Modelling and controller design for position tracking of pneumatic actuator system

Abstract

To control a pneumatic actuator precisely is not a simple task, as the actuator system consists of many unknown and unpredictable non-linearity. In the worst case, it is prone to disturbances where the position tracking will be inaccurate. This thesis proposes an adaptive controller design for position tracking of the pneumatic actuator system, validated via a simulation study. Considering the mentioned problem, the proposed solution is an Extremum Seeking (ES) based PI(D) controller. This decision is based on the favour of current industry trends, that is, to maintain PID control structure while providing a model-free, data-driven approach and smart control scheme. A preliminary progression has been made to confirm the simulation model for the future controller design stage. The modelling methodology is carried out using the empirical method, and the selected model is AAAA331 structure polynomial model. This is due to the consideration of the parsimonious / theoretical structure of the model and high validation accuracy. Additionally, the model is subjected only for simulation purposes and not for model-based controller design. The benchmark controller tuning using Ziegler-Nichols and IMC-PID approach is carried out to investigate the relationship of PID parameters to the performance of the pneumatic actuator system in position tracking. The proposed model-free online parameter optimization scheme using extremum seeking algorithm is applied on Proportional and Integral paths because it is found that these 2 paths are the primary source of affecting the transient performance response based on the investigation from the benchmark controller. The simulation validation test is carried out using 2 different types of input: step input of 50 mmmm, 100 mmmm, 150 mmmm, and square wave input of ±50 mmmm and ±75 mmmm. The learning ability of the proposed controller scheme was observed when the parameters of the ES-PI(D) controller would automatically tune without properly initialisation and provides a consistent response behaviour when subjected to different inputs. Besides, consistency in transient response and settling time is one of the most significant advantages of the proposed controller.