Adaptive Observer for Nonlinearly Parameterized Hammerstein System With Sensor Delay—Applied to Ship Emissions Reduction

Abstract

Taking offspring in a problem of ship emission reduction by exhaust gas recirculation control for large diesel engines, an underlying generic estimation challenge is formulated as a problem of joint state and parameter estimation for a class of multiple-input single-output Hammerstein systems with first order dynamics, sensor delay and a bounded time-varying parameter in the nonlinear part. The paper suggests a novel scheme for this estimation problem that guarantees exponential convergence to an interval that depends on the sensitivity of the system. The system is allowed to be nonlinear parameterized and time dependent, which are characteristics of the industrial problem we study. The approach requires the input nonlinearity to be a sector nonlinearity in the time-varying parameter. Salient features of the approach include simplicity of design and implementation. The efficacy of the adaptive observer is shown on simulated cases, on tests with a large diesel engine on test bed and on tests with a container vessel.Index Terms-Nonlinear control systems, joint state and parameter observer, sensor delay.

Marine diesel engine control to meet emission requirements and maintain maneuverability

“…The authors of the present paper first proposed a simplified scavenge oxygen model and nonlinear adaptive EGR controller in [34]. A control-oriented scavenge oxygen model was analytically derived from the MVEM in [11], and a joint state and parameter estimator for this model was presented in [35] along with a proof of exponential convergence. An adaptive feedforward EGR controller based on an inversion of the control-oriented oxygen model was presented in [2] along with convergence proofs and results from a sea trial that showed significant improvement compared to a PI controller.…”

“…The adaptive feedforward (AFF) EGR controller presented in [2] and one of the extensions presented in this paper are based on a control-oriented model (COM) of the molar scavenge oxygen fraction that was presented in [11] and a nonlinear parameter estimator presented in [35]. The COM is a first-order Hammerstein model with three molar flows as inputs…”

Adaptive feedforward control of exhaust recirculation in large diesel engines

“…A similar COM had earlier been presented in (Nielsen et al, 2015) along with a nonlinear controller. A joint state and parameter observer for the COM was presented in (Nielsen et al, 2017a).…”

“…The nonlinear parameter estimator from (Nielsen et al, 2017a) is used for estimating θ θ = k τ y(t) + y(t) − g(θ(t), d(t − ∆t), u(t − ∆t)) dt (19) where k > 0. Note the direct gain from measurement y to estimated parameterθ.…”

“…The estimator described by (19) includes an implicit state due to the integral. In (Nielsen et al, 2017a) the parameter estimate bounds were proven to be…”