Forecasting of photovoltaic power using regularized ensemble Extreme Learning Machine

Teo, TT; Logenthiran, T; Woo, WL; Abidi, K

doi:10.1109/TENCON.2016.7848040

Forecasting of photovoltaic power using regularized ensemble Extreme Learning Machine

Lookup NU author(s): Tiong Teck Teo, Dr Thillainathan Logenthiran, Dr Wai Lok Woo, Dr Khalid Abidi

Downloads

Full text for this publication is not currently held within this repository. Alternative links are provided below where available.

Abstract

The increasing penetration of renewable energy sources with intermittent nature generation challenges the grid operator to accurately plan and schedule their generators. In this context accurate forecasting model are vital to ensure smooth day-to-day operation with high renewable energy sources. Artificial Neural Network (ANN) have shown promising ability for accurate forecast. The ANN proposed in this paper are trained using historical dataset and training algorithm, Extreme Learning Machine (ELM). ELM requires randomly initialized parameters which affect the forecasting model. This paper propose a method to reduce the randomness of ELM by adding a regularizing term and combining multiple ELM. The ANN is implemented using MATLAB and trained using real-life data. The result shows that the randomness are greatly reduce and has a higher forecasting accuracy than a single ELM.