The lean NOx trap (LNT) has been proposed as a NOx after-treatment device for diesel engines. In order to successfully apply a LNT to diesel engines, it has thus far been necessary for the LNT to have intermittent rich operation (so-called rich spike, RS), and the stored NOx is purged and reduced by RS to maintain NOx conversion performance. However, since RS tends to increase fuel consumption, optimized RS control is necessary to keep the fuel consumption increase to a minimum.
In order to do that, a suitable LNT control system is necessary, like one that has the functionality to accurately estimate the amount of stored NOx, and indicate the proper RS timing according to this value. In this research a model-based control method was adopted to accurately estimate the amount of stored NOx present in the LNT. The LNT parameters were characterized by polynomial fits and modeled. Moreover, using a NOx sensor mounted downstream of the LNT, an adaptive estimation method was adopted. Even if the LNT characteristics change due to aging, for example, it is still possible to adjust parameters in the LNT model and continue to accurately estimate the amount of stored NOx.
A LNT control system using these methods was applied to diesel engines, followed by transient mode tests under various engine operating conditions. Results showed that with this control methodology, the fuel consumption was improved by 5% at the same NOx conversion rate, compared to the constant RS interval control method. In addition, even when LNT performance decreased, the NOx conversion rate was found to remain fairly constant due to adaptive estimation effects.