Development of an optimum flow rate control method for a hybrid system with PVT and heat pump using a clustering-based regression model

Addressing greenhouse gas emissions is crucial for enhancing energy efficiency in buildings. This study proposes an optimized flowrate control strategy for a hybrid Photovoltaic-Thermal (PVT) and Air-to-Water Heat Pump (AWHP) system to improve performance and reduce costs. While Neural Network (NN) models are effective for complex control, their computational burden limits real-world application. To overcome this, we employed a computationally efficient regression model and enhanced its accuracy using a clustering-based approach. Various clustering algorithms, including Gaussian Mixture Models (GMM), K-Means, and Agglomerative Clustering, were tested with regression models such as Linear, Ridge, Lasso, and Polynomial Regression. The GMM-based Polynomial Regression model achieved the highest predictive accuracy. The developed PVT and AWHP simulation models were validated using real-scale experimental data, ensuring high accuracy based on CV(RMSE) and R-squared metrics. The optimized control strategy led to a 4 % cost reduction and a 4.4 % improvement in Coefficient of Performance (COP). Correlation analysis revealed strong relationships between environmental variables and optimal flowrate, confirming the model's adaptability to different conditions. This approach provides a practical, efficient control solution for hybrid PVT-AWHP systems, enhancing energy efficiency and operational performance in buildings.