Model-based simultaneous diagnosis of ammonia injection failure and catalyst ageing in [formula omitted] engine after-treatment systems

Abstract

After-treatment systems (ATS) in vehicle powertrains are developed to achieve maximum pollutant abatement efficiency under design operating conditions. However, anomalies in the ATS may appear during real operation, such as reducing agent supply failure and catalyst ageing. As in the case of an ATS composed of a selective catalytic reduction (SCR) system plus an ammonia slip catalyst (ASC), injection failure leads to insufficient ammonia levels to reduce nitrogen oxide (NOx), and ageing decreases the ammonia storage capacity and NOx conversion efficiency, these drawbacks drive both pollutants to levels above those expected and allowed by governmental regulations. To address this problem, a novel methodology was proposed that simultaneously estimates the operational state of the ATS in terms of ammonia injection failure and ASC catalyst ageing. The approach through control-oriented models based on an artificial neural network (ANN) and sensor signal analysis (SSA), as well as an extended Kalman filter (EKF) observer, compares the current emission levels estimated by the observer in a two-dimensional probabilistic plan and a variable time-window, with those expected if the ATS were operating under normal conditions. The proposed methodology was evaluated in real driving cycles with the ATS subjected to several undesired operating conditions. As a result, the model was able to accurately estimate the levels of ammonia injection failure and catalyst ageing state. Therefore, enabling the control system to modify the ammonia injection strategy to achieve desired levels of NOx and NH3 emissions.