The article presents the development of a lean approach for virtual electronic control unit (ECU) calibration. In this calibration method, virtual models are used to improve the calibration quality or reduce the calibration effort. Unlike state-of-the-art approaches, no dedicated engine simulation models for hardware-in-the-loop (HiL) operation are utilized. The developed engine simulation consists of physical ECU real-time capable 0D models. Major benefit of this approach is the multiple use of the developed models for virtual calibration of customer ECUs and vehicle operation using rapid-control-prototyping-ECUs (RCP-ECUs). The engine model consists of a physical air path, an air charge model, a gas exchange and a torque model as well as a novel mathematical combustion and exhaust gas temperature model. The configuration of the engine model was done for a turbocharged four-cylinder gasoline reference engine.
For the virtual calibration, the sensor values and actuator set points of the customer ECU are exchanged with a HiL real-time computer. Besides the ECU and a wiring harness, no additional hardware components are required. The virtual calibration was performed using only a proportion of the available experimental data. The result shows a satisfying accuracy of the nominal and actual path, compared to a conventional calibration dataset created with all available experimental test bench data. Applying the generated maps for air path actuator set points, air charge, spark timing, and efficiency, the required effort in terms of test bench time can be reduced significantly.
For the operation of prototype vehicles, the models can be executed in real-time with a RCP-ECU. Due to the physical and mathematical modeling approach, the required calibration effort for engine and vehicle operation can be reduced significantly as well.
