论文信息 - Short-Term Load Forecasting Using Parametric and Non-parametric Approaches

Short-Term Load Forecasting Using Parametric and Non-parametric Approaches

Electricity sector of any country contributes a good amount in its GDP. Hence, big losses or inefficiency in this sector can not be tolerated. Hence, load forecasting is the need of hour. Since, nowadays, there is much variety in load consumptions, short-term load forecasting is gaining more attention. This paper will focus on an accurate and reliable short-term load forecasting (STLF) system which will reduce system operation cost and make system more reliable. Time series models are applied on GEFCom 2014 dataset. Two types of approaches, parametric and non-parametric, have been used and compared. In this paper, state-of-the-art models (XGBoost and LSTM) from both approaches are used.

[1] Hassan Soltan,et al. A methodology for Electric Power Load Forecasting , 2011 .

[2] Souhaib Ben Taieb,et al. A gradient boosting approach to the Kaggle load forecasting competition , 2014 .

[3] Rob J Hyndman,et al. Probabilistic energy forecasting: Global Energy Forecasting Competition 2014 and beyond , 2016 .

[4] Heeyoung Kim,et al. A new metric of absolute percentage error for intermittent demand forecasts , 2016 .

[5] Hesham K. Alfares,et al. Electric load forecasting: Literature survey and classification of methods , 2002, Int. J. Syst. Sci..

[6] Xiaohua Li,et al. Electric load forecasting in smart grids using Long-Short-Term-Memory based Recurrent Neural Network , 2017, 2017 51st Annual Conference on Information Sciences and Systems (CISS).