Modeling of Fluid Powered Actuators Using Auto Regressive with Exogenous Input Model

Abstract

System identification has importance in modeling and control of industrial systems. The main task of system identification is to build a suitable model that represents the relationship between input, output and disturbances of a real system. The thesis presents identification and discrete time linear modeling of a hydraulic actuator. This thesis demonstrates how to formulate hydraulic functions for both normal and faulty conditions with internal leakage using both offline and on-line measurements. Least square and recursive least square methods are used to estimate the model parameters based on the Auto Regressive technique with Exogenous input (ARX) model. For the offline case, square and sine wave signals are used as input control signals. For the online case, random input control signal is applied. Prediction error criterion is used for model validation based on experimental data. It is shown that the ARX model is capable of representing a valve-controlled hydraulic system dynamics.