Steady state adaptive fueling control of internal combustion engines

Presented in this thesis is a novel steady-state adaptive fueling strategy for an internal combustion engine. The proposed strategy utilizes a feedforward controller to determine the necessary correction in fuel pulsewidth such that after a throttle transient, the air-fuel ratio is regulated to stoichiometry. To determine the correction in fuel pulsewidth necessary for stoichiometric air-fuel ratio, the feedforward controller utilizes a novel nonlinear steady-state gain model for the air-path and the fuel-path. It is shown that engine speed is not an important factor for model development if fuel pulsewidth perturbation and relative LOAD perturbation are considered as inputs to the system. From an experimental testing standpoint, it is shown that the data required for model development can be obtained by performing perturbation tests at a single operating point. The ramification is that perturbation tests at different operating points within the engine operating envelope need not be performed. To achieve stoichiometric air-fuel ratio at steady-state, a feedforward fueling strategy is developed. By using the steady-state gain models, the feedforward fueling strategy directly accounts for the nonlinearities in the air-path and the fuel-path of a spark ignition engine. Since the performance of the feedforward controller is contingent upon the accuracy of the air-path and the fuel-path model, an adaptive algorithm is employed to recover modeling accuracy should the performance of the feedforward strategy deteriorate. The adaptive algorithm adjusts the air-path model via feedback error based adaptation and recursive least squares to account for errors in the air-path and/or the fuel-path model. The proposed adaptive strategy is not contingent upon a particular type of exhaust gas oxygen sensor. In fact, a relay type exhaust gas oxygen sensor can be utilized. Since adaptation is limited to the air-path model, traversing the relative LOAD input space plays a vital role in recovering model accuracy. The requirement is that the driver input must result in a wide range of relative LOAD perturbations to complete the adaptation process. The proposed steady-state adaptive fueling strategy is experimentally verified on the Ford 4.6L-2 valve V-8 sequential fuel injected engine