TK193 : Vertical Position Identification of Plasma in Damavand Tokamak by means of Magnetic sensors as Measuring Device
Thesis > Central Library of Shahrood University > Electrical Engineering > MSc > 2011
Authors:
Abstarct: Vertical movement of Plasma column in Tokomak is substantially unstable and the instability causes its strike against chamber’s wall and disruption. So control system can provide necessary power to control Plasma by creating variable magnetic fields.
Control system of Plasma vertical position in Damavand tokomak, until now, worked according to measured magnetic field flux by saddle loop sensors, which in this thesis, in order to increase accuracy, the real Plasma vertical position and directly plasma vertical position is examined. If we want to start up the control system, related circuit for creating plasma vertical position signal (Zp), signal supporter circuit (PRBS), main refrence entrance and controller circuit PD, are designed and built in an electronic board integrally
In order to analyze polar fields to measure plasma position and recognition of effective physical factors for producing suitable model of system vertical position, the exact position of sensors is defined. So, accuracy of position of magnetic sensors which has been presented by Russian, examined and it was cleared that it was not accurate and the correct one was prepared. After that caliber coefficient of sensors would be defined to analyze polar fields measured by magnetic sensors and it was defined by designing a caliber index experiment.
For the purpose of analyze and design control systems, it’s necessary to have a mathematical model of the device that can describe system dynamic as much as possible. In order to prepare dynamic model of plasma vertical position in tokomak, we tried to use experimental data and recognition system technique and designed predictor model and nonlinear simulator for plasma vertical position. In this case, necessary experiments designed and performed. The method and condition of experimental data has been presented and by creating a data baxse, identification possibility and different modeling are provided for future researches. One of the most important problems in identifying this device is impossibility of total and one time applying signal PRBS, which is as a result of plasma resistance time and hardware limitation in starting up control current. Because of this some part of the signal is applied.
predictor model was prepared by means of one laxyer neural network MLP, with NARX construction and training by LM that output results of experimental data and model presented a good credit. After that, a simulator model is prepared by means of one laxyer neural network MLP, with NOE construction and training by LM that output results and experimental data shows a good credit for the model. Finally, output results of modeling was compared with past projects in this case, and presented a progress in modeling process.
Keywords: