Offline Simulation and Automated Optimization of a Function Model for Embedded Engine Management Systems

Virtual sensor models are frequently used for embedded engine management systems (EMS) in an ECU to represent specific system behavior of the engine and replace actual sensor values with modeled ones. The virtual sensor models are physically motivated and contain complex structures with numerous calibration parameters in order to describe engine system behavior accurately. The air charge function model, which is one of the complex virtual sensor models for EMS, has many calibration parameters to be optimally calibrated in the model in order to ensure good model quality for replacing actual sensors. Due to model complexity and many parameters in the model, a manual calibration may not be practical or even impossible. Instead, an offline optimization tool can help speed up this time-consuming calibration process. This paper covers automated calibration and results by a generic optimization tool, ETAS ASCMO-MOCA (MOdel CAlibration). Different model formats of the air charge function referring to the EMS software specification were created and adopted into MOCA for offline simulation and automated optimization. The difference between the measured actual value and the modeled output was minimized by the MOCA optimization algorithm, which was implemented to find optimal calibration parameters in the model. Optimization results were significantly satisfactory enough to replace the conventional manual calibration and supplier’s internal optimization tools. This automated calibration performed offline on a personal computer (PC) by a generic optimization tool helps engineers secure reliability for calibration quality by improving the accuracy of calibration parameters and increase calibration efficiency by reducing development time and costs compared to the conventional manual calibration. This methodology has been considered an efficient way for many calibration tasks in an automotive manufacturer. Therefore, it is widely used for the other calibration tasks of engine or vehicle development processes.