Volume 48

Ensemble-Tree Model Based on Bayesian Optimization for Solar Energy Generation Prediction in Smart Homes Oluwatoyosi Bamisile, Dongsheng Cai, Chukwuebuka J. Ejiyi, Chiagoziem C. Ukwuoma, Qi Huang, Olusola Bamisile



Smart homes use devices that automate tasks like security, lighting, and temperature control. These homes let people control appliances remotely through the Internet of Things (IoT), adjusting to their schedules for better energy use. But as energy use rises, it causes more pollution, and climate problems, and puts more strain on energy sources. Therefore, it’s important to track energy use closely as the world moves into the use of renewable energy to avoid power outages, save money, and protect the environment especially because of the intermittent nature of renewable energy. This paper proposed an Ensemble-Tree Model Based on Bayesian Optimization or Solar Energy Generation Prediction in Smart Homes. First, three tree-like machine learning models training hyperparameters were optimized using the Bayesian optimization technique. Secondly, their output was concatenated based on Mean Aggregation Methods in mathematics. Lastly, the prediction was done based on k-fold cross-validation. The ‘smart-home-dataset with weather-information dataset is used while using the R-squared (R2), Mean Squared Error (MSE), and Mean Absolute Error (MAE) to observe how accurate the predictions are. Results show that the proposed model outperforms other machine learning models with R2 value of 0.988 as compared in this paper.

Keywords Solar Generation, Smart Homes, Internet of Things, Machine Learning, Prediction

Copyright ©
Energy Proceedings