Non-dominated sorting genetic algorithms are recognized for their robustness and flexibility in optimizing renewable energy systems, surpassing traditional methods by handling multiple objectives and generating diverse Pareto-optimal solutions. However, inefficiencies due to random initial populations and mutations can impact processing times and error rates. This study introduces the controlled non-dominated sorting genetic algorithm, which enhances optimization with controlled population initialization and mutation mechanisms. Compared to the conventional non-dominated sorting genetic algorithms, the controlled version shows superior performance, achieving a 2.4% error reduction, a 117% lower task violation rate, and a 157% faster processing time at high energy demands. A case study in Ifrane, Morocco—a tourism village with significant seasonal energy demand—illustrates the application of the algorithm. Results show optimal scenarios for standalone and grid-connected systems, considering potential grid export opportunities. Standalone configurations generate 271 MWh surplus energy annually, with 15 MWh unmet demand, requiring 125 kW power converters. Real scenarios synchronize lower rated power with grid imports, reducing net present costs by 18% and levelized costs by 24%. Hypothetical scenarios demonstrate potential revenue generation with negative net present and levelized costs if export prices match import costs. Grid-connected and thermal energy storage systems are more cost-effective despite higher emissions.