NO$_x$ Sensor Reading Correction in Diesel Engine Selective Catalytic Reduction System Applications

In this paper, we investigate NO
_x
sensor reading correction in diesel engine selective catalytic reduction (SCR) system applications. It is well known that the urea-based SCR system is promising to remove engine-out NO
_x
emissions and achieve strict legislative emission regulations. However, if the urea is overdosed, the system also leads to ammonia slip in the tailpipe. In order to control the NO
_x
emissions and constrain the ammonia slip, a feedback loop with NO
_x
and ammonia sensors may be needed. An interesting phenomenon in the closed loop is that a well calibrated NO
_x
sensor does not work well in the urea-based SCR systems. For this typical application, there may be several sources for the NO
_x
sensor reading mismatching issues. Due to the existence of the gaseous ammonia, the NO
_x
sensor reading is approximately a combination of the actual NO
_x
concentration and the ammonia concentration. Another phenomenon is that the influence of the ammonia concentration to the NO
_x
sensor reading is significantly different with respect to different temperatures. Based on such observations and the importance of the NO
_x
concentration reading, we devote to develop an algorithm to correct the NO
_x
sensor reading. With considerable experimental data which cover most of the typical SCR operating temperature range, we obtain the relationship between the cross-sensitivity factor and the temperature by employing the adaptive-network-based fuzzy inference system (ANFIS). The developed fuzzy inference system is then applied to experimental validation tests. Compared with the existing extended Kalman filter (EKF), the developed fuzzy inference system has prominent advantages on the reading correction performance.