Evaluating the Interplay of Model Complexity and Parameter Diversity in Fuzzy Logic-Based Electric Vehicle Regenerative Braking

As electric vehicles (EVs) become increasingly popular worldwide, optimizing energy regeneration and overall driving range remains critically important. Although incorporating multi-dimensional vehicle parameters into control architectures captures complex real-world dynamics, the structural trade-off between expanding input parameter dimensionality and maintaining a fixed-capacity rule base remains under-explored. This paper evaluates the effect of fuzzy logic control complexity on the regeneration process and overall driving range by comparing three MATLAB-based models under custom testing scenarios designed to isolate complexity impacts: Basic (4 inputs), Advanced (5 inputs), and Realistic Parameterized (10 inputs). Across the models, output ranges were set to 0–10 km, 0–20 km, and 0–35 km, respectively, incorporating variables such as vehicle speed, battery state of charge (SoC), braking intensity, road grade, braking duration, component temperatures, wheel slip, vehicle load, and battery state of health (SoH). To establish a baseline for evaluating parameter expansion, the number of rules was kept constant at 25 across all models. The results indicate that expanding parameter dimensionality under a fixed rule-base capacity reduces rule coverage efficiency, with the Basic, Realistic Parameterized, and Advanced Models utilizing 51.84%, 48.62%, and 38.66% of their operational capacity, respectively—demonstrating that parameter expansion requires proportional adaptations in rule-base complexity. Overall, the Realistic Parameterized Model achieved the highest performance, delivering an average range increase of 17.03 km due to the inclusion of realistic temperature and load dynamics. Consequently, this model is recommended for industrial applications and testing advanced EV energy management systems.