20 th Annual International Conference on Mechanical Engineering-ISME2012 15-17 May, 2012, School of Mechanical Eng., Shiraz University, Shiraz, Iran 1 ISME2012-3479 A Neural Network Control for 3-D Overhead Gantry Crane System With Uncertain Load Disturbance Yousef Bazargan - Lari 1 , Mohammad Eghtesad 2 , Ahmadreza Ghahramani 3 , Ehsan Zakeri 4 , Kimia Bazargan – Lari 5 , 1 Department of mechanical engineering, Shiraz branch, Islamic Azad University, Shiraz, Iran; bazargan@iaushiraz.ac.ir 2 School of Mechanical Engineering, Shiraz University, Shiraz, Iran; eghtesad@shirazu.ac.ir 3 Department of mechanical engineering, Shiraz branch, Islamic Azad University, Shiraz, Iran; rezaghahramani87@gmail.com 4 Department of mechanical engineering, Shiraz branch, Islamic Azad University, Shiraz, Iran; ehsan8631@gmail.com 5 CSE and IT department, Shiraz, University, Shiraz, Iran; kimia.bazargan@gmail.com Abstract In the present work, the adaptive neural network controlling technique is considered in order to have tracking control of a 3-D overhead gantry crane system which uses in industry to transport heavy loads. The dynamic equation used in this paper is based on close form equations of motion, made by lagrangian method. To control this system, a proper control law was conceived and used. With the aim of having an experimental condition in common with the real condition, the controller is designed in the appearance of load disturbance. The Simulation results substantiate the accuracy and the similarity between the desired values and the tracked ones. Keywords: adaptive neural network, gantry crane, uncertainty Introduction Since different industries are on the track of daily progress, human needs of accurate various supplies and equipments has increased. Certainly, with the high precision and high performance tools, their controlling and transport is also much more complicated and requires more meticulousness. One of the devices which play an indispensable role in this movement, are cranes, in particular, the three dimensional ones. In this issue, overhead crane is one of the devices which are used in factories and harbors, in order to carry heavy loads. The main problems that exist in this type of crane and reduce the efficiency and productivity are changes in speed or direction of movement, the efficiency decrease due to environmental disturbances, or lack of accurate control because of human error. Hence, as observed, in addition to careful design, the creation and use of a suitable controller for an optimal control of these devices is inevitable. In past few years, many efforts have been made to maintain an ideal design for the crane. Development of modeling and control [1] and determine the optimal speed in order to minimal the load swing and anti- swinging control [3, 4, 5] are samples of these efforts. Raja Esmail et al (2009) had presented a generalized modeling structure that provides a closed form dynamic equation of motion of a 3-D overhead gantry system [2]. The research object of this article is to tracking control a 3-D overhead gantry system in the appearance of an uncertain lead disturbance. Since the system must be exceedingly accurate, the adaptive neural network control method has been used to control it. A main feature of this control law is the stability properties of the system algorithm being independent of the mechanism used to obtain the tracking performance with a prescribed attention [6]. The control function is programmed in the environment of MATLAB/SIMULINK, based on the adaptive neural network method. Elements of a Paper The gantry crane is a 3-D overhead system; where in its main variables are , φθ , signifying the swing angels of the rope, l denotes its length and F as the driving force. The whole system’s schema is shown in figure1. Figure1: the schema of the model In order to drive the dynamic equation of this systems motion, the lagrangian method must be use to figure out the total energy. In the following, the Euler-lagrange formulation must be considered in order to depict the systems performance. To characterize the crane’s motion, the rail, cart and payload position vectors are presented by [ ] 0 c r x y = (1) [ ] 0 0 r r x = (2) [ ] sin sin sin cos cos p r x l y l l θ φ θ φ φ = + − (3)