Bayesian optimization in a support vector regression model for short-term electricity load forecasting

Main Article Content

Su Wutyi Hnin Chawalit Jeenanunta

Abstract

Purpose: Support vector regression (SVR) has long been known as a great tool in forecasting. SVR combined with evolutionary algorithms have been used in many remarkable applications. Bayesian optimization (BO) has the ability to find good points in a search space without many function evaluations. This paper presents short-term load forecasting (STLF) in Thailand using SVR with Bayesian optimization (BO). The purpose of this paper is to improve forecasting accuracy by optimizing the hyperparameters of SVR.


        Design/methodology/approach: The Electricity Generating Authority of Thailand (EGAT) provides 30 minute load data for Bangkok and the metropolitan region. The data from August 2015 to July 2017 is used for training and testing. Mean absolute percentage error (MAPE) and tracking signal (TS) are used to measure the performance of the proposed model. The hyperparameters of SVR are optimized using three algorithms, the genetic algorithm (GA), particle swarm optimization (PSO), and Bayesian optimization (BO). Findings: By comparing the MAPE results, the SVR-BO outperforms the other two algorithms.

Keywords

Article Details

How to Cite
Hnin, S. W., & Jeenanunta, C. (2019). Bayesian optimization in a support vector regression model for short-term electricity load forecasting. Engineering and Applied Science Research, 46(3), 267-275. Retrieved from https://www.tci-thaijo.org/index.php/easr/article/view/166962
Section
TECHNICAL

References

[1] Fan GF, Peng LL, Zhao X, Hong WC. Applications of hybrid EMD with PSO and GA for an SVR-based load forecasting model. Energies. 2017;10(11):1-22.

[2] Smola AJ, Schölkopf B. A tutorial on support vector regression. Stat Comput. 2004;14(3):199-222.

[3] Vapnik VN, Vapnik V. Statistical learning theory. New York: Wiley; 1998.

[4] Vapnik V. The nature of statistical learning theory. New York: Springer; 2013.

[5] Li YC, Fang TJ, Yu EK. Study of support vector machines for short-term load forecasting. Proceedings of the CSEE. 2003;23(6):55-9.

[6] Lin SW, Ying KC, Chen SC, Lee ZJ. Particle swarm optimization for parameter determination and feature selection of support vector machines. Expert Syst Appl. 2008;35(4):1817-24.

[7] Fan GF, Peng LL, Hong WC. Short term load forecasting based on phase space reconstruction algorithm and bi-square kernel regression model. Appl Energ . 2018;224:13-33.

[8] Gardner ES. Exponential smoothing: The state of the art. J Forecast. 1985;4(1):1-28.

[9] Papalexopoulos AD, Hesterberg TC. A regression-based approach to short-term system load forecasting. IEEE Trans Power Syst. 1990;5(4):1535-47.

[10] Akaike H. Fitting autoregressive models for prediction. Ann Inst Stat Math. 1969;21(1):243-7.

[11] Huang SJ, Shih KR. Short-term load forecasting via ARMA model identification including non-Gaussian process considerations. IEEE Trans Power Syst. 2003;18(2):673-9.

[12] Taylor JW. Triple seasonal methods for short-term electricity demand forecasting. Eur J Oper Res. 2010;204(1):139-52.

[13] Conejo AJ, Plazas MA, Espinola R, Molina AB. Day-ahead electricity price forecasting using the wavelet transform and ARIMA models. IEEE Trans Power Syst. 2005;20(2):1035-42.

[14] Mbarek MB, Feki R. Using fuzzy logic to renewable energy forecasting: a case study of France. Int J Energ Tech Pol. 2016;12(4):357-76.

[15] Li W, Yang Y, Yang Z, Zhang C. Fuzzy system identification based on support vector regression and genetic algorithm. Int J Model Ident Contr. 2011;12(1-2):50-5.

[16] Draidi A, Labed D. A Neuro-fuzzy approach for predicting load peak profile. Int J Electr Comput Eng. 2015;5(6): 1304-10.

[17] Jeenanunta C, Abeyrathna KD. Combine particle swarm optimization with artificial neural networks for short-term load forecasting. Int Sci J Eng Tech. 2017;1(1):25-30.

[18] Basak D, Pal S, Patranabis DC. Support vector regression. Neural Inform Process-Lett Rev. 2007;11(10):203-24.

[19] Drucker H, Burges CJ, Kaufman L, Smola AJ, Vapnik V. Support vector regression machines. In: Mozer MC, Jordan MI, Petsche T, editors. Advances in neural information processing systems. USA: MIT Press; 1997. p. 155-61.

[20] Chawalit J, Su WH. A Support vector regression model(SVR) for short term electricity forecasting in Thailand. 2016 International Conference on Applied Statistics; 2016 Jul 13-15; Phuket, Thailand.

[21] Su WH, Chawalit J. Short-term electricity load forecasting in Thailand: an analysis on different input variables. IOP Conf Ser: Earth Environ Sci. 2018;192(2018):1-8.

[22] Erişen E, Iyigun C, Tanrısever F. Short-term electricity load forecasting with special days: an analysis on parametric and non-parametric methods. Ann Oper Res. 2017:1-34.

[23] Barman M, Choudhury ND, Sutradhar S. A regional hybrid GOA-SVM model based on similar day approach for short-term load forecasting in Assam, India. Energy. 2018;145:710-20.

[24] Sousa JC, Jorge HM, Neves LP. Short‐term load forecasting based on support vector regression and load profiling. Int J Energ Res. 2014;38(3):350-62.

[25] Mohandes M. Support vector machines for short‐term electrical load forecasting. Int J Energ Res. 2002;26(4):335-45.

[26] Hong WC. Hybrid evolutionary algorithms in a SVR-based electric load forecasting model. Int J Electr Power Energ Syst. 2009;31(7-8):409-17.

[27] Hong WC, Dong Y, Chen LY, Lai CY. "Taiwanese 3G mobile phone demand forecasting by SVR with hybrid evolutionary algorithms. Expert Syst Appl. 2010;37(6):4452-62.

[28] Hong WC, Dong Y, Zhang WY, Chen LY, Panigrahi EP. Cyclic electric load forecasting by seasonal SVR with chaotic genetic algorithm. Int J Electr Power Energ Syst. 2013;44(1):604-14.

[29] Ju FY, Hong WC. Application of seasonal SVR with chaotic gravitational search algorithm in electricity forecasting. Appl Math Model. 2013;37(23):9643-51.

[30] Abeyrathna KD, Jeenanunta CJ. Hybrid particle swarm optimization with genetic algorithm to train artificial neural networks for short-term load forecasting. Int J Swarm Intell Res. 2019;10(1):1-14.

[31] Sadati N, Zamani M, Mahdavian HRF. Hybrid particle swarm-based-simulated annealing optimization techniques. IECON 2006 - 32nd Annual Conference on IEEE Industrial Electronics; 2006 Nov 6-10; Paris, France. USA: IEEE; 2006. p. 644-8.

[32] Niu D, Wang Y, Wu DD. Power load forecasting using support vector machine and ant colony optimization. Expert Syst Appl. 2010;37(3):2531-9.

[33] Gopi G, Dauwels J, Asif MT, Ashwin S, Mitrovic N, Rasheed U, et al. Bayesian support vector regression for traffic speed prediction with error bars. 16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013); 2013 Oct 6-9; The Hague, Netherlands. USA: IEEE; 2013. p. 136-41.

[34] Chu W, Keerthi SS, Ong CJ. Bayesian support vector regression using a unified loss function. IEEE Trans Neural Network. 2004;15(1):29-44.

[35] Klein A, Falkner S, Bartels S, Hennig P, Hutter F. Fast bayesian optimization of machine learning hyperparameters on large datasets. arXiv:1605.07079. 2016:1-9.

[36] Snoek J, Larochelle H, Adams RP. Practical bayesian optimization of machine learning algorithms. In: Pereira F, Burges CJC, Bottou L, Weinberger KQ, editors. Proceeding Advances in neural information processing systems; 2012 Dec 3-6; Nevada, USA. USA: NIPS. p. 2951-9.

[37] Martinez-Cantin R. Bayesopt: a bayesian optimization library for nonlinear optimization, experimental design and bandits. J Mach Learn Res. 2014;15(1):3735-9.

[38] Gao JB, Gunn SR, Harris CJ, Brown M. A probabilistic framework for SVM regression and error bar estimation. Mach Learn. 2002;46(1-3):71-89.

[39] Law T, Shawe-Taylor J. Practical bayesian support vector regression for financial time series prediction and market condition change detection. Quant Finance. 2017;17(9):1-14.

[40] Jeenanunta C, Abeyrathna KD, Dilhani MS, Hnin S W, Phyo PP. Time series outlier detection for short-term electricity load demand forecasting. Int Sci J Eng Tech. 2018;2(1):37-50.

[41] Cao LJ, Tay FEH. Support vector machine with adaptive parameters in financial time series forecasting. IEEE Trans Neural Network. 2003;14(6):1506-18.