Research On Intelligent Identification And Positioning System Of Nuclear Fuel Rods

The assembly of nuclear fuel rods refers to the process of assembling nuclear fuel rods into their corresponding bundle clamps.Due to the huge number,variety of types,small size,and high assembly accuracy required,automated assembly has always been a major challenge in nuclear power field.Because nuclear fuel rods have certain radiation and potential safety hazards to the human body,achieving intelligent identification and assembly of nuclear fuel rods has become an urgent requirement in the nuclear power industry.With the development of computer and image processing technology,the intelligent recognition and assembly of nuclear fuel rods based on machine vision has become possible.In this thesis,a set of nuclear fuel rods intelligent identification and positioning system has been developed to meet the needs of automated assembly of nuclear fuel rods,realized the fuel rod type recognition,positive and negative judgments,calculation of angular,d istribution of isolated blocks,posture recognition and adjustment.Based on the support of ZB Nuclear Fuel Components Co.,Ltd.,this thesis conducts corresponding research on key technologies such as the visual module of nuclear fuel rod intelligent identification and positioning system.The main tasks are as follows:?1?In the analysis of functional requirements and accuracy basically,the overall scheme for nuclear fuel rod intelligent identification and positioning system was designed.Based on the characteristics of nuclear fuel rods,an hypercentric optical imaging system was designed based on an hypercentric optical image acquisition scheme,supplemented by the reflective mode of stripe-shaped reflective strips on an open diffuser reflector.Taking the hypercentric optical imaging system as the design basis,combining the system accuracy and functional requirements,the working model and hardware structure of the nuclear fuel rod intelligent identification and positioning system were designed.?2?Based on the morphological edge extraction,an adaptive weighted multi-scale,multi-direction morphological edge detection algorithm is proposed.Five kinds of directional and multi-scale structure operators are designed to extract the edges of the image,and the weights of the structural factors of each scale are adaptively determined according to the noise and sharpness information of the image.The experimental results show that the edge image detected by our algorithm has the characteristics of continuous,complete,less burrs,no false edges,etc.The standard rod test results show that the average position error of the edge of our algorithm is0.12°,and the maximum positioning error is 0.28°,which meets the engineering requirements.?3?For the needs of the nuclear fuel rod intelligent identification and positioning system,a rotating telecentric correction algorithm for nuclear fuel rods,a straightness evaluation algorithm for nuclear fuel rods,and a nuclear fuel rod identification and positioning algorithm were designed.On the basis of prior knowledge of the isolation block thickness,a center-of-nut compensation algorithm for nuclear fuel rods was developed.The algorithm makes the center of the extraction closer to the true center of the circle,further improving the positioning accuracy of nuclear fuel rods.Aiming at the measurement of nuclear fuel rods's straightness,a linear evaluation algorithm based on the two-point connection method was designed.The algorithm evaluates the straightness of the nuclear fuel rods by extracting the center of the two-port and the center of the middle spacer block.According to the characteristics of the nuclear fuel rod image,a feature extraction and location algorithm was designed to realize the functions of nuclear fuel rod type identification,positive and negative judgment,and angle calculation.The experimental results show that the nuclear fuel rod visual recognition and positioning system can accurately and efficiently complete the nuclear fuel rod type identification,positive and negative judgments,posture adjustment and other work.And the system's process capabilityC_pk=1.1432,positioning qualification rate_sP=99.95%,type recognition accuracy rate of 100%,positive and negative recognition accuracy rate of 100%,average single root recognition and positioning time of 8.92 s.