Reinforcement Neural Network for the Stabilization of a Furuta Pendulum

被引：3

作者：

Allotta, Benedetto ^{[1
]}

Pugi, Luca ^{[1
]}

Bartolini, Fabio ^{[1
]}

机构：

[1] Univ Firenze, Dipartimento Energet Sergio Stecco, Sez Meccan Applicata Macchine, I-50100 Florence, FI, Italy

来源：

PROCEEDINGS OF EUCOMES 08, THE SECOND EUROPEAN CONFERENCE ON MECHANISM SCIENCE | 2009年

关键词：

Furuta pendulum; Control system; Reinforcement neural networks;

D O I：

10.1007/978-1-4020-8915-2_35

中图分类号：

O3 [力学];

学科分类号：

08 ; 0801 ;

摘要：

The Furuta pendulum is a well known mechatronic system in which the goal of the control system is to stabilize the pendulum in the upright position. The control strategy is split into two different stages: the "swing-up," aimed at rising the pendulum near the upright position and the "stabilization" that stabilises the upright equilibrium point by actuating the horizontal arm. The researchers of the University of Florence have realized two prototypes of the Furuta pendulum in order to enlarge the didactical offer of the Mechatronics Laboratory and the Complex Dynamics and Control Systems laboratory and to supply a test bed for control techniques. The features of the prototype are described in this paper as well as the application of a Reinforcement Neural Network to the stabilization phase.

引用

页码：287 / 294

页数：8