Search Results

This paper describes an in-cylinder lean-rich combustion (no-post-injection for rich) switching control approach for modern diesel engines equipped with exhaust-treatment systems. No-post-injection rich combustion is desirable for regeneration of engine exhaust-treatment systems thanks to its less fuel penalty compared with regeneration approaches using post-injections and / or in-exhaust injections. However, for vehicle applications, it is desirable to have driver-transparent exhaust-treatment system regenerations, which challenge the in-cylinder rich-lean combustion transitions. In this paper, a nonlinear in-cylinder condition control system combined with in-cylinder condition guided fueling control functions were developed to achieve smooth in-cylinder lean-rich switching control at both steady-state and transient operation. The performance of the control system is evaluated on a modern light-duty diesel engine (G9T600).

A new diesel engine system adopting alternative combustion with rich and near rich combustion, and an airflow-dominant control system for precise combustion control was used with a 4-way catalyst system with LNT (lean NOx trap) to achieve Tier II Bin 5 on a 2.2L TDI diesel engine. The study included catalyst temperature control, NOx regeneration, desulfation, and PM oxidation with and without post injection. Using a mass-produced lean burn gasoline LNT with 60,000 mile equivalent aging, compliance to Tier II Bin 5 emissions was confirmed for the US06 and FTP75 test cycles with low NVH, minor fuel penalty and smooth transient operation.

An airflow-dominant control system was developed to provide precise engine and exhaust treatment control with low air fuel ratio alternative combustion. The main elements of the control logic include a real-time state observer for in-cylinder oxygen mass estimation, a simplified packaging scheme for all air-handling and fueling parameters, a finite state machine for control mode switching, combustion control models to maintain robust alternative combustion during transients, and smooth rich/lean switching during lean NOx trap (LNT) regeneration without post injection. The control logic was evaluated on a passenger car equipped with a 4-way catalyst system with LNT and was instrumental in achieving US Tier II Bin 5 emission targets with good drivability and low NVH.

This paper describes a hybrid robust nonlinear control approach for modern diesel engines running low temperature combustion and conventional diesel combustion modes. Using alternative combustion modes has become a promising approach to reduce engine emissions. However, due to very different in-cylinder conditions and fueling parameters for different combustion modes, control of engines operating multiple combustion modes is very challenging. It becomes difficult for conventional calibration / mapping based approaches to produce satisfactory results in terms of engine torque responses and emissions. Advanced control techniques are then demanded to accomplish the tasks. An innovative hybrid control system is designed to track different key engine operating variables at different combustion modes as well as avoid singularity which is inherent for turbocharged diesel engines running multiple combustion modes.

Tremendous amount of useful information can be extracted from the cylinder pressure signal for engine combustion control. However, the physical cylinder pressure sensors are undesirably expensive and their health need to be monitored for fault diagnostic purpose as well. This paper presents the results of the development of a virtual cylinder pressure sensor (VCPS) with individual variable-oriented independent estimators. Two neural network-based independent cylinder pressure related variable estimators were developed and verified at steady state. The results show that these models can predict the variables correctly compared with the extracted variables from the measured physical cylinder pressure sensor signal. Good generalization capabilities of the developed models are observed in the sense that the models work well not only for the training data set but also for the new inputs that they have never been exposed to before.