Abstract
Abstract
The trade-off between torque and system efficiency in electric motorcycles poses a challenge because improvements in dynamic performance do not always align with energy efficiency. This study optimizes electric motorcycle controller settings by integrating experiments, surrogate modeling, multi-objective optimization, and compromise decision-making. The investigated variables were battery current (BCR), nominal battery-pack voltage (BVG), throttle curve acceleration (TCA), and power limit (PLM). Torque and system efficiency were measured experimentally, while their Signal-to-Noise Ratios (SNR T and SNR η) were used as the primary modeling and optimization objectives. A Box–Behnken Design (BBD) with 27 conditions and three replicates was used, and the resulting SNR responses were modeled using Response Surface Methodology (RSM). The RSM surrogate models showed strong overall fit, with R² values of 99.34% for SNR T and 95.99% for SNR η. Multi-Objective Particle Swarm Optimization generated the Pareto front, and fuzzy decision-making selected BCR 80 A, BVG 66 V, TCA 1 V, and PLM 67% as the best compromise. Experimental confirmation yielded 60.00 N·m torque, 0.92 system efficiency, 35.56 dB SNR T, and − 0.72 dB SNR η. Confirmation at the selected compromise and torque-maximized Pareto configurations supported predictive agreement within the investigated domain. Internal Grey Relational Analysis benchmarking ranked the proposed MOPSO–fuzzy approach first, with a Grey Relational Grade of 0.70. The integrated framework provides a practical strategy for balancing dynamic performance and energy efficiency through controller tuning.