Conference Paper

Very short-term electricity load demand forecasting using support vector regression

DOI: 10.1109/IJCNN.2009.5179063 Conference: International Joint Conference on Neural Networks, IJCNN 2009, Atlanta, Georgia, USA, 14-19 June 2009
Source: DBLP


In this paper, we present a new approach for very short term electricity load demand forecasting. In particular, we apply support vector regression to predict the load demand every 5 minutes based on historical data from the Australian electricity operator NEMMCO for 2006-2008. The results show that support vector regression is a very promising approach, outperforming backpropagation neural networks, which is the most popular prediction model used by both industry forecasters and researchers. However, it is interesting to note that support vector regression gives similar results to the simpler linear regression and least means squares models. We also discuss the performance of four different feature sets with these prediction models and the application of a correlation-based sub-set feature selection method.

8 Reads
  • Source
    • "Setiawan et al (Setiawan, A. et al, 2009) have presented a new approach for the very short-term electricity load demand forecasting using SVR. Support vector regression was applied to predict the load demand every 5 minutes based on historical data from the Australian electricity operator NEMMCO for 2006-2008. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Electric load forecasting is a real-life problem in industry. Electricity supplier’s use forecasting models to predict the load demand of their customers to increase/decrease the power generated and to minimize the operating costs of producing electricity. This paper presents the development and the implementation of three new electricity demand-forecasting models using the adaptive neuro-fuzzy inference system (ANFIS) approach in parallel load series. The input-output data pairs used are the real-time quart-hourly metropolitan France electricity load obtained from the RTE website and forecasts are done for lead-time of a 1 h ahead. Results and forecasting performance obtained reveal the effectiveness of the third proposed approach and shows that 56 % of the forecasted loads have an APE (absolute percentage error) under 0.5, and an APE under one was achieved for about 80 % of cases. Which mean that it is possible to build a high accuracy model with less historical data using a combination of neural network and fuzzy logic.
    Computational Intelligence Applications in Modeling and Control, 2015 edited by Ahmad Taher Azar, Sundarapandian Vaidyanathan, 12/2014: chapter One-Hour Ahead Electric Load Forecasting Using Neuro-fuzzy System in a Parallel Approach: pages 95-121; Springer International Publishing., ISBN: 978-3-319-11017-2
  • Source
    • "It can be seen in the literature that the projections of electricity demand are basic for power system studies such as expansion, planning and operation. However, it can be held on different time horizons like long-term (1–10 years), medium-term (several months to a year), short-term (a day ahead) and very short-term (hours and minutes ahead), depending on the planning objectives [1]. The estimation of future behavior of electric loads is directly linked to the task of decision making and achievement of management actions on both the demand and supply side. "
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a new methodology for electricity demand forecasting on very short-term horizon based on a discrete probabilistic model (Markov Chain). The modeling process is automated by a feature extraction tool, the Self-Organizing Map, considering historical data of climate variables (air temperature, relative humidity and wind speed) and load behavior, related through the thermal discomfort index and wind chill. Thus, it is possible to estimate the probability of a certain demand level occur given a current climatic condition, as well as the number of time intervals (hours) until this occurs. The forecast is then used to control the decentralized dispatch of a small hydroelectric power plant, aiming to minimize overtaking the transmission contract.
    Electric Power Systems Research 07/2014; 112:27–36. DOI:10.1016/j.epsr.2014.03.005 · 1.75 Impact Factor
  • Source
    • "Several methodologies for forecasting electricity demand on different projection horizons have been proposed in recent years [1]-[9]. However, the implementation of this methodologies are limited to medium and large electrical systems and their performance is associated to the reliability of historical data from different variables and their forecast ability depends heavily on accurate knowledge of the load behavior of the electric systems under study. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Climate changes are related to air temperature, relative humidity, wind speed, precipitation level, among others. These variables have a large effect on very short-term electricity demand (minutes or hour ahead). In this context, this paper presents a methodology for very short-term load forecasting considering the climatic change impact. Therefore, air temperature and relative humidity are related through the Discomfort Index, and the first with the wind speed are linked through the Wind Chill. These climate indexes are used in this paper for electric load forecasting together with a multiple regression model. The proposed projection methodology was tested computationally and compared with real date obtained from a distribution utility located in southern Brazil. The results demonstrate the strong dependence of the evolution of electricity demand with climatic change in the very short-term.
Show more