Designe of an adaptive neuro-fuzzy position controller for a pneumatic system

Abstract

Pneumatic actuators are recently used in many application for robotics and automation such as drilling, sawing, spraying, and gripping. This utilization is due to high speed, low cost, and safety characteristics of these actuators. However, the use of pneumatic systems in position control applications is somewhat restricted due to significant system nonlinearities, caused by air compressibility, friction effects and variations of process parameters with time. In recent years, many researchers tried to solve these restrictions by applying adaptive control mechanism. In this work we used a hybrid structure of fuzzy control and neural networks for the control of a pneumatic actuator. Here the neural network used to adjust all of the output membership functions in the fuzzy logic controller and works as a neural adaption mechanism. The steepest descent learning procedure is implemented and the output from a conventional feedback controller is used as an error by neural network for tuning the parameters of the fuzzy controller. Then, we compared our approach with the fuzzy adaption mechanism. The results obtained by simulation, show better performance of the proposed Adaptive Neuro-Fuzzy Control structure.