Design of a neural network based intelligent PI controller for a pneumatic system

Pneumatic actuation systems are widely used in industrial automation, such as drilling, sawing, squeezing, gripping, and spraying. Also, they are used in motion control of materials and parts handling, packing machines, machine tools, food processing industry and in robotics; e.g. two-legged robot. In this paper, a Neural Network based intelligent PI controller is designed and simulated to increase the position accuracy in a pneumatic servo actuator where the pneumatic actuator consists of a proportional directional control valve connected with a pneumatic rodless cylinder. In this design, a well-trained Neural Network provides the PI controller with suitable gains depending on feedback representing changes in position error and changes in external load force. These gains should keep the positional response within minimum overshoot, minimum rise time and minimum steady state error. A comparison between this type of controller with a conventional PI type shows that the position of cylinder using a conventional PI controller keeps jittering even when the cylinder reaches the required steady state. This is because of nonlinearities that exist in the pneumatic actuator. This jitter does not persist when a Neural Network based Intelligent PI type controller is used.