Medium Term Power Load Forecasting for Java and Bali Power System Using Artificial Neural Network and SARIMAX
Authors:
- 1 November 2018
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Abstract
Power load forecasting is an important part of electrical company operations. An accurate forecast can help the company makes various important decisions. Two known models for power load forecasting are ARMA model and its variants and Artificial Neural Network (ANN). The ARMA model has been used for decades while ANN can be considered as a more recent approach. In this paper MLP and SARIMAX are proposed to model the power load of Java and Bali power system. Both models can be used to forecast the load of Java and Bali power system with MAPE of 2.4% for SARIMAX and 2.7% for MLP. The time needed to build a SARIMAX model is shorter compared to MLP. In general, SARIMAX performs better compared to MLP. An application is also developed to facilitate data transformation, model training, and forecasting based on the proposed models.Keywords
This publication has 9 references indexed in Scilit:
- Short-term electricity load forecasting using time series and ensemble learning methods2015
- A Review of Short Term Load Forecasting using Artificial Neural Network ModelsProcedia Computer Science, 2015
- A methodology for Electric Power Load ForecastingAlexandria Engineering Journal, 2011
- Load Forecasting2006
- Survey of hybrid fuzzy neural approaches to electric load forecasting2002
- Identification of ARMAX model for short term load forecasting: an evolutionary programming approach2002
- Electric load forecasting: Literature survey and classification of methodsInternational Journal of Systems Science, 2002
- A neural network short term load forecasting model for the Greek power systemIEEE Transactions on Power Systems, 1996
- An implementation of a neural network based load forecasting model for the EMSIEEE Transactions on Power Systems, 1994