This study focuses on the operational optimization of a 1000 kW high-speed diesel generator (Mitsubishi S16R) located at PLTD Muara Wahau, a remote power station in East Kalimantan, Indonesia. The generator operates with Biodiesel B35, a national renewable fuel standard containing 35% biodiesel and 65% petroleum diesel. While B35 offers environmental benefits, its lower heating value and distinct combustion characteristics result in an 18% reduction in generator output and increased specific fuel consumption (SFC), posing challenges to performance and fuel efficiency in isolated areas. To address these issues, a hybrid modeling and optimization framework is proposed, combining response surface methodology (RSM), artificial neural networks (ANN), and multi-objective genetic algorithm (MOGA). A multi-criteria decision-making approach using TOPSIS is applied to evaluate alternative operating scenarios. The study investigates two modes: base load (cos φ = 0.96, load = 698 kW) and load share (cos φ = 0.97, load = 829 kW). The RSM model in base load mode achieves a fuel consumption of 0.21 l/kWh and efficiency of 42.78%, while the ANN-MOGA model in load share mode records 0.24 l/kWh and 39.42% efficiency. The results demonstrate that parameter optimization can significantly improve the performance of B35-fueled generators. The integrated methodology provides a practical solution for enhancing operational efficiency and sustainability in remote, off-grid power systems, with potential for broader application in similar decentralized energy contexts