A comparative analysis of optimized gear shifting controls for minimizing fuel consumption and engine emissions using neural networks, fuzzy logic, and rule-based approaches

Reducing vehicle emissions and fuel consumption is a crucial sustainable objective for global authorities, prompted by the adverse impacts of air pollution on the economy, environment, and human health. Gear shifting control represents an appropriate approach to reducing exhaust emissions and fuel consumption while maintaining vehicle acceleration, particularly in conventional vehicles powered solely by combustion engines, which still dominate the commercial market. This study aims to compare three gear shifting control strategies for a conventional vehicle: a rule-based mapping approach, fuzzy logic control, and artificial neural networks. The strategies are designed to minimize fuel consumption and emissions based on the engine’s operating conditions. Employing an interactive weight adaptive genetic algorithm, the controllers are optimized under identical driving conditions and constraints, involving a combined driving cycle with various driving styles. The results are subsequently evaluated and compared for their robustness under different driving conditions, including real-world scenarios, thereby addressing a research gap about the comprehensive comparison of these gear shifting controllers and determining the optimal parameters for each. Ultimately, the study reveals that the optimized controllers, compared to the standard gear shifting strategy, achieved a reduction of 4.02% and 46.93% in fuel consumption and emissions for the rule-based strategy, 5.06% and 46.01% for the fuzzy logic control, and 5.52% and 33.23% for the artificial neural network. Furthermore, the artificial neural network demonstrated the most favorable results in terms of a single objective solution, while the fuzzy logic control exhibited the best trade-off between fuel savings and emission reduction.