“An Ensemble Approach based on Numerical Weather Prediction and Random Forest for Solar Energy Forecasting”
Country of Study
Kenya
Institution
Jomo Kenyatta University of Agriculture and Technology
Expected Year of Completion
2023
Thematic Area
Information and Technology
Education
Dorcas is pursuing her PhD in Computing, at Jomo Kenyatta University of Agriculture and Technology and works as an Administration Director at Code Pamoja Institute of Professional Studies. She has an MSc in IT from Masinde Muliro University and is due to finish her studies in 2023.
Research Summary
As countries scramble to meet rising energy demands and actualise SDG number seven to provide clean and affordable energy, the adoption and use of renewable energy, particularly, solar energy, has shot up. The main challenge in solar energy production is its intermittent nature. Integrating solar power farms to the national electricity grid has the potential to make the grid unstable, exposing it to likely interruptions. Dorcas’ research focuses on accurately forecasting the amount of solar power a specific earth surface can receive over a given period thus enabling an informed decision. There exists research on how to effectively forecast the amount of solar irradiance, but none has attempted to use a technique where all of the three solar forecasting techniques, namely, Numerical Weather Prediction (NWP), satellite imagery and machine learning algorithms were considered. This study aims at integrating the NWP and ML techniques and then use data obtained from a satellite-to-irradiance model to train and test the developed algorithm with the intention of developing a better performing and optimised algorithm for solar radiation forecasting.
Publications:
1. An Ensemble Model Based on Deep Semantic Segmentation Network and Graph Convolutional Network for Cloud Detection
2. Mitigating Preconception in Machine Learning Classifiers
