IEEE Digital Library
Graduate Research School
Graduate Research School / Centre for Communications Engineering Research / School of Engineering
Developing a good demand forecasting model is the art of identifying the best modelling parameters. Improving the forecasting performance needs to study the input/output parameters of the system to identify the effective forecasting variables. In this paper, the energy demand of Joondalup Campus of Edith Cowan University (ECU) in Western Australia has been selected as a case study for the design and verification of a suitable forecasting model. Fuzzy Subtractive Clustering Method (FSCM) based Adaptive Neuro Fuzzy Inference System (ANFIS) is used as a proposed modelling network in this paper. Basically, three-input forecasting models have been developed based on 12-month models to perform ECU energy demand forecasting. The input/output parameters selection was made after analysing the historical demand pattern in ECU energy system. Generally, increasing the number inputs in model network may have wider training scope and better forecasting accuracy. However, the wrong choice of the additional input would deteriorate the forecasting accuracy. From analysing the historical operation of ECU energy system, four and five-input variables could be identified and modelling has been performed. The result show that four-input models were the best in the prediction performance among 12-month models of the annual demand predicion of ECU.
This is an Author's Accepted Manuscript of: Mahmoud, T.S. , Habibi, D. , Bass, O. , & Lachowicz, S. (2011). Load demand forecasting: model inputs selection. Paper presented at the 2011 IEEE Innovative Smart Grid Technologies Asia. Perth, Australia. Available here
© 2011 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.