Anti-Windup Adaptive Look-Up Table Algorithms With Application to Data-Driven Engine Controls

Abstract

Abstract This technical brief presents anti-windup adaptation algorithms for a look-up table (LUT), widely used in data-driven engine control systems to accurately model complex features while minimizing computational demand. Engine control systems are prone to uncertain variations due to aging, faults, and manufacturing tolerances, which can impact performance and emissions unless effectively managed. Therefore, there is a growing demand for adaptive features in these systems to maintain robust performance and emissions over their lifespan. This study develops computationally efficient adaptive look-up table (ALUT) algorithms using anti-windup recursive parameter estimation and covariance matrix resetting, ensuring robust and rapid adaptation under various operating conditions. The effectiveness of these algorithms is demonstrated through adapting an engine-out nitrogen oxides (NOx) concentration map, which is crucial for tailpipe emission controls in compression-ignition (CI) engines.