Research On Keytechnologies For Deformationmeasurement Of Nuclear Fuelassemblies Based On Binocularstructured Light

Nuclear fuel assemblies are used on pressurized water reactor(PWR)in the nuclear power plants,withstanding high temperature,high pressure and high radiation.Nuclear fuel assemblies are prone to deformation over long periods of time.Therefore,it is important to monitor the in-situ deformation of nuclear fuel assemblies.In this paper,the key technology of nuclear fuel assembly deformation measurement was investigated by means of non-contact 3D measurement with binocular structured light projection.A prototype of binocular structured light measurement was built based on system calibration,error correction and point cloud measurement.The main research contents of this paper are as follows:Firstly,the binocular stereo vision imaging model was studied,including monocular pinhole imaging model,imaging aberration model,binocular stereo correction and stereo matching algorithm.For the stereo matching algorithm,an artificial feature phase stereo matching method based on multi-threaded parallel accelerated computation was used.The improved complementary Gray code unwrapping method was used to improve the decoupling accuracy for the step error analysis in the phase unwrapping process of N-step phase-shifted wrapping.Secondly,a nonlinear phase error model was established for the analysis of the gamma error due to the nonlinear response of the system in the phase-shifted stripe projection.It was demonstrated experimentally that the increase of the phase shift step number N can effectively reduce the nonlinear phase error.On this basis,an active correction method with projection streak precoding was proposed,which can effectively reduce the phase error,but the robustness to the shooting environment was poor.A passive correction method based on AG-UNet network was proposed to establish the pixel-level mapping relationship from the phase of low phase-shifted streaks to the phase of high phase-shifted streaks through a deep neural network.The pixel-by-pixel phase error correction was realized and the high-precision phase without increasing the number of streak projections was solved.Then,the component deformation measurement method was investigated based on the point cloud of nuclear fuel components acquired in underwater conditions.Corrections were made for the bending distortion of the point cloud due to refraction of the optical path.By combining RANSAC plane fitting,point cloud coordinate rigid transformation and point cloud filtering algorithms,the problem of noisy and non-measurement target point clouds was solved.A highly robust extraction of clear target point clouds to be measured was achieved.Based on the extracted target point clouds,the upper tube seat gap measurement method and the grill size measurement method wese designed.Finally,based on the Win11 64-bit operating system,the cross-platform QT framework and VS2013 development tools were used.The Open Cv image processing library and PCL point cloud library were used as the programming language to realize the integrated software development of camera acquisition,image processing,point cloud processing and dimensional measurement functions.The simulation nuclear fuel assembly measurement experiments were completed,and the experimental results showed that the platform can meet the measurement accuracy requirement of error in 0.5mm.