Photovoltaic (PV) System Output Power Forecasting Using Support Vector Machines (SVM) Technique

Wong, Wai Leong (2019) Photovoltaic (PV) System Output Power Forecasting Using Support Vector Machines (SVM) Technique. Project Report. UTeM, Melaka. (Submitted)

Text Photovoltaic (PV) System Output Power Forecasting Using Support Vector Machines (SVM) Technique - Wong Wai Leong - 24 Pages.pdf - Submitted Version Download (350kB)

Abstract

The use of Solar Photovoltaic (PV) system for power generation have expanded rapidly for the past few years. However, the growth of the solar PV system is also causing problems for the management of the power distribution as the operator have to always maintain the stability of the power grid between power generation and power distribution. Therefore, solar power output forecasting have become an important task to focus on to overcome the problems of using solar PV system for power generation. A solar power output prediction model is developed in this project to predict the day ahead hourly power output by using the Support Vector Machines (SVM) method. The prediction model is developed based on the data and module technology of the Solar Lab of Faculty of Electrical Engineering (FKE) in University Teknikal Malaysia Melaka (UTeM). The prediction model is designed by training the prediction model using local data with regression learner application in MATLAB software version R2017b. The results indicate that using SVM model to forecast solar power output is valid and the accuracy of the prediction is satisfied. The predictor variables used to trained the predictive model is analyzed. Irradiance and Module Temperature are the most dominant variables that will give a large impact to the accuracy of the trained predictive model to perform day ahead solar PV power output forecating.

Actions (login required)

View Item