| Title |
An Intelligent Photovoltaic Power Forecasting Model based on Bagged-Boosted STACK Support Vector Regression with Kernel Linear |
| Authors |
제라르도 온도 미하(Gerardo Ondo Micha) ; 김철환(Chul-Hwan Kim) |
| DOI |
https://doi.org/10.5370/KIEE.2021.70.11.1633 |
| Keywords |
Distribution energy sources; ensemble regression; intermittency of renewable energy. power forecasting; Support Vector Machine. |
| Abstract |
Due to the intermittent nature of Renewable Energy Sources (RES) and complexed pattern of time series data, increasing the forecasting accuracy by conventional methods and single-learner based Machine Learning (ML) algorithms is becoming increasingly challenging. Therefore, appropriate combination of ML schemes and data processing techniques is very important to increase the prediction accuracy of RES. This paper presents an intelligent Photovoltaic (PV) forecasting model using an ensemble learner approach based on a combination of bagging, boosting, and stacking algorithms and a Support Vector Regression with Kernel Linear (SVRL) meta-learner. The proposed approach pre-processes time series data using Correlation Matrix Analysis (Corr) and Principal Component Analysis (PCA). In level 0 of the STACK model, seven hybrid models are used as base learners and their individual predictions are used as input for the SVRL meta-learner in level 1. To evaluate the performance of our model, data collected from a 350KW 3rd PV power plant in Gyeongnam, South Korea, were used for simulations, and results were compared with bagging, boosting and bagging-boosting algorithms used separately, and results show higher forecasting accuracy of the proposed algorithm. |