Short-term forecast of electricity load before and during the pandemic Covid-19 in Malaysia by using least square support vector machine (LSSVM)

Jaffry, Saiful Adlan (2020) Short-term forecast of electricity load before and during the pandemic Covid-19 in Malaysia by using least square support vector machine (LSSVM). Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

	Text (24 Pages) Short-term forecast of electricity load before and during the pandemic Covid-19 in Malaysia by using least square support vector machine (LSSVM).pdf - Submitted Version Download (362kB)
	Text (Full text) Short-term forecast of electricity load before and during the pandemic Covid-19 in Malaysia by using least square support vector machine (LSSVM).pdf - Submitted Version Restricted to Repository staff only Download (2MB)

Abstract

When there is demand for electricity, it is a must to generate it. Therefore, it is a necessity for the electric power utilities that the load on their systems should be estimated in advance. This estimation of load in advance is commonly known as load forecasting, it is essential for power system planning. Load forecast is vitally important for the electric industry and not only for deregulated economy. Since the Corona Virus Disease 2019 (COVID-19) pandemic occurs, it gave a big impact towards the world’s economy mainly in Malaysia. Load forecast is essential to predict the demand after the pandemic happens. But it is found that there are no research papers had been done yet for this event and simultaneously, there has no previous work on using LSSVM with optimize Genetic Algorithm (GA) in Malaysia. The previous work for short-term load forecast in Malaysia needs to be improvise by achieving the smallest error. Thus, this project testing is held on June 2019 and June 2020 and focused on short-term forecast where it is trained on hourly data. The load data was attained from Grid System Operator (GSO) website. The goal of the result is to achieve the smallest error by using LSSVM approach by using MATLAB software. The collected data has been categorised into raw and normalized data, and has been set up into input and output forecast. The input has 24 hours of the specific day-type for three consecutive weeks starting from the 1st week of the specific month while the output will be on the 4th week. The function of GA is basically a search-based algorithm that have been used to solve optimization problems in machine learning where could solve difficult and time-consuming problems. GA optimization is to find the most optimum value of gamma and sigma of LSSVM. For the overall results, the average MAPE by using LSSVM stand-alone for 2019 and 2020 is 6.39268% and 6.63344% simultaneously. While the average MAPE for LSSVM-GA for 2019 and 2020 is 6.03614% and 4.73508% simultaneously. This shows that by using GA optimization, better accuracy is resulted.

Actions (login required)

View Item