Gain-scheduled air path system control and compressor air mass flow estimation in diesel engines

Abstract

In this dissertation, we deal with the independent problems of air path system control and compressor air mass flow estimation in diesel engines. In this regard, the thesis can be divided into two main parts: control and estimation. In the first part, control part, we consider regulation of the air path system in diesel engines. To that end, an extensively used mean-value engine model in the literature is considered. The underlying nonlinear model is converted into a rational linear parameter-varying (LPV) form and a gain scheduled control approach is used. The main contribution here is that the commonly used simplifying assumptions in the literature (like constant engine speed, use of engine charts, assuming exhaust manifold pressure to be equal to intake manifold pressure plus some constant value, taking constant manifold temperatures, etc.) are avoided and a better engine model with a better subsequent controller design are achieved. As a result, the control system design is covering a wide range of engine operating points. In the estimation part, we present two different estimation approaches, each having advantages over the other. In the first approach, we develop a general deterministic state estimation method for states appearing linearly in nonlinear systems. The es- timation method is based on representation of input-output pairs by a linear moving model and the use of recursive-least squares with an adaptive forgetting factor. The method is on-line applicable and very easy to apply. It is shown that the developed estimation method outperforms the popular extended kalman filter (EKF) method on some case studies. Next, the developed estimation method is used for a diesel engine model to estimate compressor air mass flow, whose measurement is difficult or unreliable in some situations and therefore its estimation is important in these cases. Again, a comparison with EKF shows the advantage of the developed method. The second estimation method is based on the use of a state estimation method for affine parameter-dependent linear systems. To use this method, the underlying nonlinear engine model is first transformed into an a±ne parameter-dependent form and then the method is used. This approach has the advantage of having an asymptotic convergence nature, when compared to the first one. Again, this method is used for compressor air mass flow estimation and the results are compared with those of EKF to demonstrate the superiority of the approach.