17. Internationales Stuttgarter Symposium

Uncertainties in Model-Based Diesel Particulate Filter Diagnostics Using a Soot Sensor

“…An improvement in sensor accuracy and a more accurate soot model will lead to a significant increase in the OBD accuracy of diagnosis. This agrees with the conclusion of a previous publication for a resistive soot sensor, which suggests that the highest effort should be put on the soot model calibration [36]. The accuracy of the sensor and DPF sub-models have a lower impact on the result compared to the previous factors (Figure 16) but can also be crucial for future tighter limits.…”

“…Nevertheless, the transient correction demands a lot of calibration effort for every application. Another possible future alternative for accuracy improvement is the calibration of the model based on several data from real driving operation of the vehicle [36]. Also, the utilization of additional input parameters could enhance the accuracy without the need for exhaustive calibration.…”

“…Although the Axisuite model is hard to be implemented in the current ECUs, it is a valuable tool for offline calculations to demonstrate the feasible accuracy of a DPF model. Also, a compromise between these two solutions could be models which are based on the exhaust flow or temperature for the calculation of the second-by-second efficiency [36] but are not included in the current analysis.…”

“…On the other hand, an indirect comparison at every time step can increase the frequency of detection. In this case, the OBD index is calculated at every time step (usually dt = 1 s) as in [36]:OBDIndex(t)=∫0t1τ(t)dt, where,t = duration from the start of accumulation (input from the sensor, 1st branch in Figure 6);τ(t) = estimated response time (input from the models, 2nd branch in Figure 6). …”

“…Several studies have presented the development of prototypes based on resistive technology [30,31,32,33,34,35]. Nowadays, despite the high maturity level, the low cost and the broad adoption of resistive sensors in passenger car applications, the DPF diagnosis based on this technology and the necessary sub-models, is still a complex solution with a lot of challenges [36]. This inevitably leads to limited accuracy and capability of identifying different DPF failure levels which might not be a significant issue for the already used OBD applications but makes it prohibitive to use the sensor and the model for future more stringent regulations or other approaches such as on-board monitoring (OBM).…”