Research On Inspection Techniques Of Grid Spacer Defect Based On Machine Vision

Grid spacers are critical component of nuclear fuel assembly, used to store and locate nuclear fuel rods. During the procedure of welding and assembly, defects may appear with grid spacers on solder joints, steel projections, springs and guid tubes, and this may cause serious safety problems. Grid spacers must be inspected before being p ut into use. By now, the inspection way of grid spacer defects is manual inspection in China. However, there are many disadvantages of manual inspection, such as high subjectivity, easy to be tired, low efficiency, high cost and so on. With the development of computer and image processing technology, automatic optical inspection technology based on machine vision has replaced manual inspection gradually. It is analyzed and detected finally through extracting and processing image captured from optical imaging system, and the technology has also become the future development direction of grid spacer defect inspection.Therefore, it has important significance for the research of grid spacer defect inspection technology and development of inspection system. In this dissertation, the design of grid spacer defect inspection system, related theory of defect recognition and the key technology of the defect inspection are studied. The main achievements and contents are as follows.1. Through analyzing the index and the principle of grid spacer defect inspection system, the inspection system including image acquisition module, light source module, motion module and image processing module is put forward. Through plane array camera with high-resolution and telecentric imaging system with high depth of field, clear and stable image can be captured. By moving mechanism and image process algorithms, grid spacer is divided into blocks for detecting, and high precision and high efficiency automatic optical defect inspection system is realized finally.2. In preprocessing unit, median filter, mean filter and Gaussian filter have been studied to reduce grid spacer image noise(the camera and lighting equipment cause). Experimental results show that Gauss filter is the best as the first choice considering the computational efficiency and the processing effect. Image segmentation algorithms including gray mean value, OTSU, iterative threshold method are studied. Considering large contrast between target and background of grid spacer image and computational complexity, the effect and the efficiency of gray mean value algorithm for the segmentation are the best. Finally in the preprocessing part, Gauss filter and gray mean value algorithm are used for denoising and image segmentation, which can improve inspection accuracy and system performance.3. The relative positions of targets to be detected are different, so a method through building position template combing with the mechanical platform movement position to locate steel projections, springs and guid tubes is proposed. Three positioning algorithms of solder joints based on Hough transform, template matching and morphology are studied, and the experiment show that the positioning algorithm based on morphology is not affected by the influence of solder joint defects, which is the most effective through using the strip feature of solder joint. In view of the defects of solder joints, steel projections, springs and guide tubes on grid spacers, defect inspection algorithms are proposed based on Canny edge detection algorithm, used to identify and detect different defects. The actual test data indicates that defects of steel projections, springs and guide tubes can be detected quickly with high accuracy by application of position template positioning, morphology analysis positioning, Canny edge detection and defects inspection algorithms.4. The software system mainly includes three modules: image acquisition module, graphic user module and data storage module. Through software controlling, grid spacer inspection can be realized. Experiment results for 100 grid spacers show that omission ratio is less than 1%, false inspection rate is less than 5%, and inspection time of one grid spacer is less than 10 minutes.Through the research of grid spacer defect inspection technology, the design scheme of defect inspection system, defects recognition and inspection algorithms are put forward. Grid spacer defect inspection can be realized via software finally, and the accuracy and efficiency satisfy the index of system. As the first appearance defect inspection system in C hina, the system has been successfully applied in the actual product inspection.