WEAP031 (Poster)

Presenter: Giuliano Buceti (ENEA)
email: arizzo@dees.unict.it
Review Status: Proceedings Ready - 02/06/02
FullText: pdf
Eprint: physics/0111193

An Automatic Validation System for Interferometry Density Measurements in the ENEA-FTU Tokamak Based on Soft- Computing {*}

G.Buceti (ENEA), L.Fortuna (Univ. Catania), A.Rizzo (Univ. Catania), M.G. Xibilia (Univ. Messina)

With the continuous growth of the number of sensors installed in tokamaks and other big experimental physics plants, reliability of measurements has become a fundamental issue, both for feeding the control systems with reliable measurements and for physicians to analyse the real physics of the experiment. At present, most of the work is carried out manually by the experts responsible of the single diagnostic, which results in a very time-consuming activity. Due to the uncertain knowledge of the phenomena, the huge amount of data stored and the heuristic involved, a suitable approach for the automatic validation of measurements has been individuated in the soft-computing- based techniques. Some results have recently been achieved by the authors at JET, concerning the validation of the measurements of the vertical stresses on the bottom legs of the vacuum vessel during Vertical Displacement Events (VDEs) occurring at disruptions [1,2]. In this paper, a sensor validation strategy for the measurements of plasma line density in the ENEA-FTU tokamak is presented. Density measurements are performed by a 5-channel DCN interferometer. The approach proposed is based on the design of a neural model of the observed system., i.e. a model able to emulate the behavior of a fault-free sensor and of a two-stage fuzzy system able to detect the occurence of a fault by using a set of suitable indicators. The validation strategy has been implemented and embedded in an interactive software tool installed at FTU. Statistics concerning the rate of fault detection agree with the rate of uncertainty usually achieved in the post-pulse manual validation.
[1] L. Fortuna, A. Gallo, A. Rizzo, MG Xibilia, 'An Innovative Intelligent System for Fault Detection in Tokamak Machines', Proceedings of ICALEPCS 99, International Conference on Accelerators and Large Experimental Physics Control Systems, Trieste, Italy, Oct. 99
[2] L. Fortuna, V. Marchese, A. Rizzo, M.G. Xibilia, 'A Neural Networks Based System for Post Pulse Fault Detection and Data Validation in Tokamak Machines', Proceedings ISCAS 99 conference, IEEE International Symposium on Circuits and Systems, Orlando, FL, june 1999
{*} Paper supported by MURST Project 'Fault Detection and Diagnosis, Supervision and Control Reconfiguration in Industrial Process Automation'
ICAL Footer




Last updated: September 13, 2002

Website Problem? Contact Technical Publications