Bio-Inspired Explainable Evolutionary Rule Mining for Thermodynamic Performance Assessment of a Solar Greenhouse Dryer

Abstract

This study investigates the thermodynamic and drying performance of a greenhouse dryer integrated with a parabolic trough solar collector (PTSC) and develops interpretable operating rules using a bio-inspired explainable artificial intelligence framework. Outdoor apple-drying experiments were conducted, and system performance was evaluated in terms of energy, drying, and exergy efficiencies. The experimental results indicated that energy efficiency ranged from 17.7% to 29.2%, drying efficiency from 1.0% to 9.7%, and exergy efficiency from 5.6% to 8.4%. Measured variables, including temperature, relative humidity, product weight, and solar radiation, were used to classify the efficiencies into low, medium, and high categories using the Chaotic Rule-based Strength Pareto Evolutionary Algorithm 2 (CRb-SPEA2). As a bio-inspired evolutionary computing approach, CRb-SPEA2 employs population-based search, selection, Pareto dominance, and multi-objective optimization mechanisms inspired by natural evolutionary processes. In contrast to conventional black-box machine learning models, the proposed method extracts explicit decision rules that define physically meaningful operating ranges. The maximum recall values were 0.952, 1.000, and 0.971 for the high-energy-, drying-, and exergy-efficiency classes, respectively. The extracted rules identified solar radiation, temperature, relative humidity, and product weight as dominant factors affecting dryer performance.