Controller design of hydraulic actuator system using self-tuning and model reference adaptive control

Abstract

Nowadays, hydraulic actuator system has become a major drive system in industrial sector especially when involving motion control or position tracking applications. However, due to its natural behaviour which is highly nonlinear, associated with many uncertainties and having parameters that change with timevariation, handling and controlling a hydraulic actuator system is a challenging task. The purpose of this study is to model and to design a controller for hydraulic actuator system. Thus, in order to develop a system that meets the desired performance such as a highly-accurate trajectory tracking, a special knowledge about the system togather with a suitable modelling and control design for the system is mandatory. In this research, Self-tuning Controller using Generalized Minimum Variance Control Strategy and Model Reference Adaptive Controller using Gradient Method has been designed to improve the performance of hydraulic actuator system. System Identification technique with the aid of System Identification Toolbox in MATLAB is used to estimate the mathematical model of the system. System Identification is chosen because it only requires a set of input and output data without the prior knowledge about the system, in order to obtain the system’s transfer function. Auto Regressive with exogeneous input (ARX) model was selected as system’s model structure and the best model among ARX orders was selected based on the analysed result of fitting percentage, loss function and Akaike’s Final Prediction Error. The obtained model was then used to develop the controller for hydraulic actuator system. The output performance was analysed and it has been shown that the output of controlled system successfully tracked the given input signal for both simulation and experimental modes. It has also been observed that Model Reference Adaptive Controller using Gradient Method demonstrates a better output performance compared to Self-tuning Controller using Generalized Minimum Variance Control Strategy in terms of having a minimum phase lagging and a better transient response in terms of rise time, settling time and steady state error.