图像边缘检测技术对偏滤器X点的识别研究  

Identification of X-point of divertor based on image technology for edge detection

作　　者：翟文延 黄渊[1] 高金明[1] 冯震[1] 郭文平 刘亮[1] 刘春华[1] 侯智培 ZHAI Wen-yan;HUANG Yuan;GAO Jin-ming;FENG Zhen;GUO Wen-ping;LIU Liang;LIU Chun-hua;HOU Zhi-pei(Southwestern Institute of Physics,Chengdu 610041)

机构地区：[1]核工业西南物理研究院,成都610041

出　　处：《核聚变与等离子体物理》2023年第4期462-468,共7页Nuclear Fusion and Plasma Physics

基　　金：国家自然科学基金项目(11775072);四川省科技计划项目(2020YFSY0046,2021YFSY0044)。

摘　　要：利用计算机视觉技术,对HL-2A偏滤器放电X点进行了识别研究。首先根据HL-2A偏滤器放电图的分析需求,基于直观的等离子体标准下单零放电位形下的边缘可见光辐射模型建立了Canny边缘检测的处理模型,主要对图像中等离子体边缘进行了预处理。结合Harris角点检测的算法,在提取出图像特征点后,利用多边形逼近的方式确定了特征点在图像中的坐标值。分别拟合出强、弱场侧的等离子体边界曲线及其渐近线,渐近线交点近似为X点,边界曲线向中心逼近所包围的区域可近似为X点存在区域。通过实际的HL-2A装置第37638次放电偏滤器放电图验证,经过简化条件后,拟合出的X点与EFIT反演对比误差范围在3cm以内。The identification of X-point in the HL-2A divertor discharge based on image technology was promoted.Firstly,according to the analysis requirements of HL-2A divertor discharge image,the Canny edge detection processing model is established based on the edge visible light radiation model under the single zero discharge configuration under the more intuitive plasma standard.The plasma edge in the image is mainly preprocessed.After extracting the image feature points with Harris corner detection algorithm,the coordinate value of feature points in the image is determined by polygon approximation.The plasma boundary curves and their asymptotes on the strong and weak field sides were fitted respectively.The intersection of the asymptotes is approximately X-point,and the region surrounded by the approximation of the boundary curve to the center can be approximately X-point region.Through the HL-2A 37638th divertor discharge diagram verification,after simplified conditions,the fitting X-point and EFIT inversion error range is within 3cm.

关 键 词：CANNY边缘检测 HARRIS角点检测 X点 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]