Cafer Bal

A virtual electrical drive control laboratory: Neuro-fuzzy control of induction motors

Neural and fuzzy courses are widely offered at graduate and undergraduate level due to the successful applications of neural and fuzzy control to nonlinear and unmodeled dynamic systems, including electrical drives. However, teaching students a neuro‐fuzzy controlled electrical drive in a laboratory environment is often difficult for schools with limited access to expensive equipment facilities. Therefore, computer simulations and dedicated software are needed to assist the students in visualizing the concepts and to provide graphical feedback during the learning process. In this article, an educational software is proposed for the neuro‐fuzzy control of induction machine drives. The tool helps students learn the application of neuro‐fuzzy control of electrical drives. The software has a flexible structure and graphical user interface. The neuro‐fuzzy architecture, the motor and load parameters can be easily changed in the developed software. Neuro‐fuzzy control performance of induction motors can be monitored graphically for various control structures and current controllers Comput Appl Eng Educ 14: 211–221, 2006; Published online in Wiley InterScience (www.interscience.wiley.com); DOI 10.1002/cae.20082

Development and Implementation of a Fuzzy-Neural Network Controller for Brushless DC Drives

Abstract In this paper, aProportional-Derivative and Integral (PD-I) type Fuzzy-Neural Network Controller (FNNC) based on Sugeno fuzzy model is proposed for brushless DC drives to achieve satisfied performance under steady state and transient conditions. The proposed FNNC uses the speed error, change of error and the error integral as inputs. While the PD-FNNC is activated in transient states, the PI-FNNC is activated in steady state region. A transition mechanism between the PI and PD type fuzzy-neural controllers modifies the control law adaptively. The gradient descent algorithm is used to train the FNN in direct adaptive control scheme. Presented experimental results show the effectiveness of the proposed control system, by comparing the performance of vazious control approaches including PD type FNNC, PI type FNNC and conventional PI controller, under nonlinear loads and parameter variations of the motor.

A hybrid neuro-fuzzy controller for brushless DC motors

In this paper, a hybrid neuro-fuzzy controller (NFC) is presented for the speed control of brushless DC motors to improve the control performance of the drive under transient and steady state conditions. In the hybrid control system, proportional-derivative (PD) type neuro-fuzzy controller (NFC) is the main tracking controller, and an integral compensator is proposed to compensate the steady state errors. A simple and smooth activation mechanism described for integral compensator modifies the control law adaptively. The presented BLDC drive has fast tracking capability, less steady state error and robust to load disturbance, and do not need complicated control method. Experimental results showing the effectiveness of the proposed control system are presented.

Modeling and implementation of a biomimetic robotic fish

In this study, design and implementation of a remote-controlled, 4-joints flapping mechanism and autonomous-swimming biomimetic robotic fish are presented. The propulsive model of the robotic fish is given considering the biological fish structure. The motion control of the robotic fish is performed by using speed and position control. The forward speed of the robotic fish can be adjusted by changing oscillation frequency, oscillation amplitude and length of the oscillation mechanism. Its position is controlled by implementing different joints angles.

DYNAMIC MODEL AND SIMULATION OF ONE ACTIVE JOINT ROBOTIC FISH

This study considers the dynamic model of one active joint robotic fish by using Lagrange method and simulation of the robotic fish model in MATLAB/SimMechanics environment. Compared results of these two different models are given in the study. The mathematical model of the system is derived from Lagrange energy equations of the robotic fish inspired from a real carangiform fish. The Computer Aided Design (CAD) model of the robotic fish is designed by using SolidWorks and it is transferred to the SimMechanics environment. The hydrodynamic effects, which are linear and nonlinear drag force, are also adapted and head motion, one active joint, and one passive joint angles found by using MATLAB Simulink environment. Obtained results for joint angles from both dynamic and SimMechanics models are compared and proved with animation video of the robotic fish.

Modeling of inverted pendulum on a cart by using Artificial Neural Networks

In this study, mathematical model of the single bar inverted pendulum on a cart, which exhibits similar behavior to the dynamics of a robotic arm, is derived by using Lagrange method. At the same time, nonlinear model of the inverted pendulum is simulated in MATLAB environment by using Multilayer Artificial Neural Networks (MANN). MANN is trained by back propagation learning algorithm and dynamic NARX network model is selected for MANN. This model is the basis for dynamics of the robotic arm.

Motion Control of Three-Rotor Unmanned Underwater Vehicle

This paper presents the design of a three-rotor Unmanned Underwater Vehicle (UUV) and generating nonlinear mathematical model. The main propose of this design is to adapt to the UUV effective propellers. For this purpose, three-rotor model is aligned to the Center of Gravity (CoG) and UUV can move in three dimensional space. The derived mathematical model includes the kinematics, dynamics and hydrodynamics effects acting on the body. In order to achieve 6-DoF motion control, three-layered control structure is designed and generalized linear/angular positions and velocities are examined. In addition, three dimensional circular trajectory tracking is performed to achieve effective motion responses.

Design and Control of Diving Mechanism for the Biomimetic Robotic Fish

This paper focuses on generating nonlinear mathematical model of the robotic fish and design a novel diving mechanism for the robotic fish prototype. For this purpose, diving behavior of a real fish is analyzed by using Kineova 8.20. The designed diving mechanism model is constructed by using moving of the sliding mass and dynamic model of robotic fish behavior is performed completely. The diving mechanism is also rearranged by adapting to the robotic fish prototype in order to obtain simple and better design performance. Hence, three dimensional motions results are obtained by using the nonlinear mathematical model and diving mechanism of the robotic fish with different diving angles.

FARKLI KUYRUK MODELLERİNE SAHİP BİR ROBOT BALIĞIN FSI ANALİZİ

Bu calismada, akiskan icerisinde hareket edebilen ve farkli kuyruk yapilarina sahip olan robot baliga akiskan tarafindan etki eden hiz, basinc, kinetik enerji ve girdap degerlerinin bulunmasi amaclanmistir. Belirtilen amaca ulasmak icin Hesaplamali Akiskanlar Dinamigi (HAD) yazilimlarindan biri olan ANSYS paket programi kullanilmistir. Analizler icin carangiform yuzus moduna sahip baliklar ornek alinarak bir robot balik modeli ve 3 farkli kuyruk modeli SolidWorks paket programinda tasarlanmistir. Olusturulan robot balik modelleri icin Akiskan-Yapi Analizi (FSI) yontemi tercih edilmistir. Analizlerin sonuclarina gore kuyruklarin yuzey alanlarinin akiskanda olusturdugu hiz, basinc, kinetik enerji ve girdap degerlerini nasil etkiledigi incelenmistir. FSI analizlerindeki basinc, girdap ve kinetik enerji degerlerine bakildiginda olusturulan robot balik modeli icin en uygun kuyruk yapisinin girintili kuyruk yapisi oldugu sonucuna varilmistir.

Link length optimization of a biomimetic robotic fish based on Big Bang — Big Crunch algorithm

This paper is concerned with the link length optimization method of a biomimetic Carangiform robotic fish with multi-link propulsion mechanism. Motion characteristic of a real fish depends on the body traveling wave and propulsion mechanism of the robotic fish should imitates the traveling wave function. In order to imitate the body traveling wave with minimum error, intersection method is used and Big Bang - Big Crunch (BB-BC) optimization algorithm is adapted to this method. BB-BC algorithm is a heuristic and evolutionary optimization method. BB-BC is preferred in many nonlinear engineering problems because of the low computation time and very fast convergence speed. As a results, optimum link lengths and endpoints of the each joint are determined by using this combined method. Numerical results show that precise fitting effects and link length optimization can improve the propulsion efficiency of the robotic fish. Also, optimum links are proportioned according to actual size of a real Carangiform carp fish in the main axis and free swimming gaits of the real carp are proved.