Fault Detection and Diagnosis for DC Motor in Robot Movement System using Neural Network. A. Che Soh, M.Sc.* and R.Z. Abdul Rahman, M.Eng. Department of Electrical and Electronic Engineering, Universiti Putra Malaysia, Serdang,Malaysia. E-mail: azura@eng.upm.edu.my * ABSTRACT Most of intelligent control in movement control involves fuzzy logic and neural network systems. In this research, a neural network is used to detect and diagnose the faults that may occur in a DC motor system during robot operations. The DC motor system is constructed using the SIMULINK ® toolbox. This system provides the normal and faulty data that has been used for training purpose in the neural network system to get the normal and faulty models. Finally, from the simulation results, the neural network is able to recognize the system characteristic whether in normal conditions or faulty conditions. (Keywords: intelligent control, neural network, digital controller, DC motor, robot movement) INTRODUCTION A fault diagnosis system should perform two tasks, namely fault detection and fault isolation. The purpose of the former is to determine whether a fault has occurred in the system. To achieve this goal, all the available information from the system should be collected and processed to detect any changes from the nominal behavior of the process. The second task is devoted to locate the fault source [4], [5]. Artificial intelligence systems are one among the newest scientific fields in the world that can be used in fault diagnosis. A neural network is one of the branches of the artificial intelligence besides the fuzzy logic system, experts system and multivariate regression [1], [2], [3]. There is lot of research which is related to neural networks and some have been applied in real life. Design and development in robotic systems has become an important field in engineering industries due to the requirement of high performance autonomous machines used to increase production and handle duty under tough environments. In this research, the system is specifically designed to define robotic movement. Generally, the design task can be divided into five parts: movement systems, sensor systems, power systems, controller circuits, and mechanical construction of the robot. However, only movement, sensor, and controller system designs are involved in this simulation. A robot movement system can be designed using DC servo motors, AC motors, or pneumatic systems. Motors are normally used for joints and the robot’s wheel construction because of their small size and ease of control. This study focus on the detection and diagnosis for DC motors in the robot design. This is because the motor is the main part of the robot design. The neural network is used to detect and diagnosis the fault during the operation of DC servo motors, so that the designed system can be improved. The problem will be overcome quickly if the fault can detected early. Simulation and analysis of this robotic movement control system provides designers with information on how the neural network method can affect the output speed and stability of the system. From this information, designers can proceed with a larger design and with more complex controller systems. ROBOT MOVEMENT SYSTEM The robot movement system designed for this project is based on a few subsystems normally available in most digitally controlled movement design technology. The speed system is the DC The Pacific Journal of Science and Technology –35– http://www.akamaiuniversity.us/PJST.htm Volume 10. Number 1. May 2009 (Spring)