Competition of Fuzzy Logic Controllers Applied on Flexible Manipu Mustafa TINKIR Department of Mechanical Engineering, Faculty of Eng. and Arch.,University of Selçuk, Konya, TURKEY e-mail: mtinkir@selcuk.edu.tr Mete KALYONCU Department of Mechanical Engineering, Faculty of Eng. and Arch.,University of Selçuk, Konya, TURKEY e-mail: mkalyoncu@selcuk.edu.tr Ümit ÖNEN Department of Mechanical Engineering, Faculty of Eng. and Arch.,University of Selçuk, Konya, TURKEY e-mail: uonen@selcuk.edu.tr FATH MEHMET BOTSALI Department of Mechanical Engineering, Faculty of Eng. and Arch.,University of Selçuk, Konya, TURKEY e-mail: fbotsali@selcuk.edu.tr Abstract— This paper presents the performances of different type fuzzy logic controllers which are adaptive neural-network based fuzzy logic (ANNFL) controller, hierarchical adaptive neural-network based fuzzy logic (HANNFL) controller and adaptive neural-network based interval type2fuzzy logic (ANNIT2FL) controller. ANNFL, HANNFL and ANNIT2FL controllers are applied on flexible manipulator for both position and tip deflection control to learn which one satisfy us with it’s performance according to desired goals.The performances of the proposed controllers are evaluated on the basis of the experimental results. Keywords- Adaptive neural-network, hierarchical, interval type2, fuzzy logic controller, flexible manipulator, position and tip deflection control. I. I NTRODUCTION Controller design solutions to minimize the effects of the flexible displacements in light robots are highly demanded in the industrial and space applications which require accurate trajectory control. Controlled robot manipulators are usually designed to reach a target or to follow a trajectory. In the first case, a short settling time is expected while in the tracking condition a high speed robot arm displacement is planned. Thus, strong control actions are applied and, as a result, undesired controlled system features could appear if hidden vibrating modes are excited enough. Existing studies [1-7] on flexible robot manipulator can be divided into two groups: those on dynamic, and those on control. Fuzzy Logic Controller (FLC) is credited with being an adequate methodology for designing robust controllers that are able to deliver a satisfactory performance in applications where the inherent uncertainty makes it difficult to achieve good results using traditional methods [8]. As a result the FLC has become a popular approach to flexiblerobot control in recent years [9]. Some of these investigations on the control of flexible robot manipulators consider adaptive neural-network based fuzzy logic control and hierarchical fuzzy logic control [9-14] due to their several advantages over othercontrol techniques. Adaptive neural-network based interval type2 fuzzy logic (ANNIT2FL) control w not taken into consideration for single link flexible rob manipulator in most of the cited investigations despite some advantage to be indicated in this study. In this paper, comparison of different type fuzzy lo controllers applied on a single flexible manipulator system proposed to see which one has better performance ov position and tip deflection control. For this purpose adapt neural-network basedfuzzy logic (ANNFL) controller, hierarchical adaptive neural-network basedfuzzy logic (HANNFL) controller and adaptive neural-network base interval type2 fuzzy logic (ANNIT2FL) controller are designed by using MATLAB/ANFIS toolbox.Control applications are realized via MATLAB/Simulink softwar and experimental set up of a single flexible manipulator an the performances of the proposed controllers are eval on the basis of the experimental results. II. D YNAMIC MODELLING OF THE FLEXIBLE ROBOT MANIPULATOR The schematic diagram of the flexible robot manipulator is given in Fig. 1. This considered robot arm consists flexible beam with a distributed mass. The mass and flexib properties are assumed to be distributed uniformly along t flexible arm. The flexible arm is assumed as an Euler- Bernoulli beam. θ represents the angular position of the equivalent rigid link with respect to the fixed frame X represents the position of the end point on the flexible arm with respect to the equivalent rigid link. Assuming sm deflection of the link, an approximate linear time inva dynamic model is derived using Lagrangian formulation an the dynamic equation is represented in matrix form as: F K C M = + + q q q (1) where the vector q is generalized coordinates. q=[ θ α ] T , M is the mass matrix, C is the damping matrix, K is th stiffness matrix, and F=[T 0] T . T is the input torque applied at the joint. The variable α ( α =D/L) represents the slope at the free end of the flexible link. ___________________________________ 978-1-61284-840-2/11/$26.00 ©2011 IEEE