Catwalk line dynamic prediction and automatic control method based on real-time monitoring data

Abstract

This application discloses a method for dynamic prediction and automatic adjustment of catwalk alignment based on real-time monitoring data. The method includes: real-time acquisition of the vertical coordinates of each monitoring point; dividing the load-bearing cable into several cable segments according to the sag adjustment rope support points; calculating the equivalent vertical stiffness coefficient based on the catenary mechanical parameters of each cable segment; constructing an influence matrix between the change in rope length at the support points and the change in vertical displacement at the monitoring points based on the equivalent vertical stiffness coefficient; after sag adjustment, updating the equivalent vertical stiffness coefficient and recalculating the influence matrix using a recursive estimation method based on the actual change in rope length, the measured change in vertical displacement, and the predicted residual; comparing the real-time alignment with the target alignment; and solving a regularized least squares problem when the deviation exceeds an allowable threshold to obtain the coordinated adjustment amount for each support point. This application can improve the accuracy of catwalk alignment prediction and adjustment, and reduce repeated trial adjustments.

Description

Technical Field

[0001] This application relates to the field of bridge construction monitoring and control technology, specifically to a method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data. Background Technology

[0002] When constructing the main cable of a long-span suspension bridge using the overhead cable-spinning method, the catwalk is located below the designed position of the main cable and supported by the load-bearing cables. As the steel wires are laid in batches, the load on the load-bearing cables continuously increases, and their sag, tension, and overall alignment of the catwalk change accordingly. To effectively control the difference in wire length within the same strand, it is necessary to adjust the sag adjustment rope in a timely manner so that the position of the catwalk meets the quality control requirements of each construction stage.

[0003] Current methods for adjusting catwalk alignment primarily rely on manual measurement and point-by-point trial adjustments. This involves shortening or lengthening the sag adjustment rope based on deviations at local measuring points, followed by repeated measurements and corrections. Because the load-bearing cable is a continuous, flexible cable structure, any change in the position of any support point (i.e., the cable strand forming device) will cause a redistribution of tension in adjacent cable segments and affect displacement in other locations, easily leading to a situation where local deviations decrease while deviations in adjacent areas increase. For ultra-long span bridges, with numerous support points and high cable segment flexibility, multiple rounds of adjustments are typically required, resulting in lengthy construction times.

[0004] Furthermore, the static model established based on the initial design parameters is insufficient to reflect the actual changes in condition caused by factors such as increased wire load, temperature variations, installation deviations, and slippage at the anchor ends. If the influence relationships between each support point cannot be corrected in a timely manner using measured data, subsequent adjustments are prone to errors; if all influence relationships are directly identified, extensive experimental adjustments are required, which is difficult to meet the requirements of construction progress and safety control. Therefore, how to predict changes in the catwalk alignment based on real-time monitoring data and implement coordinated adjustments while considering the mutual influence of each support point remains an urgent problem to be solved in catwalk construction control. Summary of the Invention

[0005] To overcome the problems existing in related technologies, this specification provides a method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data.

[0006] According to a first aspect of the embodiments of this application, a method for dynamic prediction and automatic adjustment of catwalk alignment based on real-time monitoring data is provided, including: The vertical coordinates of each monitoring point are collected in real time; the load-bearing cable is divided into several cable segments according to the position of each support point of the sag adjustment rope, and the equivalent vertical stiffness coefficient of each cable segment is calculated based on the catenary mechanical parameters of each cable segment. An influence matrix is ​​constructed based on the equivalent vertical stiffness coefficient. The influence matrix represents the linear mapping relationship between the change in rope length at each support point and the change in vertical displacement at each monitoring point, with the equivalent vertical stiffness coefficient vector as the parameter. After each adjustment of the sag adjustment rope, the actual rope length change at each support point and the measured vertical displacement change at each monitoring point are recorded. Based on the prediction residual between the measured value and the predicted value of the influence matrix, the equivalent vertical stiffness coefficient vector is updated using a recursive estimation method, and the influence matrix is ​​recalculated. The vertical coordinates of each monitoring point in real time are compared with the target line shape to obtain the line shape deviation vector. When the deviation exceeds the allowable threshold, the regularized least squares problem is solved on the line shape deviation vector using the current influence matrix to obtain the coordination adjustment amount of each support point and output it.

[0007] As an optional approach, the calculation of the equivalent vertical stiffness coefficient of each cable segment based on the catenary mechanical parameters of each segment includes: linearizing the catenary equation of each cable segment at the current equilibrium position with small displacements, and calculating the equivalent vertical stiffness coefficient based on the horizontal component force, horizontal projected span, and vertical load per unit horizontal projected length of the cable segment; the horizontal component force is determined by iteratively solving a system of multiple catenary equilibrium equations for the entire load-bearing cable.

[0008] As an optional approach, the construction of the influence matrix based on the equivalent vertical stiffness coefficient includes: applying a unit vertical displacement to each support point, using the zero displacement of the fixed support points at both ends of the load-bearing cable as the boundary condition, and establishing vertical force balance equations for the remaining support points; in the vertical force balance equations, the vertical restoring forces of the left and right cable segments acting on each support point are represented by the product of the equivalent vertical stiffness coefficient of the corresponding cable segment and the displacement difference between adjacent nodes, and the sum of the two is zero; the vertical force balance equations constitute a tridiagonal linear equation system, which is solved using the chasing method; the monitoring points located between adjacent support points obtain the corresponding influence coefficients by linear interpolation of the displacement responses of the support points at both ends of the cable segment.

[0009] As an optional approach, the recursive estimation method is the recursive Gauss-Newton method; the Jacobian matrix in the update process is obtained by taking the partial derivatives of both sides of the equation system of the influence matrix with respect to the equivalent vertical stiffness coefficient of each cable segment, and the coefficient matrix decomposition result of the original equation system is reused to solve the problem.

[0010] As an optional approach, in the recursive Gauss-Newton method, the observation noise covariance matrix is ​​a diagonal matrix, and the diagonal elements are the squares of the vertical measurement standard deviations of the monitoring sensors; the initial error covariance matrix of the parameter estimation is a diagonal matrix, and each diagonal element is calculated based on the estimated value of the initial equivalent vertical stiffness coefficient of the corresponding cable segment and the relative uncertainty coefficient. The relative uncertainty coefficient is determined by propagating the coefficient of variation of the load and tension of the load-bearing cable through the stiffness formula.

[0011] As an optional approach, the method also includes adjustment verification and parameter feedback steps: After the sag adjustment rope corresponding to the coordinated adjustment amount is adjusted, the steady-state vertical coordinates of each monitoring point after adjustment are collected, and the measured vertical displacement change of each monitoring point before and after adjustment is calculated; the measured vertical displacement change is subtracted from the predicted displacement change calculated by the influence matrix based on the actual rope length change to obtain the prediction residual; when the L2 norm of the prediction residual exceeds the update trigger threshold, the recursive estimation method is executed using the actual rope length change and the measured vertical displacement change as observation data to update the equivalent vertical stiffness coefficient vector and recalculate the influence matrix; after the update, the linear deviation vector is recalculated, and if the deviation still exceeds the allowable threshold, the regularized least squares problem is solved again using the updated influence matrix to obtain the supplementary coordinated adjustment amount.

[0012] As an optional approach, the average value of the vertical coordinate data of each continuously collected monitoring point is taken as the average value of a sliding window for a set time period; the steady-state vertical coordinates before and after the adjustment operation are taken as the coordinate values ​​when the difference between the average values ​​of two adjacent sliding windows is less than the set stability criterion.

[0013] As an alternative, when the catwalk is a multi-span continuous structure, a vertical stiffness constraint term at the top of the tower is added to the vertical force balance equation at the support point where the cable saddle is located. The solution range of the influence matrix is ​​then extended from a single span to adjacent spans.

[0014] According to a second aspect of the embodiments of this application, a catwalk linear dynamic prediction and automatic control system based on real-time monitoring data is also provided, comprising: The monitoring and stiffness calculation module is used to collect the vertical coordinates of each monitoring point in real time; the load-bearing cable is divided into several cable segments according to the position of each support point of the sag adjustment rope, and the equivalent vertical stiffness coefficient of each cable segment is calculated based on the catenary mechanical parameters of each cable segment. An influence matrix construction module is used to construct an influence matrix based on the equivalent vertical stiffness coefficient. The influence matrix represents the linear mapping relationship between the change in rope length at each support point and the change in vertical displacement at each monitoring point, using the equivalent vertical stiffness coefficient vector as a parameter. The parameter update module is used to record the actual rope length change at each support point and the measured vertical displacement change at each monitoring point after each sag adjustment rope adjustment. Based on the prediction residual between the measured values ​​and the predicted values ​​of the influence matrix, the equivalent vertical stiffness coefficient vector is updated using a recursive estimation method, and the influence matrix is ​​recalculated. The coordination and control module is used to compare the real-time vertical coordinates of each monitoring point with the target line shape to obtain the line shape deviation vector. When the deviation exceeds the allowable threshold, the module uses the current influence matrix to solve the regularized least squares problem on the line shape deviation vector to obtain the coordination adjustment amount of each support point and output it.

[0015] According to a third aspect of the embodiments of this application, an electronic device is provided, including: a processor; a memory for storing a computer program executable by the processor; wherein the processor is configured to execute the computer program in the memory to implement the method described in the first aspect.

[0016] This application determines the equivalent vertical stiffness coefficient based on the catenary mechanical parameters of each cable segment of the load-bearing cable, and constructs an influence matrix between the rope length change at the support point and the vertical displacement change at the monitoring point using this coefficient. This allows the mutual influence relationship between the support points of the catwalk to be characterized with fewer physical parameters, avoiding the need for multiple experimental adjustments required to directly identify a large number of matrix elements. During the spinning process, the equivalent vertical stiffness coefficient is recursively updated using the rope length change and measured vertical displacement change formed by each actual adjustment. This allows the influence matrix to be corrected in a timely manner according to changes in the load, tension, and on-site conditions of the load-bearing cable, thereby improving the consistency between the predicted alignment and the actual response. When the real-time alignment deviation exceeds the allowable range, the coordinated adjustment amount of each support point is solved based on the current influence matrix. The adjustment direction and adjustment range can be determined under the condition of comprehensively considering the displacement transmission relationship of the entire span, reducing repeated corrections caused by point-by-point trial adjustments. After the adjustment is completed, the measured residual is used to determine whether to update the parameters and calculate the supplementary adjustment amount, which helps to improve the accuracy and stability of the catwalk alignment control and shorten the on-site adjustment time.

[0017] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Furthermore, no embodiment in this disclosure is required to achieve all the effects described above. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this specification and, together with the description, serve to explain the principles of this specification.

[0019] Figure 1 This is a flowchart of a method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data in one embodiment of this application.

[0020] Figure 2 This is a flowchart of the recalculation of the influence matrix in one embodiment of this application.

[0021] Figure 3 This is a schematic block diagram of a catwalk linear dynamic prediction and automatic control system based on real-time monitoring data, as described in this application.

[0022] Figure 4 This is a structural diagram of an electronic device according to one embodiment of this application. Detailed Implementation

[0023] 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.

[0024] This embodiment is applicable to long-span suspension bridge projects using the Air Spinning (AS) method for erecting the main cable, especially suitable for ultra-long spans exceeding 1500 meters. The catwalk is erected below the designed position of the main cable, suspended and supported by the load-bearing cable. Sag adjustment ropes are installed at several locations along the bridge span direction, connecting the catwalk to the upper fixed components. The vertical position of the catwalk is adjusted by changing the rope length at each support point, matching the alignment of the load-bearing cable with the designed alignment of the air cable at each stage. The load-bearing cable is a continuous flexible cable along the bridge span direction. The cable segments at each support point are mechanically interconnected; that is, adjusting the rope length at a certain support point causes a redistribution of tension in adjacent cable segments, resulting in changes in the vertical displacement of nearby and even distant support points. During the spinning process, steel wires are placed on the catwalk in batches, the load on the load-bearing cable continuously increases, and the mutual influence between the support points also changes with the load state. In projects with a main span exceeding 1500 meters, the number of support points is large and the cable segments have high flexibility, which dramatically increases the number of adjustments and the time required by the traditional point-by-point trial-and-error method.

[0025] Please see Figure 1 , Figure 1 This is a flowchart of a method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data, as described in one embodiment of this application. Figure 1 As shown, the method includes steps 101-104: Step 101: Collect the vertical coordinates of each monitoring point in real time; divide the load-bearing cable into several cable segments according to the position of each support point of the sag adjustment rope, and calculate the equivalent vertical stiffness coefficient of each cable segment based on the catenary mechanical parameters of each cable segment.

[0026] In some embodiments, monitoring sensors are deployed along the catwalk support cable, and GNSS receivers are installed at each sag adjustment rope support point and at selected intermediate positions between adjacent support points to collect the vertical coordinates of each monitoring point in real time. Let the total number of sag adjustment rope support points be... The total number of monitoring points is , The sampling frequency of the GNSS receiver can be configured, for example, to 1 Hz; in order to reduce the instantaneous displacement fluctuations caused by wind vibration, the average value of the continuously acquired vertical coordinates is taken as a sliding window of a set duration, for example, 30 to 60 seconds, to filter out the low-frequency oscillation components of the catwalk and retain the quasi-static trend of the linear change.

[0027] Before and after each sag adjustment, the coordinate value at which the difference between the average values ​​of two adjacent sliding windows is less than the set stability criterion (e.g., 2 mm) is taken as the steady-state vertical coordinate. Optionally, a dual-axis tilt sensor can be installed between adjacent support points, with a sampling frequency configured, for example, 5 to 10 Hz. The average value of the sliding window is also used to supplement local slope information when GNSS signal accuracy is insufficient. Data from each sensor is transmitted to the field data processing terminal via wired or wireless means. The target vertical coordinates of each monitoring point at different spinning stages are calculated based on the design alignment of the main cable and the design clearance from the catwalk panel to the main cable, and are denoted as the target alignment. The process is updated according to the predetermined plan as the spinning stage progresses.

[0028] The load-bearing cable is divided into sections based on the support points of each sag adjustment rope. Section, of which Segment connection number The node and the first Nodes ( ), the 0th node and the Each node represents a fixed support point for the load-bearing cable at either end of the bridge span, such as the anchorage at the top of the tower or the cable saddle. Each section of the load-bearing cable, under its own weight and the weight of the existing steel wires, forms a catenary shape. Horizontal projection span of the segment (Unit: meters) Determined by the difference in horizontal coordinates between two adjacent nodes; vertical load per unit horizontal projected length (Unit: Newtons per meter) equals the sum of the self-weight of the load-bearing cable and the weight of the steel wire already laid within that section, divided by [missing value]. Horizontal component force (Unit: Newton) The parameters are determined by iteratively solving a system of equilibrium equations for multiple catenary segments along the entire load-bearing cable. This is a standard practice in the field of cable structure calculation. All parameters can be obtained based on design data and the current load condition before spinning begins.

[0029] In one embodiment, when the vertical position of a support point changes slightly, the tension of the cable segments on both sides of that support point changes accordingly, generating a vertical restoring force. To quantitatively describe the proportional relationship between this restoring force and displacement, an equivalent vertical stiffness coefficient is introduced. (Unit: Newtons per meter) refers to the magnitude of the vertical restoring force generated by the cable segment when one end is fixed and the other end undergoes a unit vertical displacement. By linearizing the catenary equation with small displacements at the current equilibrium position, we obtain: ; This correction term reflects the contribution of cable sag to vertical stiffness: the larger the sag-to-span ratio, the larger the correction term, meaning the cable segment is softer and the displacement recovery ability is weaker, consistent with the basic laws of flexible cable mechanics. When the sag-to-span ratio is small, the correction term is much smaller than 1. Approximately As an example, if a certain segment Newton rice, Newtons per meter, with a correction term of approximately , Newtons per meter.

[0030] Step 102: Construct an influence matrix based on the equivalent vertical stiffness coefficient. The influence matrix represents the linear mapping relationship between the change in rope length at each support point and the change in vertical displacement at each monitoring point, using the equivalent vertical stiffness coefficient vector as a parameter.

[0031] Specifically, define the influence matrix. ,element This indicates that only the first... With the sag adjustment rope at the first support point being 1 meter and the rope lengths at the other support points remaining unchanged, the first... Vertical displacement change at each monitoring point (positive upwards). Vector composed of the rope length changes at each support point. (Positive values ​​indicate shortening, negative values ​​indicate lengthening) and the vertical displacement change at each monitoring point. The linear mapping relationship between them is expressed as: ; This linear relationship is a first-order Taylor expansion approximation of the catenary equation at the current equilibrium position. The single adjustment amount of the sag adjustment rope is usually on the order of millimeters to centimeters, which is a small amount relative to the overall span and sag of the load-bearing cable.

[0032] The influence matrix is ​​represented by the equivalent vertical stiffness coefficient vector. Let it be a parameter, denoted as .all The matrix elements are composed of The stiffness parameters of each segment are uniquely determined, with , For example, 2400 elements are controlled by 41 parameters, reducing the number of parameters by nearly two orders of magnitude. The physical basis for this is that each segment of the load-bearing cable generates strict mechanical constraints through a continuous flexible cable structure. In this embodiment, the segment stiffness parameter refers to the equivalent vertical stiffness coefficient corresponding to each cable segment, that is, a parameter characterizing the proportional relationship between the vertical restoring force generated when the cable segment undergoes a small vertical displacement in the current equilibrium state and the vertical displacement itself, expressed in Newtons per meter. The segment stiffness parameters of each cable segment collectively constitute the equivalent vertical stiffness coefficient vector, which serves as the parameter to be updated in the influence matrix.

[0033] Specifically, the displacement generated when adjusting a certain support point decreases progressively to both sides, with the decrease in magnitude at each level determined by the corresponding cable segment. Decision; multiple [items] on the same decay path Share the same group Change any section This will simultaneously change multiple influence coefficients in the entire row and column related to that position. Therefore, there are correlations between matrix elements given by the mathematical structure of a tridiagonal linear equation system, and the amount of independent information is much smaller than the total number of elements.

[0034] Initial influence matrix Calculate column by column before spinning begins; specifically, for the first... List( ), in the Applying a unit vertical displacement to each support point The displacement of the fixed support points at both ends is zero. , ), the remaining free support points ( The vertical force equilibrium condition for the left cable segment is: the left cable segment relative to the support point The vertical restoring force generated is The restoring force generated by the right cable segment is The sum of the two is zero: ; This system of equations is composed of A system of three diagonal linear equations with unknown displacements at three free support points is solved using the Thomas algorithm, a standard method in numerical analysis for solving systems of three diagonal linear equations. Solving for the unknowns only requires approximately The time complexity of the next floating-point operation is... Proportional.

[0035] For monitoring points located between adjacent support points, their displacement response is obtained by linear interpolation of the displacements of the support points at both ends of the cable segment: Let the first... The monitoring point at the 1st Within the segment, the horizontal coordinate distance from the first The distance between each support point and the horizontal projection span of this segment The ratio is ( ),but ; The physical basis for this interpolation is that the displacement distribution within the cable segment is approximately linear within a small disturbance range, consistent with the linearization conclusion of the catenary. (Traversal) to Complete the calculations for all columns to get the full result. .by For example, if a system of 40 tridiagonal linear equations is required, with 39 unknowns each time, a typical industrial control computer can complete all the calculations within milliseconds.

[0036] in, It has an approximately banded structure, the first The maximum value of the column appears in the first position. OK( Approaching 1, meaning the adjustment point itself has the maximum response), the following numbers to both sides and The difference increases and then rapidly decreases. The cable tension is high near the tower end. Large, with a rapid attenuation rate, so adjusting one support point has minimal impact on the distance; the mid-span area has a large sag. Small size and slow decay rate mean that adjusting one support point can affect monitoring points over a wide range. This is the mechanical reason why point-by-point adjustments across the mid-span area are prone to repeated failures.

[0037] Step 103: After each sag adjustment rope adjustment, record the actual rope length change at each support point and the measured vertical displacement change at each monitoring point. Based on the prediction residual between the measured values ​​and the predicted values ​​of the influence matrix, update the equivalent vertical stiffness coefficient vector using a recursive estimation method and recalculate the influence matrix.

[0038] Vertical load per unit horizontal projection length during the spinning process Constantly increasing, and As it changes, the initial influence matrix The data gradually deviates from the actual situation. If the influence matrix is ​​recalculated according to the entire process of the mechanical model, the real-time loads and tensions of each section are difficult to obtain accurately due to factors such as temperature changes, uneven distribution of steel wires, and slippage of the anchorage end of the load-bearing cable at the construction site. The deviation between the calculated results and the actual values ​​will cause systematic errors in subsequent coordination and adjustment. If all data is identified entirely from the observation data... At least matrix elements are required. Sublinearly independent adjustment operations, with For example, this would mean at least 40 tentative adjustments, which is not feasible given the construction schedule. This embodiment utilizes the parameterized relationships of the aforementioned steps. The identification object is compressed from all matrix elements to The stiffness parameters of each segment allow sufficient observational information to update the parameters with only a few actual adjustments. Number of cable segments Equals the total number of support points plus 1, and the total number of matrix elements. It equals the product of the total number of monitoring points and the total number of support points, with the former being much smaller than the latter.

[0039] Specifically, Figure 2 This is a flowchart of the recalculation of the influence matrix in one embodiment of this application, as follows: Figure 2 As shown in step 1031, after each adjustment of the sag adjustment rope, the actual change in rope length at each support point and the measured change in vertical displacement at each monitoring point are recorded.

[0040] After each sag adjustment rope adjustment, the system records two sets of data: the actual change in rope length at each support point. (Unit: meters, obtained from actuator displacement records or rope length measuring devices) and measured vertical displacement changes at each monitoring point. (Unit: meters, obtained by subtracting the steady-state vertical coordinates before and after adjustment). Subscript To adjust the operation sequence number, starting from 1 and incrementing, these two sets of data constitute a set of observation samples regarding the segment stiffness parameters.

[0041] In step 1032, the prediction residuals between the measured values ​​and the predicted values ​​of the influence matrix are calculated. Specifically, this can be expressed by the formula: ; ; For the first The parameter estimates before the next update, where... This refers to the number of recursive steps in the recursive Gauss-Newton method. The prediction residual contains two parts: the random components of GNSS measurement noise and residual wind vibration, and the systematic bias caused by the deviation of the current parameters from the true values. The goal of parameter updates is to gradually reduce the systematic bias component so that the prediction residual eventually tends to a level determined solely by measurement noise.

[0042] In step 1033, the equivalent vertical stiffness coefficient vector is updated using a recursive estimation method, and the influence matrix is ​​recalculated. Specifically, the parameter update employs the Recursive Gauss-Newton method, a standard method for nonlinear least squares recursive estimation. The core idea is to incrementally correct the parameters each time new observation data is obtained, using the first-order linearized approximation of the current model at the parameter estimates. (Data pairs are then obtained.) Next, calculate the Jacobian matrix. ( ), the elements are: ; Indicates the first When the stiffness parameter of segment changes by 1 unit (Newtons per meter), the first The rate of change of the predicted displacement value at each monitoring point is expressed in meters divided by (Newtons per meter), i.e., meters squared per Newton. By considering the two sides of the tridiagonal linear equations in the aforementioned steps... Taking the partial derivatives, we find that the coefficient matrix of the partial derivative equation is the same as the coefficient matrix of the original system of equations, where... Is only The corresponding rows of two adjacent support points have sparse matrices with non-zero elements, so the existing chasing method decomposition results can be reused to solve the problem, with a single additional computation cost of... Proportional to all After traversing the values, by The complete result is obtained by weighting and combining each component. .

[0043] The parameter update formula is as follows: ; ; To observe the noise covariance matrix, it is configured as a diagonal matrix, with diagonal elements... . To monitor the standard deviation of the sensor's vertical measurements, its value is based on the nominal vertical positioning accuracy of the GNSS receiver, combined with the effect of sliding window mean processing on suppressing random errors. If the window contains... Individual sampling (e.g., 30-second window, 1 Hz sampling time) After filtering, the standard deviation is reduced to 1 / 3 of the original nominal accuracy. Times. Based on the original nominal accuracy of 0.05 meters, For example, Meters, in practice, for example, 0.01 meters.

[0044] For the first The error covariance matrix of the parameter estimates before the next update, where... It is the number of recursive steps in the Gauss-Newton method, and its initial value is... The initial error covariance matrix for parameter estimation is taken as a diagonal matrix, and the first... The diagonal elements are , The estimated initial equivalent vertical stiffness coefficient of the corresponding cable segment calculated in the aforementioned steps. The relative uncertainty coefficient is determined by propagating the coefficients of variation of the load and tension of the load-bearing cable using the stiffness formula. Before construction, statistical analysis is performed on the manufacturing tolerances and installation deviations of the load-bearing cable to estimate the uncertainty. The coefficient of variation (e.g., 5%) and The coefficient of variation (e.g., 10%) is determined by ( (Since this is a measured value, the uncertainty is very small and can be ignored.) Therefore, we can obtain... The relative standard deviation is approximately equal to The coefficient of variation, after considering the safety margin of the model simplification error, is taken as... Values, for example, 0.2.

[0045] Updated Substitute the values ​​into the column-by-column solution process described above and recalculate the influence matrix. As an example, if the third paragraph The initial value is Newtons per meter, updated after accumulating data from three adjustment operations. Newtons per meter corresponds to the physical reality that the increased sag and 10% decrease in vertical stiffness in this section due to the increased spinning load. Meanwhile... The diagonal elements of the matrix (the error covariance matrix of parameter estimation) decrease after each update, reflecting the reduction in parameter uncertainty, which causes the incremental correction magnitude of subsequent updates to gradually converge.

[0046] Step 104: Compare the real-time vertical coordinates of each monitoring point with the target line shape to obtain the line shape deviation vector. When the deviation exceeds the allowable threshold, use the current influence matrix to solve the regularized least squares problem on the line shape deviation vector to obtain the coordination adjustment amount of each support point and output it.

[0047] Specifically, the system will display the real-time vertical coordinates of each monitoring point. (Steady-state value after sliding window averaging) and the target linearity at the current stage By comparison, the linear deviation vector is obtained: ; Positive values ​​indicate values ​​above the target, and negative values ​​indicate values ​​below the target. The absolute value of any component exceeds the allowable threshold. The calculation of the coordination adjustment amount is triggered at that time. The design tolerance is determined based on the allowable deviation index of the main cable alignment in the design documents: the design tolerance is usually given in the form of the vertical coordinate tolerance of each control point. Take half of the minimum value to the full range, for example, it can be configured to be 20 to 50 mm. The purpose of taking half is to reserve a margin for adjusting the execution deviation.

[0048] Based on the current influence matrix Solving a regularized least squares problem for a linear deviation vector, the objective function of which consists of two terms: the first term is the squared L2 norm of the adjusted expected residual deviation. ,when When this term is zero, the deviation is completely eliminated; the second term is the square of the L2 norm of the adjustment amount at each support point. This is used to limit excessive adjustments at individual support points. The analytical solution is: ; It is an identity matrix; the negative sign makes the deviation positive (the monitoring point is higher than the target) and the corresponding adjustment amount negative (the cat path is lowered by extending the rope). for Symmetric positive definite matrices can be solved using Cholesky decomposition. Cholesky decomposition decomposes a symmetric positive definite matrix into the product of a lower triangular matrix and its transpose. Then, a system of linear equations can be solved through two steps: forward substitution and backward substitution. The calculation time is in the millisecond range under operating conditions of 30 to 100.

[0049] Regularization coefficient For dimensionless quantities, take the ratio of the square of the standard deviation of the vertical measurement of the monitoring sensor to the square of the standard deviation of the expected adjustment of a single support point: ; This is the statistical standard deviation or engineering experience value of the single adjustment amount at a single support point during construction, for example, 0.05 meters; When taking 0.01 meters From the perspective of Bayesian estimation, This is equivalent to the ratio of the observation noise variance to the prior parameter variance: The larger the value, the more conservative the adjustment of each support point, but the deviation is not fully eliminated; The smaller the value, the more thorough the deviation correction, but the adjustment amount for individual support points may be too large. In practice... The value can be adjusted within the range of 0.01 to 1. In some embodiments, a larger value is used in the initial stage where the matrix accuracy is relatively low. (e.g., 0.5) to reduce the risk of over-adjustment due to matrix inaccuracy; gradually decrease to, for example, 0.05 as parameter identification accuracy improves to enhance correction adequacy.

[0050] In one embodiment, the calculation is obtained Then, a constraint check is performed on each component. If the absolute value of the adjustment amount at a certain support point exceeds the single allowable upper limit, the check will be performed. (For example, 100 mm, determined by the sag adjustment rope winch stroke or safety limit), cut this component to After truncating, substitute the truncated value into... Reassess the expected residual bias; if any component of the residual bias still has an absolute value exceeding If so, the marker needs to be adjusted after this round of adjustments is implemented.

[0051] Final coordination adjustment amount The output is sent to the construction control terminal to display the rope length adjustment direction and value for each support point in a list or graphical format. On catwalks equipped with electric winch-type sag adjustment ropes, displacement control commands can be directly issued to each winch; on manually adjusted catwalks, workers operate according to the displayed values.

[0052] As an example, suppose that among the deviations corresponding to the 40 support points, the deviations at points 18 to 22 in the middle of the span are approximately meters, approximately on both sides to The length of the support points is approximately 30 to 40 millimeters, close to zero near the tower top. After coordinated calculations, the shortening at the mid-span support points is approximately 10 to 20 millimeters, on both sides approximately 10 to 20 millimeters, and only a few millimeters at the tower top. The adjustment amount for each support point is not simply equal to the deviation value at the corresponding position, but rather a globally optimal allocation obtained by integrating the mutual influence between all support points through an influence matrix. The adjustment amount for some support points will be slightly larger than their local deviation to compensate for the reverse influence of the adjustment on adjacent positions.

[0053] In some embodiments, after the sag adjustment rope corresponding to the coordinated adjustment amount is adjusted, and the vertical coordinates of each monitoring point reach a steady state, the system collects the adjusted steady-state vertical coordinates. Compared with the records before adjustment Subtraction yields the measured vertical displacement change. Simultaneously record the actual rope length changes at each support point. (The actual execution value may differ from the instruction value) There are deviations, which may originate from factors such as the accuracy of manual operation or the positioning accuracy of the winch.

[0054] Will Based on the influence matrix Calculated predicted displacement change Subtraction yields the predicted residual And calculate its L2 norm. And the update trigger threshold Compare, This can be expressed by a formula: ; by , Taking rice as an example, Meters. This means that if the influence matrix is ​​completely accurate, the prediction residuals are caused only by measurement noise, and each component is independent with a standard deviation of . The statistical expectation of the predicted residual L2 norm is approximately this value. The predicted residuals contain systematic biases exceeding the noise level, indicating an identifiable difference between the segment stiffness parameters of the influence matrix and their actual values. Therefore, based on the adjusted values... As the observed data, the Jacobian matrix is ​​calculated and updated according to the recursive formula described in the preceding steps. and The influence matrix was then recalculated. .

[0055] After the update is complete, the linear deviation vector will be recalculated. If the deviation still exceeds the allowable threshold The updated influence matrix is ​​then used to repeat the aforementioned steps to solve for the supplementary coordination adjustment. As adjustment operations accumulate, the estimated values ​​of the segment stiffness parameters gradually approach the true values. As the diagonal elements gradually decrease, the uncertainty decreases, the incremental correction magnitude of parameter updates converges, and the prediction residuals of subsequent coordinated adjustments decrease accordingly. Typically, after 2 to 4 cycles of adjustment, verification, and update, the accuracy of the influence matrix is ​​sufficient to control the full-span linear deviation within the allowable range in one round of adjustment.

[0056] In one embodiment, the above steps are performed cyclically during the spinning process: continuous monitoring of linear deviation during the spinning stage; triggering calculation and coordination adjustment when the deviation exceeds the limit; triggering verification and updating parameters as needed after execution; then returning to the monitoring state to wait for the next triggering, and this cycle runs through the entire spinning process.

[0057] Alternatively, when installing a cable force sensor on the load-bearing cable, the measured cable force data can be directly substituted into the calculation steps described above. This yields an initial equivalent vertical stiffness coefficient that more closely resembles reality. This reduces the number of adjustment operations required for online identification. For multi-span continuous catwalks (such as three-span suspension bridges with intermediate towers), the load-bearing cables of adjacent spans are connected by cable saddles at the top of the towers; during modeling, a vertical stiffness constraint term at the top of the tower is added to the vertical force balance equation at the support point where the cable saddle is located. , The value is determined by the tower structure analysis, with units of Newtons per meter, which expands the solution range of the influence matrix from a single span to adjacent spans, reflecting the transmission effect of inter-span displacement.

[0058] in, The vertical displacement of the saddle point at the top of the tower relative to the initial equilibrium position, with upward as positive, in meters; The constraint stiffness of the tower body at the top of the tower to the vertical displacement of the cable saddle is the vertical restoring force provided by the tower body when the cable saddle produces a unit vertical displacement. The unit is N / m, and it is determined by tower structure analysis or on-site measurement. This represents the restoring effect of the tower's elastic constraints on the vertical displacement of the cable saddle, and is added to the vertical force balance equation at the suspension point in the form of a restoring force opposite to the displacement, in a uniform positive direction. If the vertical stiffness at the tower top is very large, Take the larger value to approximate the fixed constraint; for multi-span continuous catwalks, the saddle displacement affects two adjacent spans at the same time, so this term expands the solution range of the influence matrix from a single span to adjacent spans.

[0059] Therefore, by quantitatively expressing the interrelationships of displacement between support points in the form of an influence matrix and parameterizing them with segment stiffness parameters, the amount of observation data required for online identification is significantly reduced. Sufficient information can be accumulated and parameters updated with only a few actual adjustment operations. The recursive Gauss-Newton method is used to instantly correct the estimated value of the segment stiffness parameters using the input and output data of each adjustment operation. The influence matrix can track the continuous changes in the load state of the load-bearing cable during spinning. When the deviation exceeds the limit, regularized least squares are used to solve for the coordinated adjustment amount of all support points in one go. The values ​​of each support point comprehensively consider all interrelationships, and the full-span alignment deviation can be controlled within the allowable range in one round of adjustment, significantly reducing the number of repeated adjustments and time required by the traditional point-by-point trial-and-error method.

[0060] Please see Figure 3 , Figure 3 This is a schematic block diagram of a catwalk alignment dynamic prediction and automatic control system based on real-time monitoring data, as described in this application. Figure 3 As shown, the system includes: The monitoring and stiffness calculation module 3001 is used to collect the vertical coordinates of each monitoring point in real time; the load-bearing cable is divided into several cable segments according to the position of each support point of the sag adjustment rope, and the equivalent vertical stiffness coefficient of each cable segment is calculated based on the catenary mechanical parameters of each cable segment. The influence matrix construction module 3002 is used to construct an influence matrix based on the equivalent vertical stiffness coefficient. The influence matrix represents the linear mapping relationship between the change in rope length at each support point and the change in vertical displacement at each monitoring point, using the equivalent vertical stiffness coefficient vector as a parameter. The parameter update module 3003 is used to record the actual rope length change at each support point and the measured vertical displacement change at each monitoring point after each sag adjustment rope adjustment. Based on the prediction residual between the measured value and the predicted value of the influence matrix, the equivalent vertical stiffness coefficient vector is updated using a recursive estimation method, and the influence matrix is ​​recalculated. The coordination and control module 3004 is used to compare the real-time vertical coordinates of each monitoring point with the target line shape to obtain the line shape deviation vector. When the deviation exceeds the allowable threshold, the module uses the current influence matrix to solve the regularized least squares problem on the line shape deviation vector to obtain the coordination adjustment amount of each support point and output it.

[0061] Based on the same concept as the above method, this application proposes an electronic device, see [link to previous application]. Figure 4As shown, it includes: a processor 4001 and a machine-readable storage medium 4002, the machine-readable storage medium 4002 storing machine-executable instructions that can be executed by the processor 4001; the processor 4001 is used to execute the machine-executable instructions to implement the catwalk linear dynamic prediction and automatic control method based on real-time monitoring data disclosed in the above example of this application.

[0062] Based on the same concept as the methods described above, this application also provides a machine-readable storage medium storing a plurality of computer instructions. When executed by a processor, these computer instructions can implement the catwalk linear dynamic prediction and automatic control method based on real-time monitoring data disclosed in the examples above. The machine-readable storage medium can be any electronic, magnetic, optical, or other physical storage device, and can contain or store information such as executable instructions, data, etc.

[0063] Based on the same application concept as the above method, this application embodiment also provides a computer program product, the computer program product including a computer program, which, when executed by a processor, implements the catwalk linear dynamic prediction and automatic control method based on real-time monitoring data disclosed in the above examples of this application.

[0064] In the above embodiments, the descriptions of each embodiment have different focuses. Parts not described in detail in a certain embodiment can be referred to in the relevant descriptions of other embodiments. The above descriptions are merely preferred embodiments of this application and explanations of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to the technical solutions formed by specific combinations of the above technical features, but should also cover other technical solutions formed by arbitrary combinations of the above technical features or their equivalent features without departing from the inventive concept.

Claims

1. A method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data, characterized in that, include: The vertical coordinates of each monitoring point are collected in real time; the load-bearing cable is divided into several cable segments according to the position of each support point of the sag adjustment rope, and the equivalent vertical stiffness coefficient of each cable segment is calculated based on the catenary mechanical parameters of each cable segment. An influence matrix is ​​constructed based on the equivalent vertical stiffness coefficient. The influence matrix represents the linear mapping relationship between the change in rope length at each support point and the change in vertical displacement at each monitoring point, with the equivalent vertical stiffness coefficient vector as the parameter. After each adjustment of the sag adjustment rope, the actual rope length change at each support point and the measured vertical displacement change at each monitoring point are recorded. Based on the prediction residual between the measured value and the predicted value of the influence matrix, the equivalent vertical stiffness coefficient vector is updated using a recursive estimation method, and the influence matrix is ​​recalculated. The vertical coordinates of each monitoring point in real time are compared with the target line shape to obtain the line shape deviation vector. When the deviation exceeds the allowable threshold, the regularized least squares problem is solved on the line shape deviation vector using the current influence matrix to obtain the coordination adjustment amount of each support point and output it.

2. The method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data according to claim 1, characterized in that, The calculation of the equivalent vertical stiffness coefficient of each cable segment based on the catenary mechanical parameters of each segment includes: linearizing the catenary equation of each cable segment at the current equilibrium position with small displacements, and calculating the equivalent vertical stiffness coefficient based on the horizontal component force, horizontal projected span, and vertical load per unit horizontal projected length of the cable segment; the horizontal component force is determined by iteratively solving a system of multiple catenary equilibrium equations for the entire load-bearing cable.

3. The method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data according to claim 1, characterized in that, The construction of the influence matrix based on the equivalent vertical stiffness coefficient includes: applying a unit vertical displacement to each support point, using the zero displacement of the fixed support points at both ends of the load-bearing cable as the boundary condition, and establishing vertical force balance equations for the remaining support points; in the vertical force balance equations, the vertical restoring forces of the left and right cable segments on each support point are represented by the product of the equivalent vertical stiffness coefficient of the corresponding cable segment and the displacement difference of the adjacent nodes, and the sum of the two is zero; the vertical force balance equations constitute a tridiagonal linear equation system, which is solved using the chasing method; the monitoring points located between adjacent support points obtain the corresponding influence coefficients by linear interpolation of the displacement response of the support points at both ends of the cable segment.

4. The method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data according to claim 1, characterized in that, The recursive estimation method is the recursive Gauss-Newton method; the Jacobian matrix in the update process is obtained by taking the partial derivatives of both sides of the equation system of the influence matrix with respect to the equivalent vertical stiffness coefficient of each cable segment, and the solution is obtained by reusing the coefficient matrix decomposition result of the original equation system.

5. The method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data according to claim 4, characterized in that, In the recursive Gauss-Newton method, the observation noise covariance matrix is ​​a diagonal matrix, and the diagonal elements are the squares of the vertical measurement standard deviations of the monitoring sensors; the initial error covariance matrix of parameter estimation is a diagonal matrix, and each diagonal element is calculated based on the estimated value of the initial equivalent vertical stiffness coefficient of the corresponding cable segment and the relative uncertainty coefficient. The relative uncertainty coefficient is determined by propagating the coefficient of variation of the load and tension of the load-bearing cable through the stiffness formula.

6. The method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data according to claim 1, characterized in that, The process also includes adjustment verification and parameter feedback steps: After the sag adjustment rope corresponding to the coordinated adjustment amount is adjusted, the steady-state vertical coordinates of each monitoring point after adjustment are collected, and the measured vertical displacement change of each monitoring point before and after adjustment is calculated; the measured vertical displacement change is subtracted from the predicted displacement change calculated by the influence matrix based on the actual rope length change to obtain the prediction residual; when the L2 norm of the prediction residual exceeds the update trigger threshold, the recursive estimation method is executed using the actual rope length change and the measured vertical displacement change as observation data to update the equivalent vertical stiffness coefficient vector and recalculate the influence matrix; after the update is completed, the linear deviation vector is recalculated, and if the deviation still exceeds the allowable threshold, the regularized least squares problem is solved again using the updated influence matrix to obtain the supplementary coordinated adjustment amount.

7. The method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data according to claim 1, characterized in that, The average value of the vertical coordinate data of each continuously collected monitoring point is taken as the average value of a sliding window for a set time period; the steady-state vertical coordinates before and after the adjustment operation are taken as the coordinate values ​​when the difference between the average values ​​of two adjacent sliding windows is less than the set stability criterion.

8. The method for dynamic prediction and automatic control of catwalk alignment based on real-time monitoring data according to claim 1, characterized in that, When the catwalk is a multi-span continuous structure, a vertical stiffness constraint term at the top of the tower is added to the vertical force balance equation at the support point where the cable saddle is located at the top of the tower. The solution range of the influence matrix is ​​extended from a single span to adjacent spans.

9. A catwalk alignment dynamic prediction and automatic control system based on real-time monitoring data, characterized in that, include: The monitoring and stiffness calculation module is used to collect the vertical coordinates of each monitoring point in real time; the load-bearing cable is divided into several cable segments according to the position of each support point of the sag adjustment rope, and the equivalent vertical stiffness coefficient of each cable segment is calculated based on the catenary mechanical parameters of each cable segment. An influence matrix construction module is used to construct an influence matrix based on the equivalent vertical stiffness coefficient. The influence matrix represents the linear mapping relationship between the change in rope length at each support point and the change in vertical displacement at each monitoring point, using the equivalent vertical stiffness coefficient vector as a parameter. The parameter update module is used to record the actual rope length change at each support point and the measured vertical displacement change at each monitoring point after each sag adjustment rope adjustment. Based on the prediction residual between the measured values ​​and the predicted values ​​of the influence matrix, the equivalent vertical stiffness coefficient vector is updated using a recursive estimation method, and the influence matrix is ​​recalculated. The coordination and control module is used to compare the real-time vertical coordinates of each monitoring point with the target line shape to obtain the line shape deviation vector. When the deviation exceeds the allowable threshold, the module uses the current influence matrix to solve the regularized least squares problem on the line shape deviation vector to obtain the coordination adjustment amount of each support point and output it.

10. An electronic device, characterized in that, include: processor; Memory for storing computer programs executable by the processor; The processor is configured to execute a computer program in the memory to implement the method as described in any one of claims 1 to 8.

Images