Distributed digital real-time control system for TCV tokamak
•A new distributed digital control system for the TCV tokamak has been commissioned.•Data is shared in real-time between all nodes using the reflective memory.•The customised Linux OS allows achieving deterministic and low latency behaviour.•The control algorithm design in Simulink together with the...
Saved in:
Published in: | Fusion engineering and design Vol. 89; no. 3; pp. 155 - 164 |
---|---|
Main Authors: | , , , , , , , , |
Format: | Journal Article |
Language: | English |
Published: |
Elsevier B.V
01-03-2014
|
Subjects: | |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | •A new distributed digital control system for the TCV tokamak has been commissioned.•Data is shared in real-time between all nodes using the reflective memory.•The customised Linux OS allows achieving deterministic and low latency behaviour.•The control algorithm design in Simulink together with the automatic code generation using Embedded Coder allow rapid algorithm development.•Controllers designed outside the TCV environment can be ported easily.•The previous control system functions have been emulated and improved.•New capabilities include MHD control, profile control, equilibrium reconstruction.
A new digital feedback control system (named the SCD “Système de Contrôle Distribué”) has been developed, integrated and used successfully to control TCV (Tokamak à Configuration Variable) plasmas. The system is designed to be modular, distributed, and scalable, accommodating hundreds of diagnostic inputs and actuator outputs. With many more inputs and outputs available than previously possible, it offers the possibility to design advanced control algorithms with better knowledge of the plasma state and to coherently control all TCV actuators, including poloidal field (PF) coils, gas valves, the gyrotron powers and launcher angles of the electron cyclotron heating and current drive system (ECRH/ECCD) together with diagnostic triggering signals. The system consists of multiple nodes; each is a customised Linux desktop or embedded PC which may have local ADC and DAC cards. Each node is also connected to a memory network (reflective memory) providing a reliable, deterministic method of sharing memory between all nodes. Control algorithms are programmed as block diagrams in Matlab-Simulink providing a powerful environment for modelling and control design. The C code is generated automatically from the Simulink block diagram and compiled, with the Simulink Embedded Coder (SEC, formerly Real-Time Workshop Embedded Coder), into a Linux shared library (“.so” file) and distributed to target nodes in the discharge preparation phase. During the TCV discharge, an application on each node is executed that dynamically loads the shared library at runtime. In order to obtain reliable and reproducible real time execution of the algorithm, all interrupts to the CPU on each node are suspended just before firing the shot and re-enabled afterwards. Since installation, the new digital control system has been used for a multitude of plasma control applications, ranging from basic experiments of coil current and density control to advanced experiments of MHD (magnetohydrodynamics) and plasma profile control, as well as real-time plasma transport simulations. Recently, a real-time version of a plasma equilibrium reconstruction code was developed and implemented, providing the future possibility to control the plasma shape and profiles directly during the discharge evolution. This paper presents the architecture of the new control system, its integration into the TCV plant and a sample of control applications used for TCV plasma discharges. |
---|---|
Bibliography: | ObjectType-Article-2 SourceType-Scholarly Journals-1 ObjectType-Feature-1 content type line 23 ObjectType-Article-1 ObjectType-Feature-2 |
ISSN: | 0920-3796 1873-7196 |
DOI: | 10.1016/j.fusengdes.2013.11.001 |