Candidate: Cláudio Luís d’Elia Machado
Program: UFSC / POSMEC
Date: June, 2003
Advisor: Raul Guenther
Abstract: This work has as focus the study of the friction of a hydraulic actuator and its compensation in closed loop with the fixed cascade controller. The hydraulic actuator is frequently employed due to characteristics as high relationship power/size and speed of response. However, among the difficulties related to closed loop, is the friction that reduces the performance of the controller. In this work, the closed loop control of the hydraulic actuator is accomplished by the fixed cascade controller, because this one presents higher performance than the classic controllers (PID and state controller) and by possibility of the friction compensation. Its control laws are based on the nonlinear model of the hydraulic actuator, interpreting this one as two interconnected subsystems: the mechanical subsystem and the hydraulic subsystem. It is shown in a theoretical way that the friction causes trajectory tracking errors with the fixed cascade controller and that these errors are reduced when the friction compensation is employed. The model of the variable viscous friction coefficient is used to represent the nonlinear friction in the dynamic model of the hydraulic actuator and, with this, it is observed in simulation, the “stick-slip” effect and the influence of the friction in the open loop control of the hydraulic actuator. The compensation strategy consists in estimating the friction power through an artificial neural network appropriately trained by an algorithm denominated backpropagation with momentum using training patterns obtained experimentally. This estimated value is employed in the control law of the mechanical subsystem to compensate the friction. Due to the architecture of the neural network and the realized training, results a satisfactory representation of the nonlinear friction of the hydraulic actuator. The experimental results show the reduction of the trajectory tracking errors when the fixed cascade controller with friction compensation is implemented.
MACHADO, C. L. d’L. Compensação de atrito em atuadores hidráulicos utilizando redes neurais. 2004. Dissertação (Mestrado em Engenharia Mecânica). Universidade Federal de Santa Catarina, Florianópolis.