인공신경회로망을 이용한 커먼레일 디젤엔진의 에어시스템 제어에 관한 연구

Abstract

In diesel engines, a variable geometry turbocharger (VGT) and an exhaust gas recirculation (EGR) system are used to increase specific power and to reduce NOx emission, respectively. Since the dynamics of the VGT and EGR are highly nonlinear and coupled each other, it is necessary to substitute an existing conventional lookup table-based linear controller with nonlinear MIMO controller for better control performance. This paper presents a coordinated control of a variable geometry turbocharger and an exhaust gas recirculation of the common rail diesel engines. The objective of the control system is to tracking both a target mass air flow and a target intake manifold pressure by adjusting the EGR valve lift and the VGT vane position. In order to accomplish the objective, we designed a nonlinear MIMO controller by using a neural network based control scheme which adopts an indirect adaptive control approach. The neural adaptive controller calculates actuator positions for tracking the target values, based on the measured values from the sensors located in engines, such as engine speed, injection quantity, intake manifold pressure and mass air flow. In this study, we used the backpropagation algorithm for online training of the neural network. The proposed air control algorithm was validated with engine experiments in transient operation. Validation results demonstrated that the proposed control system effectively tracks the target value than the conventional control system. Also, the proposed air charge control system does not requires time consuming calibration procedures.