Predictive Cruise Control with Velocity Profile Online Optimization for Energy Efficiency

Factors such as road gradient, driving style, and traffic conditions all significantly impact the energy consumption performance of heavy trucks. Predictive Cruise Control (PCC) algorithm, which based on Model Predictive Control (MPC), can utilize the upcoming road gradient to achieve energy-efficient driving by controlling the truck's longitudinal speed. However, as a type of cruise control, PCC regulates longitudinal speed within a certain range, which limits its ability to fully exploit the energy-saving potential brought by speed variations. Dynamic Programming (DP) can compute the globally optimal vehicle speed for energy-efficient driving using pre-obtained map information. However, it requires significant computational resources and cannot be used in real-time online scenarios. This paper utilizes the powerful fitting capability of neural networks to enable online use of the DP algorithm. It utilizes the suggested vehicle speed output by the network to guide the PCC algorithm in speed tracking, thereby significantly enhancing energy efficiency. The proposed method has been compared with the Adaptive Cruise Control (ACC) algorithm and the original PCC algorithm. Simulation results show that this method achieves superior energy efficiency: compared to the ACC algorithm, it improves energy savings by 14% to 18%, and compared to the original PCC, it improves energy savings by 6 % to 8 %.