Mass air flow control of common-rail diesel engines using an artificial neural network

Abstract

This paper presents a non-linear control system design for managing the mass flow rate in a common-rail direct-injection diesel engine. In diesel engines, exhaust gas recirculation systems are used to reduce the nitrogen oxide emissions. Since an exhaust gas recirculation system has highly non-linear characteristics and is coupled with a variable-geometry turbocharger, it requires a non-linear control system, which can substitute for an existing conventional lookup-table-based controller, in order to secure a control performance in the transient operating regime. In this project, the objective of the control system is to track the target mass air flow by adjusting the exhaust gas recirculation valve lift. In order to accomplish this objective, a non-linear control system is proposed that adopts a neural-network-based control scheme and an indirect adaptive control approach. The neural adaptive controller determines the position of the exhaust gas recirculation valve lift for tracking target values, based on measured values. An error back-propagation algorithm for online training of neural networks is employed. The proposed control system was validated with engine experiments under transient operating conditions. It was demonstrated from experimental results that the proposed control system shows an improved target-value-tracking performance, when compared with conventional mass air flow controllers.