一种基于高斯过程回归模型的错峰用电负荷预测方法
|
摘要
错峰用电方案是供电企业通过行政、技术、经济等手段和用电客户协商结果,但实际上用电客户是否执行错峰用电方案由多种因素决定,具有一定的随机性,即计划错峰负荷与实际错峰负荷往往不一致,其差值满足某种概率模型。错峰用电负荷预测的目的是提前了解错峰用电方案的有效性,是供电企业的重要工作内容。基于上述错峰负荷的随机特点,提出了一种基于高斯过程的错峰用电负荷预测方法。实验表明,该方法比循环神经网络和传统时间序列方法的准确性要高。
Staggering peak using electricity plan is the result of negotiation between power supply enterprises and customers through administrative, technical and economic means. But in fact, whether or not customers implement the plan is decided by a variety of factors and has a certain randomness, that is, the planned staggering peak load is often inconsistent with the actual staggering peak load, and the difference meets a certain probability model. The purpose of forecasting staggering peak using electricity load is to understand the effectiveness of staggering peak using electricity plan in advance, which is an important work content of power supply enterprises. Based on the above stochastic characteristics of staggering peak load, this paper presents a forecasting method for staggering peak using electricity load based on Gaussian process. Experiments show that the accuracy of this method is higher than that of recurrent neural network and traditional time series methods.
Staggering peak using electricity plan is the result of negotiation between power supply enterprises and customers through administrative, technical and economic means. But in fact, whether or not customers implement the plan is decided by a variety of factors and has a certain randomness, that is, the planned staggering peak load is often inconsistent with the actual staggering peak load, and the difference meets a certain probability model. The purpose of forecasting staggering peak using electricity load is to understand the effectiveness of staggering peak using electricity plan in advance, which is an important work content of power supply enterprises. Based on the above stochastic characteristics of staggering peak load, this paper presents a forecasting method for staggering peak using electricity load based on Gaussian process. Experiments show that the accuracy of this method is higher than that of recurrent neural network and traditional time series methods.
引文
[1] Fatih Tasgetiren M, Pan Quan-Ke, Suganthan P N, et al.A discrete differential evolution algorithm for the no-wait flowshop scheduling problem with total flowtime criterion[C].Proceedings of the 2007 IEEE Symposium on Computational Intelligence in Scheduling.New York, USA:ACM,2007:251-258
[2] Liu Jian, WANG Shuang-hu, MING Zheng-feng.需求侧负荷管理中最优错峰计划的生成[J].电力系统自动化,2006.30(8):47-51
[3] Lu Tingting,Gao Ciwei,Su Weihua, et al.Optimal scheduling of peak load shifting by orderly power consumption management[J].Power DSM,2013(6):15-22
[4]石坤,孔令基,李彬等.基于熵值法改进的新型电力错峰资源调度策略[J].电力建设,2017.38(4):57-63
[5]李汉巨.考虑气象因素累积效应的电力系统96点日负荷预测方法[J].电气技术,2018.19(4):28-32
[6]陈亚,李萍.基于神经网络的短期电力负荷预测仿真研究[J].电气技术,2017.18(1):26-29
[7]程卓.引入负荷温度梯度的负荷预测方法研究[J].电气技术,2015.16(10):13-16
[8]何志昆,刘光斌,赵曦晶,王明昊.高斯过程回归方法综述[J].控制与决策,2013.8:1121-1129
[9] Jozefowicz R, Zaremba W, Sutskever I. An empirical exploration of recurrent network architectures[C].Proceedings of the 32nd International Conference on Machine Learning(ICML-15),2015:2342-2350
[2] Liu Jian, WANG Shuang-hu, MING Zheng-feng.需求侧负荷管理中最优错峰计划的生成[J].电力系统自动化,2006.30(8):47-51
[3] Lu Tingting,Gao Ciwei,Su Weihua, et al.Optimal scheduling of peak load shifting by orderly power consumption management[J].Power DSM,2013(6):15-22
[4]石坤,孔令基,李彬等.基于熵值法改进的新型电力错峰资源调度策略[J].电力建设,2017.38(4):57-63
[5]李汉巨.考虑气象因素累积效应的电力系统96点日负荷预测方法[J].电气技术,2018.19(4):28-32
[6]陈亚,李萍.基于神经网络的短期电力负荷预测仿真研究[J].电气技术,2017.18(1):26-29
[7]程卓.引入负荷温度梯度的负荷预测方法研究[J].电气技术,2015.16(10):13-16
[8]何志昆,刘光斌,赵曦晶,王明昊.高斯过程回归方法综述[J].控制与决策,2013.8:1121-1129
[9] Jozefowicz R, Zaremba W, Sutskever I. An empirical exploration of recurrent network architectures[C].Proceedings of the 32nd International Conference on Machine Learning(ICML-15),2015:2342-2350