System identification and model predictive control using CVXGEN for Electro-Hydraulic Actuator

Abstract

Hydraulics have been widely used in heavy industries for decades. The demand for intelligent hydraulic control system has been increasing as tough robotic researches are getting more popular. Despite the high power to weight ratio delivery, the hydraulic actuator suffers from nonlinearity properties that cause difficulties in applying precise position control. In this paper we proposed Model Predictive Control (MPC) to control an Electro-Hydraulic Actuator (EHA) where its dynamic characteristics is obtained through system identification method. Control signal generation optimisation and constraint handling are seldom included in the conventional control system design process. Therefore we introduce CVXGEN, a Code Generator for Embedded Convex Optimization that utilises the Quadratic Programming (QP) interior-point solver for MPC optimisation problem. Predictive Functional Control (PFC) is used to validate the CVXGEN-MPC and both algorithms are implemented in simulation and experiment of EHA position control to highlight the optimisation and constraint handling problem. Control performance, control effort, constraint handling and disturbance handling of both methods are discussed.