NOx Sensor Ammonia Cross Sensitivity Factor Estimation in Diesel Engine Selective Catalytic Reduction Systems

Abstract

Since the NOx sensor is crosssensitive to gaseous ammonia, the reading of NOx sensor may not be accurate when the gaseous ammonia is involved and it is a combination of the actual NOx concentration plus the ammonia concentration multiplied by a factor. The factor is named as NOx sensor crosssensitivity factor. As the reading is inaccurate, this crosssensitivity phenomenon restricts the application of NOx sensor in the selective catalytic reduction (SCR) system. A practical and economic approach is to design an observer to estimate the crosssensitivity factor and correct the NOx sensor reading. In this work, the NOx sensor ammoniacrosssensitivity factor observer design problem for diesel engine SCR systems is investigated. To achieve the objective, first, a threestate nonlinear model for the SCR system is adopted. To establish the model of the ammonia crosssensitivity factor, it is assumed that the qthorder derivative of the factor is zero. Then, based on the nonlinear model, a proportionalmultipleintegral (PMI) observer of the factor is proposed and a nonlinear estimation error system is obtained. With the linearparametervarying (LPV) technique, the stability, and the Hâˆ‍ performance of the nonlinear estimation error system are investigated. Based on the analysis results, the design methods of the stable and the Hâˆ‍ observer gains are developed. At the end, simulations and comparisons via an experimentally validated full vehicle simulator are carried out to illustrate the efficacy and the advantages of the proposed approach over the existing methods. Since the effect of disturbance to the estimation is considered and constrained in the Hâˆ‍ observer, the designed Hâˆ‍ observer has a better estimation performance when the system is subject to disturbance.