Cascade Control With Friction Compensasion Based on Artificial Neural Network for a Hydraulic Actuator

Abstract

This paper addresses the friction compensation in hydraulic actuators using an artificial neural network combined with a suitable control technique. The proposed neural network is trained off-line and allows calculate an estimative of the friction force on-line very quickly based on the hydraulic force and on the cylinder velocity. The estimated friction force is introduced directly in the force line of the system using a cascade controller, in which the hydraulic actuator is interpreted as two interconnected subsystems: a mechanical one driven by a hydraulic one. The convergence properties of the closed loop system are established using the Lyapunov method. Experimental results validate the main theoretical results of the proposed strategy.