The tightening of the diesel pollutant emission regulations has made the performances obtainable from steady-state map controls, commonly employed in Internal Combustion Engine management, unsatisfactory. To overcome these limitations a NonLinear Quadratic Regulator (NLQR) system for the High Pressure Exhaust Gas Recirculation (HP-EGR) loop of a turbocharged diesel engine has been developed to control the intake O2 concentration and the Intake MAnifold Pressure (IMAP). This model-based control approach exploits the prediction of two dynamic Recurrent Neural Networks (RNN) to compute the command actions for the HP-EGR valve and VGT (Variable Geometry Turbocharger) rack position. Engine speed, engine load, HP-EGR and VGT valves positions together with the intake O2 concentration and IMAP feedbacks are the inputs used by the RNN to compute the predictions. In order to select the next HP-EGR and VGT control actions the effect of different command combinations, retrieved from a discretized action space, are evaluated through a quadratic objective function to be minimized. Two different transient profiles have been used to test the designed control system against the steady-state map approach. The developed control system has shown a satisfactory performance improvement over the map control. Therefore it is suitable for the subsequent assessment on the real engine.

NLQR Control of High Pressure EGR in Diesel Engine / Ventura, Loris; Malan, STEFANO ALBERTO. - ELETTRONICO. - (2020). (Intervento presentato al convegno 2020 20th International Conference on Control, Automation and Systems tenutosi a BEXCO, Busan, Corea del Sud nel Oct. 13-16, 2020).

NLQR Control of High Pressure EGR in Diesel Engine

Loris, Ventura;Stefano, Malan
2020

Abstract

The tightening of the diesel pollutant emission regulations has made the performances obtainable from steady-state map controls, commonly employed in Internal Combustion Engine management, unsatisfactory. To overcome these limitations a NonLinear Quadratic Regulator (NLQR) system for the High Pressure Exhaust Gas Recirculation (HP-EGR) loop of a turbocharged diesel engine has been developed to control the intake O2 concentration and the Intake MAnifold Pressure (IMAP). This model-based control approach exploits the prediction of two dynamic Recurrent Neural Networks (RNN) to compute the command actions for the HP-EGR valve and VGT (Variable Geometry Turbocharger) rack position. Engine speed, engine load, HP-EGR and VGT valves positions together with the intake O2 concentration and IMAP feedbacks are the inputs used by the RNN to compute the predictions. In order to select the next HP-EGR and VGT control actions the effect of different command combinations, retrieved from a discretized action space, are evaluated through a quadratic objective function to be minimized. Two different transient profiles have been used to test the designed control system against the steady-state map approach. The developed control system has shown a satisfactory performance improvement over the map control. Therefore it is suitable for the subsequent assessment on the real engine.
2020
978-89-93215-19-9
File in questo prodotto:
File Dimensione Formato  
TC6.2.pdf

non disponibili

Descrizione: Articolo principale
Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 814 kB
Formato Adobe PDF
814 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2854127