Ph.D. Proposal Presentation by Patrick Opdenbosch
Tuesday, November 29, 2005

(Dr. Nader Sadegh (Co-Chair), Dr. Wayne Book (Co-Chair))

"Auto-Calibration and Control Applied to Electro-Hydraulic Valves"

Abstract

Although modern electro-hydraulic systems offer great advantages and new alternatives compared to conventional components, there are still great challenges for the control task due to their inherent complex nonlinearities, state constraints, time varying characteristics, parametric uncertainty, payload variation, and disturbance effects often encountered. One common approach to control complex hydraulic valves is to perform steady state input-output calibration on the system at hand and employ inverse mapping with the aid of lookup tables for its control. It can be pointed that those systems possessing time varying characteristics would require undergoing periodic recalibration to perform satisfactorily throughout its operational life.

This research will concentrate on developing a general theoretical framework for online auto-calibration and control of general nonlinear systems with time varying characteristics and parametric uncertainty. More specifically, the inverse dynamics of the system are learned while in operation (online) by employing a simple neural network structure that emulates an adaptive lookup table. This is complemented by online estimation and control of the error dynamics of the nonlinear system along a desired state trajectory thus providing better performance without requiring detailed knowledge of the system at hand.

The control theory developed herein is to be applied to a Wheatstone bridge assembly of novel Electro-Hydraulic Poppet Valves (EHPV) utilized for motion control of hydraulic actuators. It is expected that this method will improve the value of using these types of valves for motion control in hydraulics, leading eventually to a self-learning and more efficient operation of hydraulic systems.