Optimal short term load forecasting using LSSVM and improved BFOA considering Malaysia pandemic disrupted situation

Zaini, Farah Anishah (2024) Optimal short term load forecasting using LSSVM and improved BFOA considering Malaysia pandemic disrupted situation. Masters thesis, Universiti Teknikal Malaysia Melaka.

	Text (24 Pages) Optimal short term load forecasting using LSSVM and improved BFOA considering Malaysia pandemic disrupted situation.pdf - Submitted Version Download (965kB)
	Text (Full Text) Optimal short term load forecasting using LSSVM and improved BFOA considering Malaysia pandemic disrupted situation.pdf - Submitted Version Restricted to Registered users only Download (4MB)

Abstract

The COVID-19 pandemic's unprecedented disruptions significantly impacted electricity demand patterns across the globe. In Peninsular Malaysia, strict lockdown measures (Movement Control Orders - MCOs) led to the closure of non-essential businesses and stay-at-home orders. These sudden and dramatic shifts in consumption patterns posed a significant challenge for power system operations, which rely heavily on accurate short-term load forecasting (STLF) for efficient and cost-effective operation. Inaccurate forecasts can have substantial economic consequences, especially during peak load periods. Due to that reason, in this study, the hybrid forecasting model based on the Least Square Support Vector Machine (LSSVM) and Improved Bacterial Foraging Optimization Algorithm (IBFOA) is developed to perform an accurate STLF and applied to load in Peninsular Malaysia during the pandemic disrupted situation. The IBFOA is proposed by modifying the chemotaxis process in BFOA using a Sine Cosine Algorithm (SCA), which improves the convergence speed and accuracy of the algorithm. The LSSVM-IBFOA model demonstrates superior performance compared to standalone LSSVM and LSSVM-BFOA based on Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Squared Error (MSE), Normalized RMSE (NRMSE), and Determination Coefficient (R²). Using the proposed hybrid method, LSSVM-IBFOA consistently achieves the most significant error reductions for each errors value based on the average of five day-types (Monday-Sunday), on the testing datasets for the years 2020 and 2021. Furthermore, the proposed method demonstrates superior generalizability, with a substantial decrease in testing error compared to validation error in both years. For instance, in 2020, MAPE, MAE, MSE, RMSE, and NRMSE all witnessed reductions of 33.02%, 32.15%, 59.39%, 33.11%, and 32.75%, respectively. Similar trends were observed in 2021.This suggests the model's ability to adapt to changing load patterns, making it a valuable tool for real-world forecasting applications. Improved forecasting accuracy empowers energy providers to optimise resource allocation, power generation scheduling, and grid management, leading to potential cost reductions and increased efficiency.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	COVID-19 pandemic impact, Load forecasting accuracy, Electricity demand patterns
Divisions:	Library > Tesis > FTKE
Depositing User:	Muhamad Hafeez Zainudin
Date Deposited:	14 Feb 2025 16:45
Last Modified:	14 Feb 2025 16:45
URI:	http://eprints.utem.edu.my/id/eprint/28544
Statistic Details:	View Download Statistic

Actions (login required)

View Item