Metaheuristic energy efficiency optimization of solar-powered absorption cooling systems under operating climactic conditions integrated with explainable AI

Abstract

The increasing energy demands, and environmental concerns necessitate the development of efficient and sustainable cooling technologies, particularly in arid regions such as Riyadh, Saudi Arabia. This study aims to optimize the energy efficiency, specifically the Coefficient of Performance (COP), of solar-powered absorption cooling systems using advanced Explainable Artificial Intelligence (XAI) and Machine Learning (ML) algorithms. For this purpose, predictive models employing Artificial Neural Networks (ANN) were developed. Subsequently, nature-inspired optimization techniques, including Chicken Swarm Optimization (CSO), Moth Flame Optimization Algorithm (MFOA), and Whale Optimization Algorithm (WOA) were explored to improve the prediction skills. The study utilized three distinct feature combinations to capture various operational and environmental parameters, and the models were evaluated using several statistical metrics. The results demonstrated significant improvements in predictive accuracy with the optimized models. Combination 1 (C1) achieved near-perfect goodness-of-fit values with marginal gains from optimization, while Combination 2 (C2) also showed high values with slight reductions in the optimized models. Similarly, Combination 3 (C3) initially had lower performance. Still, optimization techniques notably improved the goodness-of-fit values, with ANN-CSO showing an 11.46 % increase, ANN-MFOA a 10.36 % increase, and ANN-WOA a 10.54 % increase. Further feature importance using SHAP analysis indicated that the most influential parameters were heat absorption (Qa), heat exchange rate (Qe), cooling capacity (Qc), and heat dissipation (Qd), with solar irradiance (Ir) having a minor impact. These findings indicate that the optimization techniques are particularly effective for feature sets that initially underperform. The study recommends integrating these advanced optimization methods into the design and operation of solar-powered absorption cooling systems to enhance energy efficiency and reduce greenhouse gas emissions. Future research should explore the scalability and adaptability of these optimized models across different climatic regions, incorporate real-time data, and investigate the economic and long-term stability aspects under various operational scenarios to ensure comprehensive practical implementation and sustainability benefits.