DESIGN OF NEURO-FUZZY SYSTEM CONTROLLER FOR DC SERVO MOTOR BASED SATELLITE TRACKING SYSTEM

Abstract

There are increasing needs to develop control systems that can be used to automatically point Direct Current (DC) servo motor driven parabolic antennas to moving targets, notably in satellite tracking. This is in order to maintain the desired Line of Sight (LoS) and Received Signal Level (RSL) to guarantee quality communication. The problem is to design a reliable closed-loop control system with suitable algorithms to achieve accurate dish positioning and the establishment of a strong communication link between an earth station and a target over 400km in space. Conventional methods such as the three term Proportional Integral Derivative (PID) controller have been applied to this problem, mostly due to its simplicity and low cost. However, such techniques have failed to work for nonlinearities associated with servo motor such as saturation, backlash, friction, inertia, mechanical parameter variation and un-modeled dynamics which lead to lack of precise mathematical model. The aim of this research is to design a Neuro-Fuzzy System Controller (NFSC) which can be applied to a DC servomotor-based parabolic dish antenna positioning system. To achieve this, first, a Fuzzy Logic Controller (FLC) was designed using expert knowledge. Next, NFSC was designed based on the FLC algorithms by employing Neural Network (NN) learning to tune the Fuzzy Logic (FL) rule base through hybrid training technique. The architecture of the antenna control system utilized Adaptive Neuro-Fuzzy Inference System (ANFIS) design environment of MATLAB/SIMULINK. The advantage of using NFSC method is that it has a high degree of nonlinearity tolerance, learning ability and solves problems that are difficult to address with the conventional techniques such as PID. For convenience, experiments were conducted offline based on the DC servo motor other than with a live antenna load. The results obtained were compared with those of a conventional PID controller as part of analysis. It was observed that the NFSC method was able to work for the nonlinear parameters and dynamic factors involved in the original DC servo motor system. Consequently, as seen from the results obtained, it achieved the desired output DC servomotor position with reduced rise time, settling time and overshoot in comparison with the PID controller. In an online system an antenna is treated as the load which is directly coupled to the DC servomotor shaft. Therefore, the ability of the developed NFSC to accurately control the antenna azimuth and elevation positions based on the location of satellite under interest with respect to the earth station antenna site was tested.