic forecast based on hybrid PCA–LSSVM using dimensionality reducted data[J].Neurocomputing,2016(211):72-83.
• [17]Hinton G E,Salakhutdinov R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313(5786):504-507.
• [18]Sch?lkopf B,Platt J,Hofmann T.Greedy layer-wise training of deep networks[C]//International Conference on Neural Information Processing Systems.QC,Canada:MIT Press,2006:153-160.
• [19]Mohamed A R,Dahl G E,Hinton G.Acoustic modeling using deep belief networks[J].IEEE Transactions on Audio Speech&Language Processing,2011,20(1):14-22.
• [20]Dahl G E,Yu D,Deng L,et al.Application of pretrained deep neural networks to large vocabulary conversational speech recognition[J].IEEE Transactions on Audio,Speech&Language Processing,2012,20(1):30-42.
• [21]Goodfellow L,Bengio Y S,Gourville A.Deep learning[M].Mit Press,2016:240-249.
• [22]Rumelhart D E,Hinton G E,Williams R J.Learning representations by back-propagating errors[J].Parallel Distributed Processing:Explorations in the Microstructure of Cognition,1986(1):318-362.
• [23]陈凯.深度学习模型的高效训练算法研究[D].合肥:中国科学技术大学,2016:4-6.