基于图像光流法拼接的燃料组件关键几何尺寸的快速检测方法

A Fuel Assembly Deformation Fast Detection Method based on Image Optical-flow Stitching

核电燃料组件是核反应堆的核心部件,定期对燃料组件关键几何尺寸(燃料组件长度、燃料棒直径、燃料棒间距等)进行检测对保证反应堆的安全运行至关重要。为实现燃料组件关键几何尺寸的快速精确检测,本文研究适用于弱特征物体的光流全景拼接算法,采用交叉检验与背景滤除的方法优化光流场,并针对燃料组件成排均布的结构特点,提出基于空间扫略的燃料组件边缘特征精确提取算法,实现燃料组件关键几何尺寸(长度、直径、间距)的快速检测。实验结果表明,尺寸检测精度优于0.2 mm,组件长度检测误差1.25%,检测时间4.5 min,满足核电站燃料组件现场检测需求(长度检测误差<3%、检测时间<10 min)。

The nuclear power fuel assembly is the core component of a nuclear reactor.Regular inspection of the key geometric dimensions(fuel assembly length,fuel rod diameter,fuel rod spacing,etc.)of the fuel assembly is an essential part of the safe operation of the reactor.To achieve fast and accurate detection of its key deformation dimensions,the present work investigates the panoramic stitching algorithm suitable for objects with weak features based on the optical flow method,uses cross-checking and background filtering methods to optimize the optical flow field,proposes a spatially sweeping edge extracting method based on the structural characteristics of the fuel assembly.The deformation detection accuracy is better than 0.2 mm,the length detection error of failure probability is 1.25%,and the detection time is 4.5 min,which meets the needs of on-site detection of the nuclear fuel assembly.(the length detection error<3%,the detection time<10 min).