A tokamak plasma boundary identification and shape reconstruction method

By combining the U-Net network and the XGBoost model, high-precision identification and configuration reconstruction of plasma boundaries in a tokamak device were achieved, overcoming the shortcomings of magnetic measurement and traditional image processing algorithms, and achieving high-precision and stable boundary identification results.

CN115908460BActive Publication Date: 2026-03-03HEFEI INSTITUTE OF PHYSICAL SCIENCE CHINESE ACADEMY OF SCIENCES
2 Cites 0 Cited by

Patent Information

Application Number
CN202211255445.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-10-13
Publication Date
2026-03-03
Estimated Expiration
2042-10-13

AI Technical Summary

Technical Problem

In existing technologies for tokamak devices, magnetic measurement methods are difficult to achieve plasma configuration reconstruction over long periods of time, and traditional image processing algorithms are not effective in edge extraction when the brightness of the vacuum chamber changes, resulting in errors and drift problems.

Method used

A U-Net-based image segmentation method is used to extract plasma boundaries, and an XGBoost model is combined for configuration reconstruction. By optimizing the model with cross-entropy loss function and regularization term, high-precision boundary recognition and configuration reconstruction without manual ROI setting are achieved.

Benefits of technology

High-precision plasma boundary extraction on a tokamak device was achieved with an average error of 7.36 mm, avoiding errors in magnetic measurement and drift problems in image processing algorithms, and improving the accuracy and stability of boundary recognition.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application provides a tokamak plasma boundary identification and shape reconstruction method. First, the EAST plasma boundary is identified by using a U-Net full convolutional neural network, then boundary points are selected, and finally, the pixel coordinates of the boundary points are fitted with EFIT through an XGBoost model. The application can convert the boundary from the image plane to the tokamak polar surface, and realize the plasma shape reconstruction based on a monocular visible light camera. According to the experimental results, the algorithm has high reconstruction accuracy, and the average error on the test set is only 7.36 mm.
Need to check novelty before this filing date? Find Prior Art