Large-scale solar photo-voltaic (PV) power plants at transmission-level interconnection form a challenge of operation and scheduling dispatch due to the fluctuating nature of the power output. Forecasting the power output is of an increasing importance as the PV generation capacity increases in the electric grid. The main objective of this research is to propose a novel online application of a fast machine learning technique based on the Truncated-Regularized Kernel Ridge Regression (TR-KRR) algorithm to forecast the solar irradiance which is directly correlated with PV power output for very-short, short, and medium-term horizons. The proposed model takes a live stream of historical weather parameters as an input to make accurate predictions of future values to tackle the limitations of operating a large-scale PV power plant by reducing the uncertainty of solar power generation. The performance of the proposed model is compared with commonly used forecasting models like the Persistence model, the Modified 24-h Persistence model, and the Least Square model, for cloudy and clear days application.
Date of Award | May 2020 |
---|
Original language | American English |
---|
- Solar photo-voltaic
- solar irradiance
- machine learning
- kernel ridge regression.
Online Machine Learning Application for PV Power Plant at Transmission Level Interconnection
Almarzooqi, A. M. H. (Author). May 2020
Student thesis: Master's Thesis