SCIENTIA SINICA Informationis, Volume 48 , Issue 10 : 1300-1315(2018) https://doi.org/10.1360/N112018-00075

Decomposition-optimization-ensemble learning approach for electricity price forecasting

More info
  • ReceivedMar 30, 2018
  • AcceptedMay 25, 2018
  • PublishedOct 9, 2018


随着新一轮电力市场改革的持续推进, 电价作为反映市场运营状况的重要指标, 准确预测电价能够帮助电力市场博弈方进行风险规避, 达到经济收益最大化.本文提出了一种新的基于分解–优化–集成(decomposition-optimization-ensemble, DOE)的混合学习模型来预测电价,首先利用快速集成经验模态分解方法将波动性较大的电价数据分解成一系列本征模态函数和一个残差序列.然后对鲸鱼算法从收敛速度、精度和算法搜索能力3个方面进行改进, 再利用改进的鲸鱼算法优化径向基神经网络的扩展系数,采用优化后的径向基神经网络模型对分解得到的本征模态函数和残差序列进行预测.最后对分解后的子序列预测值进行集成, 得到电价的预测值. 为了验证混合学习模型的预测效果,本文对美国宾夕法尼亚–新泽西–马里兰电力市场的电价进行中长期和短期预测.实证结果显示DOE混合学习模型在水平精度和方向精度上均能获得很好的效果.

Funded by

国家自然科学基金(批准号: 61773401,61304067,11601524,61761130081,11571368)


[1] Maciejowska K, Nowotarski J, Weron R. Probabilistic forecasting of electricity spot prices using Factor Quantile Regression Averaging. Int J Forecasting, 2016, 32: 957-965 CrossRef Google Scholar

[2] Conejo A J, Plazas M A, Espinola R. Day-Ahead Electricity Price Forecasting Using the Wavelet Transform and ARIMA Models. IEEE Trans Power Syst, 2005, 20: 1035-1042 CrossRef ADS Google Scholar

[3] Xu W, Zhou H, Cheng N. Internet of vehicles in big data era. IEEE/CAA J Autom Sin, 2018, 5: 19-35 CrossRef Google Scholar

[4] Zhang C S. Challenges in machine learning. Sci Sin Inform, 2013, 43: 1612--1623. Google Scholar

[5] Itaba S, Mori H. An Electricity Price Forecasting Model with Fuzzy Clustering Preconditioned ANN. IEEJ TransPE, 2018, 138: 90-98 CrossRef ADS Google Scholar

[6] Saadaoui F. A seasonal feedforward neural network to forecast electricity prices. Neural Comput Applic, 2017, 28: 835-847 CrossRef Google Scholar

[7] Xiao Q, Zeng Z. Scale-Limited Lagrange Stability and Finite-Time Synchronization for Memristive Recurrent Neural Networks on Time Scales.. IEEE Trans Cybern, 2017, 47: 2984-2994 CrossRef PubMed Google Scholar

[8] Wu A, Liu L, Huang T. Mittag-Leffler stability of fractional-order neural networks in the presence of generalized piecewise constant arguments.. Neural Networks, 2017, 85: 118-127 CrossRef PubMed Google Scholar

[9] ?zgüner E, T?r O B, Güven A N. Probabilistic day-ahead system marginal price forecasting with ANN for the Turkish electricity market. Turk J Elec Eng Comp Sci, 2017, 25: 4923-4935 CrossRef Google Scholar

[10] Zhu S, Yang Q, Shen Y. Noise further expresses exponential decay for globally exponentially stable time-varying delayed neural networks.. Neural Networks, 2016, 77: 7-13 CrossRef PubMed Google Scholar

[11] Zhang W, Zhang H, Liu J. Weather prediction with multiclass support vector machines in the fault detection of photovoltaic system. IEEE/CAA J Autom Sin, 2017, 4: 520-525 CrossRef Google Scholar

[12] Ma Z, Zhong H, Xie L. Month ahead average daily electricity price profile forecasting based on a hybrid nonlinear regression and SVM model: an ERCOT case study. J Mod Power Syst Clean Energy, 2018, 6: 281-291 CrossRef Google Scholar

[13] Shao Z, Yang S L, Gao F. A new electricity price prediction strategy using mutual information-based SVM-RFE classification. Renew Sustain Energy Rev, 2017, 70: 330-341 CrossRef Google Scholar

[14] Yan X, Chowdhury N A. Mid-term electricity market clearing price forecasting using multiple least squares support vector machines. IET Generation Transmission Distribution, 2014, 8: 1572-1582 CrossRef Google Scholar

[15] Mareda T, Gaudard L, Romerio F. A parametric genetic algorithm approach to assess complementary options of large scale windsolar coupling. IEEE/CAA J Autom Sin, 2017, 4: 260-272 CrossRef Google Scholar

[16] Gu W, Yu Y, Hu W. Artificial bee colony algorithmbased parameter estimation of fractional-order chaotic system with time delay. IEEE/CAA J Autom Sin, 2017, 4: 107-113 CrossRef Google Scholar

[17] Heydari A, Keynia F. A new intelligent heuristic combined method for short-term electricity price forecasting in deregulated markets. Australian J Electrical Electron Eng, 2016, 13: 258-267 CrossRef Google Scholar

[18] Liu H, Mi X, Li Y. An experimental investigation of three new hybrid wind speed forecasting models using multi-decomposing strategy and ELM algorithm. Renew Energy, 2018, 123: 694-705 CrossRef Google Scholar

[19] Sun S, Wang S, Zhang G. A decomposition-clustering-ensemble learning approach for solar radiation forecasting. Sol Energy, 2018, 163: 189-199 CrossRef ADS Google Scholar

[20] Mirjalili S, Lewis A. The Whale Optimization Algorithm. Adv Eng Software, 2016, 95: 51-67 CrossRef Google Scholar

[21] Wang J, Du P, Niu T. A novel hybrid system based on a new proposed algorithm-Multi-Objective Whale Optimization Algorithm for wind speed forecasting. Appl Energy, 2017, 208: 344-360 CrossRef Google Scholar

[22] Bento P M R, Pombo J A N, Calado M R A. A bat optimized neural network and wavelet transform approach for short-term price forecasting. Appl Energy, 2018, 210: 88-97 CrossRef Google Scholar

[23] Wang D, Luo H, Grunder O. Multi-step ahead electricity price forecasting using a hybrid model based on two-layer decomposition technique and BP neural network optimized by firefly algorithm. Appl Energy, 2017, 190: 390-407 CrossRef Google Scholar

[24] Shayeghi H, Ghasemi A, Moradzadeh M. Day-ahead electricity price forecasting using WPT, GMI and modified LSSVM-based S-OLABC algorithm. Soft Comput, 2017, 21: 525-541 CrossRef Google Scholar

[25] Wang Y H, Yeh C H, Young H W V. On the computational complexity of the empirical mode decomposition algorithm. Physica A-Statistical Mech its Appl, 2014, 400: 159-167 CrossRef ADS Google Scholar

[26] Gu J, Gu M, Cao C. A novel competitive co-evolutionary quantum genetic algorithm for stochastic job shop scheduling problem. Comput Operations Res, 2010, 37: 927-937 CrossRef Google Scholar

[27] Zhan S, Huo H. Improved PSO-based task scheduling algorithm in cloud computing. J Inf Computat Sci, 2012, 9: 3821--3829. Google Scholar

[28] Lewis C. International and Business Forecasting Methods. London: Butter-Worths, 1982. Google Scholar

  • Figure 1

    (网络版彩图) FEEMD-IWOA-RBF混合学习框架

  • Figure 2

    (网络版彩图) PJM电力市场原始电价数据

  • Figure 3

    (网络版彩图) FEEMD算法分解后的电价序列

  • Figure 4

    (网络版彩图) 预测模型中长期预测相对误差

  • Figure 7

    (网络版彩图) 混合模型短期预测相对误差

  • Figure 8

    (网络版彩图) 模型的RMSE结果对比图

  • Figure 13

    (网络版彩图) 模型的$D_s$结果对比


    Algorithm 1 IWOA算法

    Initialize the whales' population $X_i~(i=1,2,\ldots,n)$ and calculate the fitness of each whale;

    $X^*_{\rm~prey}=$ the fittest whale (a similar position of the prey);

    while $l<$ maximum number of iterations

    for each whale

    Update $a,~A,~C,~t$ and $p$;

    if1 $p<0.5$

    if2 $|A|<1$

    update the position of the current whaleby 13;

    else if2 $|A|>1$

    Select a random whale $X_{\rm~rand}$;

    Update the position of the current whale by 5;

    end if2

    else if1 $p>0.5$

    Update the position of the current whale by 7;

    end if1

    Using quantum method to update the positions of all the whales by 10;

    Choose the better updating method by comparing the fitness of 5, 7, 13 and 10;

    end for

    Check if any whale goes beyond the search space and amend it;

    Calculate the fitness of each whale;

    Update $X^*_{\rm~prey}$ if there is a better solution;


    end while

    return $X^*_{\rm~prey}.$

  • Table 11  
  • Table 22  
  • Table 33  
  • Table 44  
  • Table 55  
  • Table 66  
  • Table 77  

Copyright 2020  CHINA SCIENCE PUBLISHING & MEDIA LTD.  中国科技出版传媒股份有限公司  版权所有

京ICP备14028887号-23       京公网安备11010102003388号