DIII-D RESEARCH OPPORTUNITIES FORUM FOR THE 2013 EXPERIMENTAL CAMPAIGN Review | Direct submission with log-in | Request submission without log-in
Print this page

Print this page
Title	217: PID (proportional integral derivative)control of Error Field
Name:	Egemen Kolemen ekolemen@pppl.gov	Affiliation:	Princeton Plasma Physics Laboratory
Research Area:	Stability & Disruption Avoidance	Presentation time:	Requested
Co-Author(s):		ITPA Joint Experiment :	No
Description:	Current Error Field control algorithm is proportional only. Control Theory states that for a linear single input single output system PID (proportional integral derivative) gives as good as a control as any other more complicated system. In our case we have two inputs phase and magnitude but it is highly likely that PID for this case is very close to the most optimal control. The underlying reasoning behind this observation is that a big portion of the control effort is about the power supply and coil dynamics not the EF dynamics. The PID control have been avoided is due to the possible numerical noise in obtaining the derivative and felling that integral term is not of the greatest importance. There many control methods to overcome the issues related to derivative term. As for the integral term being not that important, it the combination of P, I and D terms that gives the optimal control effort. This control can be designed for static or dynamic EF correction. A more ambitious experiment would use real-time PID auto-tuning via relay-feedback. In this case the tuning algorithm is turned on during various stages of the shot and the control is auto-tuned in real-time. This type of control tuning is standard practice in real life control systems such as chemical factories where the process can not be stopped and the system dynamically evolves.	ITER IO Urgent Research Task :	No
Experimental Approach/Plan:	Use automatic relay-feedback experimental PID tuning method. Unlike previously used control development methods, this is a single shot tuning method without user interface (i.e. automatic). Under normal circumstances, we will be running a single for a tuning shot for regime of interest. For real-time PID auto-tuning via relay-feedback, the tuning algorithm is turned on during various stages of the shot and the control is auto-tuned in real-time. Background: There are great number of experimental and data mining based methods to tune the PID control for systems. While, I will be developing PID controls based on data mining methods to get a first level answer, experience with previous experiments from NSTX show that the best approach is to use the automatic-experimental tuning.
Background:
Resource Requirements:	1/2 day.
Diagnostic Requirements:
Analysis Requirements:
Other Requirements: