oa A High Performance Engine Control Scheme With A Delay-compensation Component

In this paper, a high performance control system is designed for air to fuel ratio of a F-150 Ford truck with a V8 4.6L lean-burn SI engine, is reported as a process with considerable time varying delay. Two approaches have been widely employed to control processes with time-varying delays, (1) designing a feedback robust control system that maintains stability for the whole range of probable delays, which leads to the loss of performance, and (2) designing a feedback control system for a linear model approximated based on Pade formula, which increases the system's order by one and makes the system non-minimum-phase. This article presents a third approach. The basis of the proposed control method is a feedforward-feedback control system which is stable in both continuous and discrete domains for any high control gain in the absence of time delay. That is, theoretically, control gain can be arbitrarily high while maintaining closed loop stability; as a result, the sole source of performance restriction is the actuator limitation to attain high gains. However, in the presence of a time delay of Td, stability and performance analysis is valid only if the system output is known Td seconds in advance. A predictive algorithm with adaptive horizon is proposed to make system's output known upfront. This algorithm relies on the fact that time-varying delay can be identified in real-time for the investigated lean-burn engine as a function of its intake air mass flow rate and rotational speed. This algorithm forms the delay-compensation component of the control system. A filtered PID control system, designed for the same engine and reported in 2012, is compared with the proposed control system. In order to have realistic test scenarios, engine operating conditions are based on a typical Federal Test Procedure (FTP) results. Three different measurement noises are used in simulations. In all simulations with fixed and time-varying delays and with different noise scenarios, the proposed control system clearly outperform filtered PID control system with 22% to 48% less mean of absolute control error. The key success factor of the proposed control system is the employed predictive algorithm. This algorithm predicts the control error and let the control system act to avoid the error in advance. It is an advantage over all feedback control systems needing the measured signals to generate the feedback error, where a delay in measurement results in a delay in feedback error generation and a delay in action. The predictive algorithm largely relies on mathematical models; thus, the influence of parameter identification inaccuracies on overall control system performance may be a concern. In order to investigate this matter, parameters with intentionally wrong values were used in control system design to simulate parameter identification inaccuracies, tests showed an error up to ±20% in parameters identification of engine mathematical model increases the mean of absolute error by only 1%.