基于卷积神经网络与纵横交叉算法的二维组合短期负荷预测方法研究
|
摘要
随着电力系统的飞速发展,电力负荷数据的规模也愈加庞大,以往基于小规模数据的电力负荷预测算法可能无法容纳大量数据集。为改善预测模型的工程实用性,本文提出了一种新型的机器学习模型,该模型将卷积神经网络(CNN)与纵横交叉优化算法(CSO)结合起来,应用于短期负荷预测。从大规模的负荷数据中收集到横向相邻日和纵向的相似日负荷数据,设置横向预测和纵向预测的权值系数,再用CSO优化算法去找最优系数,得到最后的二维组合预测结果,并与其它机器学习算法比较。通过实验证明,模型可以快速有效地处理大规模的负荷数据,具有较强的泛化能力。
With the rapid development of power system,the scale of power load data is becoming larger and larger.In the past,power load forecasting algorithms based on small-scale data may not be able to accommodate large data sets.In order to improve the engineering practicability of the predictive model,a new machine learning model is proposed which combines the Convolutional Neural Network(CNN) with the Cross and Cross Optimization Algorithm(CSO) for short-term load forecasting.Collecting similar daily load data from horizontally adjacent days and verticals from large-scale load data,setting the weight coefficients of lateral prediction and longitudinal prediction,and then using CSO optimization algorithm to find the optimal coefficient to obtain the final two-dimensional combined prediction,the result is compared to other machine learning algorithms.Experiments show that the model can quickly process large-scale load data and has strong generalization ability.
With the rapid development of power system,the scale of power load data is becoming larger and larger.In the past,power load forecasting algorithms based on small-scale data may not be able to accommodate large data sets.In order to improve the engineering practicability of the predictive model,a new machine learning model is proposed which combines the Convolutional Neural Network(CNN) with the Cross and Cross Optimization Algorithm(CSO) for short-term load forecasting.Collecting similar daily load data from horizontally adjacent days and verticals from large-scale load data,setting the weight coefficients of lateral prediction and longitudinal prediction,and then using CSO optimization algorithm to find the optimal coefficient to obtain the final two-dimensional combined prediction,the result is compared to other machine learning algorithms.Experiments show that the model can quickly process large-scale load data and has strong generalization ability.
引文
[1]廖旎焕,胡智宏,马莹莹,等.电力系统短期负荷预测方法综述[J].电力系统保护与控制,2011,39(1):147-152.
[2]FAN S,CHEN L.Short-Term Load Forecasting Based on an Adaptive Hybrid Method[J].IEEE Transactions on Power Systems,2006,21(1):392-401.
[3]Diamantoulakis P D,Kapinas V M,Karagiannidis G K.Big Data Analytics for Dynamic Energy Management in Smart Grids[J].Big Data Research,2015,2(3):94-101.
[4]李龙,魏靖,黎灿兵,等.基于人工神经网络的负荷模型预测[J].电工技术学报,2015,30(8):225-230.
[5]吴倩红,高军,侯广松,等.实现影响因素多源异构融合的短期负荷预测支持向量机算法[J].电力系统自动化,2016,40(15):67-72+92.
[6]杨再鹤,向铁元,郑丹.基于小波变换和SVM算法的微电网短期负荷预测研究[J].现代电力,2014,31(3):74-79.
[7]Hernández,Luis,Baladrón,Carlos,Aguiar J M,et al.Artificial neural networks for short-term load forecasting in microgrids environment[J].Energy,2014,75(1):252-264.
[8]王岗,姜杰,唐昆明,等.基于自适应双向加权最小二乘支持向量机的超短期负荷预测[J].电力系统保护与控制,2010,38(19):142-146.
[9]吴润泽,包正睿,王文韬,等.Hadoop架构下基于模式匹配的短期电力负荷预测方法[J].电工技术学报,2018,33(7):1542-1551.
[10][7]Hernández,Luis,Baladrón,Carlos,Aguiar J M,et al.Artificial neural networks for short-term load forecasting in microgrids environment[J].Energy,2014,75(1):252-264.
[11]X.Wang,Y.Shi,J.Zhang,L.Bo,C.Cheng.Computation services and applications of electricity big data based on Hadoop Power Syst Technol,2015,39(11):3128-3133.
[12]GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[C]//Computer Vision and Pattern Recognition(CVPR),2014 IEEE Conference on.S.l.:s.n.,2014:580-587.
[13]Geoffrey E.Hinton,Nitish Srivastava,Alex Krizhevsky,et al.Improving neural networks by preventing co-adaptation of feature detectors[J].CoRR,2012.
[14]Xishuang Dong,Lijun Qian and Lei Huang.[IEEE 2017IEEE International Conference on Big Data and Smart Computing(BigComp)-Jeju Island,South Korea(2017.2.13-2017.2.16)]2017 IEEE International Conference on Big Data and Smart Computing(BigComp)-Short-term load forecasting in smart grid:A combined CNN and K-means clustering approach[C]//IEEE International Conference on Big Data&Smart Computing.IEEE,2017:119-125.
[15]孟安波,胡函武,刘向东.基于纵横交叉算法优化神经网络的负荷预测模型[J].电力系统保护与控制,2016,44(7):102-106.
[2]FAN S,CHEN L.Short-Term Load Forecasting Based on an Adaptive Hybrid Method[J].IEEE Transactions on Power Systems,2006,21(1):392-401.
[3]Diamantoulakis P D,Kapinas V M,Karagiannidis G K.Big Data Analytics for Dynamic Energy Management in Smart Grids[J].Big Data Research,2015,2(3):94-101.
[4]李龙,魏靖,黎灿兵,等.基于人工神经网络的负荷模型预测[J].电工技术学报,2015,30(8):225-230.
[5]吴倩红,高军,侯广松,等.实现影响因素多源异构融合的短期负荷预测支持向量机算法[J].电力系统自动化,2016,40(15):67-72+92.
[6]杨再鹤,向铁元,郑丹.基于小波变换和SVM算法的微电网短期负荷预测研究[J].现代电力,2014,31(3):74-79.
[7]Hernández,Luis,Baladrón,Carlos,Aguiar J M,et al.Artificial neural networks for short-term load forecasting in microgrids environment[J].Energy,2014,75(1):252-264.
[8]王岗,姜杰,唐昆明,等.基于自适应双向加权最小二乘支持向量机的超短期负荷预测[J].电力系统保护与控制,2010,38(19):142-146.
[9]吴润泽,包正睿,王文韬,等.Hadoop架构下基于模式匹配的短期电力负荷预测方法[J].电工技术学报,2018,33(7):1542-1551.
[10][7]Hernández,Luis,Baladrón,Carlos,Aguiar J M,et al.Artificial neural networks for short-term load forecasting in microgrids environment[J].Energy,2014,75(1):252-264.
[11]X.Wang,Y.Shi,J.Zhang,L.Bo,C.Cheng.Computation services and applications of electricity big data based on Hadoop Power Syst Technol,2015,39(11):3128-3133.
[12]GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[C]//Computer Vision and Pattern Recognition(CVPR),2014 IEEE Conference on.S.l.:s.n.,2014:580-587.
[13]Geoffrey E.Hinton,Nitish Srivastava,Alex Krizhevsky,et al.Improving neural networks by preventing co-adaptation of feature detectors[J].CoRR,2012.
[14]Xishuang Dong,Lijun Qian and Lei Huang.[IEEE 2017IEEE International Conference on Big Data and Smart Computing(BigComp)-Jeju Island,South Korea(2017.2.13-2017.2.16)]2017 IEEE International Conference on Big Data and Smart Computing(BigComp)-Short-term load forecasting in smart grid:A combined CNN and K-means clustering approach[C]//IEEE International Conference on Big Data&Smart Computing.IEEE,2017:119-125.
[15]孟安波,胡函武,刘向东.基于纵横交叉算法优化神经网络的负荷预测模型[J].电力系统保护与控制,2016,44(7):102-106.