A rating system for post pulse data validation

Abstract

The aim of an automatic data validation system in a fusion experiment is to account—after every shot—for any occurrence of faulty sensors and unreliable measurements, thus preventing the proliferation of poor pulse data. In the past years a prototype has been successfully developed at Frascati Tokamak Upgrade (FTU) on a small set of density measurements. The results have shown that the model can be further extended to plant and diagnostic data, and that the same system can be used to assign to raw data a quality factor, to be stored in the archive and to be used in the post-shot elaboration phase as a selection criterion. In this way, a data validation system can also provide data analysts with an useful tool to be used as a key—together with other significant parameters, like plasma current, or magnetic field—to search the archive for quality data. This paper will describe how, using soft computing techniques, both these functions have been implemented on FTU, providing the users with a simple interface for fault detection developed in an open source environment (PHP–MySQL), to be finalised into the realisation of an overall rating system for FTU data.