الفهرس 
Pneumatic Models
MATHEMATICAL MODELLING OF A PNEUMATIC ACTUATOROnly 14 pages are availabe for public view
 |  | from | 75 |  |  |
| | | from | 75 | | |
Abstract
Experimental and theoretical study is carried out to control the position of
a pneumatic cylinder using online Self-Tuning PID controller (denoted as STNPID controller)based on neural network with back propagation scheme.
The inputs to the STNPID are the tracking error, set point and control action, while the outputs of the neural network are the optimized gains kp, ki and kd.
The STNPID uses neural networks for online tuning of the PID controller gains depending on the environmental and system requirements in order to make the nonlinear system unaffected by the unpredictability of system’s parameters and disturbances such as noise.
High speed digital control valves (operating at frequencies up to 150 Hz) are implemented to accurately control the position of the cylinder piston using Pulse Width Modulation (denoted as PWM) technique.
Results show that the STNPID controller is able to track both constant and variable set point trajectories efficiently by the pneumatic actuator system.
Comparison of the results of STNPID with the conventional PID controller shows that the self-tuning of the PID gains can cope with the nonlinearity of the pneumatic system which is incompatible with the linear behavior of the conventional PID controller. The tracking response for STNPID controller is enhanced with less overshoot and less steady state errors compared to the conventional PID. For square wave, the maximum overshoot is improved from18.83 % for PID to 2.67 % for STNPID. The steady state error is also improved from 9.42 % to 2.67 %.
The oscillation behavior for the sinusoidal response of the conventional PID controller is damped by the STNPID controller and better response is observed. Also, the actuator acts smoothly and uniformly compared to the conventional PID controller with less oscillatory actuator response.