Three-axis turntable device for automatically identifying radioactive solids and decision-making level fusion method

Abstract

The invention discloses a three-axis turntable device for automatically identifying radioactive solids and a decision-making level fusion method.The three-axis turntable device comprises a console, a radiation detector, a depth camera and a three-axis rotating mechanism, and the three-axis rotating mechanism comprises an inner-axis rotating mechanism, a middle-axis rotating mechanism and an outer-axis rotating mechanism; the control table comprises a core processing unit, a USB interface, an RS422 interface, a DO interface and an SVDC interface, the USB interface is connected with the depth camera and the radiation detector, the RS422 interface is in communication connection with motors of the three-axis rotating mechanism, and the DO interface and the SVDC interface are used for being connected with external equipment. The system is integrated with other detection equipment, so that the automatic radioactive solid identification process is realized, the identification efficiency and accuracy are improved, and the risk of manual operation is reduced at the same time.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of automatic identification of radioactive solids, and specifically discloses a three-axis rotary table device for automatically identifying radioactive solids and a decision-level fusion method. BACKGROUND

[0002] With the wide application of nuclear technology in many fields such as energy, medicine, and industry, the amount of radioactive solid waste is increasing. The treatment and disposal of these radioactive solid wastes need to strictly follow relevant safety specifications, and accurate identification of their radioactive characteristics is one of the key steps. However, traditional radioactive solid identification mostly relies on manual operation, and workers need to closely contact radioactive substances, which not only increases the risk of radiation exposure, but also may lead to misoperation and affect the identification accuracy. Moreover, manual identification is inefficient, and it needs to detect and analyze each radioactive solid, which takes a long time and is difficult to meet the needs of large-scale treatment. Existing automatic identification equipment also has the problem of single detection angle, and can only detect radioactive solids at a fixed angle, which cannot fully obtain their radioactive distribution characteristics. Moreover, the current automatic identification equipment usually only has a single function, such as radioactive intensity detection or radionuclide identification, and lacks integration with other functional modules.

[0003] Therefore, in order to accurately identify the radioactive characteristics of radioactive solids, a device capable of multi-angle detection is needed. SUMMARY

[0004] The three-axis rotary table device for automatically identifying radioactive solids and the decision-level fusion method can rotate and adjust the radioactive solids in all directions, so that they can be detected at different angles to obtain more comprehensive radioactive distribution information. The three-axis rotary table device can also be integrated with other detection equipment, data processing systems, etc. to realize an automatic radioactive solid identification process, improve the identification efficiency and accuracy, and reduce the risk of manual operation.

[0005] To achieve the above purpose, the technical solution adopted by the application is as follows:

[0006] The three-axis rotary table device for automatically identifying radioactive solids comprises a control console, a radiation detector, a depth camera, and a three-axis rotary mechanism,

[0007] The three-axis rotary mechanism comprises an inner shaft rotary mechanism, a middle shaft rotary mechanism, and an outer shaft rotary mechanism,

[0008] The inner shaft rotary mechanism comprises an inner frame, an inner shaft system, an inner shaft system fixed support structure, and an inner shaft motor, the inner shaft system and the inner shaft motor are respectively installed on the inner shaft system fixed support structure, the inner frame is installed above the inner shaft system, and the output shaft of the inner shaft motor is connected to the inner shaft system.

[0009] The middle shaft rotating mechanism comprises a middle frame, a middle shaft system, a middle shaft system fixing support structure and a middle shaft motor, the middle shaft system fixing support structure is fixed on the middle frame, the middle shaft system and the middle shaft motor are respectively installed on the middle shaft system fixing support structure, the middle frame is installed on the middle shaft system, and the output shaft of the middle shaft motor is connected with the middle shaft system.

[0010] The outer shaft rotating mechanism comprises an outer frame, an outer shaft system, an outer shaft system fixing support structure and an outer shaft motor, the outer shaft system fixing support structure is fixed on the outer frame, the outer shaft system and the outer shaft motor are respectively installed on the outer shaft system fixing support structure, the outer frame is installed on the outer shaft system, and the output shaft of the outer shaft motor is connected with the outer shaft system.

[0011] The console, the radiation detector and the depth camera are arranged above the inner frame, the console comprises a core processing unit, a USB interface, an RS422 interface, a DO interface and an SVDC interface, the USB interface of the console is connected with the depth camera and the radiation detector respectively, the console is in communication connection with the inner shaft motor, the middle shaft motor and the outer shaft motor through the RS422 interface, and the DO interface and the SVDC interface of the console are used for connecting external equipment.

[0012] The radiation detector is used for identifying the type and radiation intensity of the radionuclide.

[0013] The depth camera captures three-dimensional images or depth information by emitting a detection signal and receiving a reflection signal reflected from an object, and is used for identifying or positioning a radioactive solid.

[0014] As a preferred technical scheme of the present application, the base comprises a mounting platform and supporting legs fixed below the mounting platform, the outer shaft system fixing support structure is fixed on the mounting platform, and supporting seats are arranged below the supporting legs.

[0015] As a preferred technical scheme of the present application, the inner shaft system comprises an inner shaft and an inner bearing, the inner shaft system fixing support structure is internally provided with an inner shaft motor controller and an inner shaft transmission structure for controlling the inner shaft motor, one end of the inner shaft is connected with the output shaft of the inner shaft motor through the inner shaft transmission structure, and the other end is connected with the inner frame, the inner bearing is used for supporting the inner shaft, and the inner shaft motor is fixed on the left side of the inner shaft system fixing support structure through bolts.

[0016] As a preferred technical scheme of the present application, the inner shaft motor controller is in communication connection with the console through the RS422 interface.

[0017] As a preferred technical scheme of the present application: the middle shaft system comprises a middle shaft and a middle shaft bearing, the middle shaft system fixing support structure is internally provided with a middle shaft motor controller for controlling a middle shaft motor and a middle shaft transmission structure, one end of the middle shaft is connected with the output shaft of the middle shaft motor through the middle shaft transmission structure, one end is connected with the middle frame, the middle shaft bearing is located between the middle shaft system fixing support structure and the middle frame, the left side of the middle frame is rotatably connected with the outer frame through a rotating shaft, the middle is connected with the inner shaft system fixing support structure, and the right side is connected with the middle shaft, the middle shaft system fixing support structure is vertically fixed on the side surface of the outer frame of the outer shaft rotating mechanism through bolts, and the middle shaft motor is fixed on the front side surface of the middle shaft system fixing support structure through bolts.

[0018] As a preferred technical scheme of the present application: the middle shaft motor controller is in communication connection with the control console through an RS422 interface.

[0019] As a preferred technical scheme of the present application: the outer shaft system comprises an outer shaft and an outer shaft bearing, the outer shaft system fixing support structure is internally provided with an outer shaft motor controller for controlling an outer shaft motor and an outer shaft transmission structure, the left side of the outer frame is rotatably connected with the middle frame through a rotating shaft, the right side is connected with the middle shaft system fixing support structure through bolts, and the lower side is connected with the outer shaft of the outer shaft system, one end of the outer shaft is connected with the output shaft of the outer shaft motor through the outer shaft transmission structure, and one end is connected with the outer frame, the outer shaft bearing is located between the outer shaft system fixing support structure and the outer frame, and the bottom surface of the outer shaft system fixing support structure is fixed on the base through bolts, and the outer shaft motor is fixed on the side surface of the outer shaft system fixing support structure through bolts.

[0020] As a preferred technical scheme of the present application: the outer shaft motor controller is in communication connection with the control console through an RS422 interface.

[0021] As a preferred technical scheme of the present application: the inner shaft system fixing support structure, the middle shaft system fixing support structure and the outer shaft system fixing support structure are all in the shape of a hollow cuboid.

[0022] The decision-level fusion method is characterized in that it comprises the following steps:

[0023] S1, the paired image detection data and radiation intensity data transmitted by the depth camera and the radiation detector are respectively detected and analyzed by corresponding YOLOv8 image detection models and radiation field data analysis modules to obtain respective output results;

[0024] S11. The image detection data is successively extracted by the feature extraction module in the YOLOv8 image detection model, fused by the improved feature fusion module, and predicted by the prediction output module. At this time, the prediction output module will perform position regression prediction on the input feature image so that the output detection frame can surround the target object. In this process, detection frames with different confidence scores will be generated. Then, these target detection frames will be screened according to their respective confidence scores through the non-maximum suppression method. The ones with high scores will be retained and the ones with low scores will be eliminated. Finally, a unique target detection frame with the final highest score is generated. The same is true for the radiation intensity data detection results.

[0025] S2, YOLOv8 image detection model and radiation field data analysis module will obtain the location of the predicted target on their respective input images and , and output the detection scores of the detection results of the two candidate targets at the same time and , that is, output the prediction box position and category detection score of the same target respectively and ;

[0026] S3. After obtaining the above two pairs of parameters, compare the target detection scores of the same target in the two images. and The target position detection frame and detection score corresponding to the higher score are selected as the final recognition result and displayed on the fused image. ≥ When the detection target based on the image detection result is selected as the output target, the candidate box position based on the radiation data detection result is retained. and the detection score of the target prediction category As the final detection result; otherwise, select the candidate target calibrated based on the radiation data detection result as the final output target, and retain the candidate box position based on the image detection result and the detection score of the target prediction category as the final test result.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] The three-axis turntable device for automatically identifying radioactive solids proposed in the present invention integrates mechanical structure design, hardware circuit design, and control system. Through control system design, the present invention simplifies the operating process and reduces the professional skill requirements for operators, making operation easier and faster. It also reduces the risk of direct contact with radioactive materials and protects the health and safety of workers. The three-axis turntable device structure designed by the present invention is capable of full-scale rotation. Combined with high-precision detection equipment and a solid waste target detection algorithm based on decision-level fusion, it can quickly and continuously detect radioactive solids and more comprehensively and accurately identify the characteristics of radioactive solids. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a schematic diagram of the overall structure of the three-axis turntable device for automatically identifying radioactive solids in the present invention;

[0030] Figure 2 This is a control system block diagram of the three-axis turntable device for automatically identifying radioactive solids in the present invention;

[0031] Figure 3 It is the target detection flow chart of the decision-level fusion method;

[0032] Figure 4 It is a flow chart of the decision-level fusion method.

[0033] List of reference numerals:

[0034] 1. Control console; 2. Radiation detector; 3. Depth camera; 4. Inner axis rotation mechanism; 41. Inner frame; 42. Inner axis system; 43. Inner axis system fixed support structure; 44. Inner axis motor; 50. Middle axis rotation mechanism; 51. Middle frame; 52. Middle axis system; 53. Middle axis system fixed support structure; 54. Middle axis motor; 6. Outer axis rotation mechanism; 61. Outer frame; 62. Outer axis system; 63. Outer axis system fixed support structure; 64. Outer axis motor; 7. Base; 71. Mounting platform; 72. Legs; 73. Support seat. DETAILED DESCRIPTION

[0035] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0036] like Figure 1 As shown, in the embodiment of the present application, the three-axis turntable device for automatically identifying radioactive solids provided by the present invention includes a console 1, a radiation detector 2, a depth camera 3 and a three-axis rotation mechanism.

[0037] The three-axis rotation mechanism includes an inner axis rotation mechanism 4 , a middle axis rotation mechanism 5 , an outer axis rotation mechanism 6 and a base 7 .

[0038] The inner shaft rotating mechanism 4 comprises an inner frame 41, an inner shaft system 42, an inner shaft system fixed support structure 43 and an inner shaft motor 44. The inner shaft system 42 comprises an inner shaft and an inner bearing. The inner shaft, as a rotating shaft, is the core part of the inner shaft rotating mechanism 4. One end of the inner shaft is connected to the output end of the inner shaft motor 44 through an inner shaft transmission structure, and the other end is connected to the inner frame 41. The inner bearing is used to support the inner shaft, reduce friction and wear, and ensure the stability and accuracy of rotation. The inner frame 41 is in the shape of a disc, used to fix and support the console 1, the radiation detector 2 and the depth camera 3 above, and can be rotated by the inner shaft to realize angle control of the radiation detector 2 and the depth camera 3 above. The inner shaft system fixed support structure 43 is in the shape of a hollow cuboid, with a supporting function around, and has an inner shaft motor controller and an inner shaft transmission structure inside. The inner shaft motor 44 is fixed on the left side of the inner shaft system fixed support structure 43 by bolts.

[0039] The middle shaft rotating mechanism 5 comprises a middle frame 51, a middle shaft system 52, a middle shaft system fixed support structure 53 and a middle shaft motor 54. The middle shaft system 52 comprises a middle shaft and a middle bearing. One end of the middle shaft is connected to the rotating shaft of the middle shaft motor 54 through a middle shaft transmission structure, and the other end is connected to the middle frame 51. The middle bearing is located between the middle shaft system fixed support structure 53 and the middle frame 51, used to reduce friction of the middle shaft during rotation, improve rotation efficiency, and ensure stable operation of the middle shaft. The left side of the middle frame 51 is connected to the outer frame 61 through a rotating shaft, the middle is connected to the inner shaft rotating mechanism 4, and the right side is connected to the middle shaft, with a supporting and transmission function. The middle shaft system fixed support structure 53 is also in the shape of a hollow cuboid, fixed vertically on the side of the outer frame 61 of the outer shaft rotating mechanism 6 by bolts, with a supporting function on the other four sides, and has a middle shaft motor controller and a middle shaft transmission structure inside. The middle shaft motor 54 is fixed on the front side of the middle shaft system fixed support structure 53 by bolts.

[0040] The outer shaft rotating mechanism 6 comprises an outer frame 61, an outer shaft system 62, an outer shaft system fixing support structure 63 and an outer shaft motor 64. The outer frame 61 is connected to the middle frame 51 through a rotating shaft on the left side, connected to the middle shaft rotating mechanism 5 through a bolt on the right side, and connected to the outer shaft of the outer shaft system 62 at the bottom. The outer frame 61 as a whole plays a supporting and transmission role. The outer shaft system 62 is parallel to the horizontal plane and comprises an outer shaft and an outer bearing. One end of the outer shaft is connected to the rotating shaft of the outer shaft motor 64 through an outer shaft transmission structure, and the other end is connected to the outer frame 61. The outer bearing is located between the outer shaft system fixing support structure 63 and the outer frame 61, used for supporting and reducing the friction of the outer shaft during rotation, and improving the rotation efficiency. The outer shaft system fixing support structure 63 is also a hollow cuboid shape, with the bottom surface fixed on the plane of the base 7 through a bolt, the other four surfaces play a supporting role, and the inside has an outer shaft motor controller and an outer shaft transmission structure. The outer shaft motor 64 is fixed on the side of the outer shaft system fixing support structure 63 through a bolt. The base 7 comprises a plane and four legs, capable of bearing the weight of the whole three-axis turntable and providing stable support for the three-axis rotating mechanism.

[0041] The console 1 comprises a core processing unit, two USB interfaces, an RS422 interface, two DO interfaces and an SVDC interface, responsible for receiving data from the depth camera 3 and the radiation detector 2, sending control instructions to the inner, middle and outer shaft motor controllers, and connecting other external devices. The core processing unit selects GD32F450 chip according to the control system requirements and actual operation needs. GD32F450 is a domestic 32-bit general-purpose microcontroller (MCU) based on ARM Cortex-M4 core launched by Semco, with a processing speed of up to 168MHz, rich peripheral interfaces, high-precision analog and digital signal processing capability, and built-in support for multiple communication protocols. One of the USB interfaces of the console 1 is connected to the depth camera 3, and the other is connected to the radiation detector 2, which can realize data transmission. The console 1 communicates with the inner, middle and outer shaft motor controllers through the RS422 interface and sends control instructions. The two DO interfaces and one SVDC interface of the console 1 can connect other external devices to realize remote control and communication.

[0042] The radiation detector 2 adopts a high-sensitivity scintillation crystal detector, integrating X-ray detection, automatic energy spectrum analysis and automatic nuclide identification functions, which can quickly and accurately identify the type and radiation intensity of radioactive nuclides, and has search, detection and alarm functions.

[0043] The depth camera 3 integrates traditional color imaging (RGB) and depth perception functions of image acquisition function, which emits a detection signal and uses a sensor to receive the signal reflected from the object, captures three-dimensional images or depth information, and is used for identifying or positioning radioactive solids.

[0044] The base 7 is used for overall support and installation of the device, and includes a mounting platform 71 and supporting legs 72 fixed below the mounting platform 71, and the outer shaft system fixed support structure 63 is fixed on the mounting platform 71, and a supporting seat 73 is installed below the supporting legs 72, which not only plays a role in preventing slipping but also increases the contact area with the ground to achieve better support effect.

[0045] As shown in Figure 2 In the embodiment of the present application, the entire device coordinates the work of each part through the console to realize the automatic identification and operation of the radioactive solid. Among them, the depth camera is connected with the console through the USB interface, which is used to capture three-dimensional images or depth information to identify and locate the radioactive solid. The console as the core of the system is responsible for receiving data from the depth camera and the radiation detector, and sending control instructions to the inner, middle and outer shaft motor controllers, and can also connect other external devices through the DO2 and 5VDC1 interfaces to realize remote control. The radiation detector is connected with the console through the USB interface, which is used to detect the radiation emitted by the radioactive solid to identify the type and intensity of the radioactive substance. The inner, middle and outer shaft motor controllers receive RS422 communication instructions from the console, which are responsible for controlling the connected motors to realize multi-angle detection and identification of the radioactive solid. The driving motor is responsible for providing power to drive the mechanical structure to perform physical operation and adjust the motor for precise positioning or fine tuning.

[0046] As shown in Figures 3-4 The present application provides a decision-level fusion method, which is as follows:

[0047] First, the paired image detection data and radiation intensity data transmitted by the depth camera 3 and the radiation detector 2 are respectively detected and analyzed by the corresponding YOLOv8 image detection model and radiation field data analysis module to obtain respective output results;

[0048] The image detection data is first subjected to feature extraction in the YOLOv8 image detection model, then subjected to feature fusion in the improved feature fusion module, and finally subjected to prediction in the prediction output module. At this time, the prediction output module will perform position regression prediction on the input feature image to make the output detection frame surround the target object. In this process, detection frames with different confidence scores will be generated, and then these target detection frames will be screened according to their respective confidence scores by the non-maximum suppression (NMS) method, and the ones with high scores will be retained while the ones with low scores will be eliminated, finally generating a unique target detection frame with the highest final score. The same applies to the radiation intensity data detection result;

[0049] Finally, each model will obtain the position of the predicted target on the input image and , and output the detection scores of the detection results of the two candidate targets at the same time and , that is, output the prediction box position and category detection score of the same target respectively and ;

[0050] After obtaining the above two pairs of parameters, compare the target detection scores of the same target in the two images. and The target position detection frame and detection score corresponding to the higher score of the two are selected as the final recognition result and displayed on the fused image. ≥ When the detection target based on the image detection result is selected as the output target, the candidate box position based on the radiation data detection result is retained. and the detection score of the target prediction category As the final detection result; otherwise, select the candidate target calibrated based on the radiation data detection result as the final output target, and retain the candidate box position based on the image detection result and the detection score of the target prediction category as the final test result.

[0051] The decision-level fusion method provided by the present invention combines visual information and radiation information to achieve automatic identification of radioactive solids. Image detection data comes from the depth camera 3 as the first data input. The image detection data is input into the YOLOv8 image detection model, which processes the image data, identifies and locates the object from the image, and outputs the recognition result. Radiation sensing data comes from the radiation detector 2 as the second data input, which is used to capture the radiation information emitted by the radioactive solid. The radiation sensing data is input into the radiation field data analysis module, which is responsible for processing and analyzing the radiation data, extracting useful information, and outputting the recognition result. The recognition results obtained from the YOLOv8 image detection model and the radiation field data analysis module are output separately and sent to the decision-level fusion module, which is responsible for integrating and analyzing the recognition results of the two data sources to produce a more accurate and comprehensive fusion recognition result. This fusion method utilizes the intuitiveness of image data and the physical properties of radiation data to improve the reliability and accuracy of the recognition system, and is suitable for scenarios where accurate identification and analysis of radioactive materials are required.

[0052] The YOLOv8 image detection model of the application is improved according to the latest generation of YOLO series target detection YOLOv8 model developed by Ultralytics Company, mainly including three parts of backbone, feature fusion part (Neck) and detection head (Head). In the backbone network, the idea of cross-stage partial connection (CSP) in YOLOv5 is still adopted, and on this basis, the C2f module designed by the efficient long-range attention network (ELAN) structure is used to replace the original C3 module in YOLOv5, which can achieve the two purposes of lightweight and rich gradient flow. And continue to use the SPPF module proposed in the previous YOLO series work, which can realize the fusion of feature maps of local features and global features in the backbone network. In the feature fusion part, the feature fusion method based on PAN combined with FPN is still used. In the detection head part, the original coupled head is replaced by the current mainstream decoupled head, which separates the classification branch and the regression branch. Among them, the loss function used in the classification branch is binary cross-entropy loss (BCE), and the loss function used in the regression branch is complete intersection over union loss (CIoU) combined with distribution focal loss (DFL). Another big feature of the YOLOv8 image detection model described in the application is to abandon the classic anchor-based sample matching strategy and use an anchor-free sample matching strategy. It uses a dynamic TaskAligned allocator for matching strategy, which can guide the network to focus on high-quality samples according to the classification score and the intersection over union (IoU) value. In this strategy, the following formula can be used to calculate the score t of each sample, which will be used to guide the selection of samples and the calculation of the loss function:

[0053]

[0054] Where, alpha and beta are weight hyperparameters, used to balance the importance of classification score and IoU value. The classification score s reflects the confidence of the prediction, while the IoU value u reflects the matching degree of the predicted box and the real box.

[0055] The YOLOv8 image detection model of the application has a good balance between high precision and high speed. Through the improvement of the network structure and the introduction of advanced training strategies, the small target recognition ability is greatly improved while maintaining extremely fast detection speed.

[0056] The radiation field data analysis module of the present application is a system component specifically designed to process and analyze data captured by radiation detectors. The module takes into account the directionality of radiation intensity, which is crucial for accurate assessment of the radiation field. Traditional heat maps are primarily used to display data density or intensity variations in a two-dimensional plane, but when dealing with radiation intensity, considering that the radiation source can come from different directions, it is necessary to introduce directional considerations. This can be achieved by mapping the data into a polar coordinate system, where the position of each point is determined not only by its distance but also by its azimuth angle (0° to 360°). Considering that the actual sampling points are usually limited, it is essential to use appropriate interpolation algorithms to estimate the radiation intensity of unmeasured points. Inverse Distance Weighting (IDW) is a commonly used technique that estimates the value of an unknown point by weighted averaging of the values of surrounding known points, where the weight is inversely proportional to the distance. Among them, IDW estimates the value of the unknown point by weighted averaging, and the weight is inversely proportional to the distance:

[0057]

[0058] wherein, is the weight, is the distance between the unknown point and the i-th known point, p is an adjustable parameter that controls the degree of influence of distance on weight, is the estimated value of the unknown point, is the value of the i-th known point, and n is the total number of known points.

[0059] Through IDW interpolation, a continuous radiation intensity distribution map can be obtained, which takes into account the directionality of radiation and can more accurately reflect the true situation of the radiation field. Moreover, the radiation intensity distribution can be displayed in the form of a polar coordinate graph, which helps to intuitively understand the directionality and intensity variation of the radiation source. In the polar coordinate graph, a heat map is used to represent the radiation intensity of different regions, with color or shade representing the intensity. The general practice is to use cool tones (such as blue) to represent low radiation areas and warm tones (such as red) to represent high radiation intensity areas. This color coding method not only conforms to people's visual habits, but also helps to highlight the hotspot areas that need to be focused on. In addition to basic radiation intensity data, geographical information such as terrain features and building layout can also be integrated.

[0060] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in any other form, and any modification or equivalent change made in accordance with the technical essence of the present application still falls within the scope of the present application.

Claims

1. A three-axis turntable device for automatically identifying radioactive solids, characterized by: It includes a control console (1), a radiation detector (2), a depth camera (3) and a three-axis rotation mechanism. The three-axis rotation mechanism includes an inner axis rotation mechanism (4), a middle axis rotation mechanism (50) and an outer axis rotation mechanism (6). The inner shaft rotation mechanism (4) comprises an inner frame (41), an inner shaft system (42), an inner shaft system fixed support structure (43), and an inner shaft motor (44); the inner shaft system (42) and the inner shaft motor (44) are respectively mounted on the inner shaft system fixed support structure (43); the inner frame (41) is mounted above the inner shaft system (42); and the output shaft of the inner shaft motor (44) is connected to the inner shaft system (42); The central axis rotating mechanism (50) comprises a central frame (51), a central axis system (52), a central axis system fixed support structure (53) and a central axis motor (54); the central axis system fixed support structure (53) is fixed on the central frame (51); the central axis system (52) and the central axis motor (54) are respectively mounted on the central axis system fixed support structure (53); the central frame (51) is mounted on the central axis system (52); and the output shaft of the central axis motor (54) is connected to the central axis system (52); The outer shaft rotation mechanism (6) comprises an outer frame (61), an outer shaft system (62), an outer shaft system fixed support structure (63) and an outer shaft motor (64); the outer shaft system fixed support structure (63) is fixed on the outer frame (61); the outer shaft system (62) and the outer shaft motor (64) are respectively mounted on the outer shaft system fixed support structure (63); the outer frame (61) is mounted on the outer shaft system (62); and the output shaft of the outer shaft motor (64) is connected to the outer shaft system (62); The console (1), the radiation detector (2) and the depth camera (3) are arranged above the inner frame (41); the console (1) comprises a core processing unit, a USB interface, an RS422 interface, a DO interface and an SVDC interface; the USB interface of the console (1) is respectively connected to the depth camera (3) and the radiation detector (2); the console (1) is respectively connected to the inner shaft motor (44), the middle shaft motor (54) and the outer shaft motor (64) via the RS422 interface; the DO interface and the SVDC interface of the console (1) are used to connect to external devices; The radiation detector (2) is used to identify the type and radiation intensity of radioactive nuclides; The depth camera (3) captures three-dimensional images or depth information by emitting detection signals and receiving reflection signals reflected from objects, which are used to identify or locate radioactive solids.

2. The three-axis turntable device for automatically identifying radioactive solids according to claim 1, characterized in that: The invention also includes a base (7), wherein the base (7) includes a mounting platform (71) and a support leg (72) fixed below the mounting platform (71); the outer shaft fixed support structure (63) is fixed on the mounting platform (71); and a support seat (73) is installed below the support leg (72).

3. The three-axis turntable device for automatically identifying radioactive solids according to claim 1, characterized in that: The inner shaft system (42) includes an inner shaft and an inner bearing. An inner shaft motor controller and an inner shaft transmission structure for controlling the inner shaft motor (44) are provided inside the inner shaft system fixed support structure (43). One end of the inner shaft is connected to the output shaft of the inner shaft motor (44) through the inner shaft transmission structure, and the other end is connected to the inner frame (41). The inner bearing is used to support the inner shaft. The inner shaft motor (44) is fixed to the left side of the inner shaft system fixed support structure (43) by bolts.

4. The three-axis turntable device for automatically identifying radioactive solids according to claim 3, characterized in that: The inner shaft motor controller is in communication connection with the control console (1) via an RS422 interface.

5. The three-axis turntable device for automatically identifying radioactive solids according to claim 1, characterized in that: The middle shaft system (52) includes a middle shaft and a middle bearing. A middle shaft motor controller and a middle shaft transmission structure for controlling the middle shaft motor (54) are provided inside the middle shaft system fixed support structure (53). One end of the middle shaft is connected to the output shaft of the middle shaft motor (54) through the middle shaft transmission structure, and the other end is connected to the middle frame (51). The middle bearing is located between the middle shaft system fixed support structure (53) and the middle frame (51). The left side of the middle frame (51) is rotatably connected to the outer frame (61) through a rotating shaft, the middle is connected to the inner shaft system fixed support structure (43), and the right side is connected to the middle shaft. The middle shaft system fixed support structure (53) is vertically fixed to the side of the outer frame (61) of the outer shaft rotating mechanism (6) by bolts, and the middle shaft motor (54) is fixed to the front side of the middle shaft system fixed support structure (53) by bolts.

6. The three-axis turntable device for automatically identifying radioactive solids according to claim 5, characterized in that: The central axis motor controller is connected to the control console (1) via an RS422 interface.

7. The three-axis turntable device for automatically identifying radioactive solids according to claim 1 or 2, characterized in that: The outer shaft system (62) includes an outer shaft and an outer bearing. An outer shaft motor controller and an outer shaft transmission structure for controlling the outer shaft motor (64) are provided inside the outer shaft system fixed support structure (63). The left side of the outer frame (61) is rotatably connected to the middle frame (51) through a rotating shaft, the right side is connected to the middle shaft system fixed support structure (53) through bolts, and the bottom is connected to the outer shaft of the outer shaft system (62). One end of the outer shaft is connected to the output shaft of the outer shaft motor (64) through the outer shaft transmission structure, and the other end is connected to the outer frame (61). The outer bearing is located between the outer shaft system fixed support structure (63) and the outer frame (61). The bottom surface of the outer shaft system fixed support structure (63) is fixed to the base (7) by bolts, and the outer shaft motor (64) is fixed to the side of the outer shaft system fixed support structure (63) by bolts.

8. The three-axis turntable device for automatically identifying radioactive solids according to claim 7, characterized in that: The external shaft motor controller is connected to the control console (1) via an RS422 interface.

9. The three-axis turntable device for automatically identifying radioactive solids according to claim 1, characterized in that: The inner shaft system fixed support structure (43), the middle shaft system fixed support structure (53) and the outer shaft system fixed support structure (63) are all in the shape of a hollow cuboid.

10. The decision-level fusion method according to any one of claims 1 to 9, characterized in that: The steps include: S1, detecting and analyzing the paired image detection data and radiation intensity data transmitted by the depth camera (3) and the radiation detector (2) using the corresponding YOLOv8 image detection model and radiation field data analysis module, respectively, to obtain respective output results; S11. The image detection data is successively extracted by the feature extraction module in the YOLOv8 image detection model, fused by the improved feature fusion module, and predicted by the prediction output module. At this time, the prediction output module will perform position regression prediction on the input feature image so that the output detection frame can surround the target object. In this process, detection frames with different confidence scores will be generated. Then, these target detection frames will be screened according to their respective confidence scores through the non-maximum suppression method. The ones with high scores will be retained and the ones with low scores will be eliminated. Finally, a unique target detection frame with the final highest score is generated. The same is true for the radiation intensity data detection results. S2, YOLOv8 image detection model and radiation field data analysis module obtain the location of the predicted target on their respective input images and , and output the detection scores of the detection results of the two candidate targets at the same time and , that is, output the prediction box position and category detection score of the same target respectively and ; S3. After obtaining the above two pairs of parameters, compare the target detection scores of the same target in the two images. and The target position detection frame and detection score corresponding to the higher score are selected as the final recognition result and displayed on the fused image. ≥ When the detection target based on the image detection result is selected as the output target, the candidate box position based on the radiation data detection result is retained. and the detection score of the target prediction category As the final detection result; otherwise, select the candidate target calibrated based on the radiation data detection result as the final output target, and retain the candidate box position based on the image detection result and the detection score of the target prediction category as the final test result.

Images