Optimizing geothermal systems with genetic algorithm using deep artificial neural networks as a fitness function

This study presents a comprehensive analysis of geothermal power plants and provides a systematic review of the optimization methodologies employed in their operation. Particular emphasis is placed on heuristic optimization techniques, detailing their advantages and the outcomes derived from the developed models. The inherent challenges associated with geothermal system optimization specifically, the constraints arising from system complexity and the necessity of integrating numerous parameters are critically examined. Furthermore, heuristic approaches, including artificial neural networks and genetic algorithms, which have gained significant traction in recent research, are systematically outlined. Within the scope of this research, an advanced optimization methodology was developed utilizing operational data from an Organic Rankine Cycle (ORC) geothermal power plant in İzmir. By incorporating artificial neural networks as a fitness function within a genetic algorithm, the proposed computational framework successfully enhanced the gross power output from 4.943 kW to 6.624 kW, thereby determining the optimal operating conditions. Consequently, a 39.41% increase in net power output was achieved. Additionally, multiple operational scenarios were analyzed, demonstrating the efficacy and applicability of the proposed computational approach in optimizing geothermal energy systems