Real-time control environment for the RFX experiment

https://doi.org/10.1016/j.fusengdes.2005.06.146Get rights and content

Abstract

A comprehensive set of control schemes can be presently implemented on RFX due to the enhanced load assembly and renewed power supply system. The schemes include: plasma equilibrium control and resistive wall mode stabilization, aiming at controlling actively the discharge when the passive action of the shell vanishes; the rotation of the localised helical deformation to minimize the enhanced plasma–wall interaction; the MHD mode control and the ‘intelligent shell’, aiming at achieving a better comprehension of the underlying physics.

To the purpose, an integrated, distributed, digital system has been developed consisting of a set of computing nodes. Each node can act either as pre-processing or control station, the former acquiring raw data and computing intermediate control parameters, the latter executing control algorithms and driving the power amplifiers. An overview of the system architecture is presented in the paper with reference to the software real-time environment providing both basic functions, such as data read-out and real-time communication, and useful tools to program control algorithms, to perform simulations and to commission the system.

To simulate the control schemes, the real-time environment is extended to include a so called ‘simulation mode’, in which the real-time nodes exchange their input/output signals with one station running a suitable model of the experiment, for instance the two dimensional FEM code MAXFEA in the case of the equilibrium control. In this way the control system can be tested offline and the time needed for the commissioning of algorithms reduced.

Section snippets

Feedback control

The RFX experiment has undergone major modifications in these years to provide the machine with control windings, enhanced power supply systems, modified load assembly and real-time control systems, thus making feasible an extensive set of innovative, active control schemes by impressed external magnetic field [1], [2], [3], [4], [5].

Some control schemes, such as the equilibrium control and the stabilisation of the MHD resistive wall modes (RWM) are required, due to the time constant of the new

System overview

The control algorithms described above need sampling frequencies of up to some kHz for their digital implementation. The system of actuators and signals is, in general, not diagonal. For these reasons, an integrated, distributed system has been designed and implemented, consisting of computing nodes connected to exchange data in real-time. A single architecture is technically feasible, due to the common time requirements of the systems, and is particularly suitable for the control applications

Software environment

The system software has been developed to run under VxWorks, one of the proprietary real-time operating systems suitable for hard real-time [7]. The system software developed in-house implements basic functions such as optimised data read-out from ADC modules, real-time digital offset correction, data conversion, storage of raw data and intermediate control parameters for post-pulse diagnostics and analysis, real-time data exchange, and download of output references into DAC modules. The

Control algorithms

Time critical algorithms are coded directly in C language, using routines highly optimised to use at best the parallel architecture of the processors. The knowledge of the internal structure of the processor allows writing optimised code in terms of execution time, with the disadvantage that the resulting code is hardly portable. Portability must not be underestimated in physics experiments, as Single Board Computers become obsolete in few years, whereas real-time control systems are required

Testing and commissioning

The design of the RFX feedback control system is based on extensive and accurate modeling activities, due to the fact that the magnetic structure of the machine is being assembled and, for this reason, no direct measurements are possible till completion of assembly. To prepare at best the commissioning phase, in order to minimise the time required and to anticipate the experimental phase, two activities have been carried out in parallel, aiming at establishing an appropriate testbed for the

Conclusions

For the real-time control requirements of the RFX experiment a modular approach has been followed to implement the hardware architecture and the software environment. A subset of the system has been installed at Extrap-T2R and used to control the MHD modes. The experimentation is still underway, and the first results are very encouraging. This has allowed testing the real-time system before the completion of the RFX load assembly. A simulation mode has also been developed for the test of the

References (9)

There are more references available in the full text version of this article.

Cited by (0)

View full text
Copyright © 2005 Elsevier B.V. All rights reserved.