Remote control method for nuclear power plant valves

Abstract

The application provides a nuclear power plant valve remote control method, which is applied to a valve control device, the valve control device comprises an execution component, the end of the execution component is provided with a docking and locking unit, and the method comprises the following steps: acquiring image information and distance information of a valve wheel, identifying characteristic parameters and spatial poses of the valve wheel according to the image information and the distance information; according to the characteristic parameters and the spatial poses, the execution component is controlled to move to the valve wheel and to be mechanically docked and locked with the valve wheel through the docking and locking unit; the locked execution component is driven to move to control the valve wheel to perform a preset operation, and the locking is released and the execution component is controlled to exit after the operation is completed. The image information and the distance information of the valve wheel are acquired to automatically identify the characteristic parameters and the spatial poses, and then the execution component automatically completes the whole-process operation with the valve wheel, the radiation exposure risk of personnel is fundamentally avoided, and adaptive identification and operation of different specifications of valve wheels are realized.

Description

Technical Field

[0001] This invention relates to the field of remote intelligent control technology for nuclear power plant valves, and more specifically, to a method for remote control of nuclear power plant valves. Background Technology

[0002] The internal piping system of nuclear power plants contains a large number of manually operated valves, covering both routine work areas and high-radiation-dose areas. Routine maintenance, operational adjustments, and emergency response all require frequent valve opening and closing operations. Currently, manual operation on-site is the primary method. However, due to the long travel distance and limited rotational speed of most valves, completing a single opening or closing operation often takes tens of minutes or even longer. This results in operators being exposed to high radiation for extended periods without shielding, significantly increasing their cumulative radiation dose. For certain emergency intervention valves requiring rapid response (such as ARE valves), the response time for manual on-site arrival and operation is too long, failing to meet emergency timeliness requirements and easily leading to serious consequences such as reactor tripping or turbine shutdown. While some scenarios utilize simple remote assistance tools, a complete, universal, and fully automated remote intervention operation solution is lacking.

[0003] The existing operating mode has significant shortcomings in several dimensions: First, personnel must enter high-radiation areas to directly contact the valves, posing a high risk of radiation exposure. Second, in emergency situations, it relies on on-site personnel, resulting in long response times and hindering rapid response. Third, existing remote operation methods are mostly based on dedicated logic for single valves, requiring manual adaptation to specifications and calibration of postures in advance, which cannot achieve adaptive operation for different valve wheels, resulting in poor versatility and low efficiency. Fourth, the entire operation process cannot achieve a closed-loop process of valve wheel autonomous identification, automatic docking and locking, and automatic unlocking and disengagement after operation, still requiring manual intervention in docking, unlocking, and disengagement steps, making it difficult to meet the development needs of intelligent and minimally manned operation and maintenance in nuclear power plants. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a remote control method for valves in nuclear power plants, addressing the problems existing in the prior art.

[0005] The technical solution adopted by this invention to solve its technical problem is: constructing a remote control method for nuclear power plant valves, applied to a valve control device, wherein the valve control device includes an actuator, and the end of the actuator is provided with a docking and locking unit; the method includes the following steps: Step S1: Acquire image information and distance information of the valve wheel, and identify the feature parameters and spatial pose of the valve wheel based on the image information and distance information; Step S2: Based on the characteristic parameters and the spatial pose, control the execution component to move to the valve wheel and mechanically dock and lock it with the valve wheel through the docking and locking unit; Step S3: Drive the locked actuator to move to control the valve wheel to perform a preset operation, and release the lock and control the actuator to exit after the operation is completed.

[0006] Preferably, the valve control device further includes an industrial camera and multiple laser rangefinders disposed at the end of the actuating component, and in step S1, acquiring the image information and distance information of the valve wheel includes: The industrial camera acquires image information of the valve wheel; The distance information of the valve wheel surface is acquired by the laser rangefinder; The image information is normalized to scale the image to a uniform pixel size; The coordinates of each ranging point in the collected distance information are transformed to the camera coordinate system, and the coordinates of each ranging point are unified to the same reference coordinate system through a homogeneous transformation matrix.

[0007] Preferably, the image information includes images of the spoke edges, outer contour, and center hole of the valve wheel; the distance information includes the spatial coordinates of multiple distance measuring points on the valve wheel surface; the feature parameters include the number of spokes, outer diameter, and center hole diameter of the valve wheel; and the spatial pose includes the mounting tilt angle and spoke tilt angle of the valve wheel plane. In step S1, identifying the feature parameters and spatial pose of the valve wheel based on the image information and the distance information includes: The number of spokes is obtained by counting the number of spokes based on the spoke edges; The outer diameter is calculated based on the pixel size of the outer contour and the distance information; The center hole of the valve wheel is detected based on the center hole image, and the center hole diameter is calculated based on the pixel size of the center hole and the distance information. The valve wheel plane is fitted based on the spatial coordinates of each ranging point, and the angle between the normal vector of the valve wheel plane and the axial direction of the end of the actuator is calculated to obtain the installation tilt angle. The spoke tilt angle is obtained by fitting the height variation of each ranging point relative to the valve wheel plane.

[0008] Preferably, the step of calculating the outer diameter based on the pixel size of the outer contour and the distance information includes: Obtain the pixel diameter of the outer contour Combined with camera pixel size Lens focal length and object distance The outer diameter was calculated using a pinhole imaging model. The calculation formula is: Wherein, the object distance The tilt angle is determined based on the distance information collected from three laser ranging points, and corrected according to the following formula: in, It is the average of the distance measurements from the three laser ranging points. The installation tilt angle is as described.

[0009] Preferably, the step of fitting the valve wheel plane based on the spatial coordinates of each ranging point, calculating the angle between the normal vector of the valve wheel plane and the axial direction of the end of the actuating component, and obtaining the installation tilt angle includes: Based on the spatial coordinates of three non-collinear distance measuring points , , Construct direction vector , Through cross product The normal vector of the valve wheel plane is obtained, and the three coordinate components of the normal vector are ( , , ), calculate the angle between the normal vector and the axial direction of the end of the actuator, and obtain the mounting tilt angle. The calculation formula is: in, , , The normal vectors are respectively The three coordinate components.

[0010] Preferably, the step of fitting the spoke tilt angle based on the height variation of each ranging point relative to the valve wheel plane includes: Control a measuring point to move radially along the valve wheel spokes, and continuously collect the spatial coordinates of multiple measuring points; The slope value is obtained by linear fitting based on the height and radial position of each measuring point relative to the valve wheel plane. The spoke tilt angle is calculated based on the slope value.

[0011] Preferably, step S1 further includes: A feature vector is constructed consisting of the number of spokes, the outer diameter, the center aperture, and the spoke tilt angle; and each feature in the feature vector is normalized. The K-nearest neighbor algorithm is used to calculate the weighted Euclidean distance between the normalized feature vector and each template in the preset valve wheel feature library. A preset number of templates with the smallest distance are selected, and the matching results and confidence scores are output by majority voting. When the confidence level is greater than or equal to the preset threshold, the identification is deemed successful, and the valve wheel number of the successful match and the corresponding preset valve parameters are output.

[0012] Preferably, step S2 includes: Step S21: Based on the valve preset parameters, the number of spokes, the outer diameter, the center bore diameter, the installation tilt angle, and the spoke tilt angle, determine the docking posture and docking parameters of the docking locking unit, and plan the movement path of the actuator from the current position to the valve wheel; Step S22: Control the execution component to move according to the motion path, so that the docking locking unit contacts the valve wheel; when the docking locking unit contacts the valve wheel, perform mechanical locking between the docking locking unit and the valve wheel.

[0013] Preferably, the docking locking unit includes a chuck, which is disposed at the end of the actuating component, and step S21 includes: Step S211: Generate a chuck rotation control command based on the number of spokes of the valve wheel to control the rotation of the chuck so that the chuck's claws are aligned with the spokes of the valve wheel in spatial orientation; Step S212: Obtain the preset kinematic parameters of the actuator, take the center position of the valve wheel as the target position and the parallelism between the axial direction of the actuator end and the normal vector of the valve wheel plane as the target attitude, use inverse kinematics to separate the position solution and attitude solution, calculate multiple sets of inverse kinematic solutions, and select the solution with the minimum sum of motion of each joint as the optimal solution; Step S213: Based on the optimal solution, a sequence of intermediate points is generated between the current position and the target position using a linear interpolation algorithm, and a velocity planning is performed using a polynomial interpolation method to calculate and generate the motion path of the execution component from the current position to the target position.

[0014] Preferably, the actuator is a multi-joint robotic arm. In step S212, the inverse kinematics is used to separate the position and attitude solutions, and multiple sets of inverse kinematics solutions are calculated, including: Calculate the motion angles of the first three joints of the multi-joint robotic arm based on the target position to determine the position of the end effector of the multi-joint robotic arm. Based on the target posture and the calculated motion angles of the first three joints, the motion angles of the last three joints of the multi-joint robotic arm are calculated to determine the posture of the end effector of the multi-joint robotic arm. By combining the motion angles of the first three joints with those of the last three joints, multiple sets of inverse kinematic solutions are obtained.

[0015] The beneficial effects of this invention are that it provides a remote control method for nuclear power plant valves, applied to a valve control device. The valve control device includes an actuator, and the end of the actuator is provided with a docking and locking unit. The method includes the following steps: Step S1: Acquire image information and distance information of the valve wheel, and identify the characteristic parameters and spatial pose of the valve wheel based on the image information and the distance information; Step S2: Based on the characteristic parameters and the spatial pose, control the actuator to move to the valve wheel and mechanically dock and lock it with the valve wheel through the docking and locking unit; Step S3: Drive the locked actuator to move to manipulate the valve wheel to perform a preset operation, and release the lock and control the actuator to exit after the operation is completed. This invention automatically identifies the characteristic parameters and spatial pose of the valve wheel by acquiring its image and distance information, and then controls the actuator to automatically complete the docking, locking, operation, and unlocking / disengagement with the valve wheel. This fundamentally avoids the risk of personnel radiation exposure, shortens emergency operation time, and realizes adaptive identification and operation of valve wheels of different specifications. It solves the problems of poor universality and low automation of existing methods, and meets the needs of safe operation and maintenance and emergency response in high-irradiation scenarios. Attached Figure Description

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a schematic diagram of the valve control device. Figure 2 This is a logic flowchart of a remote control method for valves in a nuclear power plant. Detailed Implementation

[0017] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the invention are now described in detail with reference to the accompanying drawings. In the following description, specific details such as particular structures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known devices, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0018] like Figure 1 As shown, Figure 1 This is a schematic diagram of the structure of the nuclear power plant valve control device of the present invention. The valve control device is arranged in a three-layer architecture of "local sensing, local execution, and remote monitoring", and mainly includes a remote control terminal, valve wheel identification component, data processing component, local execution component, signal transmission component, and safety protection component.

[0019] The remote control terminal is installed in the main control room. Local execution components are installed at the valve end, including a mobile execution unit (in some embodiments, this includes a robotic arm 1, used to drive the docking and locking unit to move in space); a drive unit (in some embodiments, this includes a motor and reducer 2, installed at the end of the mobile execution unit, used to output rotational power); and a docking and locking unit (in some embodiments, this includes a chuck 3, fixedly connected to the output end of the drive unit, used to grip and lock the spokes of the valve wheel). A valve wheel identification component is fixed to the end of the execution components, including a radiation-resistant image acquisition unit and a multi-point laser ranging unit. In some embodiments, the radiation-resistant image acquisition unit includes an industrial camera and an infrared supplementary light 4, used to acquire image information of the valve wheel; the multi-point laser ranging unit in some embodiments includes three point laser ranging sensors 5, used to acquire distance information from the valve wheel surface. A contact switch 6 is provided at the center of the chuck 3 to confirm whether the chuck 3 is in contact with the valve wheel 8. A cylinder 7 is also located near the chuck 3. The cylinder 7 is connected to the contact switch 6 via a bracket and is used to drive the contact switch 6 to move away from the chuck 3. An industrial camera and infrared fill light 4 are mounted at the foremost end of the robotic arm 1, with the lens facing the valve wheel 8. The infrared fill light is coaxially arranged in a ring structure around the camera lens. Three point laser rangefinders 5 are arranged in an equilateral triangle around the camera's optical axis and fixed to the end of the robotic arm 1. The signal transmission components adopt a wired and wireless dual-redundancy communication scheme, and the safety protection components include a torque monitoring module and a communication fault monitoring module.

[0020] like Figure 2 As shown, Figure 2 This is a logic flowchart of the nuclear power plant valve remote control method of the present invention. The present invention provides a nuclear power plant valve remote control method, applied to the aforementioned valve control device, the method comprising the following steps: Step S1: Acquire image information and distance information of valve wheel 8, and identify the feature parameters and spatial pose of valve wheel 8 based on the image information and distance information; In this step, it should be noted that the valve wheels 8 at nuclear power plants vary in specifications. Different valve wheels 8 have different numbers of spokes, outer diameters, center bore diameters, mounting angles, and spoke inclination angles. Only by accurately obtaining these parameters can a suitable operating plan be generated based on the actual situation of the valve. This step uses multi-source sensor fusion to simultaneously acquire visual images and spatial distance data of the valve wheels 8, providing a complete data foundation for subsequent feature recognition and automatic docking.

[0021] In an optional embodiment, before performing this step, the system first performs a power-on self-test process: the remote control terminal starts up and sends self-test commands to the local execution component, valve wheel 8 identification component, and data processing component. After each component completes its status self-test, it feeds back the self-test results to the remote control terminal. Only after all components are normal will the system proceed to step S1; otherwise, an alarm will be triggered and the operation will be terminated.

[0022] Step S2: Based on the characteristic parameters and spatial pose, control the actuator to move to the valve wheel 8 and mechanically dock and lock it with the valve wheel 8 through the docking and locking unit; In this step, it should be noted that this step is divided into two sub-stages: first, the docking posture is determined and the motion path is planned based on the feature parameters and spatial pose identified in step S1; then, the actuator is controlled to move to valve wheel 8 according to the planned path and complete the docking and locking.

[0023] Step S3: Drive the locked actuator to move to control the valve wheel 8 to perform a preset operation, and release the lock and control the actuator to exit after the operation is completed.

[0024] This step includes three stages: valve operation, unlocking / disengagement, and record keeping. During operation, the operator can choose either a one-button switch mode or a continuous fine-tuning mode. The actuator rotates at the set speed and torque limit, driving the valve wheel 8. During this process, the torque monitoring module monitors the output torque in real time, and executes a protection operation when the torque exceeds the limit. After operation, the locking mechanism automatically releases, the actuator returns to a safe position, and an electronic operation log is generated.

[0025] The specific implementation of the present invention will be described in detail below with reference to optional embodiments.

[0026] Furthermore, in an optional embodiment, the valve control device further includes an industrial camera and multiple laser rangefinders 5 disposed at the end of the actuator. In step S1, acquiring image information and distance information of the valve wheel 8 includes: acquiring image information of the valve wheel 8 through the industrial camera; acquiring distance information of the surface of the valve wheel 8 through the laser rangefinders 5; normalizing the image information and scaling the image to a uniform pixel size; transforming the coordinates of each ranging point in the acquired distance information to the camera coordinate system, and unifying the coordinates of each ranging point to the same reference coordinate system through a homogeneous transformation matrix.

[0027] Specifically, the industrial camera employs a radiation-resistant CMOS sensor, with the lens facing the valve wheel 8. The preset angle between the optical axis and the normal to the plane of the valve wheel 8 is 0° to 15°. An infrared supplementary light 4, arranged in a ring structure, coaxially surrounds the camera lens and consists of 8 to 12 LEDs with a wavelength of 850nm, providing uniform illumination to the valve wheel 8 and preventing glare from affecting image quality. Three point laser rangefinders 5 are arranged in an equilateral triangle around the camera's optical axis, with a spacing of 40mm to 60mm between each pair. Each laser beam has an angle of 5° to 10° with the camera's optical axis, forming a 120° circular distribution. The entire sensing module is encapsulated within a radiation-resistant protective cover and fixed to the end of the actuator.

[0028] During data acquisition, an industrial camera captures panoramic images of the valve wheel 8 at a resolution of 1920×1080 and a frame rate of at least 15fps. Three laser rangefinders 5 simultaneously acquire distance data from three different locations on the surface of the valve wheel 8. Each sensor scans the valve wheel 8's disc surface, edge, and central axis at 5 to 8 points. The 5 to 8 point scanning is implemented by a robotic arm 1 moving the end effector within a small range, allowing the fixedly mounted laser rangefinders 5 to sequentially acquire data from multiple locations on the valve wheel 8's surface. After the image data is transmitted to the data processing unit, it is first normalized, scaling the image to a uniform pixel size. For the laser ranging data, the coordinates of each ranging point are transformed from the sensor's local coordinate system to the camera's coordinate system.

[0029] Specifically, let the first The installation location of the laser rangefinder sensor 5 is as follows: The laser beam direction vector is The measured distance is Then the three-dimensional coordinates of the laser spot in the camera coordinate system. It is given by the following formula: For the multi-point scanning data after the robotic arm 1 moves, a homogeneous transformation matrix T is needed to unify all points to the same reference coordinate system to eliminate the coordinate offset caused by the movement of the robotic arm 1. After the above normalization and coordinate transformation processing, the image data and laser ranging data have been aligned to a unified coordinate system, providing standardized data input for subsequent feature recognition.

[0030] Further, in an optional embodiment, the image information includes images of the spoke edges, outer contour, and center hole of the valve wheel 8; the distance information includes the spatial coordinates of multiple ranging points on the surface of the valve wheel 8; the feature parameters include the number of spokes, outer diameter, and center hole diameter of the valve wheel 8; and the spatial pose includes the mounting tilt angle and spoke tilt angle of the valve wheel 8 plane. In step S1, identifying the feature parameters and spatial pose of the valve wheel 8 based on the image information and distance information includes: counting the number of spokes based on the spoke edges to obtain the number of spokes; calculating the outer diameter based on the pixel size of the outer contour and distance information; detecting the center hole of the valve wheel 8 based on the center hole image, and calculating the center hole diameter based on the pixel size of the center hole and distance information; fitting the valve wheel 8 plane based on the spatial coordinates of each ranging point, calculating the angle between the normal vector of the valve wheel 8 plane and the axial direction of the end of the actuator to obtain the mounting tilt angle; and fitting the spoke tilt angle based on the height change of each ranging point relative to the valve wheel 8 plane.

[0031] Specifically, the recognition of the number of spokes is achieved using image processing algorithms. First, the image is denoised using Gaussian filtering with a two-dimensional Gaussian function: Perform convolution filtering on the image, where, , The x and y coordinates of the image pixels are σ = 1.5, which eliminates sensor noise.

[0032] Then the cumulative distribution function is used: Histogram equalization is performed to enhance the contrast between the spokes and the background. Among other things, For the current target gray level, This represents the total number of pixels in the image. This represents the number of pixels with gray level j.

[0033] Next, the Sobel operator is applied to calculate the gradient magnitude: in, For the image in Gradient in the horizontal direction, For the image in Gradient in the direction (vertical direction); Calculation direction: By combining nonmaximum suppression and dual threshold hysteresis tracking (threshold set to 150 / 50), single-pixel level edges are extracted, and the Suzuki algorithm is used to filter the outermost closed contour of the outer circle of valve wheel 8.

[0034] In polar coordinate space, Hough accumulator voting is performed on edge points according to the polar coordinate transformation formula to extract the set of line segments. Neighborhood clustering is then performed based on the direction angle, with a tolerance of 15°. Lines with direction angles within 15° are grouped together, and the number of groups is counted to obtain the number of spokes N. The polar coordinate transformation formula is as follows: in, , The coordinates of the edge points in the image space are the angle between the normal of the line and the x-axis (polar angle).

[0035] The outer diameter was measured using a fusion of image-based and laser three-point concircle methods. For the image-based method, the pixel diameter of the valve wheel 8's outer contour was extracted. Calculate the physical outer diameter based on the pinhole imaging model Regarding the laser three-point concircle method, three laser rangefinders 5 are fixedly mounted on the end effector, evenly distributed in a 120° circle. After the robotic arm 1 is in position, the three laser beams naturally form three non-collinear light spots on the valve wheel 8. When the three points satisfy the triangle area threshold (≥10mm²) and the distance between them is reasonable, the outer diameter is directly calculated using the three-point concircle formula. The outer diameters measured by the two methods are fused to obtain the optimal outer diameter estimate. If higher accuracy is required, the robotic arm 1 can be controlled to move within a small radial range to collect multiple sets of three-point data. A circle is then fitted using the least squares method or the RANSAC algorithm to finally output the optimal outer diameter estimate.

[0036] Further, in an optional embodiment, the outer diameter is calculated based on the pixel size and distance information of the outer contour, including: obtaining the pixel diameter of the outer contour. Combined with camera pixel size Lens focal length and object distance The outer diameter was calculated using a pinhole imaging model. The calculation formula is: Among them, object distance The tilt angle is determined based on the distance information collected from three laser ranging points, and corrected according to the following formula: in, It is the average of the distance measurements from the three laser ranging points. The installation tilt angle is for this purpose.

[0037] Specifically, in terms of distance recognition, the distance L from the reference surface at the end of the actuator to the surface of the valve wheel 8 is calculated based on the distance data collected by the three laser rangefinders 5. Let the installation position of the i-th laser rangefinder 5 be... The laser beam direction vector is The ranging value is The vertical distance between the laser spot and the end reference plane is: The average distance from the end reference surface to the valve wheel 8 surface is obtained by averaging the three points: when And if it remains stable for 0.5 seconds, the end of the actuator is determined to be in position.

[0038] Further, in an optional embodiment, the valve wheel 8 plane is fitted based on the spatial coordinates of each ranging point, and the angle between the normal vector of the valve wheel 8 plane and the axial direction of the end of the actuator is calculated to obtain the installation tilt angle, including: based on the spatial coordinates of three non-collinear ranging points... , , The spatial coordinates of the three laser ranging points are as follows: Construct direction vector , Through cross product The normal vector of the valve wheel plane 8 is obtained, and the three coordinate components of the normal vector are ( , , ), calculate the angle between the normal vector and the axial direction of the end of the actuator to obtain the installation tilt angle. The calculation formula is: in, , , Normal vectors The three coordinate components.

[0039] Specifically, the physical meaning of this formula is: the cosine of the angle between the normal vector and the axial direction (i.e., the Z-axis) of the actuator end is equal to the absolute value of the Z-axis component of the normal vector divided by the total length of the normal vector. Taking the inverse cosine gives the installation tilt angle. When this angle is less than or equal to 20°, the system determines that automatic docking is possible and sets the end target attitude to be parallel to the normal vector of valve wheel 8; when this angle is greater than 20°, it determines that automatic intervention is not possible, the system alarms and terminates the operation. If the three points are approximately collinear, causing the normal vector magnitude to approach zero, the fitting is deemed to have failed, triggering re-acquisition or switching to a backup method.

[0040] Further, in an optional embodiment, the spoke tilt angle is obtained by fitting the height change of each measuring point relative to the plane of the valve wheel 8, including: controlling a measuring point to move radially along the spokes of the valve wheel 8, and continuously collecting the spatial coordinates of multiple measuring points; performing linear fitting based on the height and radial position of each measuring point relative to the plane of the valve wheel 8 to obtain a slope value; and calculating the spoke tilt angle based on the slope value.

[0041] Specifically, the robotic arm 1 drives the end effector to translate, causing the light spot of the fixedly installed laser rangefinder 5 to move radially along the spokes from the root to the outer edge, with a moving speed not exceeding 5 mm / s. During the movement, the laser rangefinder value and the pose feedback of the robotic arm 1 end effector are continuously collected, and the spatial three-dimensional coordinates of the light spot at each sampling point are calculated, thereby obtaining the height of each point relative to the plane of the valve wheel 8. and radial position .

[0042] After collecting at least 5 points, a linear fit was performed using the least squares method: Obtain the slope ,in, The intercept of the fitted line, For height, Where is the radial distance. Then the spoke inclination angle is: If the feature library indicates that the valve wheel has 8 spokes and a straight structure ( This scanning process can be skipped, and settings can be set directly. °.

[0043] Further, in an optional embodiment, step S1 further includes: constructing a feature vector composed of the number of spokes, outer diameter, center aperture, and spoke tilt angle; normalizing each feature in the feature vector; calculating the weighted Euclidean distance between the normalized feature vector and each template in the preset valve wheel 8 feature library using the K-nearest neighbor algorithm; selecting a preset number of templates with the smallest distance; and outputting the matching result and confidence level according to the majority voting method; when the confidence level is greater than or equal to a preset threshold, the identification is determined to be successful, and the valve wheel 8 number and the corresponding preset valve parameters are output.

[0044] Specifically, constructing a four-dimensional feature vector Subsequently, due to the different dimensions and large differences in numerical ranges of the various features (spoke count ranges from 3 to 6, outer diameter from 80 to 400 mm, center aperture from 0 to 50 mm, and spoke tilt angle from 0° to 30°), min-max normalization is required to map each feature to the [0,1] interval. The normalization formula is: in, The normalized eigenvalues ​​are mapped to the interval [0,1]. These are the original eigenvalues. The minimum value of this feature in the feature library. This is the maximum value of this feature in the feature library.

[0045] Then, the K-nearest neighbor algorithm (K=3) is used to calculate the weighted Euclidean distance between the normalized feature vector and each template in the preset valve wheel 8 feature library: in, These represent the measured number of spokes, outer diameter, center aperture, and spoke inclination angle, respectively (the apostrophe indicates normalization). This represents the feature value corresponding to the j-th template in the normalized feature library. The weights are set as follows: number of spokes. =0.4, outer diameter =0.3, center aperture =0.2, spoke tilt angle =0.1. Take the three templates with the smallest distance, and output the matching results and confidence scores using majority voting. Confidence level = number of votes / 3 × 100%. When the confidence level is ≥ 85%, the identification is considered successful, and the valve wheel number of the successfully matched valve and the corresponding preset valve parameters (such as torque limit, drive mode, etc.) are output; otherwise, remote manual calibration is triggered.

[0046] The remote manual calibration method is as follows: When automatic recognition fails, the operator performs manual calibration in the main control room via a remote control terminal. The terminal displays real-time images from the industrial camera and provides a semi-automatic calibration toolset, including functions such as center marking, outer circle fitting, spoke marking, center hole marking, and angle measurement. The operation procedure is as follows: 1. Click the center point of valve wheel 8 on the image; 2. Click 3 to 6 points along the outer edge, and the system automatically fits the outer circle and calculates the diameter D; 3. Draw a line along each spoke, and the system automatically counts the number of spokes. 4. Click on 3 points on the edge of the center hole, and the system will automatically fit and calculate the hole diameter. 5. Draw lines along the spoke direction, and the system will automatically calculate the tilt angle. 6. Confirm submission. The system will generate a temporary feature template and continue execution. After calibration, the system will automatically verify the rationality of the parameters (spoke uniformity, outer diameter range, tilt angle threshold, etc.). The entire process requires no personnel to enter the site and takes approximately 2 to 3 minutes to complete.

[0047] The principle of remote manual pixel-to-physical-size calibration is as follows: when an operator clicks or draws lines on an image, the system obtains the pixel dimensions. Simultaneously, the background system acquires the laser rangefinder distance L in real time and reads the camera's intrinsic parameters (focal length). Pixel size Object distance Substitute into the pinhole imaging model: The actual physical outer diameter and bore diameter of valve wheel 8 are calculated, and all conversions are completed automatically by the system.

[0048] Specifically, the valve wheel 8 may be installed at a certain tilt angle. If tilt angle correction is not performed, directly using the average distance as the object distance will lead to errors in the image-based diameter measurement. By introducing tilt angle correction, the object distance is corrected to... This method effectively compensates for measurement deviations caused by the tilt of valve wheel 8, improving the accuracy of outer diameter measurement. The final estimated value of outer diameter D is obtained by fusing the measurement results from the image method and the laser three-point concircle method. In the three-point concircle method, three laser rangefinders 5 are fixedly installed on the end effector and evenly distributed in a 120° circle. After the robotic arm 1 is in position, the three laser beams naturally form three non-collinear light spots on the valve wheel 8 disk. Their spatial coordinates are determined by the installation geometry and the laser rangefinder value, without the need for dynamic selection by an algorithm. If the three points meet the triangle area threshold (≥10mm²) and the distance between them is reasonable, the outer diameter can be directly calculated using the three-point concircle formula. If higher accuracy is required, the robotic arm 1 can be controlled to move in a small radial range to collect multiple sets of three-point data. The circle is then fitted using the least squares method or the RANSAC algorithm, and the optimal outer diameter estimate is finally output.

[0049] Further, in an optional embodiment, step S2 includes: step S21: determining the docking posture and docking parameters of the docking locking unit according to the valve preset parameters, number of spokes, outer diameter, center bore diameter, installation tilt angle and spoke tilt angle, and planning the movement path of the actuator from the current position to the valve wheel 8; step S22: controlling the actuator to move according to the movement path, so that the docking locking unit contacts the valve wheel 8; when the docking locking unit contacts the valve wheel 8, performing mechanical locking between the docking locking unit and the valve wheel 8.

[0050] Specifically, in step S21, the optimal docking parameters are matched based on the parameters identified in step S1: the initial posture parameters of the docking locking unit, the jaw opening, the rotary valve torque, etc. Among these, the number of spokes determines the jaw number matching of the chuck 3 (three spokes correspond to three jaws, five spokes correspond to five jaws) and the rotation alignment angle; the outer diameter and center bore diameter determine the jaw opening and centering position of the chuck 3; the installation tilt angle determines the approach angle of the end of the actuator (the end axis should be parallel to the normal vector of the valve wheel 8 plane); the spoke tilt angle determines the contact method between the jaws and spokes of the chuck 3; and the valve preset parameters provide reference values ​​such as the upper limit of torque and the drive mode.

[0051] The specific method of path planning is as follows: taking the center position of valve wheel 8 as the target position, and taking the parallelism between the end axis of the actuator and the plane normal vector of valve wheel 8 as the target posture, the motion angles of each joint of the actuator are solved by inverse kinematics.

[0052] Among them, target location The center of the circle determined by the aforementioned three laser ranging points is given, and the target attitude is set as the Z-axis of the end effector intersecting the valve wheel normal vector. Parallel to and pointing towards the valve wheel, with free X and Y axis directions at the end, which can be determined by the robotic arm configuration, typically taking the X and Y directions to maintain the current posture. The target position vector... and target attitude matrix Combined into a homogeneous transformation matrix: in, This represents the target attitude matrix of the robotic arm's end effector relative to the base coordinate system. Let the target position vector be... This is the homogeneous transformation matrix of the target pose obtained by inverse kinematics.

[0053] The target position is defined by a circle determined by three laser ranging points, with the center of this circle being the center of valve wheel 8. For the six-joint robotic arm 1, the DH parameter table for each joint is first established: The coordinate transformation matrix of adjacent joints is: in, Joint angle, The length of the link. For the link twist angle, This is the link offset.

[0054] The total transformation matrix from the base to the end is: in, This is the total transformation matrix of the robotic arm from the base to the end effector. Indicates the first The joint relative to the first The coordinate transformation matrix of each joint.

[0055] The Pieper method is used. First, the first three joint angles are calculated based on the target position. Then, the last three joint angles are calculated based on the target posture and the calculated first three joint angles. Multiple sets of inverse kinematic solutions are obtained using geometric or algebraic methods. The solution with the minimum sum of joint motions is selected as the optimal solution. in, For the currently solved... Each joint angle, For the angle values ​​of each joint at the previous moment, select the solution with the minimum total motion to reduce energy consumption and wear.

[0056] Based on the optimal solution, linear interpolation is used to generate a sequence of intermediate points between the starting point and the ending point. Let the starting point be... The destination is The number of interpolation steps is , No. The interpolation points are: Each interpolation point is converted into a joint angle through inverse kinematics, driving the robotic arm 1 to reach each point sequentially.

[0057] And a fifth-degree polynomial is used: in, , , , , , The coefficients of the fifth-degree polynomial are undetermined and are determined by the boundary conditions.

[0058] Substituting the boundary conditions, we get: The joint angle at the start of the movement; The joint angle at the end of the movement; The start time; This is the end time.

[0059] The initial velocity and acceleration are both 0, and the final velocity and acceleration are both 0. Substituting these into the boundary conditions, the coefficients are determined according to the following formula: In this embodiment, both the initial velocity and acceleration are 0, and both the final velocity and acceleration are 0. The initial time is... =0, termination time =1s, meaning the preset running time is 1s, according to the control cycle. The trajectory is generated point by point after discretization.

[0060] In step S22, the remote control terminal issues a positioning command, and the actuator moves towards the valve wheel 8 according to the planned motion path. The docking and locking unit includes a chuck 3 (three-jaw or five-jaw chuck 3), which is located at the end of the actuator. Each jaw is designed to wrap around the spokes of the valve wheel 8 on three sides. The chuck 3 adopts a combination of three-jaw and five-jaw chucks. When the valve wheel 8 is detected to be three-spoked, the three-jaw mode is activated; when it is five-spoked, the five-jaw mode is activated.

[0061] A contact switch 6 is located at the center of the chuck 3 to detect whether it is in contact with the valve wheel 8. When the actuator carries the chuck 3 close to the valve wheel 8, the end of the robotic arm 1 advances, causing the jaws of the chuck 3 to wrap around the spokes of the valve wheel 8 from three directions. When the chuck 3 is in contact with the valve wheel 8, the contact switch 6 is triggered, at which point the cylinder 7 extends. Subsequently, the cylinder 7 resets, driving the bracket of the contact switch 6 and the contact switch 6 to move towards the valve end, preventing the cable at the rear end of the contact switch 6 from being dragged or tangled during the rotation of the chuck 3. Then, it is determined that the docking locking unit has made contact with the valve wheel 8, and the mechanical locking between the chuck 3 and the spokes of the valve wheel 8 is performed. An additional locking mechanism has been eliminated on the chuck 3; the locking is completed after the chuck 3 moves into position by the motor of each joint of the robotic arm 1 holding the brake. After locking, a signal indicating that the chuck is in position is fed back to the remote control terminal.

[0062] Further, in an optional embodiment, the docking locking unit includes a chuck 3, which is disposed at the end of the actuator. Step S21 includes: Step S211: Generate a chuck 3 rotation control command according to the number of spokes of the valve wheel 8 to control the rotation of the chuck 3 so that the pawl of the chuck 3 is aligned with the spokes of the valve wheel 8 in spatial orientation; Step S212: Obtain the preset kinematic parameters of the actuator, take the center position of the valve wheel 8 as the target position, and take the parallelism between the axial direction of the actuator end and the normal vector of the plane of the valve wheel 8 as the target posture, use inverse kinematics to separate the position solution and the posture solution, calculate multiple sets of inverse kinematic solutions, and select the solution with the minimum sum of motion of each joint as the optimal solution; Step S213: Based on the optimal solution, use a linear interpolation algorithm to calculate and generate an intermediate point sequence between the current position and the target position, and use a polynomial interpolation method for velocity planning to calculate and generate the motion path of the actuator from the current position to the target position.

[0063] Specifically, in step S211, when the valve wheel 8 is identified as having three spokes, the three-jaw mode is invoked; when the valve wheel 8 is identified as having five spokes, the five-jaw mode is invoked. At the same time, after confirming the angular deviation between the current jaw of the chuck 3 and the spokes of the valve wheel 8, the end motor of the robotic arm 1 drives the chuck 3 to rotate by the corresponding angle to eliminate the angular deviation, so that the jaw of the chuck 3 and the spokes of the valve wheel 8 are precisely aligned in spatial orientation.

[0064] In step S212, the kinematic parameters of the actuator include the DH parameters of each joint. The inverse kinematics solution uses the Pieper method, first calculating the angles of the first three joints based on the target position. The last three joint angles are calculated based on the target posture and the previously calculated first three joint angles. Multiple sets of inverse kinematic solutions are obtained through geometric or algebraic methods. The solution with the minimum sum of joint motions is selected as the optimal solution. in, This represents the i-th joint angle currently being solved; For the angle values ​​of each joint at the previous moment, select the solution with the minimum total motion to reduce energy consumption and wear.

[0065] In step S213, the linear interpolation algorithm generates a sequence of intermediate points between the current position and the target position, and the fifth-order polynomial velocity planning makes the position, velocity and acceleration of each joint change continuously and smoothly, eliminating motion impact.

[0066] Further, in an optional embodiment, the execution component is a multi-joint robotic arm 1. In step S212, inverse kinematics is used to separate the position solution and the attitude solution to calculate multiple sets of inverse kinematic solutions, including: calculating the motion angles of the first three joints of the multi-joint robotic arm 1 based on the target position to determine the position of the end of the multi-joint robotic arm 1; calculating the motion angles of the last three joints of the multi-joint robotic arm 1 based on the target attitude and the calculated motion angles of the first three joints to determine the attitude of the end of the multi-joint robotic arm 1; and combining the motion angles of the first three joints and the last three joints to obtain multiple sets of inverse kinematic solutions.

[0067] Specifically, for the six-joint robotic arm 1, the inverse kinematics solution essentially involves decomposing the end-effector pose into two parts: position and orientation, which are solved separately. In the position solution stage, based on the aforementioned homogeneous transformation matrix of the target pose... It can extract the target location. and target posture In the location determination stage, based on the target location... The first three joint angles can be solved using geometric or algebraic methods. These three joints determine the position of the robotic arm's end effector in space. During the attitude determination phase, based on the target attitude: in, Indicates the first A rotation matrix for each joint axis, which is a matrix about the joint angles. Functions used to describe joints Coordinate system relative to joint The rotational transformation of the coordinate system can be derived from the homogeneous transformation matrix. Extracted from This represents the target attitude matrix of the robotic arm's end effector relative to the base coordinate system. Using the previously calculated... Solve for the last three joint angles These three joints determine the orientation of the robotic arm's end effector. By combining the motion angles of the first three joints with those of the last three joints, multiple inverse kinematic solutions can be obtained. The solution with the minimum sum of motion parameters of each joint is selected as the optimal solution. This method decomposes the six-dimensional problem into two three-dimensional problems, reducing computational complexity and improving solution efficiency.

[0068] In step S3 of controlling the opening and closing of the valve, the operator selects the control mode on the remote control terminal: one-button switch mode (automatic rotation according to the preset opening degree) or continuous fine-tuning mode (custom number of rotations or angle). The remote control terminal issues a rotation command according to the selected mode, and the drive unit (motor + reducer) rotates at the set speed and safety torque limit. The encoder provides real-time feedback on the rotation position, driving the locked actuator and valve wheel 8 to rotate, thus performing the opening or closing operation.

[0069] During operation, the torque monitoring module in the safety protection components (which can be directly measured by a torque sensor or calculated by detecting the current and speed of the drive motor) monitors the output torque in real time. When the torque exceeds a preset threshold, the system immediately stops, rotates in the opposite direction 1 / 4 turn to relieve the force, and then restarts the rotation at 50% speed. If the torque still exceeds the limit after restarting, the valve is determined to be stuck, and the system automatically jumps to the unlocking and disengagement step and reports a fault. In emergency situations where disengagement is not possible, the system remains clamped and issues the highest-level alarm, requiring manual intervention.

[0070] After the operation is completed, the unlocking and disengagement steps are executed. The remote control terminal issues an unlocking command, and the locking mechanism on the docking locking unit is released (mechanically released). The robotic arm 1 returns to the preset safe position according to the planned path. If normal unlocking fails (e.g., the locking mechanism is stuck), the emergency procedure of the safety protection component is triggered: the communication fault monitoring module detects the abnormality, automatically executes forced unlocking (e.g., cutting off the power source, activating the backup release mechanism), and records the fault.

[0071] Finally, the remote control terminal automatically records all parameters of this operation, including valve wheel number 8, operation time, operator, number of opening and closing revolutions or opening degree, maximum torque, abnormal events, etc., generates an electronic operation log and stores it. The system resets and waits for the next instruction.

[0072] In the above embodiments, the signal transmission component adopts a dual-redundant communication scheme of wired and wireless. In areas where wiring is feasible, wired communication (such as industrial Ethernet, RS-485) is used first to ensure stable and reliable transmission and strong anti-interference capabilities. In scenarios where wiring is not possible or temporary deployment is required, wireless communication (such as Wi-Fi, LoRa, 5G private network) is switched to. The two methods can be automatically switched according to the on-site operating conditions or manually selected by the remote control terminal to ensure uninterrupted transmission of commands and data in the complex electromagnetic environment of nuclear power plants.

[0073] The safety protection components include a torque monitoring module and a communication fault monitoring module, which monitor the drive torque and communication status in real time. In the event of a sudden power failure or communication interruption, the unlocking and disengagement process is automatically triggered. When the torque exceeds the limit, the machine is automatically stopped to avoid damage to the valve and ensure operational safety.

[0074] This invention integrates multi-source sensing technologies, including image recognition and laser ranging, to automatically identify the characteristic parameters and spatial pose of the valve wheel 8. Combined with an automatic docking, locking, and unlocking mechanism, it achieves adaptive adaptation to valve wheels 8 of different specifications and fully unmanned remote control. Regular valve opening and closing, as well as rapid emergency intervention, can be completed without personnel entering high-irradiation areas, fundamentally avoiding the risk of radiation exposure to personnel and significantly shortening emergency response time. It also solves the technical problems of poor versatility and low automation in existing methods, meeting the dual needs of safe operation and maintenance and emergency response in high-irradiation scenarios of nuclear power plants, and providing reliable technical support for intelligent and minimally manned operation and maintenance of nuclear power plants.

[0075] It is understood that the above embodiments only illustrate preferred embodiments of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can freely combine the above technical features without departing from the concept of the present invention, and can also make several modifications and improvements, all of which fall within the protection scope of the present invention. Therefore, all equivalent transformations and modifications made with respect to the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A remote control method for nuclear power plant valves, applied to valve control devices, characterized in that, The valve control device includes an actuating component, and the end of the actuating component is provided with a docking locking unit. The method includes the following steps: Step S1: Acquire image information and distance information of the valve wheel, and identify the feature parameters and spatial pose of the valve wheel based on the image information and distance information; Step S2: Based on the characteristic parameters and the spatial pose, control the execution component to move to the valve wheel and mechanically dock and lock it with the valve wheel through the docking and locking unit; Step S3: Drive the locked actuator to move to control the valve wheel to perform a preset operation, and release the lock and control the actuator to exit after the operation is completed.

2. The method for remote control of nuclear power plant valves according to claim 1, characterized in that, The valve control device also includes an industrial camera and multiple laser rangefinders disposed at the end of the actuator. In step S1, acquiring the image information and distance information of the valve wheel includes: The industrial camera acquires image information of the valve wheel; The distance information of the valve wheel surface is acquired by the laser rangefinder; The image information is normalized to scale the image to a uniform pixel size; The coordinates of each ranging point in the collected distance information are transformed to the camera coordinate system, and the coordinates of each ranging point are unified to the same reference coordinate system through a homogeneous transformation matrix.

3. The method for remote control of nuclear power plant valves according to claim 1, characterized in that, The image information includes images of the spoke edges, outer contour, and center hole of the valve wheel; the distance information includes the spatial coordinates of multiple distance measuring points on the valve wheel surface; the feature parameters include the number of spokes, outer diameter, and center hole diameter of the valve wheel; and the spatial pose includes the mounting tilt angle and spoke tilt angle of the valve wheel plane. In step S1, identifying the feature parameters and spatial pose of the valve wheel based on the image information and the distance information includes: The number of spokes is obtained by counting the number of spokes based on the spoke edges; The outer diameter is calculated based on the pixel size of the outer contour and the distance information; The center hole of the valve wheel is detected based on the center hole image, and the center hole diameter is calculated based on the pixel size of the center hole and the distance information. The valve wheel plane is fitted based on the spatial coordinates of each ranging point, and the angle between the normal vector of the valve wheel plane and the axial direction of the end of the actuator is calculated to obtain the installation tilt angle. The spoke tilt angle is obtained by fitting the height variation of each ranging point relative to the valve wheel plane.

4. The method for remote control of nuclear power plant valves according to claim 3, characterized in that, The step of calculating the outer diameter based on the pixel size of the outer contour and the distance information includes: Obtain the pixel diameter of the outer contour Combined with camera pixel size Lens focal length and object distance The outer diameter was calculated using a pinhole imaging model. The calculation formula is: Wherein, the object distance The tilt angle is determined based on the distance information collected from three laser ranging points, and corrected according to the following formula: in, This is the average of the distance measurements from the three laser ranging points. The installation tilt angle is as described.

5. The method for remote control of nuclear power plant valves according to claim 3, characterized in that, The step of fitting the valve wheel plane based on the spatial coordinates of each ranging point, calculating the angle between the normal vector of the valve wheel plane and the axial direction of the end of the actuator, and obtaining the installation tilt angle includes: Based on the spatial coordinates of three non-collinear distance measuring points , , Construct direction vector , Through cross product The normal vector of the valve wheel plane is obtained, and the three coordinate components of the normal vector are ( , , ), calculate the angle between the normal vector and the axial direction of the end of the actuator, and obtain the mounting tilt angle. The calculation formula is: in, , , The normal vectors are respectively The three coordinate components.

6. The method for remote control of nuclear power plant valves according to claim 3, characterized in that, The step of fitting the spoke tilt angle based on the height variation of each ranging point relative to the valve wheel plane includes: Control a ranging point to move radially along the valve wheel spokes and continuously collect the spatial coordinates of multiple ranging points; The slope value is obtained by linear fitting based on the height and radial position of each measuring point relative to the valve wheel plane. The spoke tilt angle is calculated based on the slope value.

7. The method for remote control of nuclear power plant valves according to claim 3, characterized in that, Step S1 further includes: A feature vector is constructed consisting of the number of spokes, the outer diameter, the center aperture, and the spoke tilt angle; and each feature in the feature vector is normalized. The K-nearest neighbor algorithm is used to calculate the weighted Euclidean distance between the normalized feature vector and each template in the preset valve wheel feature library. A preset number of templates with the smallest distance are selected, and the matching results and confidence scores are output by majority voting. When the confidence level is greater than or equal to the preset threshold, the identification is deemed successful, and the valve wheel number of the successful match and the corresponding preset valve parameters are output.

8. The method for remote control of nuclear power plant valves according to claim 7, characterized in that, Step S2 includes: Step S21: Based on the valve preset parameters, the number of spokes, the outer diameter, the center bore diameter, the installation tilt angle, and the spoke tilt angle, determine the docking posture and docking parameters of the docking locking unit, and plan the movement path of the actuator from the current position to the valve wheel; Step S22: Control the execution component to move according to the motion path, so that the docking locking unit contacts the valve wheel; when the docking locking unit contacts the valve wheel, perform mechanical locking between the docking locking unit and the valve wheel.

9. The method for remote control of nuclear power plant valves according to claim 8, characterized in that, The docking locking unit includes a chuck, which is disposed at the end of the actuating component. Step S21 includes: Step S211: Generate a chuck rotation control command based on the number of spokes of the valve wheel to control the rotation of the chuck so that the chuck's claws are aligned with the spokes of the valve wheel in spatial orientation; Step S212: Obtain the preset kinematic parameters of the actuator, take the center position of the valve wheel as the target position and the parallelism between the axial direction of the actuator end and the normal vector of the valve wheel plane as the target attitude, use inverse kinematics to separate the position solution and attitude solution, calculate multiple sets of inverse kinematic solutions, and select the solution with the minimum sum of motion of each joint as the optimal solution; Step S213: Based on the optimal solution, a sequence of intermediate points is generated between the current position and the target position using a linear interpolation algorithm, and a velocity planning is performed using a polynomial interpolation method to calculate and generate the motion path of the execution component from the current position to the target position.

10. The method for remote control of nuclear power plant valves according to claim 9, characterized in that, The actuator is a multi-joint robotic arm. In step S212, the inverse kinematics is used to separate the position and attitude solutions, and multiple sets of inverse kinematics solutions are calculated, including: Calculate the motion angles of the first three joints of the multi-joint robotic arm based on the target position to determine the position of the end effector of the multi-joint robotic arm. Based on the target posture and the calculated motion angles of the first three joints, the motion angles of the last three joints of the multi-joint robotic arm are calculated to determine the posture of the end effector of the multi-joint robotic arm. By combining the motion angles of the first three joints with those of the last three joints, multiple sets of inverse kinematic solutions are obtained.

Images