Implementation of Iterative Learning Control on a Pneumatic Actuator.

Abstract

Pneumatic systems play a pivotal role in many industrial applications, such as in petrochemical industries, steel manufacturing, car manufacturing and food industries. Besides industrial applications, pneumatic systems have also been used in many robotic systems. Nevertheless, a pneumatic system contains different nonlinear and uncertain behaviour due to gas compression, gas leakage, attenuation of the air in pipes and frictional forces in mechanical parts, which increase the system’s dynamic orders. Therefore, modelling a pneumatic system tends to be complicated and challenges the design of the controller for such a system. As a result, employing an effective control mechanism to precisely control a pneumatic system for achieving the required performance is essential. A desirable controller for a pneumatic system should be capable of learning the dynamics of the system and adjusting the control signal accordingly. In this study, a learning control scheme to overcome the highlighted nonlinearity problems is suggested. Many industrial processes are repetitive, and it is reasonable to make use of previously acquired data to improve a controller’s convergence and robustness. An Iterative Learning Control (ILC) algorithm uses information from previous repetitions to learn about the system’s dynamics. The ILC algorithm characteristics are beneficial in real-time control given its short time requirements for responding to input changes. Cylinder-piston actuators are the most common pneumatic systems, which translate the air pressure force into a linear mechanical motion. In industrial automation and robotics, linear pneumatic actuators have a wide range of applications, from load positioning to pneumatic muscles in robots. Therefore, the aim of this research is to study the performance of ILC techniques in position control of the rod in a pneumatic position-cylinder system. Based on theoretical analysis, the design of an ILC is discussed, showing that the controller can satisfactorily overcome nonlinearities and uncertainties in the system without needing any prior knowledge of the system’s model. The controller has been designed in such a way to even work on non-iterative processes. The performance of the ILC-controlled system is compared with a well-tuned PID controller, showing a faster and more accurate response.