OBD Algorithms: Model-based Development and Calibration

The OBD II and EOBD legislation have significantly increased the number of system components that have to be monitored in order to avoid emissions degradation. Consequently, the algorithm design and the related calibration effort is becoming more and more challenging. Because of decreasing OBD thresholds, the monitoring strategy accuracy, which is tightly related with the components tolerances and the calibration quality, has to be improved.

A model-based offline simulation of the monitoring strategies allows consideration of component and sensor tolerances as well as a first calibration optimization in the early development phase. AVL applied and improved a methodology that takes into account this information, which would require a big effort using testbed or vehicle measurements. In many cases a component influence analysis is possible before hardware is available for testbed measurements. Moreover, the number of calibration optimization loops can be significantly reduced, since the effect of different calibration datasets can also be proven on a model. Thanks to this methodology, which was verified in series development projects, the verification effort on a testbed and in vehicles can be considerably reduced.