Robust control of an integrated hydraulic actuator

Abstract

Hydraulic servos are a fast, high force and robust actuation method well suited to robotics. However, compared to electromagnetic solutions, are inefficient, difficult to integrate into systems and have significant control challenges. Moog, a manufacturer of high performance hydraulic components, have developed the Integrated Smart Actuator (ISA) which aims to address these issues by combining an efficient servovalve, actuator, sensors and control electronics into a single additively manufactured titanium package.

The aim of this work is to develop a control algorithm that utilises the known characteristics of the ISA and deals with the unknown load to create a plug and play device that requires minimal control expertise and tuning for integration into a system. The result would be a device with simplified mechanical, hydraulic, electrical and control interfaces, as well as reduced size and weight compared to a discrete component setup.

To meet the aim, a dynamic model of the ISA was developed including detailed characterisation and experimental validation of the spool flow area, spool dynamics and piston friction. The model was used for both simulation and control design. Sliding mode control was selected as a basis for the controller, with a variable boundary layer to deal with the inherent chattering phenomenon. The benefit of sliding mode control is dealing directly with modelling uncertainties and disturbances, with most of the control parameters being set in advance, and a single dominant control term that requires tuning online.

In both simulations and experimentally, the proposed controller was able to outperform the baseline linear proportional with acceleration feedback controller in terms of dynamic response and robustness to uncertainty in the load mass. Robustness to bulk modulus variation was also demonstrated in simulation. This demonstrated the novel application of varying boundary layer sliding mode control to an electrohydraulic servo and the use of an asymmetric piecewise valve flow model in a controller, including a method to linearise it and calculate the valve gain.

An issue with the proposed controller is slower response to unachievable trajectories, for example a position step demand. An online velocity, acceleration and jerk limited trajectory filter was added to shape the input demand to the controller, successfully removing the undesirable behaviour. The tuning of the trajectory filter is a trade-off between the response and the impact of high frequency components on the system.

Date of Award	26 Jun 2024
Original language	English
Awarding Institution	University of Bath
Sponsors	Moog Industrial Group
Supervisor	Ioannis Georgilas (Supervisor) & Andrew Plummer (Supervisor)