eprintid: 5363
rev_number: 2
eprint_status: archive
userid: 1
dir: disk0/00/00/53/63
datestamp: 2023-11-09 16:17:05
lastmod: 2023-11-09 16:17:05
status_changed: 2023-11-09 16:01:25
type: conference_item
metadata_visibility: show
creators_name: Ul Islam, B.
creators_name: Baharudin, Z.
creators_name: Nallagownden, P.
creators_name: Raza, M.Q.
title: A hybrid neuro-genetic approach for STLF: A comparative analysis of model parameter variations
ispublished: pub
keywords: Electric load forecasting; Electric utilities; Forecasting; Genetic algorithms; Heuristic algorithms; Neural networks; Population statistics; Soft computing, Comparative analysis; Computational time; Heuristic search technique; Load forecasting; Multi-layer perceptron neural networks; Operational decision making; Population sizes; Soft computing models, Economic analysis
note: cited By 3; Conference of 2014 IEEE 8th International Power Engineering and Optimization Conference, PEOCO 2014 ; Conference Date: 24 March 2014 Through 25 March 2014; Conference Code:105243
abstract: This paper portrays the comparison of multiple techniques applied to predict the load demand. In particular, it highlights the latest trends under new circumstances based on modern non analytical soft computing models based on Artificial Neural Network (ANN) and heuristic search technique genetic algorithm (GA), deployed in the domain of load forecasting. The prediction of future load has always been recognized as a pivotal process in the planning and operational decision making by managers of electric utilities. Multiple techniques and approaches having different engineering considerations and economic analysis are deployed for this purpose. However, ANN based methods for load forecast are found better in terms of accuracy and robustness during the past few years. This supremacy is because of the inherent ability of mapping and memorizing the relationships between inputs and outputs of ANN models during their training phase. A hybrid approach that uses ANN and GA is proposed in this research with an emphasis to study the effect of varying the model parameters of both techniques. The focus is to study the impact of varying the input variables and architecture of neural network; and population size, of GA. Further, a clear comparison is also presented that explains the results of these variations in terms of load forecast accuracy and computational time. © 2014 IEEE.
date: 2014
publisher: IEEE Computer Society
official_url: https://www.scopus.com/inward/record.uri?eid=2-s2.0-84901376518&doi=10.1109%2fPEOCO.2014.6814485&partnerID=40&md5=e220a50a30e3390d09d109986eecf9a4
id_number: 10.1109/PEOCO.2014.6814485
full_text_status: none
publication: Proceedings of the 2014 IEEE 8th International Power Engineering and Optimization Conference, PEOCO 2014
place_of_pub: Langkawi
pagerange: 526-531
refereed: TRUE
citation:   Ul Islam, B. and Baharudin, Z. and Nallagownden, P. and Raza, M.Q.  (2014) A hybrid neuro-genetic approach for STLF: A comparative analysis of model parameter variations.  In: UNSPECIFIED.