Predictive Energy Management for Hybrid Vehicles in a Virtual Cosimulation Environment
GAC Uses Route-Based Simulation to Optimize Powertrain Energy Use
“Compared with other commercial software on the market, it [RoadRunner] has relatively rich communication interfaces, which provide convenience for engineers developing model-based software when interacting between scenarios and software models.”
Key Outcomes
- MATLAB, RoadRunner, and SUMO cosimulation covers the key inputs for predictive energy management software, including gradients, speed limits, traffic flow, etc.
- Driver-in-the-loop provides effective criteria for driving style identification and speed prediction.
- Scenario generation expands algorithm test sets while reducing costs and shortening vehicle development cycles.
Modern hybrid powertrains use predictive energy management software to optimize energy use by adjusting component operating points based on data from intelligent connected vehicle (ICV) sources. Prediction accuracy depends on various factors, including vehicle speed, road conditions, and traffic flow. The GAC R&D Center, a part of China’s leading carmaker, GAC Group (Guangzhou Automobile Group), used MATLAB® and Simulink® to develop software that supports global resource planning along a predefined driving route.
The team used Deep Learning Toolbox™ to recognize driving styles and predict speed accurately, enabling correct state-of-charge (SOC) planning. With Powertrain Toolbox™, they designed an equivalent consumption minimization strategy to reduce total energy consumption by optimally splitting power between the engine and battery along the route.
For static road scene construction, the team imported map data from OpenStreetMap® to RoadRunner. They also used APIs to fetch real-time traffic conditions, configuring realistic traffic flows in SUMO. Vehicle Dynamics Blockset™ allowed the team to model the vehicle’s physical behavior. The realistic response produced by the simulated vehicle eliminated the need for physical testing.
Additionally, Automated Driving Toolbox™ helped integrate navigation modules and control algorithms to create a closed-loop simulation environment. By analyzing driving styles across multiple scenarios, the software could adapt to match different drivers’ behavior. The driver-in-the-loop capability further enabled analysis of how individual driving styles influence energy management.
