An Instrumental Variable Approach for Continuous-Time Identification of an SI Engine Model for Automatic Controller Design
Instrumental variable algorithms are popular for their favorable consistency properties and ease of implementation. In this investigation an algorithm for automatic SI engine idle speed controller design uses and instrumental variable approach to perform system identification. The instrumental variable method employed here identifies the coefficients of a continuous-time model from discretely sampled data. This continuous-time model represents the engine dynamics from the bypass air valve voltage to the engine speed. To implements the identification algorithm, filtered derivative estimators approximate engine speed derivatives and an auxiliary model generates instrumental variables. These calculations are performed at each sample instant and are passed to the recursive formulation of the instrumental variable identification algorithm. Identification is performed utilizing reference step response data. A complete consistency analysis of this algorithm is included for this realization of the instrumental variable algorithm. Automatic controller design is completed based on the identified continuous-time model coefficients. As a result, the controller is parameterized based on the model coefficients and matching the transfer function of the idle speed feedback system to an open loop transfer function which represents the desired transient and steady state performance. The controller is implemented via a bumpless transfer process. Experimental results performed on a 4.6L V-8 fuel injected SI engine demonstrate the automated controller design process and the instrumental variable identification algorithm.