Enabling Development of Power Supply System, Management and Prediction to Validate Major and Minor Changes in Co-Simulation

 Using Real World Drives and Situations in Virtual World for Power Supply System

Efficiency and stability are both major characteristics for an automotive power supply system. Both features need different operating strategies, sometimes even in the different direction, and can be improved by prediction technology. To develop prediction solutions as well as the corresponding operating strategies a deterministic test platform that enables comparison is necessary. Major requirements to the test platform are determinism, interchange of power net configuration, manner of passengers, real world driving behavior and situations and last not least parameter variation.
For this project CarMaker was extended by several individual modules. First a complete automotive power supply net, including operating strategy of E/E-components, energy storage and producer has been integrated via FMI. Part of the power supply net FMU are, vehicle system specific functions, necessary to coordinate the E/E-components. Additionally, a deterministic model of passengers was integrated into the FMI to operate customer specific vehicle functions like seat heating, etc.. Concerning the E/E components the models were developed including dependencies to vehicle status and environment parameters. Integration of management system and prediction took place via CarMaker C-Interface, as the complete system was implemented in C++.
For evaluation of the simulation the whole system was integrated into a test vehicle. The drives made with the test vehicle are converted to Road-Format and Speed-Profile to import them into CarMaker. In this way a fitting of the simulation is possible, if necessary. With a variety of drives different management or prediction functions can be compared. To check robustness and limits of the system, various parameters like, temperature, SoC at departure, prediction mode, etc. can be changed. System internal fitting can be trained as well by recalculating different drives.
At this point different use-cases can already be simulated, like changes in power net hardware, new function in E/E components, inventions in power management or prediction, training and testing learning strategy.
Further improvements of the existing concept are import of real-world traffic to replace the Speed-Profile and use the CarMaker internal driver model, connect FAS sensors to add support for power prediction or including a feedback loop to CarMaker to lead back actual power requirements from the power net to the physical CarMaker simulation (e.g. alternator).