Energy-Efficient Driving Model for Intelligent Connected Vehicles Using Multi-Objective Optimization

This study endeavors to devise an energy-saving driving model for intelligent connected vehicles by amalgamating Deep Convolutional Neural Network (DCNN), refining Non-dominated Sorting Genetic Algorithm II (NSGA-II), and employing Model Predictive Control (MPC) algorithms, with the aim of bolstering driving sensitivity, the adaptability of intelligent solutions, and mitigating vehicle energy consumption.The study impetus stems from the pressing necessity to enhance vehicle energy efficiency, safety, and comfort.Methodologically, DCNN is employed to meticulously extract driving scene features, the NSGA-II algorithm, enhanced with adaptive crossover probability and elite preservation mechanism, is utilized to optimize multi-objective driving performance, while the MPC algorithm is tasked with formulating real-time control strategies.The study is centered on enhancing the NSGA-II algorithm and its application in energy-saving driving, questing for optimal solutions in multidimensional space across common driving scenarios through a multi-objective optimization framework.The findings indicate that: (1) In urban congestion environments, optimized vehicles demonstrate heightened energy efficiency.Specifically, the average vehicle speed increases by 25.42%, underscoring vehicles' enhanced navigability in congested environments; the average traction power consumption decreases by 11.37%, markedly curbing energy wastage.During high-speed cruising, both traction power and braking power are reduced, with the number of emergency braking instances dropping from 12 to 9, bolstering both safety and energy efficiency.(2) During parking and start-stop dynamic stages, model optimization leads to a reduction in idle total time from 60 minutes to 45 minutes, and idle fuel consumption rate decreases from 0.75 L/h to 0.62 L/h, augmenting both economy and environmental compliance.(3) In mountainous winding road environments, the optimized model exhibits enhanced climbing efficiency, decreased instances of emergency braking, and improved maneuverability, enabling drivers to experience safe and comfortable driving in complex terrains.From a practical perspective, this study proffers novel solutions for energy-saving driving of intelligent connected vehicles, contributing to reductions in energy consumption, emissions, and enhanced driving experiences.Theoretically, this study delves into the fusion of deep learning, optimization algorithms, and control theory in the realm of energy-saving driving, furnishing valuable insights for further research in related domains.Additionally, the entire study process also reflects a positive contribution to environmental protection, energy conservation, and intelligent travel, aligning with the contemporary imperatives of sustainable development era.