Image processing for Plasma Facing Components protection

Type Start End
European Jan 2019 Dec 2020
Responsible URL
Josep R. Casas EUROfusion website

Description

This project is the result of a research collaboration with the Max Planck Institute of Plasma Physics (IPP) located in Greifswald (Germany) within the H2020 EUROfusion project where UPC is a LTP of CIEMAT.

Wendelstein 7-X is the largest prototype of a fusion reactor of the "Stellarator" family (see this piece of news for general information). The first operation phase (OP1) started in 2017, and IPP offers now a large amount of the resulting data (videos, images) for research.

View of the Weldenstein 7-X coil System
Fig.1  View of the Weldenstein 7-X coil System

The collaboration of GPI in this project is included in EUROfusion WPS1. The main purpose of WPS1 is the preparation, execution and exploitation of Wendelstein 7-X campaigns. In particular, the Project Project Management Plan 2019/20, foresees GPI contribution within Key milestone 1.3 in order to prepare and develop control and data acquisition systems and diagnostics addressing objectives for safe steady-state operation and employ surveillance diagnostics for the detection of wall loads, hot spots. The relevant objectives are grouped within task S1.P2.T6 Image Processing and Object Detection:

  • S1.P2.T6 (a) Develop image processing and machine learning tools for detection, tracking and classification of thermal events on the Plasma Facing Components or PFCs (strike-lines, leading edges, hot spots, surface layers, delaminations, reflections),  from thermal images that observe the plasma during operation.
                
    Fig.2a  Thermal events on PFCs (IR image)                      Fig.2b  Animation sequence of a strike line
  • S1.P2.T6 (b) Develop a real-time NN-based heat flux THEODOR estimation for the real-time image analysis system for PFC protection.
  • S1.P2.T6 (c) Develop image processing tools for video and Ha data analysis for automatic plasma events detection and their correlation with the magnetic topology. In this case, a different set of 10 visible camera sensors (EDICAMs @400fps) have observed anomalies in the plasma behavior that IPP has the challenge to study, classify and detect in the 50h of footage they have available. A first ontology of anomalies is: falling debris, MARVEs, Pellets, UFOs and hotspots.

    Fig. 3 Anomaly in the reactor core: falling debris (EDICAM)

The results of the collaboration will be reported in two reports due at the end of 2019 (S1.P.2.D6.1: Interim report on S1.P.2.T6) and 2020 (S1.P.2.D6.2: Final report S1.P.2.T6).