Experimental-based dynamic calibration of the diesel air-path

Abstract

An experimental study is presented of a novel dynamic calibration methodology applied to the air-path of a Jaguar Land Rover turbocharged diesel engine. The calibration is obtained in a one-shot process solely from data obtained from dynamic dynamometer testing, without any in-vehicle tuning. Although limited here to control of only boost pressure and exhaust gas recirculation rate with constraints on NOx and particle emissions, the methodology has the potential for a complete engine calibration including emission constrained optimisation of fuel consumption. The approach combines state space neural network (SSNN) modelling and optimisation to determine a feedforward Hammerstein-Wiener control map from the synthesised optimal control behaviour. The controller performance was verified by vehicle testing at the Jaguar Land Rover Test Track, Gaydon, UK. The outcome demonstrates a method for systematically obtaining dynamic control calibrations for good driveability and reduced emissions from limited dynamometer testing time without the need for manual tuning.