A method for detecting and correcting deviation of forked ear type wing-body joint assembly

Abstract

The application discloses a kind of for fork ear type aircraft wing-body docking assembly deviation detection and correction method, comprising the following steps: S100: build multi-view camera detection system;S200: using multi-view camera each coordinate system conversion principle completes stereo vision calibration, obtains multi-view camera internal and external parameters;S300: construct fork ear feature point coordinate detection model, solve fork ear position deviation;S400: using space elliptical view cone principle constructs fork ear space posture detection model, solves fork ear space posture deviation;S500: construct wing posture adjustment mathematical model, solve wing posture adjustment control point theoretical coordinate, obtain wing posture adjustment state equation.S600: fusion multi-view camera and laser tracker detection data, using local neighborhood adaptive artificial fish school optimization particle filtering carries out deviation correction.The application builds multi-view camera detection system and obtains the space position and space posture deviation of fork ear;Using LAFSA-PF algorithm effectively corrects the wing position and posture fine adjustment deviation caused by the longer detection time of multi-view camera.

Description

Technical Field

[0001] This invention relates to machine vision technology and aircraft digital assembly technology, and in particular to a method for detecting and correcting deviations in the assembly of wing-body joints of fork-type aircraft. Background Technology

[0002] With the continuous development of image processing and machine vision technology, visual recognition technology has been widely used in various fields such as industrial production and daily life. Furthermore, the detection accuracy of visual measurement technology has reached the micrometer level, and the processing speed has exceeded 100 frames / s. As a result, high-precision aircraft docking and assembly measurement methods based on industrial cameras have begun to be widely used.

[0003] However, using a single camera for fork ear feature recognition during the alignment stage can lead to the problem that deviations on a certain alignment end face of the fork ear cannot be detected, resulting in reduced alignment accuracy and efficiency. While using two cameras to perform feature recognition on the two alignment end faces of the fork ear separately solves the problem of undetectable deviations on a certain alignment end face, there is still ambiguity in the detection of alignment position and posture deviations of the fork ear, leading to poor detection results and reduced alignment accuracy.

[0004] Based on the above analysis, this invention uses machine vision inspection technology and takes the fork-ear wing-body docking assembly as the research object. After completing the multi-view camera stereo vision calibration, a fork-ear docking pose deviation solution model is established. An elliptical space view cone is used to construct a fork-ear spatial attitude deviation solution model. Subsequently, an adaptive fusion algorithm is used to fuse the detection data of the multi-view camera system and the laser tracker. A local neighborhood adaptive artificial fish swarm optimization particle filter algorithm is proposed to track and predict the coordinates of the wing attitude control point to correct the assembly deviation. Summary of the Invention

[0005] The purpose of this invention is to solve the technical problems existing in the prior art and to provide a method for detecting and correcting the misalignment quality of the fork ears of a multi-view camera.

[0006] To achieve the above objectives, the technical solution provided by this invention is: a method for detecting and correcting assembly deviations in wing-body docking of fork-type aircraft, comprising the following steps: S100: A laser tracker is used to assist in the horizontal adjustment of the fuselage and the adjustment of the wing position. A multi-camera detection system is built according to the requirements for the mating quality of the fork lugs. The multi-camera detection system adopts a triangular pattern distribution, including a vertical eye camera, a left eye camera, and a right eye camera (the angular position parameters between each camera are obtained through stereo calibration). S200: Utilizes the coordinate system transformation principle of multi-view cameras to complete stereo vision calibration and solve for the intrinsic and extrinsic parameters of each camera; S300: Construct a model for detecting the coordinates of the fork ear feature points, obtain the spatial coordinates of the fork ear feature points, and solve for the deviation of the fork ear alignment position. S400: Using the principle of spatial elliptical cone of vision, a spatial pose state detection model for the fork ear is constructed to solve the spatial pose deviation of the fork ear; S500: Establish a mathematical model for wing attitude adjustment, solve for the theoretical coordinates of the wing attitude adjustment control points, and construct the wing attitude adjustment state equation; S600: The improved Artificial Fish Swarm Particle Filter (LAFSA-PF) algorithm is used to track and estimate the trajectory of the fork ear feature points. An adaptive weighted fusion algorithm is introduced to fuse the detection data of multi-view cameras and laser trackers to build a wing attitude fine-tuning deviation correction model and complete the deviation correction.

[0007] Further, the specific method for building the multi-camera detection system in step S100 is as follows: the mid-fuselage is horizontally adjusted and its position is locked using a laser tracker, and then the wing position is adjusted so that the wing-end lug connectors enter the camera recognition and detection range; the second stage is to fix the camera to the fuselage position using a bracket according to the detection requirements of the position deviation and attitude angle deviation of the lugs, and set up the three cameras in a triangular distribution, with the vertical eye camera set up in the vertical direction of the lugs, and the left and right eye cameras set up in the direction of the intersection hole of the lugs, to collect and obtain the target feature information of the gap end of the lugs and the intersection hole end of the lugs.

[0008] Furthermore, the specific method for stereo vision calibration in step S200 is as follows: S201: Set the coordinate systems for the vertical eye and the left and right eye cameras as follows: , , Set up points Let point be any point in the world coordinate system of the multi-view camera system. The projected coordinates in each camera coordinate system are as follows: , and The homogeneous coordinates of the imaging points in each camera image coordinate system are as follows: , and Utilizing the results Solve for the corresponding camera coordinates from the image coordinates: ; In the formula These represent the focal lengths of each camera.

[0009] S202: Rotation and translation transformations between camera coordinate systems can be performed using homogeneous transformation matrices. To represent, therefore, we can derive the point. Relationships between different camera coordinate systems: ; In the formula These represent the rotation matrix and translation vector between the vertical, left, and right cameras, respectively.

[0010] The aforementioned camera imaging model is an ideal imaging model free from external interference. However, various errors can occur during the manufacturing and installation of the camera, causing various nonlinear distortions that can affect detection accuracy in practical applications. Therefore, camera calibration considers the nonlinear distortion factors of the camera and corrects the ideal camera imaging model to obtain more accurate calibration results. Considering lens distortion, there is an offset between the actual imaging point and the theoretical imaging point. Offset in direction , It can be expressed by the following formula: ; In the formula To input distorted image coordinates into the camera, and These are the radial and longitudinal distortion coefficients, respectively.

[0011] S203: After obtaining multiple sets of calibration board images taken by cameras at different locations, and after grayscale preprocessing, each circular pattern is segmented and then... The matrix is ​​used for sub-pixel edge extraction and circle center feature calculation.

[0012] Finally, by combining Zhang Youzheng's calibration method, stereo calibration of a multi-camera system is achieved.

[0013] Furthermore, the specific method for solving the fork lug mating position deviation in step S300 is as follows: S301: After completing the stereo vision calibration, the following can be obtained by combining the left and right eye cameras: ; The homogeneous transformation matrix between the left-eye camera and the world coordinate system of the multi-eye camera system is: ; In the formula , Representing the two coordinate systems around axis, Axial rotation angle; the point can be calculated. In the three-dimensional coordinates of the world coordinate system of the multi-camera system : ; In a multi-camera system, the world coordinates can be obtained by combining any two cameras using the above method. .

[0014] S302: The homogeneous transformation matrix between the multi-camera world coordinate system and the assembly global coordinate system is: ; In the formula Representing the two coordinate systems respectively axis, axis, Axial rotation angle. Coordinate transformation relationships can be used to further solve for the spatial coordinates in the assembly global coordinate system. .

[0015] ; S303: Obtained by solving the above formula Coordinates of the center of the intersection hole of the fork-type connector and the lug-type connector in the assembly global coordinate system , Due to the pose lock of the fork-type connector, the coordinates... For a constant value, take for Therefore, the axis of the intersection hole of the fork-shaped joint is used as the reference axis to solve the problem of the end face of the intersection hole of the fork lug. Coaxiality deviation at time : ; Solvable Spatial position deviation of the center of the intersection hole of the ear-type connector : ; In the formula The target position coordinates of the mating holes of the ear-type connector are given, and the spatial position deviation of the intersecting hole center of the ear-type connector is represented by the spatial position deviation of the ear mating position. .

[0016] Furthermore, the specific method for solving the spatial attitude deviation of the fork ear in step S400 is as follows: S401: Using the left and right eye camera combination described in step S300, observe the intersection aperture of the fork simultaneously, and the optical centers of the two cameras... , The projection of the fork-shaped aperture into the image coordinate system and the fork-shaped aperture together form two spatial elliptical view cones. To simplify the equation of the spatial elliptical view cones, the coordinate systems of the left and right cameras are transformed using a transformation matrix. After rotation, an elliptical cone coordinate system is obtained. .

[0017] Combining formula (1), we can obtain the equation of the spatial elliptical cone: ; Representing the above equation using a matrix: ; because , is a real symmetric matrix, and can be obtained through orthogonal matrices. Transform it into a diagonal matrix: ; Rewritten in standard form: ; S402: The left and right cameras can each calculate two different sets of normal vectors in the spatial elliptical view cone coordinate system: ; Then the normal vectors in the world coordinate system are: ; The true solution for the intersection of the fork lugs can be obtained by the following formula. : ; In the formula The angle between the two vectors, It is the minimum threshold of the vector angle, and its normal vector in the assembly global coordinate system can be obtained by combining it with S203. .

[0018] S403: Due to the aircraft fuselage attitude lock, the fork-type connector's attitude remains fixed. Normal vectors of fork-shaped and ear-shaped connectors in the global coordinate system , The solution can be obtained. Rotation angle between fork-type connectors and lug-type connectors : ; Two-vector rotation axis : ; The solution can be obtained using Rodriguez's formula. Rotary matrix of fork-type and ear-type connectors : ; In the formula Given a 3x3 identity matrix, the yaw angle of the fork-ear coupling is obtained using the rotation matrix of the fork-type and ear-type connectors. Pitch angle Roll angle .

[0019] Furthermore, the specific method for constructing the wing attitude adjustment state equation in step S500 is as follows: S501: Set the initial wing attitude as follows: After attitude adjustment, the wing attitude is as follows: Record the wing attitude adjustment time as .

[0020] S502: Let the trajectory prediction equation be... : ; Substituting the constraints, we get: ; S503: Order We can obtain: ; Using the solution Using the same method, we can obtain the prediction equation for the attitude adjustment trajectory of the fore-wing: ; It can be obtained Timing fork ear feature points, wing attitude control points The theoretical coordinates are: ; In the formula , For point In the wing coordinate system, .

[0021] S504: Establishing the wing attitude adjustment state equations: ; In the formula ; ; It is a 3x3 noise matrix; ; for Dynamic noise at any given moment.

[0022] Furthermore, the specific methods for correcting assembly quality dynamic detection deviations and verifying the measurement accuracy of the multi-view camera inspection system using LAFSA-PF and data fusion in step S600 are as follows: S601: Assembly quality dynamic detection deviation correction using LAFSA-PF and data fusion. First, in the initial stage of fork lug engagement, importance sampling of wing attitude control points is obtained using the wing attitude adjustment state equation. Particles As the initial sample, the importance weight of each particle at the wing attitude control point is calculated; S602: Constructing the objective function: ; In the formula, For the crowding factor, This is the latest measurement value of the wing attitude control point. To predict measured values, a local neighborhood, adaptive sensing distance, and step size optimization algorithm for artificial fish swarms are introduced.

[0023] S603: Optimize particle filtering using a local neighborhood adaptive artificial fish swarm algorithm, and calculate the optimized sampled result. The weights of the particles at each wing attitude control point are determined and normalized.

[0024] ; S604: Regarding the wing attitude control points... Estimate the state at time: ; And output the estimated value; S605: Perform comprehensive deviation correction: First, it is necessary to implement data fusion bias correction and construct a joint linear bias data fusion model for the camera and laser tracker: ; In the formula These represent the observation matrices of each camera and laser tracker, respectively; This represents the estimated position deviation of the wing attitude control point; Each device represents the state of the device. The measurement noise vector at time step. An adaptive weighted fusion algorithm is used to optimize the fused data from the camera and laser tracker to obtain the unfiltered optimized time step. The best estimate of the wing attitude control point coordinates at any given time. Then, the positioner attitude adjustment deviation is corrected, and the wing attitude adjustment state equation is solved to obtain... The wing pose estimation at any given time is obtained by comparing it with the theoretical pose. Wing position deviation ;pass By combining inverse kinematics, we can obtain Positioner pose adjustment deviation correction value generated within a time period .

[0025] Beneficial effects of this invention: This invention establishes a multi-view camera detection system to obtain the spatial position and spatial attitude deviation of the fork lugs; it solves the ambiguity problem of fork lug intersection hole pose measurement by using the spatial elliptical cone detection method; and it effectively corrects the wing pose fine-tuning deviation caused by the long detection time of the multi-view camera by using the LAFSA-PF algorithm, thereby improving the alignment accuracy. Attached Figure Description

[0026] The accompanying drawings, which are provided to further illustrate the invention and constitute a part of this invention, are illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention.

[0027] Figure 1 This is a flowchart of a method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the overall system for detecting the quality of the fork-ear mating assembly of the present invention; Figure 3 This is the specific content of the quality inspection requirements for the mating assembly of the fork lugs in this invention; Figure 4 This is a layout diagram of the multi-view camera used in this invention to monitor the engagement process of the fork lugs; Figure 5 This is a schematic diagram showing the relationship between the coordinate system of the multi-view camera system and the spatial point position coordinates of the present invention; Figure 6 This is a flowchart of the deviation correction process for LAFSA-PF and data fusion in this invention. Detailed Implementation

[0028] This section will describe in detail specific embodiments of the present invention. Preferred embodiments of the present invention are shown in the accompanying drawings. The purpose of the drawings is to supplement the textual description with graphics, so that people can intuitively and vividly understand each technical feature and overall technical solution of the present invention, but they should not be construed as limiting the scope of protection of the present invention.

[0029] This invention solves the problem that when using two cameras to monitor the fork lug mating stage during the automatic wing-body docking process of an aircraft, it is impossible to accurately obtain the spatial position of the fork lug feature points and the outline features of the fork lug intersection hole, making it difficult to fully describe the position and attitude deviation of the fork lug, which leads to a decrease in the assembly accuracy of the wing-body docking. This invention provides a new method for quality inspection of fork lug type aircraft wing-body docking.

[0030] like Figures 1-6 As shown, in accordance with the requirements for fork-lug mating quality inspection, a method for detecting and correcting assembly deviations in wing-body docking of fork-lug aircraft is proposed. This method aims to improve the mating quality of the fork-lugs and includes the following steps: S100: A laser tracker is used to assist in the horizontal adjustment of the fuselage and the adjustment of the wing position. A multi-camera detection system is built according to the requirements for the mating quality of the fork lugs. The multi-camera detection system adopts a triangular pattern distribution, including a vertical eye camera, a left eye camera, and a right eye camera (the angular position parameters between each camera are obtained through stereo calibration). S200: Utilizes the coordinate system transformation principle of multi-view cameras to complete stereo vision calibration and solve for the intrinsic and extrinsic parameters of each camera; S300: Construct a model for detecting the coordinates of the fork ear feature points, obtain the spatial coordinates of the fork ear feature points, and solve for the deviation of the fork ear alignment position. S400: Using the principle of spatial elliptical cone of vision, a spatial pose state detection model for the fork ear is constructed to solve the spatial pose deviation of the fork ear; S500: Establish a mathematical model for wing attitude adjustment, solve for the theoretical coordinates of the wing attitude adjustment control points, and construct the wing attitude adjustment state equation; S600: The improved Artificial Fish Swarm Particle Filter (LAFSA-PF) algorithm is used to track and estimate the trajectory of the fork ear feature points. An adaptive weighted fusion algorithm is introduced to fuse the detection data of multi-view cameras and laser trackers to build a wing attitude fine-tuning deviation correction model and complete the deviation correction.

[0031] In this example, the specific method for building the multi-view camera detection system in step S100 is as follows: S100: The mid-fuselage horizontal attitude is adjusted and locked using a laser tracker, followed by wing attitude adjustment to bring the wing-end lug connector into the camera's recognition and detection range; the second stage is to fix the camera to the fuselage position using a bracket according to the detection requirements of the position deviation and attitude angle deviation of the lugs, and set up three cameras in a triangular distribution, with the vertical eye camera set up in the vertical direction of the lugs, and the left and right eye cameras set up in the direction of the intersection hole of the lugs, to collect and obtain target feature information of the gap end of the lugs and the intersection hole end of the lugs.

[0032] In this example, the specific method for stereo vision calibration in step S200 is as follows: S201: Set the coordinate systems for the vertical eye and the left and right eye cameras as follows: , , Set up points Let point be any point in the world coordinate system of the multi-view camera system. The projected coordinates in each camera coordinate system are as follows: , and The homogeneous coordinates of the imaging points in each camera image coordinate system are as follows: , and Utilizing the results Solve for the corresponding camera coordinates from the image coordinates: ; In the formula These represent the focal lengths of each camera.

[0033] S202: Rotation and translation transformations between camera coordinate systems can be performed using homogeneous transformation matrices. To represent, therefore, we can derive the point. Relationships between different camera coordinate systems: ; In the formula These represent the rotation matrix and translation vector between the vertical, left, and right cameras, respectively.

[0034] The camera imaging model described above is an ideal imaging model free from external interference. However, various errors can occur during the manufacturing and installation of the camera, causing various nonlinear distortions that can affect detection accuracy in practical applications. Therefore, the nonlinear distortion factors of the camera should be considered during camera calibration, and the ideal camera imaging model should be corrected to obtain more accurate calibration results. Considering lens distortion, there is a certain offset between the actual imaging point and the theoretical imaging point. Offset in direction , It can be expressed by the following formula: ; In the formula To input distorted image coordinates into the camera, and These are the radial and longitudinal distortion coefficients, respectively.

[0035] S203: After obtaining multiple sets of calibration board images taken by cameras at different locations, and after grayscale preprocessing, each circular pattern is segmented and then... The matrix is ​​used for sub-pixel edge extraction and circle center feature calculation. Finally, combined with Zhang Youzheng's calibration method, stereo calibration of the multi-camera system is achieved.

[0036] In this example, the specific method for solving the fork lug mating position deviation in step S300 is as follows: S301: After completing the stereo vision calibration, the following can be obtained by combining the left and right eye cameras: ; The homogeneous transformation matrix between the left-eye camera and the world coordinate system of the multi-eye camera system is: ; In the formula , Representing the two coordinate systems around axis, Rotation angle along the axis. The point can be calculated. In the three-dimensional coordinates of the world coordinate system of the multi-camera system : ; In a multi-camera system, the world coordinates can be obtained by combining any two cameras using the above method. .

[0037] S302: The homogeneous transformation matrix between the multi-camera world coordinate system and the assembly global coordinate system is: ; In the formula Representing the two coordinate systems respectively axis, axis, Axial rotation angle. Coordinate transformation relationships can be used to further solve for the spatial coordinates in the assembly global coordinate system. .

[0038] ; S303: Obtained by solving the above formula Coordinates of the center of the intersection hole of the fork-type connector and the lug-type connector in the assembly global coordinate system , Due to the pose lock of the fork-type connector, the coordinates... For a constant value, take for Therefore, the axis of the intersection hole of the fork-shaped joint is used as the reference axis to solve the problem of the end face of the intersection hole of the fork lug. Coaxiality deviation at time : ; Solvable Spatial position deviation of the center of the intersection hole of the ear-type connector : ; In the formula The target position coordinates of the mating holes of the ear-type connector are given, and the spatial position deviation of the intersecting hole center of the ear-type connector is represented by the spatial position deviation of the ear mating position. .

[0039] In this example, the specific method for solving the spatial attitude deviation of the fork ear in step S400 is as follows: S401: Using the left and right eye camera combination described in step S300, observe the intersection aperture of the fork simultaneously, and the optical centers of the two cameras... , The projection of the fork-shaped aperture into the image coordinate system and the fork-shaped aperture together form two spatial elliptical view cones. To simplify the equation of the spatial elliptical view cones, the coordinate systems of the left and right cameras are transformed using a transformation matrix. After rotation, an elliptical cone coordinate system is obtained. .

[0040] Combining formula (1), we can obtain the equation of the spatial elliptical cone: ; Represented in matrix form as follows: ; because , is a real symmetric matrix, and can be obtained through orthogonal matrices. Transform it into a diagonal matrix: ; Convert to standard form: ; S402: The left and right cameras can each calculate two different sets of normal vectors in the spatial elliptical view cone coordinate system: ; Then the normal vectors in the world coordinate system are: ; The true solution for the intersection of the fork lugs can be obtained by the following formula. : ; In the formula The angle between the two vectors, It is the minimum threshold of the vector angle, and its normal vector in the assembly global coordinate system can be obtained by combining it with S203. .

[0041] S403: Due to the aircraft fuselage attitude lock, the fork-type connector's attitude remains fixed. Normal vectors of fork-shaped and ear-shaped connectors in the global coordinate system , The solution can be obtained. Rotation angle between fork-type connectors and lug-type connectors : ; Two-vector rotation axis : ; The solution can be obtained using Rodriguez's formula. Rotary matrix of fork-type and ear-type connectors : ; In the formula Given a 3x3 identity matrix, the yaw angle of the fork-ear coupling is obtained using the rotation matrix of the fork-type and ear-type connectors. Pitch angle Roll angle .

[0042] In this example, the specific method for constructing the wing attitude adjustment state equation in step S500 is as follows: S501: Set the initial wing attitude as follows: After attitude adjustment, the wing attitude is as follows: Record the wing attitude adjustment time as .

[0043] S502: Let the trajectory prediction equation be... : ; Substituting the constraints, we get: ; S503: Order We can obtain: ; Using the solution Using the same method, we can obtain the prediction equation for the attitude adjustment trajectory of the fore-wing: ; It can be obtained Timing fork ear feature points, wing attitude control points The theoretical coordinates are: ; In the formula , For point In the wing coordinate system, .

[0044] S504: Establishing the wing attitude adjustment state equations: ; In the formula ; ; It is a 3x3 noise matrix; ; for Dynamic noise at any given moment.

[0045] In this embodiment, the specific methods for correcting assembly quality dynamic detection deviations and verifying the measurement accuracy of the multi-view camera inspection system using LAFSA-PF and data fusion in step S600 are as follows: S601: Assembly quality dynamic detection deviation correction using LAFSA-PF and data fusion. First, in the initial stage of fork lug engagement, importance sampling of wing attitude control points is obtained using the wing attitude adjustment state equation. Particles As the initial sample, the importance weight of each particle at the wing attitude control point is calculated.

[0046] S602: Constructing the objective function: ; In the formula, For the crowding factor, This is the latest measurement value of the wing attitude control point. To predict measured values, a local neighborhood, adaptive sensing distance, and step size optimization algorithm for artificial fish swarms are introduced.

[0047] S603: Optimize particle filtering using a local neighborhood adaptive artificial fish swarm algorithm, and calculate the optimized sampled result. The weights of the particles at each wing attitude control point are determined and normalized.

[0048] ; S604: Regarding the wing attitude control points... Estimate the state at time: ; And output the estimated value; S605: Perform comprehensive deviation correction: First, it is necessary to implement data fusion bias correction and construct a joint linear bias data fusion model for the camera and laser tracker: ; In the formula These represent the observation matrices of each camera and laser tracker, respectively; This represents the estimated position deviation of the wing attitude control point; Each device represents the state of the device. The measurement noise vector at time step. An adaptive weighted fusion algorithm is used to optimize the fused data from the camera and laser tracker to obtain the unfiltered optimized time step. The best estimate of the wing attitude control point coordinates at any given time. Then, the positioner attitude adjustment deviation is corrected, and the wing attitude adjustment state equation is solved to obtain... The wing pose estimation at any given time is obtained by comparing it with the theoretical pose. Wing position deviation ;pass By combining inverse kinematics, we can obtain Positioner pose adjustment deviation correction value generated within a time period .

[0049] This invention, based on multi-view vision and combining multi-view stereo vision calibration methods, spatial elliptical cone solving methods, local neighborhood adaptive artificial fish swarm optimization particle filtering, and adaptive fusion algorithms, proposes a method for detecting and correcting misalignment in wing-body docking of fork-type aircraft. This method, constrained by the quality inspection requirements of fork-type aircraft wing-body docking, constructs a multi-view camera measurement coordinate network. It can not only detect the coaxiality and mating clearance of the fork-type wing but also measure the docking position and attitude deviations. Furthermore, it can correct attitude adjustment deviations caused by the long detection time of the multi-view cameras, providing a new inspection method for fork-type wing-body docking.

[0050] This invention establishes a multi-view camera detection system to obtain the spatial position and spatial attitude deviation of the fork lugs; it solves the ambiguity problem of fork lug intersection hole pose measurement by using the spatial elliptical cone detection method; and it effectively corrects the wing pose fine-tuning deviation caused by the long detection time of the multi-view camera by using the LAFSA-PF algorithm, thereby improving the alignment accuracy.

[0051] Without causing conflict, those skilled in the art can freely combine and use the above-mentioned additional technical features.

[0052] The above description is only a preferred embodiment of the present invention. Any technical solution that achieves the purpose of the present invention by essentially the same means is within the protection scope of the present invention.

Claims

1. A method for detecting and correcting assembly deviations in wing-fuselage docking of fork-type aircraft, characterized in that: Includes the following steps: S100: A laser tracker is used to assist in the horizontal adjustment of the fuselage and the adjustment of the wing position. Based on the requirements for detecting the position deviation and attitude angle deviation of the fork lugs, a multi-view camera detection system is built. The multi-view camera detection system adopts a triangular pattern distribution, including a vertical eye camera, a left eye camera, and a right eye camera. The angular position parameters between each camera are obtained through stereo calibration. S200: Utilizes the coordinate system transformation principle of multi-view cameras to complete stereo vision calibration and solve for the intrinsic and extrinsic parameters of each camera; S300: Construct a model for detecting the coordinates of the fork ear feature points, obtain the spatial coordinates of the fork ear feature points, and solve for the deviation of the fork ear alignment position. S400: Using the principle of spatial elliptical cone of vision, a spatial pose state detection model for the fork ear is constructed to solve the spatial pose deviation of the fork ear; S500: Establish a mathematical model for wing attitude adjustment, solve for the theoretical coordinates of the wing attitude adjustment control points, and construct the wing attitude adjustment state equation; S600: The improved artificial fish swarm particle filter algorithm is used to track and estimate the trajectory of the fork ear feature points. An adaptive weighted fusion algorithm is introduced to fuse the detection data of multi-view camera and laser tracker to build a wing attitude fine-tuning deviation correction model and complete the deviation correction. The specific method for solving the spatial attitude deviation of the fork ear in step S400 is as follows: S401: Using the left and right eye camera combination described in step S300, observe the intersection aperture of the fork simultaneously, and the optical centers of the two cameras... , The projection of the fork-shaped aperture onto the image coordinate system, along with the fork-shaped aperture itself, forms two spatial elliptical view cones. To simplify the equations of these spatial elliptical view cones, the coordinate systems of the left and right cameras are transformed using a transformation matrix. After rotation, an elliptical cone coordinate system is obtained. , ; Combining formula (1), we can obtain the equation of the spatial elliptical cone: ； Convert to matrix form: ； because , is a real symmetric matrix, and can be obtained through orthogonal matrices. Transform it into a diagonal matrix: ； Convert to standard form: ； S402: The left and right cameras can each calculate two different sets of normal vectors in the spatial elliptical view cone coordinate system. ； Then the normal vectors in the world coordinate system are: ； The true solution for the intersection of the fork lugs can be obtained by the following formula. : ； In the formula The angle between the two vectors, It is the minimum threshold of the vector angle, and its normal vector in the assembly global coordinate system can be obtained by combining it with S203. ; S403: Due to the aircraft fuselage attitude lock, the fork-type connector's attitude remains fixed. Normal vectors of fork-shaped and ear-shaped connectors in the global coordinate system , The solution can be obtained. Rotation angle between fork-type connectors and lug-type connectors : ； Two-vector rotation axis : ； The solution can be obtained using Rodriguez's formula. Rotary matrix of fork-type and ear-type connectors : ； In the formula Given a 3x3 identity matrix, the yaw angle of the fork-ear coupling is obtained using the rotation matrix of the fork-type and ear-type connectors. Pitch angle Roll angle .

2. The method for detecting and correcting assembly deviations in wing-body docking of a fork-ear aircraft according to claim 1, characterized in that: The specific method for building the multi-camera detection system in step S100 is as follows: the fork-type aircraft wing-body docking process is divided into two stages. The first stage is to use a laser tracker to assist in completing the fuselage horizontal attitude adjustment and wing attitude adjustment. The second stage is to build a multi-camera detection system to meet the requirements for detecting the position deviation and attitude angle deviation of the fork-type docking.

3. The method for detecting and correcting assembly deviations in wing-body docking of a fork-type aircraft according to claim 1, characterized in that: The specific method for stereo vision calibration in step S200 is as follows: S201: Set the coordinate systems for the vertical eye, left eye camera, and right eye camera as follows: , , Set up points Let point be any point in the world coordinate system of the multi-view camera system. The projected coordinates in each camera coordinate system are as follows: , and The homogeneous coordinates of the imaging points in each camera image coordinate system are as follows: , and ; utilizing the results Solve for the corresponding camera coordinates from the image coordinates: ； In the formula These represent the focal lengths of each camera; S202: Rotation and translation transformations between camera coordinate systems can be performed using homogeneous transformation matrices. To represent, therefore, we can derive the point. Relationships between different camera coordinate systems: ； In the formula These represent the rotation matrix and translation vector between the vertical, left, and right cameras, respectively. The aforementioned camera imaging model is an ideal imaging model free from external interference; however, various errors occur during the manufacturing and installation of the camera, causing various nonlinear distortions that can affect detection accuracy in practical applications. Therefore, camera calibration considers the nonlinear distortion factors of the camera and corrects the ideal camera imaging model to obtain more accurate calibration results. Considering lens distortion, there is an offset between the actual imaging point and the theoretical imaging point. Offset in direction , It can be expressed as follows: ； In the formula To input distorted image coordinates into the camera, ; and These are the radial and longitudinal distortion coefficients, respectively; S203: After obtaining multiple sets of calibration board images taken by cameras at different locations, and after grayscale preprocessing, each circular pattern is segmented and then... The matrix is ​​used for sub-pixel edge extraction and circle center feature calculation; finally, combined with Zhang Youzheng calibration method, stereo calibration of multi-camera system is realized.

4. The method for detecting and correcting assembly deviations in wing-body docking of a fork-type aircraft according to claim 1, characterized in that: The specific method for solving the fork lug mating position deviation in step S300 is as follows: S301: After completing the stereo vision calibration, the following solution can be obtained by combining the left and right eye cameras: ； The homogeneous transformation matrix between the left-eye camera and the world coordinate system of the multi-eye camera system is: ； In the formula , Representing the two coordinate systems around axis, Axial rotation angle; the point can be calculated. In the three-dimensional coordinates of the world coordinate system of the multi-camera system : ； In a multi-camera system, the world coordinates can be obtained by combining any two cameras using the above method. ; S302: The homogeneous transformation matrix between the multi-camera world coordinate system and the assembly global coordinate system is: ； In the formula Representing the two coordinate systems respectively axis, axis, Axial rotation angle; the spatial coordinates in the assembly global coordinate system can be further solved using coordinate transformation relationships. ; ； S303: Obtained by solving the above formula Coordinates of the center of the intersection hole of the fork-type connector and the lug-type connector in the assembly global coordinate system , ; where the fork-type joint pose is locked, the coordinates For a constant value, take for Therefore, the axis of the intersection hole of the fork-shaped joint is used as the reference axis to solve the problem of the end face of the intersection hole of the fork lug. Coaxiality deviation at time : ； Solvable Spatial position deviation of the center of the intersection hole of the ear-type connector : ； In the formula The target position coordinates of the mating holes of the ear-type connector are given, and the spatial position deviation of the intersecting hole center of the ear-type connector is represented by the spatial position deviation of the ear mating position. .

5. The method for detecting and correcting assembly deviations in wing-body docking of a fork-type aircraft according to claim 1, characterized in that: The specific method for constructing the wing attitude adjustment state equation in step S500 is as follows: S501: Set the initial wing attitude as follows: After attitude adjustment, the wing attitude is as follows: Record the wing attitude adjustment time as ; S502: Let the trajectory prediction equation be... : ； Substituting the constraints, we get: ； S503: Order We can obtain: ； Using the solution Using the same method, we can obtain the prediction equation for the attitude adjustment trajectory of the fore-wing: ； It can be obtained Timing fork ear feature points, wing attitude control points The theoretical coordinates are: ； In the formula , For point In the wing coordinate system, ; S504: Establishing the wing attitude adjustment state equation: ； In the formula ; ; It is a 3x3 noise matrix; ; for Dynamic noise at any given moment.

6. The method for detecting and correcting assembly deviations in wing-body docking of a fork-ear aircraft according to claim 1, characterized in that: The specific method for correcting assembly quality dynamic detection deviations based on LAFSA-PF and data fusion in step S600 is as follows: S601: Utilizing LAFSA-PF and data fusion for dynamic detection and deviation correction of assembly quality. Firstly, in the initial stage of fork lug engagement, the importance of wing attitude control points is sampled using the wing attitude adjustment state equation. Particles As the initial sample, the importance weight of each particle at the wing attitude control point is calculated; S602: Constructing the objective function: ； In the formula, For the crowding factor, This is the latest measurement value of the wing attitude control point. To predict measured values, a local neighborhood, adaptive sensing distance, and step size optimization algorithm for artificial fish swarms are introduced. S603: Optimize particle filtering using a local neighborhood adaptive artificial fish swarm algorithm, and calculate the optimized sampled result. The weights of the particles at each wing attitude control point are determined and their weights are normalized. ； S604: Regarding the wing attitude control points... Estimate the state at time: ； And output the estimated value; S605: Perform comprehensive deviation correction: First, it is necessary to implement data fusion bias correction and construct a joint linear bias data fusion model for the camera and laser tracker: ； In the formula These represent the observation matrices of each camera and the laser tracker, respectively. This represents the estimated position deviation of the wing attitude control point; Each device represents the state of the device. The measurement noise vector at time step; an adaptive weighted fusion algorithm is used to optimize the fused data from the camera and laser tracker to obtain the unfiltered optimized time step. The best estimate of the wing attitude control point coordinates at any given time. ; Then, the positioner attitude adjustment deviation is corrected, and the wing attitude adjustment state equation is solved to obtain the result. The wing pose estimation at any given time is obtained by comparing it with the theoretical pose. Wing attitude deviation ;pass By combining the inverse kinematics solution, we can obtain Positioner pose adjustment deviation correction value generated within a time period .

Images