Predictive Control Realization in Delta Domain for Automotive Powertrain System and Finite Word Length Representation

Abstract

The automotive systems contain fast processes that imply fast sampling rate. Moreover, the existing digital hardware contains processing registers that have finite precision and therefore, the representation can be restricted to a finite number of bits. Thus, the controller must be implemented on these limited word length control units in the context of fast sampling rate. Unfortunately, the classical discrete time representation may become unstable due to finite word lengths calculations. A new solution proposed in this paper is to design a predictive control strategy in the discrete d-domain. State Space d Generalized Predictive Control, in particular, referred as SS d GPC algorithm, can bring significant improvements, especially when is implemented on embedded systems. In this paper, the proposed solution of predictive controller designed in d-domain was successfully applied to automotive powertrain system and demonstrates a large potential in finite word length implementation.