AI-Driven Multi-Objective Optimization of Floating Photovoltaic-Thermal Systems with MATLAB-HOMER Pro Validation

Abstract

The shift towards renewable energy has increased interest in Floating Photovoltaic Thermal (FPVT) systems for their capacity to produce electricity and heat with a low land footprint. However, improvement of the FPV T scheme is difficult due to the multi-dimensional relationship between the electrical, thermal, and economic aspects. This research introduces an Artificial Intelligence (AI)-based multi-objective optimization and techno-economic analysis approach for FPV T systems based on ANN, GA, PSO, and RL System models. ANN surrogate predictors and optimization algorithms were implemented in MATLAB. The outcome of simulation results at STC yielded an active power and thermal output of 64.8 kW and 260 kW, respectively, resulting in a useful energy output of 324.8 kW. The conventional PV baseline system, however, produced 56.2 kW of output, at 15.3% gain in performance, higher than 480% gain in total energy utilization with the addition of the thermal recovery. The ANN predictor was highly accurate with an error of less than 0.01%, and the GA and PSO algorithms produced consistent optimal results. The reinforcement learning algorithm effectively controlled the flow rate of the coolant to achieve the desired outlet temperatures. The findings showed that a hybrid AI approach enhances electrical and thermal efficiency, speeds up the optimization process with surrogate modeling, and recommends low LCOE and attractive NPC configurations. The techno-economic modeling with HOMER Pro also validates the AI-optimized solutions. This technology advances renewable energy research through the use of multi-objective optimization, surrogate modeling, and techno-economic validation.