LNT NO _x Storage Modeling and Estimation via NARX

In recent years, due to more and more stringent government regulations on diesel emissions, diesel aftertreatment systems have attracted great deal of attention from both academia and diesel engine industries. Many different devices and approaches, such as Urea SCR, LNT, engine control related EGR and in-cylinder post injection, have been developed and applied to reduce nitrogen oxides (NO_x) emissions. Among those solutions, Lean NO_x Trap (LNT)-based emission reduction control system is one of the common approaches.

The NO_x storage capacity of an LNT depends on many different factors and operating conditions. Accurate and real-time estimation of NO_x storage is quite important for efficient system controls, particularly for enhancing system lifespan and reducing overall fuel consumption. A more precise modeling of NO_x storage has more significant impact for overall system performance. In this paper, we will discuss a few linear system models, and extend the results into a new approach for LNT NO_x storage estimation by using system identification of a nonlinear autoregressive with exogenous input (NARX) model. Adaptation and on-line training features are also proposed, and experimental data have been used for validation and verification of the methodology.