Quick calibration method and system for angle measurement accuracy of laser seeker

Abstract

The application provides a kind of laser seeker angle measurement precision fast calibration method and system.Therein, the method includes: by laser seeker emitting laser beam to the calibration rod array, and by position sensitive device collection each calibration rod surface reflected laser spot displacement amount;The laser spot displacement amount is input into spot displacement feature extraction network, and the angle and displacement time sequence containing time stamp are generated;Abnormal sampling points in angle and displacement time sequence are identified;Using spatial geometric constraint optimization algorithm, combined with fixed space angle constraint relationship, abnormal sampling points are geometrically reconstructed, and the angle measurement error calibration parameter of laser seeker is obtained.The technical scheme provided by the application realizes high-precision dynamic calibration of laser seeker angle measurement error through cooperative analysis of calibration rod space coordinates and laser spot displacement, combined with neural network and geometric constraint optimization.

Description

Technical Field

[0001] This application relates to the field of laser seeker calibration technology, and in particular to a rapid calibration method and system for the angular measurement accuracy of a laser seeker. Background Technology

[0002] In the dynamic calibration scenario of vehicle-mounted laser seekers, the continuous vibration and attitude changes during vehicle movement can cause dynamic drift of the calibration reference. Therefore, high-precision dynamic calibration is urgently needed to compensate for motion errors in real time. At the same time, the laser spot displacement data is prone to abnormal jumps due to complex environmental factors such as strong light interference and rain and fog reflection. It is necessary to establish a stable and reliable anomaly filtering mechanism to ensure the system's anti-interference capability. In addition, the multi-degree-of-freedom coupling effect of pitch angle, yaw angle and roll angle can significantly affect the angle measurement accuracy. Therefore, it is necessary to accurately separate the spatial geometric error parameters through decoupling calibration technology to meet the accurate measurement requirements under complex working conditions.

[0003] The existing solution employs a multi-sensor fusion Kalman filter dynamic calibration method, simultaneously deploying an inertial measurement unit (IMU) and position-sensitive devices on the vehicle platform. The IMU acquires attitude data such as vehicle pitch angle, yaw angle, and roll angle in real time. A kinematic model of laser spot displacement and vehicle attitude angle is constructed, and the Kalman filter algorithm is used to fuse displacement data collected by multiple frames of position-sensitive devices and real-time attitude data. Finally, the displacement sequence is smoothed and optimized based on the state prediction values ​​to output dynamic calibration parameters.

[0004] However, the existing scheme has the following two inherent defects: First, because the Kalman filter algorithm relies on the linear Gaussian assumption, when strong light interference causes continuous jumps in the displacement data of position-sensitive devices, the state prediction error will continue to accumulate and eventually lead to the divergence and inaccuracy of the calibration parameters; Second, because the inherent spatial angle constraint relationship between the calibration rods is not utilized, it is impossible to correct the systematic measurement error caused by the multi-degree-of-freedom coupling of pitch angle, yaw angle and roll angle through the geometric consistency principle, thereby reducing the reliability of the calibration parameters. Summary of the Invention

[0005] This application provides a rapid calibration method and system for the angle measurement accuracy of a laser seeker, which solves the problems of inaccurate dynamic calibration parameters under abnormal interference and the inability to correct multi-degree-of-freedom coupling errors in the prior art.

[0006] In a first aspect, this application provides a rapid calibration method for the angle measurement accuracy of a laser seeker, including:

[0007] Obtain the pitch angle, yaw angle, and roll angle of each calibration rod in the calibration rod array fixed to the vehicle platform in the vehicle coordinate system, and generate the three-dimensional spatial coordinates of each calibration rod;

[0008] A laser beam is emitted to the calibration rod array through a laser guide head, and the displacement of the laser spot reflected from the surface of each calibration rod is collected by a position-sensitive device.

[0009] The laser spot displacement is input into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps;

[0010] Identify anomalous sampling points in the angle and displacement time series;

[0011] A spatial geometric constraint optimization algorithm is used, combined with a fixed spatial angle constraint relationship, to perform geometric reconstruction on abnormal sampling points and obtain the angle measurement error calibration parameters of the laser seeker.

[0012] Optionally, the step of employing a spatial geometric constraint optimization algorithm, combined with a fixed spatial angle constraint relationship, to perform geometric reconstruction on abnormal sampling points, and obtain the angle measurement error calibration parameters of the laser seeker, including:

[0013] Based on the three-dimensional spatial coordinates of each calibration rod in the calibration rod array, the fixed spatial angle constraint relationship between any two calibration rods is extracted, and the fixed spatial angle constraint equation is established.

[0014] Based on the fixed spatial angle constraint equation, a geometric reconstruction objective function containing two different deviation terms is constructed, wherein the geometric reconstruction objective function uses the reconstruction value as the optimization variable;

[0015] A spatial geometric constraint optimization algorithm is adopted. The fixed spatial angle constraint equation and the geometric reconstruction objective function are input into the solver. The parameters of the normal sampling points adjacent to the abnormal sampling points are used as the initial iteration values. The gradient direction of the optimization variables is calculated by the constraint Jacobian matrix. The reconstruction value is iteratively optimized by combining the fixed spatial angle constraint equation, so that the value of the geometric reconstruction objective function gradually decreases.

[0016] When the geometric reconstruction objective function value satisfies the preset convergence condition, the reconstruction value that satisfies the fixed spatial angle constraint relationship will be used as the geometric reconstruction result of the abnormal sampling point.

[0017] Based on the geometric reconstruction results of all abnormal sampling points, the angular measurement error calibration parameters of the laser seeker are calculated.

[0018] Optionally, the step of using the parameters of the normal sampling points adjacent to the abnormal sampling points as initial iteration values, calculating the gradient direction of the optimization variables through the constraint Jacobian matrix, and iteratively optimizing the reconstructed values ​​in combination with the fixed spatial angle constraint equation includes:

[0019] The parameters of the normal sampling points adjacent to the abnormal sampling point are selected, and corresponding weights are assigned according to the timestamp interval. The initial iterative value of the reconstructed value of the abnormal sampling point is obtained by weighted calculation.

[0020] Based on the fixed spatial angle constraint equation and the iteration value, the partial derivative of the reconstructed value is obtained to construct the constraint Jacobian matrix. The elements in the constraint Jacobian matrix are used to reflect the sensitivity of each reconstructed value to the fixed spatial angle constraint equation. The iteration value is the initial iteration value in the first generation and the reconstructed value adjusted in the current generation in other generations.

[0021] By combining the gradient vector of the geometric reconstruction objective function with the constraint Jacobian matrix, the search direction of the optimization variables is determined by the Lagrange multiplier method;

[0022] The adjustment direction of the reconstructed value is determined according to the search direction, and a single-step adjustment amount is generated using an adaptive step size strategy. The reconstructed value is then adjusted based on the single-step adjustment amount.

[0023] Substitute the adjusted reconstruction value into the fixed spatial angle constraint equation, and calculate the geometric reconstruction objective function value corresponding to the adjusted reconstruction value based on the geometric reconstruction objective function, provided that the constraint is satisfied.

[0024] Calculate the deviation between the current generation's geometric reconstruction objective function value and the previous generation's geometric reconstruction objective function value;

[0025] If the deviation of three consecutive iterations is less than the preset threshold, the preset convergence condition is satisfied. If the deviation of three consecutive iterations is not less than the preset threshold, the construction of the constraint Jacobian matrix is ​​repeated and the iteration continues, with the reconstructed value after the current generation as the new iteration starting point.

[0026] Alternatively, if the number of iterations exceeds the maximum threshold, the value of the single-step adjustment is reduced, and the adjustment of the reconstructed value is repeated based on the reduced single-step adjustment value, and the number of iterations is reset to continue iterating.

[0027] Optionally, the step of constructing a geometric reconstruction objective function containing two different deviation terms based on the fixed spatial angle constraint equation includes:

[0028] Based on the reconstructed values ​​of the abnormal sampling points, the instantaneous spatial angle pointing to any two calibration rods is calculated, and the fixed spatial angle of the corresponding calibration rod pair is analyzed using the fixed spatial angle constraint equation.

[0029] Calculate the absolute difference between the instantaneous spatial angle and the fixed spatial angle to generate a single set of basic deviations;

[0030] Based on the spatial distribution density of the calibration rod pairs, constraint weight coefficients for the calibration rod pairs are generated;

[0031] Multiply the constraint weight coefficient by the single set of basic deviations of the corresponding calibration rod pair, and sum the multiplication results of all calibration rod pairs to obtain the basic deviation term;

[0032] The temporal deviation between the reconstructed value of the abnormal sampling point and the parameters of the adjacent normal sampling points is calculated, and different temporal smoothing weights are assigned to different normal sampling points to generate a temporal smoothing deviation term.

[0033] The basic deviation term and the temporal smoothing deviation term are superimposed to construct the geometric reconstruction objective function.

[0034] Optionally, the step of inputting the laser spot displacement into a spot displacement feature extraction network to generate an angle and displacement time sequence including timestamps includes:

[0035] The horizontal displacement component and the vertical displacement component are separated from the original signal of the laser spot displacement.

[0036] The instantaneous yaw angle is calculated based on the horizontal displacement component, and the instantaneous pitch angle is calculated based on the vertical displacement component.

[0037] The straight-line distance between the laser guide head and the surface of the calibration rod is taken as the radial displacement value;

[0038] The timestamps corresponding to the instantaneous yaw angle value, the instantaneous pitch angle value, and the radial displacement value are collected synchronously to obtain a timestamp set;

[0039] Based on the timestamp set, the instantaneous yaw angle value, instantaneous pitch angle value, and radial displacement value are integrated in the order of the timestamps to generate an angle and displacement time sequence containing timestamps.

[0040] Optionally, generating the constraint weight coefficients for the calibration pole pairs based on their spatial distribution density includes:

[0041] The number of adjacent normal sampling points within the spatial neighborhood is calculated to obtain the spatial distribution density value, wherein the spatial neighborhood is a circular area with the calibration rod pair as the geometric center and a preset vibration characteristic radius as the radius;

[0042] Based on the spatial distribution density value, a density influence factor is generated through a piecewise mapping function;

[0043] The density influence factor is linearly mapped to the corresponding preset weight interval to generate the constraint weight coefficients of the calibration rod pair.

[0044] Optionally, generating the density influence factor based on the spatial distribution density value using a piecewise mapping function includes:

[0045] The vibration density threshold is determined based on the material stiffness coefficient of the vehicle platform and the spacing of the calibration rod array;

[0046] The first and second material response constants are determined based on the sampling frequency of the position-sensitive device and the surface reflectivity of the calibration rod.

[0047] When the spatial distribution density value is less than the vibration density threshold, the spatial distribution density value is multiplied by the first material response constant to generate a density influence factor;

[0048] Alternatively, when the spatial distribution density value is greater than or equal to the vibration density threshold, the natural logarithm of the spatial distribution density value is multiplied by the second material response constant to generate a density influence factor.

[0049] Secondly, this application provides a rapid calibration system for the angle measurement accuracy of a laser seeker, comprising:

[0050] The acquisition module is used to acquire the pitch angle, yaw angle and roll angle of each calibration rod in the calibration rod array fixed to the vehicle body platform in the vehicle body coordinate system, and generate the three-dimensional spatial coordinates of each calibration rod;

[0051] The acquisition module is used to emit a laser beam to the calibration rod array through a laser guide head, and to acquire the displacement of the laser spot reflected from the surface of each calibration rod through a position-sensitive device;

[0052] The generation module is used to input the laser spot displacement into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps;

[0053] The identification module is used to identify abnormal sampling points in the angle and displacement time sequence;

[0054] The reconstruction module is used to perform geometric reconstruction of abnormal sampling points by using a spatial geometric constraint optimization algorithm combined with a fixed spatial angle constraint relationship, so as to obtain the angle measurement error calibration parameters of the laser seeker.

[0055] Thirdly, this application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are to be invoked and executed by the processing component to implement a rapid calibration method for the angle measurement accuracy of a laser seeker as described in any of the first aspects.

[0056] Fourthly, this application provides a computer storage medium storing a computer program, which, when executed by a computer, implements a rapid calibration method for the angle measurement accuracy of a laser seeker as described in any of the first aspects.

[0057] This application provides a rapid calibration method for the angle measurement accuracy of a laser seeker. The method includes: acquiring the pitch angle, yaw angle, and roll angle of each calibration rod in a calibration rod array fixed to a vehicle platform in the vehicle coordinate system, generating the three-dimensional spatial coordinates of each calibration rod; emitting a laser beam from the laser seeker to the calibration rod array, and collecting the laser spot displacement reflected from the surface of each calibration rod using a position-sensitive device; inputting the laser spot displacement into a spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps; identifying abnormal sampling points in the angle and displacement time sequence; and using a spatial geometric constraint optimization algorithm, combined with a fixed spatial angle constraint relationship, to geometrically reconstruct the abnormal sampling points to obtain the angle measurement error calibration parameters of the laser seeker.

[0058] This application has the following advantages: By establishing a collaborative mechanism between the rigid spatial reference of the calibration rod array and the dynamic detection of laser displacement, combined with intelligent temporal feature extraction and spatial geometric constraint optimization, this application effectively isolates abnormal data jumps caused by environmental interference, systematically decouples multi-degree-of-freedom coupling errors, and achieves high-precision and robust calibration of laser seeker angle measurement parameters under complex vibration environments.

[0059] Furthermore, the embodiments of this application achieve accurate geometric reconstruction of abnormal sampling points by co-optimizing the fixed spatial angle constraint equation and the dual-bias objective function, combined with gradient search and adaptive step size strategies, effectively eliminating multi-degree-of-freedom coupling errors and improving the anti-interference capability and systematic error correction capability of the laser seeker angle measurement calibration parameters.

[0060] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 A flowchart illustrating a rapid calibration method for the angle measurement accuracy of a laser seeker, provided as an embodiment of this application;

[0063] Figure 2 A schematic diagram of a rapid calibration system for the angle measurement accuracy of a laser seeker provided in this application embodiment;

[0064] Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0065] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0066] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 11, 12, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0067] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0068] To address the issues of inaccurate dynamic calibration parameters under abnormal interference and the inability to correct multi-degree-of-freedom coupling errors in existing technologies, this application provides a rapid calibration method for the angular measurement accuracy of a laser seeker. This method employs the following concept: by establishing a collaborative mechanism between a physical spatial benchmark and dynamic displacement detection, combined with intelligent feature extraction and geometric constraint optimization, highly robust dynamic calibration under complex interference environments is achieved. Specifically, the three-dimensional spatial coordinates of the calibration rod array serve as the rigid geometric constraint basis, and the laser spot displacement is used to capture microscopic deformation features in real time. A neural network automatically extracts spatiotemporal correlation sequences and identifies anomalies, avoiding the subjectivity of manual threshold setting. Finally, based on a fixed spatial angle relationship, the anomalies are geometrically reconstructed, transforming the decoupling problem of multi-degree-of-freedom coupling errors into a spatial constraint optimization problem, thereby systematically improving calibration accuracy while ensuring the algorithm's generalization ability.

[0069] Figure 1 A flowchart illustrating a rapid calibration method for the angle measurement accuracy of a laser seeker, as provided in this application embodiment, is shown below. Figure 1 As shown, the method includes:

[0070] S11. Obtain the pitch angle, yaw angle and roll angle of each calibration rod in the calibration rod array fixed to the vehicle body platform in the vehicle body coordinate system, and generate the three-dimensional spatial coordinates of each calibration rod.

[0071] The vehicle platform refers to the vehicle chassis structure that supports the calibration rod array, including a rigid frame and mounting interfaces, used to fix the calibration system components. The calibration rod array consists of multiple metal rods fixed to the vehicle platform in a specific spatial layout, with a laser-reflective coating on the surface to provide geometric constraint references. Each calibration rod is a single metal rod in the calibration rod array, and its spatial position is defined by pitch, yaw, and roll angles. The vehicle coordinate system is a right-handed three-dimensional Cartesian coordinate system established with the vehicle's center of gravity as the origin, with the X-axis pointing towards the front of the vehicle, used to uniformly describe the spatial attitude of the calibration rods. The pitch angle is the rotation angle around the Y-axis in the vehicle coordinate system, reflecting the forward and backward tilt of the calibration rod. The yaw angle is the rotation angle around the Z-axis in the vehicle coordinate system, reflecting the left and right yaw of the calibration rod. The roll angle is the rotation angle around the X-axis in the vehicle coordinate system, reflecting the lateral roll of the calibration rod. The three-dimensional spatial coordinates are transformed into X-axis, Y-axis and Z-axis coordinate values ​​through coordinate transformation, which accurately describes its position in the vehicle coordinate system.

[0072] In this embodiment, firstly, the pitch angle, yaw angle, and roll angle of each calibration rod in the calibration rod array fixed to the vehicle platform are obtained in the vehicle coordinate system by the inertial measurement unit on the vehicle platform; secondly, based on the spatial geometric relationship between the vehicle coordinate system and the calibration rods, the pitch angle, yaw angle, and roll angle are converted into the three-dimensional spatial coordinates of each calibration rod by the coordinate transformation algorithm to establish a spatial reference frame.

[0073] S12. A laser beam is emitted to the calibration rod array through the laser guide head, and the displacement of the laser spot reflected from the surface of each calibration rod is collected through the position sensitive device.

[0074] The laser seeker is a device integrating a laser emitter and an optical receiver module, used to project a laser beam onto the surface of the calibration rod and receive the reflected signal. The laser beam is the directional coherent light emitted by the laser seeker, used to generate a detectable spot displacement. The position-sensitive device is a two-dimensional optical sensor that detects the displacement of the laser spot on the photosensitive surface through the photoelectric effect, with a resolution down to the micrometer level. The laser spot displacement refers to the electrical signal output by the position-sensitive device, quantifying the horizontal and vertical offset distance of the spot center relative to a reference position.

[0075] In this embodiment, a laser beam is first emitted by a laser seeker to the surface of the calibration rod array; then, the displacement of the laser spot reflected from the surface of each calibration rod is captured in real time by a position-sensitive device. This displacement includes micron-level offset information of the spot in the horizontal and vertical directions.

[0076] S13. Input the laser spot displacement into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps.

[0077] The laser spot displacement feature extraction network refers to a feature extraction model built on a convolutional neural network, which outputs instantaneous angle and radial displacement values ​​after inputting the displacement amount. The timestamp refers to a millisecond-level time stamp generated by a high-precision timer, used to associate the acquisition time of displacement and angle data. The angle and displacement time sequence refers to a data structure sorted by timestamps, with each row containing instantaneous yaw angle, pitch angle, radial displacement value, and corresponding timestamp. The laser spot displacement feature extraction network inputs the two-dimensional displacement of the laser spot center in the image coordinate system obtained through preprocessing into the network. In this embodiment, the laser spot displacement is first input into the network; the network uses a convolutional neural network to separate the horizontal and vertical displacement components, and calculates the instantaneous yaw angle and pitch angle based on these components; simultaneously, it combines the straight-line distance from the laser seeker to the calibration rod surface to generate the radial displacement value; finally, it synchronously associates the timestamps and integrates all angle and displacement data in chronological order to generate the angle and displacement time sequence.

[0078] S14. Identify abnormal sampling points in the angle and displacement time series.

[0079] Among them, outlier sampling points refer to data outliers identified using time series data analysis algorithms, whose values ​​deviate from the normal fluctuation range by more than three standard deviations.

[0080] In this embodiment, the statistical distribution characteristics of data points in the angle and displacement time series are first analyzed; then, the isolated forest algorithm is used to detect sampling points that deviate from the normal fluctuation range and mark them as abnormal sampling points.

[0081] S15. Using a spatial geometric constraint optimization algorithm, combined with a fixed spatial angle constraint relationship, the abnormal sampling points are geometrically reconstructed to obtain the angle measurement error calibration parameters of the laser seeker.

[0082] Among them, the spatial geometric constraint optimization algorithm refers to a numerical optimization algorithm that integrates fixed spatial angle constraints, achieving geometric reconstruction by iteratively solving for the minimum objective function. The fixed spatial angle constraint relationship is the theoretical value of the spatial angle between any two calibration rods in the calibration rod array, calculated from three-dimensional spatial coordinates. Geometric reconstruction refers to the process of correcting abnormal sampling points using spatial constraint relationships, ensuring that the corrected data conforms to physical geometric consistency. The angle measurement error calibration parameters refer to the compensation coefficient matrix describing the error of the laser seeker angle measurement system, including correction parameters for yaw and pitch angles. Specifically, the fixed spatial angle refers to the spatial angle formed between the center point of the reflecting surface of any two calibration rods and the optical center of the laser seeker.

[0083] In this embodiment, firstly, the fixed spatial angle constraint relationship between any two calibration rods is extracted based on the three-dimensional spatial coordinates of the calibration rod array; secondly, a spatial geometric constraint optimization algorithm is used, with the weighted parameters of adjacent normal sampling points of abnormal sampling points as initial values, to construct an objective function containing position deviation terms and temporal smoothing terms; the gradient direction is calculated by constraining the Jacobian matrix to iteratively optimize the reconstructed value until the fixed spatial angle constraint relationship is satisfied; finally, the angle measurement error calibration parameters of the laser seeker are calculated based on the reconstruction results.

[0084] Here's a specific example: First, three calibration rods arranged in a triangle are deployed on the moving vehicle platform. An inertial measurement unit (IMU) collects the pitch, yaw, and roll angles of each rod in real time and converts them into three-dimensional spatial coordinates. Next, a laser seeker emits a laser beam onto the surface of the calibration rods, and position-sensitive devices simultaneously capture the displacement of the reflected light spot. The displacement is then input into a convolutional neural network to separate the horizontal and vertical components, calculate the instantaneous angle values ​​and radial distances, and associate them with timestamps to form a time-series sequence. Then, statistical outlier detection identifies abnormal sampling points in the sequence. Finally, an objective function is constructed based on the fixed angle constraint between the calibration rods. The iterative process is initialized with data from adjacent normal points, and outlier points are reconstructed through gradient optimization, outputting the angle measurement error compensation parameters for the laser seeker.

[0085] By executing S11~S15, this embodiment of the application establishes a collaborative mechanism between the rigid spatial reference of the calibration rod array and the dynamic detection of laser displacement. Combined with intelligent temporal feature extraction and spatial geometric constraint optimization, it effectively isolates abnormal data jumps caused by environmental interference, systematically decouples multi-degree-of-freedom coupling errors, and achieves high-precision and robust calibration of laser seeker angle measurement parameters under complex vibration environments.

[0086] In one possible embodiment, S15, a spatial geometric constraint optimization algorithm is used, combined with a fixed spatial angle constraint relationship, to geometrically reconstruct the abnormal sampling points and obtain the angle measurement error calibration parameters of the laser seeker, including:

[0087] Step 151: Based on the three-dimensional spatial coordinates of each calibration rod in the calibration rod array, extract the fixed spatial angle constraint relationship between any two calibration rods and establish the fixed spatial angle constraint equation.

[0088] The fixed spatial angle constraint equation refers to the constant relationship between the vector angles calculated based on the three-dimensional spatial coordinates of any two calibration rods in the calibration rod array. This equation is used to force the geometric reconstruction results to conform to physical space consistency. The formula for the fixed spatial angle constraint equation is: ,in, The coordinates of the optical center of the laser seeker. To determine the center coordinates of the reflecting surface of calibrator i, To determine the center coordinates of the reflecting surface of the calibration rod j, Let be the fixed spatial angle between calibrating rods i and j.

[0089] In this embodiment, firstly, based on the three-dimensional spatial coordinates of each calibration rod in the calibration rod array, the spatial vector between any two calibration rods is calculated; secondly, the cosine value of the angle between the two vectors is derived through the vector dot product formula, and the theoretical spatial angle is determined by combining the fixed geometric relationship of the calibration rod array; finally, a fixed spatial angle constraint equation is established, which forces the reconstructed outlier data to meet the geometric constraint conditions defined by the equation.

[0090] Step 152: Based on the fixed spatial angle constraint equation, construct a geometric reconstruction objective function containing two different deviation terms. The geometric reconstruction objective function uses the reconstruction value as the optimization variable.

[0091] The two distinct bias terms are the basic bias term and the temporal smoothing bias term: the basic bias term reflects the degree to which the reconstructed value deviates from the fixed spatial angle; the temporal smoothing bias term reflects the temporal continuity between the reconstructed value and adjacent normal data. The geometric reconstruction objective function is an optimization function with the reconstructed value as the independent variable. It is a linear superposition of the basic bias term and the temporal smoothing bias term. Minimizing this function ensures that the reconstructed value simultaneously satisfies spatial constraints and temporal smoothness. The reconstructed value refers to the corrected data of the outlier sampling points to be optimized, including the pitch angle, yaw angle, and radial displacement to be corrected. The optimization variable refers to the set of independent variables to be solved in the geometric reconstruction objective function, i.e., the reconstructed values ​​of the outlier sampling points. The formula for the geometric reconstruction objective function is: ,in, For reconstructed values, To reconstruct the objective function value for geometry, For calibrating the rod pair set, It is the instantaneous spatial angle. To fix the included angle in space, To constrain the weighting coefficients, For hyperparameters, For the set of adjacent normal sampling points, These are parameters for normal sampling points. For time-series smoothing weights.

[0092] In this embodiment, a basic deviation term is first defined based on the fixed spatial angle constraint equation. The absolute difference between the instantaneous spatial angle corresponding to the reconstructed value of the abnormal sampling point and the theoretical spatial angle is calculated and weighted according to the spatial distribution density of the calibration rod pair. Secondly, a temporal smoothing deviation term is defined. The temporal deviation between the reconstructed value and the parameters of the adjacent normal sampling points is calculated and given a time decay weight. Finally, the basic deviation term and the temporal smoothing deviation term are added to construct a geometric reconstruction objective function with the reconstructed value as the optimization variable.

[0093] Step 153: Using a spatial geometric constraint optimization algorithm, the fixed spatial angle constraint equation and the geometric reconstruction objective function are input into the solver. The parameters of the normal sampling points adjacent to the abnormal sampling points are used as the initial iteration values. The gradient direction of the optimization variables is calculated through the constraint Jacobian matrix. The reconstructed value is iteratively optimized in combination with the fixed spatial angle constraint equation, so that the value of the geometric reconstruction objective function gradually decreases.

[0094] The solver refers to a computational engine employing numerical optimization algorithms, which outputs the optimal solution satisfying the conditions after inputting the objective function and constraint equations. Normal sampling points refer to data points confirmed by time-series analysis to be undisturbed; their parameters are used to initialize the reconstruction process. Parameters refer to the pitch angle, yaw angle, and radial displacement measurements included in the normal sampling points. The initial iteration value refers to the starting point of the reconstruction optimization process, generated by weighted calculation of parameters from adjacent normal sampling points. The constraint Jacobian matrix is ​​a matrix formed by taking the partial derivatives of the fixed spatial angle constraint equations with respect to the reconstructed values; its elements reflect the sensitivity of each reconstructed value's variation to the constraint conditions. The gradient direction refers to the direction of the fastest descent of the geometric reconstruction objective function, calculated using the Lagrange multiplier method combined with the constraint Jacobian matrix. The geometric reconstruction objective function value refers to the scalar value output by the objective function in each iteration; its degree of decrease reflects the optimization progress.

[0095] In this embodiment, firstly, parameters of adjacent normal sampling points before and after the abnormal sampling point are selected, and weighted according to the timestamp interval to generate initial iteration values; secondly, based on the fixed spatial angle constraint equation, partial derivatives of the reconstructed values ​​are calculated to construct a constraint Jacobian matrix that reflects the sensitivity of the reconstructed value changes to the constraint equation; then, combined with the gradient vector of the geometric reconstruction objective function, the search direction of the optimization variables is determined by the Lagrange multiplier method; subsequently, an adaptive step size strategy is used to generate single-step adjustment amounts to update the reconstructed values; finally, it is verified whether the updated values ​​satisfy the constraint equation and the objective function value is calculated, and the iteration is repeated until convergence.

[0096] Step 154: When the geometric reconstruction objective function value satisfies the preset convergence condition, the reconstruction value that satisfies the fixed spatial angle constraint relationship is taken as the geometric reconstruction result of the abnormal sampling point.

[0097] The preset convergence condition refers to the criteria for determining the termination of optimization, including the change in the objective function value being less than a threshold or reaching the maximum number of iterations after three consecutive iterations. The geometric reconstruction result refers to the corrected outlier data that satisfies spatial constraints, used to replace the original outliers.

[0098] In this embodiment, the deviation between the current generation and the previous generation of geometric reconstruction objective function values ​​is first calculated; then, it is determined whether the deviation in three consecutive iterations is less than a preset threshold. If the threshold is met, the preset convergence condition is determined to be met; if the threshold is not met but the number of iterations exceeds the maximum threshold, the single-step adjustment amount is reduced and the number of iterations is reset to continue optimization; finally, the reconstruction value that satisfies the fixed spatial angle constraint equation is output as the geometric reconstruction result of the abnormal sampling point.

[0099] Step 155: Calculate the angle measurement error calibration parameters of the laser seeker based on the geometric reconstruction results of all abnormal sampling points.

[0100] In this embodiment, the geometric reconstruction results of all abnormal sampling points are first summarized and the abnormal data in the original time series are replaced; then, based on the reconstructed complete time series, the angle measurement error model of the laser seeker is fitted by the least squares method; finally, the error compensation coefficient matrix is ​​solved and the angle measurement error calibration parameters are output.

[0101] Here is a specific example: First, the theoretical angle between the three calibration rods is calculated based on their three-dimensional coordinates, establishing a fixed spatial angle constraint equation. Second, an objective function containing a basic deviation term and a temporal smoothing term is constructed, with the pitch and yaw angles of the outlier points reconstructed as optimization variables. Next, initial values ​​are generated by weighting normal data before and after the outlier point. The gradient direction is calculated using the constraint Jacobian matrix, and the reconstructed values ​​are iteratively updated with an adaptive step size to verify constraint compliance. Subsequently, when the objective function value changes less than a threshold for three consecutive times, a reconstruction result satisfying the spatial constraints is output. Finally, the reconstructed data is used to fit an error model, generating angle measurement error compensation parameters for the laser seeker.

[0102] By executing steps 151 to 155, this embodiment of the application achieves accurate geometric reconstruction of abnormal sampling points through the coordinated optimization of the fixed spatial angle constraint equation and the dual-bias target function, combined with gradient search and adaptive step size strategy. This effectively eliminates multi-degree-of-freedom coupling errors and improves the anti-interference capability and systematic error correction capability of the laser seeker angle measurement calibration parameters.

[0103] In one possible embodiment, step 153, using the parameters of the normal sampling points adjacent to the abnormal sampling points as initial iteration values, calculates the gradient direction of the optimization variables through the constraint Jacobian matrix, and iteratively optimizes the reconstructed values ​​in combination with the fixed spatial angle constraint equation, including:

[0104] Step a1: Select the parameters of the normal sampling points adjacent to the abnormal sampling point, assign corresponding weights according to the timestamp interval, and calculate the initial iterative value of the reconstructed value of the abnormal sampling point by weighted calculation.

[0105] The timestamp interval refers to the millisecond-level difference between the timestamps of abnormal sampling points and adjacent normal sampling points, used to calculate the time decay weight. The weight is a coefficient generated based on the reciprocal of the timestamp interval; the smaller the interval, the greater the weight, reflecting the principle of temporal proximity. The weighted calculation involves multiplying the parameters of each normal sampling point by their corresponding weights, summing the results, and then dividing by the total weights.

[0106] In this embodiment, the parameters of the normal sampling points adjacent to the abnormal sampling point are first selected; then, the time decay factor is calculated based on the timestamp interval between the normal sampling point and the abnormal sampling point, and the reciprocal of the timestamp interval is used as the weighting coefficient; finally, the parameters of the adjacent normal sampling points are weighted and averaged to generate the initial iterative value of the reconstructed value of the abnormal sampling point.

[0107] Step a2: Based on the fixed spatial angle constraint equation and the iteration value, take the partial derivative of the reconstructed value and construct the constraint Jacobian matrix. The elements in the constraint Jacobian matrix are used to reflect the sensitivity of each reconstructed value to the fixed spatial angle constraint equation. The iteration value is the initial iteration value in the first generation and the reconstructed value after adjustment in the current generation in other generations.

[0108] In this context, iterative values ​​refer to the dynamically updated intermediate variables during the optimization process. The first generation uses the initial iterative values, and subsequent generations use the adjusted reconstructed values. Partial derivatives are calculated for each component of the reconstructed value in the fixed spatial angle constraint equation to obtain the rate of change. Influence sensitivity refers to the values ​​of the constraint Jacobian matrix elements, quantifying the change in the constraint equation value caused by a unit change in the reconstructed value. The first generation refers to the initial stage of the optimization iteration, using the generated initial iterative values. The current generation refers to any current iteration in the process, with its iterative value being the latest adjusted result. The adjusted reconstructed values ​​are the optimized variable values ​​updated by a single-step adjustment, used for the next iteration or as the output result.

[0109] In this embodiment, the current iteration value is first obtained; then, the partial derivative of each component of the reconstructed value is calculated based on the fixed spatial angle constraint equation; finally, a constraint Jacobian matrix is ​​constructed, whose element values ​​are partial derivative values, reflecting the sensitivity of the influence of each reconstructed value change on the constraint equation satisfaction.

[0110] Step a3: Combine the gradient vector of the geometric reconstruction objective function with the constraint Jacobian matrix, and determine the search direction of the optimization variables using the Lagrange multiplier method.

[0111] Here, the gradient vector refers to the vector formed by the partial derivatives of the geometric reconstruction objective function with respect to each component of the reconstructed value, indicating the direction of descent of the function value. The Lagrange multiplier method refers to introducing constraint equations to construct a Lagrange function, and determining the search direction of constrained optimization by solving for stationary points.

[0112] The search direction refers to the optimization path jointly determined by the gradient vector and the constraint Jacobian matrix, which ensures that the objective function decreases and satisfies the constraints.

[0113] In this embodiment, the gradient vector of the geometric reconstruction objective function at the current reconstruction value is first calculated; then, the transpose of the constraint Jacobian matrix is ​​combined; finally, the constrained optimization problem is solved by the Lagrange multiplier method to determine the search direction of the optimization variables that makes the objective function decrease.

[0114] Step a4: Determine the adjustment direction of the reconstructed value based on the search direction, generate a single-step adjustment amount using an adaptive step size strategy, and adjust the reconstructed value based on the single-step adjustment amount.

[0115] Here, the adjustment direction refers to the unit vector of the search direction, which determines the positive or negative direction of the reconstructed value update. The adaptive step size strategy is a mechanism that dynamically adjusts the step size based on the rate of decrease of the objective function in historical iterations. The initial step size is relatively large, and it decreases with each iteration. The single-step adjustment amount refers to the maximum allowable change in the reconstructed value in the current generation, output by the step size strategy.

[0116] In this embodiment, the adjustment direction of the reconstructed value is first determined along the search direction; then, an adaptive step size strategy is adopted to dynamically calculate the single-step adjustment amount based on the historical rate of change of the objective function; finally, the reconstructed value is added to the product of the single-step adjustment amount and the adjustment direction to generate a new reconstructed value.

[0117] Step a5: Substitute the adjusted reconstruction value into the fixed spatial angle constraint equation. Under the condition of satisfying the constraint, calculate the geometric reconstruction objective function value corresponding to the adjusted reconstruction value based on the geometric reconstruction objective function.

[0118] In this embodiment, the adjusted reconstructed value is substituted into the fixed spatial angle constraint equation to verify whether all constraints are met. When the adjusted reconstructed value meets the constraints, the geometric reconstructed objective function value corresponding to the current reconstructed value is calculated based on the geometric reconstructed objective function formula.

[0119] Step a6: Calculate the deviation between the current generation's geometric reconstruction objective function value and the previous generation's geometric reconstruction objective function value.

[0120] The deviation refers to the absolute difference between the objective function values ​​of two adjacent generations, which is used for convergence judgment.

[0121] In this embodiment of the application, the absolute difference between the current generation geometric reconstruction objective function value and the previous generation geometric reconstruction objective function value is calculated to obtain the deviation amount.

[0122] Step a7: If the deviation of three consecutive iterations is less than the preset threshold, the preset convergence condition is satisfied. If the deviation of three consecutive iterations is not less than the preset threshold, the construction of the constraint Jacobian matrix is ​​repeated and the iteration continues, with the reconstructed value after adjustment in the current generation as the new iteration starting point.

[0123] Here, "three consecutive iterations" refers to the iteration sequence of the current generation and the two generations preceding it, used to statistically analyze convergence stability. The preset threshold is the upper limit of the deviation set according to calibration accuracy requirements, typically one-hundredth of the function value's order of magnitude. The iteration starting point refers to the initial reconstructed value of a new iteration, inherited from the previous optimization result.

[0124] In this embodiment, it is first checked whether the deviation of three consecutive iterations is less than a preset threshold; if the threshold is met, convergence is determined and the result is output; if the threshold is not met, the current reconstructed value is used as the new iteration starting point, and the constraint Jacobian matrix construction and subsequent steps are repeated.

[0125] Step a8, or, if the number of iterations is greater than the maximum number threshold, then reduce the value of the single-step adjustment, repeat the adjustment of the reconstructed value based on the reduced single-step adjustment, and reset the number of iterations to continue iterating.

[0126] Here, the iteration count refers to the cumulative number of optimization rounds starting from the first generation. The single-step adjustment after numerical reduction refers to a more conservative adjustment range generated by multiplying the current step size by 0.5. Resetting the iteration count means resetting the iteration count counter to zero, but without changing the current reconstructed value.

[0127] In this embodiment, when the number of iterations exceeds the maximum threshold, the single-step adjustment amount is first reduced proportionally; then the iteration count counter is reset; and finally, the reconstruction value adjustment step is continued based on the reduced single-step adjustment amount.

[0128] Here is a specific example: First, the angle parameters of two normal points before and after the outlier are selected, weighted according to the reciprocal of the timestamp interval, and a weighted average is used to generate the initial iteration value. Next, a constraint Jacobian matrix is ​​constructed by taking the partial derivative of the reconstructed value based on the fixed spatial angle constraint equation. Then, the gradient vector of the objective function is calculated, and the search direction is determined using the Lagrange multiplier method. Subsequently, an adaptive step-size strategy is used along this direction to calculate the single-step adjustment, update the reconstructed value, and verify constraint compliance. Then, the deviation between the current objective function value and the previous generation is calculated. When the deviation is less than a threshold for three consecutive iterations, the reconstruction result is output; if convergence is not achieved after twenty iterations, the step size is reduced and the iteration count is reset to continue optimization until the convergence condition is met.

[0129] By executing steps a1 to a8, the embodiments of this application ensure the reliability of the iteration start point through time-weighted initial values. Combined with constraint sensitivity analysis and adaptive step size control, it achieves efficient convergence under the premise of strictly satisfying spatial geometric constraints, and realizes accurate reconstruction and error correction of outlier data.

[0130] In one possible embodiment, step 152, based on the fixed spatial angle constraint equation, constructs a geometric reconstruction objective function containing two different deviation terms, including:

[0131] Step b1: Based on the reconstructed values ​​of the abnormal sampling points, calculate the instantaneous spatial angle between any two calibration rods, and use the fixed spatial angle constraint equation to analyze the fixed spatial angle between the corresponding calibration rod pairs.

[0132] The instantaneous spatial angle refers to the angle between any two calibration rod spatial vectors calculated in real time based on the reconstructed values ​​of abnormal sampling points, reflecting the dynamic geometric relationship at the current measurement moment.

[0133] In this embodiment, firstly, based on the pitch angle, yaw angle and radial displacement included in the reconstructed values ​​of the abnormal sampling points, the spatial vector pointing from the laser seeker to any two calibration rods is calculated; secondly, the cosine value of the angle between the two vectors is derived by using the vector dot product formula; finally, the instantaneous spatial angle is solved by combining the inverse trigonometric function, and the theoretical value of the fixed spatial angle of the corresponding calibration rod pair is extracted from the fixed spatial angle constraint equation.

[0134] Step b2: Calculate the absolute difference between the instantaneous spatial angle and the fixed spatial angle to generate a single set of basic deviations.

[0135] Here, a calibration rod pair refers to a combination of two calibration rods arbitrarily selected from the calibration rod array, used to establish a spatial angle constraint unit. The absolute difference refers to the absolute deviation between the instantaneous spatial angle and the fixed spatial angle, used to quantify the angular deviation of a single calibration rod pair. The single-group basic deviation refers to the absolute difference corresponding to a single calibration rod pair, serving as the basic calculation unit for spatial constraint deviation.

[0136] In this embodiment of the application, the absolute difference between the instantaneous spatial angle obtained in step b1 and the fixed spatial angle is first calculated; then, this difference is used as the single-base deviation of the current calibration rod pair to reflect the degree to which the instantaneous angle deviates from the theoretical value.

[0137] Step b3: Based on the spatial distribution density of the calibration rod pairs, generate the constraint weight coefficients of the calibration rod pairs.

[0138] The spatial distribution density refers to the number of normal sampling points within a pre-defined circular area centered on the calibration rod pair. A higher density indicates stronger vibration resistance in that area. The constraint weighting coefficient is a weighting coefficient generated based on the spatial distribution density. Higher density results in a greater weight, reflecting the differences in the reliability of spatial constraints.

[0139] In this embodiment of the application, the absolute difference between the instantaneous spatial angle obtained in step b1 and the fixed spatial angle is first calculated; then, this difference is used as the single-base deviation of the current calibration rod pair to reflect the degree to which the instantaneous angle deviates from the theoretical value.

[0140] Step b4: Multiply the constraint weight coefficient by the single-group basic deviation of the corresponding calibration rod pair, and sum the multiplication results of all calibration rod pairs to obtain the basic deviation term.

[0141] Among them, the basic deviation term refers to the sum of the results of multiplying all calibration rods, which comprehensively characterizes the degree of deviation of the reconstructed value from the overall spatial constraints.

[0142] In this embodiment, the constraint weight coefficient of each calibration rod pair is first multiplied by its single-group basic deviation; then, the multiplication results of all calibration rod pairs are accumulated; finally, the basic deviation term is obtained, which comprehensively reflects the total spatial constraint deviation of all calibration rod pairs.

[0143] Step b5: Calculate the temporal deviation between the reconstructed value of the abnormal sampling point and the parameters of the adjacent normal sampling points, and assign temporal smoothing weights to different normal sampling points to generate a temporal smoothing deviation term.

[0144] Among them, temporal deviation refers to the absolute difference between the reconstructed value of an abnormal sampling point and the parameters of adjacent normal sampling points, reflecting the degree of abrupt change in the data in the time dimension. Temporal smoothing weight refers to the time decay coefficient generated based on the timestamp interval; the smaller the interval, the greater the weight, reflecting the principle of temporal proximity. The temporal smoothing deviation term is the weighted sum of each temporal deviation multiplied by the temporal smoothing weight, used to suppress temporal jumps in the reconstruction result.

[0145] In this embodiment, the absolute difference between the reconstructed value of the abnormal sampling point and the parameters of the adjacent normal sampling points is first calculated; then, a time decay type time series smoothing weight is assigned according to the timestamp interval between the normal sampling point and the abnormal point; finally, each absolute difference is multiplied by the corresponding weight and summed to generate a time series smoothing deviation term.

[0146] Step b6: Superimpose the basic deviation term and the temporal smoothing deviation term to construct the geometric reconstruction objective function.

[0147] In this embodiment, the basic deviation term and the temporal smoothing deviation term are directly added to construct the geometric reconstruction objective function. The smaller the value of this function, the more the reconstruction result satisfies both spatial constraints and temporal continuity.

[0148] Here is a specific example: First, based on the reconstructed values, the instantaneous spatial angles pointing to the three sets of calibration pole pairs are calculated, and the corresponding fixed spatial angles are extracted. Next, the absolute difference between each set of angles is calculated as the basic deviation for a single set. Then, the number of normal points around each set of calibration poles is counted to generate a spatial distribution density, which is converted into constraint weight coefficients and multiplied by the basic deviation for a single set. Subsequently, the products of all sets are summed to obtain the basic deviation term. Simultaneously, the difference between the reconstructed values ​​and the parameters of the preceding and following normal points is calculated, and a decay weight is assigned according to the time interval to generate a time-series smoothing deviation term. Finally, the two types of deviation terms are superimposed to construct the objective function, driving the subsequent optimization process.

[0149] By executing steps b1 to b6, the embodiments of this application ensure the strength of geometric constraints through a basic deviation term weighted by spatial distribution density, and maintain data continuity by combining a temporal smoothing deviation term weighted by time decay, thereby constructing an objective function that takes into account both spatial consistency and temporal stability, effectively balancing multi-dimensional optimization objectives.

[0150] In one possible embodiment, S13, the laser spot displacement is input into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps, including:

[0151] Step 131: Separate the horizontal displacement component and the vertical displacement component from the original signal of the laser spot displacement.

[0152] The original signal refers to the analog voltage signal or digital coded signal output by the position-sensitive device, containing the position encoding information of the light spot on the two-dimensional photosensitive surface. The horizontal displacement component refers to the displacement projection of the original signal along the X-axis of the photosensitive surface after coordinate analysis, reflecting the degree of left-right offset of the light spot. The vertical displacement component refers to the displacement projection of the original signal along the Y-axis of the photosensitive surface after coordinate analysis, reflecting the degree of up-down offset of the light spot. The original signal refers to the unanalyzed physical electrical signal directly output by the photosensitive surface of the position-sensitive device. Its essence is the projection position encoding of the light spot center in the two-dimensional coordinate system of the photosensitive surface, containing mixed information of the horizontal and vertical axes but without separating the directional components. The laser spot displacement refers to the physical displacement data generated after analyzing and converting the original signal, specifically including the horizontal and vertical displacement components. Its generation process is as follows: first, the original signal is digitized through analog-to-digital conversion; second, the electrical signal is converted into physical displacement values ​​based on the calibration parameters of the photosensitive surface; and finally, the horizontal and vertical components are separated by coordinate axis projection. In short, the raw signal is the data source and unprocessed form of the displacement, while the displacement is the quantifiable physical output of the raw signal after analysis, calibration, and direction separation.

[0153] In this embodiment, the original signal of the laser spot displacement output by the position-sensitive device is first received; then, the high-frequency oscillation component and the low-frequency drift component in the original signal are separated by frequency domain filtering technology in the digital signal processing algorithm; finally, the horizontal displacement component and the vertical displacement component are extracted according to the coordinate axis definition of the photosensitive surface of the position-sensitive device.

[0154] Step 132: Calculate the instantaneous yaw angle based on the horizontal displacement component and the instantaneous pitch angle based on the vertical displacement component.

[0155] The instantaneous yaw angle is obtained by dividing the horizontal displacement component by the radial distance to calculate the arctangent, reflecting the horizontal deflection angle of the laser beam in real time. The instantaneous pitch angle is obtained by dividing the vertical displacement component by the radial distance to calculate the arctangent, reflecting the vertical tilt angle of the laser beam in real time.

[0156] In this embodiment, the horizontal displacement component is first divided by the straight-line distance from the laser seeker to the calibration rod surface to obtain the horizontal offset tangent value; then, the instantaneous yaw angle value is calculated using the arctangent function; simultaneously, the vertical displacement component is divided by the same straight-line distance to obtain the vertical offset tangent value; finally, the instantaneous pitch angle value is calculated using the arctangent function.

[0157] Step 133: Take the straight-line distance between the laser guide head and the surface of the calibration rod as the radial displacement value.

[0158] Here, the straight-line distance refers to the physical straight-line distance from the optical center of the seeker to the reflective surface of the calibration rod, as measured by the laser time-of-flight ranging module. The radial displacement value is a real-time measurement equivalent to the straight-line distance, reflecting the radial relative position change between the calibration rod and the seeker.

[0159] In this embodiment, the straight-line distance from the optical center of the laser seeker to the reflective surface of the calibration rod, measured in real time by the laser ranging module, is directly used as the radial displacement value.

[0160] Step 134: Synchronously collect the timestamps corresponding to the instantaneous yaw angle, instantaneous pitch angle, and radial displacement values ​​to obtain a timestamp set.

[0161] The timestamp set stores time stamp containers for all synchronous acquisition moments, with each timestamp associated with a set of angle and displacement data.

[0162] In this embodiment, a millisecond-level timestamp is first generated using a high-precision clock source; then, the calculation completion times of instantaneous yaw angle, instantaneous pitch angle, and radial displacement are synchronously collected; finally, the three are bound to the same timestamp to form a timestamp set.

[0163] Step 135: Based on the timestamp set, integrate the instantaneous yaw angle value, instantaneous pitch angle value, and radial displacement value in the order of the timestamps to generate an angle and displacement time sequence containing timestamps.

[0164] In this embodiment, all data points are first arranged in ascending order of timestamps in the timestamp set; then, the instantaneous yaw angle, instantaneous pitch angle, and radial displacement value corresponding to each timestamp are integrated into a data tuple; finally, all tuples are concatenated in chronological order to generate an angle and displacement time sequence.

[0165] Here is a specific example: First, the horizontal and vertical displacement components are separated from the raw signal of the position-sensitive device. Next, the horizontal component is divided by the laser ranging value to obtain the arctangent, yielding the instantaneous yaw angle; the vertical component is divided by the same ranging value to obtain the instantaneous pitch angle. Simultaneously, the laser ranging value is directly used as the radial displacement value. Then, the timestamps of the three parameters are synchronously acquired using a nanosecond-level clock to form a timestamp set. Finally, the yaw, pitch, and radial displacement values ​​corresponding to all timestamps are integrated in chronological order to generate a time series.

[0166] By executing steps 131 to 135, this embodiment of the application achieves real-time conversion of high-precision angle parameters through displacement component calculation and timestamp synchronization mechanism, constructs a complete spatiotemporally correlated observation sequence, and provides a reliable data foundation for subsequent anomaly detection and parameter calibration.

[0167] In one possible embodiment, step b3, generating constraint weight coefficients for the calibration rod pairs based on the spatial distribution density of the calibration rod pairs, includes:

[0168] Step b31: Calculate the number of adjacent normal sampling points within the spatial neighborhood to obtain the spatial distribution density value. The spatial neighborhood is a circular area with the calibration rod pair as the geometric center and the preset vibration characteristic radius as the radius.

[0169] The spatial neighborhood refers to a circular physical area centered on the geometric center of the calibration rod pair and with a preset vibration characteristic radius as its radius, used to define the density statistical range. The spatial distribution density value refers to the statistical value of the number of adjacent normal sampling points within the spatial neighborhood, reflecting the data acquisition stability of that area. The geometric center is the midpoint of the three-dimensional spatial coordinates of the two calibration rods in the calibration rod pair, serving as the density statistical reference point.

[0170] The center of the circle refers to the central point of the spatial neighborhood, which coincides with the geometric center of the calibration rod. The preset vibration characteristic radius is an empirical value of the radius set according to the typical vibration wavelength of the vehicle body, used to define the range of vibration influence. The circular region refers to a planar circular space with the geometric center as the center and the vibration characteristic radius as the radius.

[0171] In this embodiment, a circular region is first defined as the spatial neighborhood, with the geometric center point of the calibration rod pair as the center and the preset vibration characteristic radius as the length. Then, the number of all adjacent normal sampling points within the circular region is counted. Finally, the count is output as the spatial distribution density value.

[0172] Step b32: Based on the spatial distribution density value, generate the density influence factor through a piecewise mapping function.

[0173] The piecewise mapping function refers to two calculation rules based on the vibration density threshold: linear amplification is used for low-density areas, and logarithmic compression is used for high-density areas. The density influence factor is an intermediate variable after the spatial distribution density value is transformed by the piecewise function, used to eliminate dimensional differences.

[0174] In this embodiment, the vibration density threshold is first determined based on the material stiffness coefficient of the vehicle platform and the spacing of the calibration rod array; secondly, when the spatial distribution density value is less than the vibration density threshold, the density value is multiplied by the first material response constant to generate a density influence factor; when the spatial distribution density value is greater than or equal to the vibration density threshold, the natural logarithm of the density value is multiplied by the second material response constant to generate a density influence factor.

[0175] Step b33: Linearly map the density influence factor to the corresponding preset weight interval to generate the constraint weight coefficients of the calibration rod pair.

[0176] In this embodiment, firstly, the upper and lower limits of the preset weight range of the constraint weight coefficient are set; secondly, the density influence factor is subtracted from the minimum influence factor benchmark value and then divided by the influence factor variation range to obtain the normalized ratio; finally, the normalized ratio is multiplied by the weight range width and the lower limit value of the range is added to generate the constraint weight coefficient of the calibration rod pair.

[0177] Here is a specific example: First, using the center point of the calibration rod as the center, a circular area is delineated with a radius set according to the vibration wavelength. The number of normal points within the area is counted to obtain the spatial distribution density value. Second, a density threshold is set based on the stiffness of the vehicle body material: when the density value is below the threshold, it is multiplied by a linear coefficient to generate an influence factor; when it is above the threshold, the logarithm is taken and then multiplied by a compression coefficient to generate an influence factor. Finally, the influence factors are linearly mapped to a weight interval, and the constraint weight coefficients are output.

[0178] By executing steps b31 to b33, this embodiment of the application dynamically defines the statistical region through the vibration characteristic radius, and combines the segmented mapping mechanism driven by material properties to achieve adaptive conversion of spatial density to weight coefficient, thereby ensuring the strengthening of the constraint force in the high stability region.

[0179] In one possible embodiment, step b32, generating a density influence factor based on the spatial distribution density value using a piecewise mapping function, includes:

[0180] Step c1: Determine the vibration density threshold based on the material stiffness coefficient of the vehicle platform and the spacing of the calibration rod array.

[0181] The material stiffness coefficient refers to the measured value of the elastic modulus of the vehicle platform material, reflecting the material's ability to resist deformation, and is measured in gigapascals (GPa). The spacing of the calibration rod array refers to the average straight-line distance between the center points of adjacent calibration rods in the array, affecting the vibration wave propagation characteristics. The vibration density threshold is the critical value for dividing high and low density zones, obtained by multiplying the material stiffness coefficient by the calibration rod spacing by 0.25.

[0182] In this embodiment, the material stiffness coefficient of the vehicle platform is first loaded; then the measured data of the spacing of the calibration rod array is obtained; finally, the material stiffness coefficient is multiplied by the spacing of the calibration rod array and then multiplied by a preset scaling factor to generate a vibration density threshold.

[0183] Step c2: Determine the first material response constant and the second material response constant based on the sampling frequency of the position-sensitive device and the surface reflectivity of the calibration rod.

[0184] The sampling frequency refers to the number of times the position-sensitive device collects displacement data per second, determining the data temporal resolution. The surface reflectivity of the calibration rod is the measured optical reflection efficiency of the laser reflective coating on the calibration rod, affecting the signal-to-noise ratio. The first material response constant refers to the linear amplification factor in the low-density region, obtained by dividing the sampling frequency by the 1kHz reference frequency and then multiplying by the reflectivity percentage. The second material response constant refers to the logarithmic compression factor in the high-density region, obtained by dividing the sampling frequency by the 1kHz reference frequency and then dividing by the reflectivity percentage. The first material response constant is the linear amplification factor in the low-density region; when the spatial distribution density value is less than the vibration density threshold, it is multiplied by the density value to achieve linear amplification. The second material response constant is the logarithmic compression factor in the high-density region; when the spatial distribution density value is greater than or equal to the vibration density threshold, it is multiplied by the natural logarithm of the density value to achieve order-of-magnitude compression.

[0185] In this embodiment, the sampling frequency parameter of the position-sensitive device is first read; then the optical detection value of the reflectivity of the calibration rod surface is obtained; finally, the sampling frequency is divided by the reference sampling rate to obtain the frequency scaling factor, the reflectivity is divided by the standard reflectivity to obtain the reflection correction factor, the frequency scaling factor and the reflection correction factor are multiplied to generate the first material response constant, and the frequency scaling factor is divided by the reflection correction factor to generate the second material response constant.

[0186] Step c3: When the spatial distribution density value is less than the vibration density threshold, multiply the spatial distribution density value by the first material response constant to generate the density influence factor.

[0187] In this embodiment, when the spatial distribution density value is less than the vibration density threshold, the spatial distribution density value is directly multiplied by the first material response constant to output the linearly amplified density influence factor.

[0188] Step c4, or, when the spatial distribution density value is greater than or equal to the vibration density threshold, multiply the natural logarithm of the spatial distribution density value by the second material response constant to generate the density influence factor.

[0189] In this embodiment of the application, when the spatial distribution density value is greater than or equal to the vibration density threshold, the natural logarithm of the spatial distribution density value is first calculated; then the natural logarithm value is multiplied by the second material response constant to output the density influence factor after logarithmic compression.

[0190] Here is a specific example: First, the vibration density threshold is calculated based on the stiffness coefficient of the aluminum alloy vehicle body and the 20-centimeter spacing of the calibration rods. Second, based on the 2 kHz sampling frequency and 85% reflectivity of the position-sensitive device, the first and second response constants are calculated respectively. Then, when the number of normal points in a certain calibration rod pair area is lower than the threshold, the density value is multiplied by the first constant to generate an influence factor; when the number of normal points in another area exceeds the threshold, the natural logarithm of the density value is multiplied by the second constant to generate an influence factor.

[0191] By executing steps c1 to c4, the embodiments of this application achieve differentiated mapping processing of high and low density regions through the design of dynamic response constants driven by material stiffness and optical parameters, ensuring sensitivity in low density regions while avoiding oversaturation of weights in high density regions.

[0192] Figure 2 This application provides a schematic diagram of a rapid calibration system for the angle measurement accuracy of a laser seeker, as shown in the embodiments below. Figure 2 As shown, the system includes:

[0193] The acquisition module 21 is used to acquire the pitch angle, yaw angle and roll angle of each calibration rod in the calibration rod array fixed to the vehicle platform in the vehicle coordinate system, and generate the three-dimensional spatial coordinates of each calibration rod.

[0194] The acquisition module 22 is used to emit a laser beam to the calibration rod array through the laser guide head, and to acquire the displacement of the laser spot reflected from the surface of each calibration rod through the position sensitive device.

[0195] The generation module 23 is used to input the laser spot displacement into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps.

[0196] The identification module 24 is used to identify abnormal sampling points in the angle and displacement time sequence.

[0197] The reconstruction module 25 is used to perform geometric reconstruction on abnormal sampling points by using a spatial geometric constraint optimization algorithm combined with a fixed spatial angle constraint relationship, so as to obtain the angle measurement error calibration parameters of the laser seeker.

[0198] Figure 2 The aforementioned rapid calibration system for the angle measurement accuracy of a laser seeker can perform... Figure 1 The implementation principle and technical effects of the rapid calibration method for the angle measurement accuracy of a laser seeker described in the illustrated embodiment will not be repeated here. The specific operation methods of each module and unit in the rapid calibration system for the angle measurement accuracy of a laser seeker in the above embodiments have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0199] In one possible design, Figure 2 The rapid calibration system for the angle measurement accuracy of a laser seeker, as shown in the embodiment, can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32.

[0200] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0201] The processing component 32 is used to: obtain the pitch angle, yaw angle and roll angle of each calibration rod in the calibration rod array fixed to the vehicle platform in the vehicle coordinate system, and generate the three-dimensional spatial coordinates of each calibration rod;

[0202] A laser beam is emitted from the laser seeker to the calibration rod array, and the displacement of the laser spot reflected from the surface of each calibration rod is collected by a position-sensitive device. The laser spot displacement is input into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps. Abnormal sampling points in the angle and displacement time sequence are identified. A spatial geometric constraint optimization algorithm is used, combined with a fixed spatial angle constraint relationship, to geometrically reconstruct the abnormal sampling points and obtain the angle measurement error calibration parameters of the laser seeker.

[0203] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0204] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as Random Access Memory (RAM), Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0205] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0206] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0207] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0208] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0209] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown illustrates a rapid calibration method for the angular measurement accuracy of a laser seeker.

[0210] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0211] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0212] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0213] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A rapid calibration method for the angle measurement accuracy of a laser seeker, characterized in that, include: Obtain the pitch angle, yaw angle, and roll angle of each calibration rod in the calibration rod array fixed to the vehicle platform in the vehicle coordinate system, and generate the three-dimensional spatial coordinates of each calibration rod; A laser beam is emitted to the calibration rod array through a laser guide head, and the displacement of the laser spot reflected from the surface of each calibration rod is collected by a position-sensitive device. The laser spot displacement is input into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps; Identify anomalous sampling points in the angle and displacement time series; A spatial geometric constraint optimization algorithm is used, combined with a fixed spatial angle constraint relationship, to perform geometric reconstruction on abnormal sampling points and obtain the angle measurement error calibration parameters of the laser seeker.

2. The method according to claim 1, characterized in that, The aforementioned spatial geometric constraint optimization algorithm, combined with fixed spatial angle constraint relationships, performs geometric reconstruction on abnormal sampling points to obtain the angle measurement error calibration parameters of the laser seeker, including: Based on the three-dimensional spatial coordinates of each calibration rod in the calibration rod array, the fixed spatial angle constraint relationship between any two calibration rods is extracted, and the fixed spatial angle constraint equation is established. Based on the fixed spatial angle constraint equation, a geometric reconstruction objective function containing two different deviation terms is constructed, wherein the geometric reconstruction objective function uses the reconstruction value as the optimization variable; A spatial geometric constraint optimization algorithm is adopted. The fixed spatial angle constraint equation and the geometric reconstruction objective function are input into the solver. The parameters of the normal sampling points adjacent to the abnormal sampling points are used as the initial iteration values. The gradient direction of the optimization variables is calculated by the constraint Jacobian matrix. The reconstruction value is iteratively optimized by combining the fixed spatial angle constraint equation, so that the value of the geometric reconstruction objective function gradually decreases. When the geometric reconstruction objective function value satisfies the preset convergence condition, the reconstruction value that satisfies the fixed spatial angle constraint relationship will be used as the geometric reconstruction result of the abnormal sampling point. Based on the geometric reconstruction results of all abnormal sampling points, the angular measurement error calibration parameters of the laser seeker are calculated.

3. The method according to claim 2, characterized in that, The process of using the parameters of the normal sampling points adjacent to the abnormal sampling points as initial iteration values, calculating the gradient direction of the optimization variables through the constraint Jacobian matrix, and iteratively optimizing the reconstructed values ​​in combination with the fixed spatial angle constraint equation includes: The parameters of the normal sampling points adjacent to the abnormal sampling point are selected, and corresponding weights are assigned according to the timestamp interval. The initial iterative value of the reconstructed value of the abnormal sampling point is obtained by weighted calculation. Based on the fixed spatial angle constraint equation and the iteration value, the partial derivative of the reconstructed value is obtained to construct the constraint Jacobian matrix. The elements in the constraint Jacobian matrix are used to reflect the sensitivity of each reconstructed value to the fixed spatial angle constraint equation. The iteration value is the initial iteration value in the first generation and the reconstructed value adjusted in the current generation in other generations. By combining the gradient vector of the geometric reconstruction objective function with the constraint Jacobian matrix, the search direction of the optimization variables is determined by the Lagrange multiplier method; The adjustment direction of the reconstructed value is determined according to the search direction, and a single-step adjustment amount is generated using an adaptive step size strategy. The reconstructed value is then adjusted based on the single-step adjustment amount. Substitute the adjusted reconstruction value into the fixed spatial angle constraint equation, and calculate the geometric reconstruction objective function value corresponding to the adjusted reconstruction value based on the geometric reconstruction objective function, provided that the constraint is satisfied. Calculate the deviation between the current generation's geometric reconstruction objective function value and the previous generation's geometric reconstruction objective function value; If the deviation of three consecutive iterations is less than the preset threshold, the preset convergence condition is satisfied. If the deviation of three consecutive iterations is not less than the preset threshold, the construction of the constraint Jacobian matrix is ​​repeated and the iteration continues, with the reconstructed value after the current generation as the new iteration starting point. Alternatively, if the number of iterations exceeds the maximum threshold, the value of the single-step adjustment is reduced, and the adjustment of the reconstructed value is repeated based on the reduced single-step adjustment value, and the number of iterations is reset to continue iterating.

4. The method according to claim 2, characterized in that, The construction of a geometric reconstruction objective function based on the fixed spatial angle constraint equation, comprising two different deviation terms, includes: Based on the reconstructed values ​​of the abnormal sampling points, the instantaneous spatial angle pointing to any two calibration rods is calculated, and the fixed spatial angle of the corresponding calibration rod pair is analyzed using the fixed spatial angle constraint equation. Calculate the absolute difference between the instantaneous spatial angle and the fixed spatial angle to generate a single set of basic deviations; Based on the spatial distribution density of the calibration rod pairs, constraint weight coefficients for the calibration rod pairs are generated; Multiply the constraint weight coefficient by the single set of basic deviations of the corresponding calibration rod pair, and sum the multiplication results of all calibration rod pairs to obtain the basic deviation term; The temporal deviation between the reconstructed value of the abnormal sampling point and the parameters of the adjacent normal sampling points is calculated, and different temporal smoothing weights are assigned to different normal sampling points to generate a temporal smoothing deviation term. The basic deviation term and the temporal smoothing deviation term are superimposed to construct the geometric reconstruction objective function.

5. The method according to claim 1, characterized in that, The step of inputting the laser spot displacement into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps includes: The horizontal displacement component and the vertical displacement component are separated from the original signal of the laser spot displacement. The instantaneous yaw angle is calculated based on the horizontal displacement component, and the instantaneous pitch angle is calculated based on the vertical displacement component. The straight-line distance between the laser guide head and the surface of the calibration rod is taken as the radial displacement value; The timestamps corresponding to the instantaneous yaw angle value, the instantaneous pitch angle value, and the radial displacement value are collected synchronously to obtain a timestamp set; Based on the timestamp set, the instantaneous yaw angle value, instantaneous pitch angle value, and radial displacement value are integrated in the order of the timestamps to generate an angle and displacement time sequence containing timestamps.

6. The method according to claim 4, characterized in that, The step of generating constraint weight coefficients for the calibration rod pairs based on their spatial distribution density includes: The number of adjacent normal sampling points within the spatial neighborhood is calculated to obtain the spatial distribution density value, wherein the spatial neighborhood is a circular area with the calibration rod pair as the geometric center and a preset vibration characteristic radius as the radius; Based on the spatial distribution density value, a density influence factor is generated through a piecewise mapping function; The density influence factor is linearly mapped to the corresponding preset weight interval to generate the constraint weight coefficients of the calibration rod pair.

7. The method according to claim 6, characterized in that, The generation of density influence factors based on the spatial distribution density value through a piecewise mapping function includes: The vibration density threshold is determined based on the material stiffness coefficient of the vehicle platform and the spacing of the calibration rod array; The first and second material response constants are determined based on the sampling frequency of the position-sensitive device and the surface reflectivity of the calibration rod. When the spatial distribution density value is less than the vibration density threshold, the spatial distribution density value is multiplied by the first material response constant to generate a density influence factor; Alternatively, when the spatial distribution density value is greater than or equal to the vibration density threshold, the natural logarithm of the spatial distribution density value is multiplied by the second material response constant to generate a density influence factor.

8. A rapid calibration system for the angle measurement accuracy of a laser seeker, characterized in that, include: The acquisition module is used to acquire the pitch angle, yaw angle and roll angle of each calibration rod in the calibration rod array fixed to the vehicle body platform in the vehicle body coordinate system, and generate the three-dimensional spatial coordinates of each calibration rod; The acquisition module is used to emit a laser beam to the calibration rod array through a laser guide head, and to acquire the displacement of the laser spot reflected from the surface of each calibration rod through a position-sensitive device; The generation module is used to input the laser spot displacement into the spot displacement feature extraction network to generate an angle and displacement time sequence containing timestamps; The identification module is used to identify abnormal sampling points in the angle and displacement time sequence; The reconstruction module is used to perform geometric reconstruction of abnormal sampling points by using a spatial geometric constraint optimization algorithm combined with a fixed spatial angle constraint relationship, so as to obtain the angle measurement error calibration parameters of the laser seeker.

9. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement a rapid calibration method for the angle measurement accuracy of a laser seeker as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a computer, implements a rapid calibration method for the angle measurement accuracy of a laser seeker as described in any one of claims 1 to 7.

Images