Assembling method for prefabricated firewall of transformer substation

Abstract

The invention relates to the field of assembly automation and flexible manufacturing, and discloses an assembly method for a prefabricated firewall of a transformer substation, and the method comprises the following steps: setting a plurality of space positioning reference points at the edge of a wallboard; acquiring three-dimensional coordinate data to form a first point set and a second point set; calculating an initial rigid body transformation parameter between the wallboards according to the first point set and the second point set; establishing an energy functional for describing a total error; carrying out minimization solution on the energy functional by adopting a variational method to obtain an optimal disturbance vector field; according to the optimal disturbance vector field, a plurality of fine adjustment actuators of the wallboard are controlled, posture fine adjustment is conducted, and physical assembly connection of the wallboard and the target wallboard is completed. According to the invention, accurate attitude estimation and error correction in the wallboard assembly process are realized; the robustness and geometric fitting capability of the assembly process are improved; the assembly precision and efficiency are remarkably improved, and the problem that traditional redundancy control distribution is unstable is solved.

Description

Technical Field

[0001] This invention relates to the field of assembly automation and flexible manufacturing, specifically to an assembly method for prefabricated firewalls in substations. Background Technology

[0002] With the rapid construction of modern power infrastructure, the industrial prefabrication and modular assembly of substation structural components have become the mainstream trend. Especially in high-voltage or ultra-high-voltage substations, firewalls, as core isolation and safety structures, must meet requirements such as high dimensional accuracy, rapid assembly efficiency, and minimal construction intervention. To facilitate on-site construction efficiency, prefabricated firewalls often adopt standardized component forms, with major molding and pre-processing completed in the factory, requiring only docking and positioning connections on-site. However, due to transportation, foundation deformation, or local manufacturing errors, geometric errors in the on-site docking process of prefabricated wall panels become a significant challenge restricting rapid installation.

[0003] Currently, in substation wall panel assembly scenarios, positioning and docking generally rely on laser ranging, manual measurement, and mobile tool carts to achieve wall panel hoisting and rough positioning. Some engineering applications use preset anchor points or fixtures for initial attitude adjustment, and a few high-precision scenarios introduce finite-degree-of-freedom hydraulic fine-tuning devices for end-point micro-correction. However, these methods typically treat the wall panel as a rigid body, neglecting the flexible characteristics of its actual structure in attitude calculation and assembly control. Furthermore, existing assembly methods lack adaptive control mechanisms for error response paths, rely on experience for control, and lack a continuous mapping path between sensor data and adjustment actions, thus limiting both attitude adjustment efficiency and accuracy.

[0004] The biggest problem with traditional methods is that they ignore the local elastic response of the wall panel structure under stress and fail to provide a complete technical path from error identification to multi-actuator driven linkage. Once there is an asymmetric boundary deviation or local warping between the target wall panel and the current wall panel, existing rigid body registration strategies often fail to capture the true registration residual, leading to non-convergence in pose adjustment and even repeated trial assembly. In addition, the actuator layout and control quantity allocation lack mathematical mapping support, and the control system can only achieve redundancy allocation through manual trial and error. The problems of low efficiency, large errors, and strong coupling during the assembly process have not been systematically solved. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an assembly method for prefabricated firewalls in substations, which solves problems such as the lack of flexible adaptation mechanisms, discontinuous attitude adjustment, and uncoordinated control responses in existing assembly processes.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an assembly method for prefabricated firewalls in substations, comprising the following steps:

[0007] S1. Set multiple spatial positioning reference points on the edges of the wall panel to be assembled and the target wall panel respectively;

[0008] S2. Collect the three-dimensional coordinate data of spatial positioning reference points using a three-dimensional scanning device to form the first point set and the second point set;

[0009] S3. Calculate the initial rigid body transformation parameters between the wall panels based on the first point set and the second point set;

[0010] S4. Construct a perturbation vector field based on the flexible rigid body transformation model, and establish an energy functional to describe the total error;

[0011] S5. Minimize the energy functional using the variational method to obtain the optimal perturbation vector field;

[0012] S6. Based on the optimal disturbance vector field, control multiple fine-tuning actuators of the wall panel to perform attitude fine-tuning and complete the physical assembly connection between the wall panel and the target wall panel.

[0013] Preferably, the step of setting a spatial positioning reference point includes the following sub-steps:

[0014] At least four markers are evenly distributed along the edge of the wall panel, and the markers are reflective posts or metal inserts.

[0015] During the factory manufacturing stage, the three-dimensional coordinates of each reference point are calibrated using high-precision positioning equipment;

[0016] The reference point numbers are stored in the digital model for point mapping in the subsequent alignment process.

[0017] Preferably, the step of acquiring three-dimensional coordinate data includes the following sub-steps:

[0018] Start the structured light scanning system and simultaneously locate the edges of the current wall panel and the target wall panel;

[0019] Obtain the spatial coordinates of each reference point to form the first point set and the second point set respectively;

[0020] The point set is subjected to coordinate system unification and noise filtering.

[0021] Preferably, the step of calculating the initial rigid body transformation parameters includes:

[0022] Calculate the centroids of the first and second point sets and center the point sets;

[0023] Construct a covariance matrix based on a centralized point set and perform singular value decomposition on it;

[0024] Based on the decomposition results, the rotation matrix R and the translation vector T are obtained, where

[0025] T = q mean -R·p mean ;

[0026] In the formula, U and V are matrices obtained from singular value decomposition, p mean q mean These are the centroids of the two point sets, respectively.

[0027] Preferably, the initial rigid body transformation satisfies the following least squares optimization model:

[0028]

[0029] Among them, {p i} represents the current set of wall panel points, {q i Let} be the target wall panel point set, R be the rotation matrix, and T be the translation vector.

[0030] Preferably, the steps for constructing a flexible rigid body perturbation model include establishing the following form

[0031] Energy functional:

[0032]

[0033] Where ∈(x) represents the perturbation vector field, α controls rigidity preservation, β controls matching accuracy, and δ is the Dirac function, ||·|| F This represents the Frobenius norm.

[0034] Preferably, solving for the optimal perturbation vector field includes:

[0035] Variational differentiation of the energy functional E[∈] yields the perturbation control partial differential equation, which takes the form:

[0036] αΔ∈(x)-β∑ i δ(xp i )(q i -Rp i -T-∈(p i )) = 0;

[0037] Where, ∈(x) is the perturbation vector field, Δ represents the Laplace operator, and q i and p i These represent the target and the current reference point positions, respectively. i δ(xp i ) is the Dirac function;

[0038] The governing equations are discretized in the spatial region Ω, and the equations are numerically discretized using the finite difference method or the finite element method.

[0039] The optimal perturbation vector field ∈ is solved by numerical calculation.* (x), and output the adjustment amount Δx for each reference point. i This minimizes the system error.

[0040] Preferably, the optimal solution of the perturbation vector field ∈ * (x) satisfies the following governing equations:

[0041]

[0042] Where E[∈] represents the energy functional.

[0043] Preferably, the attitude fine-tuning and assembly connection steps include the following three sub-steps:

[0044] The adjustment amount of each reference point of the disturbance vector field is converted into a control signal and sent to the electric micro actuator.

[0045] Each push rod performs a corresponding displacement to correct the spatial attitude error of the wall panel;

[0046] After the error converges, the wall panel connection and fixation are completed by structural limit buckles or sliding rail structure.

[0047] An assembly system for a prefabricated firewall in a substation, applied to the assembly method for a prefabricated firewall in a substation as described in any one of claims 1-9, characterized in that it comprises:

[0048] The positioning and calibration module is used to set up spatial positioning reference points on the edge of the wall panel and store their three-dimensional coordinate information;

[0049] The three-dimensional point acquisition module is used to acquire the positions of reference points on the wall panel to be assembled and the target wall panel, forming a first point set and a second point set;

[0050] An initial attitude estimation module is used to calculate the rotation matrix and displacement vector based on the point set;

[0051] The perturbation modeling module is used to construct the energy functional of a flexible rigid body containing a perturbation vector field;

[0052] The numerical solution module is used to solve the optimal perturbation field based on the control partial differential equations.

[0053] The attitude fine-tuning module is used to control the electric push rods on the wall panel to make spatial fine-tuning according to the disturbance field;

[0054] The assembly connection module is used to position and close the wall panel after fine-tuning.

[0055] The closed-loop error verification module is used to rescan the reference point and determine whether the adjustment error is lower than the set threshold.

[0056] This invention provides an assembly method for prefabricated firewalls in substations. It offers the following advantages:

[0057] 1. This invention constructs a joint modeling mechanism of rigid body transformation and flexible perturbation, and uses point set centering and SVD decomposition to obtain initial assembly attitude parameters, achieving accurate estimation of large-scale pose errors between wall panels. Compared with existing methods that rely solely on sensor alignment or manual guidance, this scheme can stably output attitude parameters under point cloud noise and boundary irregularities, solving the problems of instability and computational redundancy in traditional rigid body matching, and exhibiting good robustness in the initial alignment stage of large-size components.

[0058] 2. This invention introduces a flexible rigid body transformation framework based on perturbation vector fields, incorporating local geometric errors during assembly into a unified model and globally modeling these errors by constructing a physically interpretable energy functional. Compared to existing rigid body assembly strategies that cannot express the adaptive behavior of non-rigid structures, this scheme can naturally express point-to-point offset distributions, compensating for the technical shortcoming of insufficient registration capability when there are small-scale distortions on the assembly contact surfaces, and improving the system's geometric fitting capability and flexible tolerance handling capability.

[0059] 3. This invention constructs a set of differentiable control partial differential equations based on a flexible error field, and uses finite difference or finite element methods to numerically solve the disturbance field, forming a precise mapping path from the assembly error field to the attitude fine-tuning control quantity. Traditional methods often lack an effective error field solution mechanism, leading to problems such as discontinuous actuator response and control solution offset. However, this scheme achieves continuous reconstruction of the disturbance-structure-drive chain, significantly improving the response coordination and physical interpretation strength of the assembly process.

[0060] 4. This invention establishes an integrated path from point cloud error field to multi-degree-of-freedom assembly control by applying the optimal disturbance vector field to the multi-point distributed fine-tuning actuator of the wall panel, realizing flexible spatial attitude correction of complex structural components. Unlike existing methods that only use single-point guidance or fixed multi-point iterative fitting, this scheme can directly generate the target offset of each control node, solving the technical obstacles of unstable allocation and unclear calculation structure in traditional redundant control, making the wall panel assembly and connection operation smoother and more efficient. Attached Figure Description

[0061] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0062] Figure 2 This is a schematic diagram of the system architecture of the present invention. Detailed Implementation

[0063] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Please see the appendix Figure 1 This invention provides an assembly method for a prefabricated firewall in a substation, comprising the following steps.

[0065] S1. Set multiple spatial positioning reference points on the edges of the wall panel to be assembled and the target wall panel respectively;

[0066] In this embodiment, regarding the problem of spatial attitude matching of wall panels during assembly, the core of step S1 lies in implementing a method for laying out spatial positioning reference points suitable for the assembly requirements of prefabricated firewalls. This step provides the geometric basis and data entry for subsequent rigid body transformation solutions and disturbance vector field construction, and is the starting point of the entire technical process.

[0067] During the wall panel manufacturing stage, to ensure high-precision spatial positioning and attitude correction during on-site assembly, multiple identifiable positioning reference points need to be placed at the edge structure of each wall panel. Preferably, the positioning reference points are markers with strong echo recognition capabilities, such as reflective columns, spherical films, or metal concave dots. They should possess mechanical and optical recognition stability and be suitable for various 3D scanning equipment such as structured light, lidar, or photogrammetry.

[0068] The layout of positioning reference points should meet the following constraints:

[0069] The distribution of reference points should cover the corner and central axis areas of the wall panel, avoiding excessive concentration or collinearity of the points, in order to ensure sufficient spatial tension.

[0070] Any three points should not be collinear. It is preferable to use a set of non-coplanar points in space consisting of no less than four points to enhance the robustness of subsequent rigid body attitude estimation.

[0071] The spacing between points should not be too small to prevent fusion or occlusion of the scanned images.

[0072] After the reference points are set up, the wall panel is scanned with structured light using a 3D scanning device to collect the spatial position data of each positioning point, forming an original 3D point cloud. At this point, each reference point has the following data structure:

[0073] p i =(x i ,y i ,z i );

[0074] in This represents the three-dimensional spatial coordinates of the i-th reference point on the wall panel to be assembled. All points form the first point set P = {p1, p2, ..., p...} n}, where n≥4.

[0075] Correspondingly, the scanning results of the target wall panel form a second point set:

[0076]

[0077] To ensure the accuracy of subsequent matching, a mapping dictionary is established for the reference point numbers during the manufacturing stage:

[0078] Map(i) = (p i ,q i );

[0079] The dictionary is output from the structured light scanner and input into the system's digital modeling module, becoming the initial rigid body matching input.

[0080] The spatial structure design of the positioning reference point also needs to consider the feasibility of machining. Preferably, the reference point structure is pre-designed by a CAD system and simultaneously sent to the CNC or numerical control machine tool machining process to achieve integrated digital manufacturing and improve structural consistency.

[0081] Ultimately, these reference points not only provide a rigid body basis in 3D recognition, but their distribution structure also determines the boundary expression capability of error control in the flexible matching process, which is a key component of the assembly accuracy control scheme of this invention.

[0082] S2. Collect the three-dimensional coordinate data of spatial positioning reference points using a three-dimensional scanning device to form the first point set and the second point set;

[0083] In this embodiment, addressing the critical requirement of precise on-site assembly of prefabricated firewalls, step S2 provides a method for acquiring three-dimensional spatial reference point data and generating point sets. This method includes multiple steps such as three-dimensional data acquisition, coordinate system transformation, data filtering, and point set standardization. This step not only fulfills the core function of spatial information input but also provides key input parameters for subsequent attitude estimation and disturbance modeling, thus playing a structural technical role.

[0084] First, a 3D spatial perception device is deployed on-site to perform structured light scanning of the reference point area. Preferably, this device is a structured light scanning system with high-density point cloud reconstruction capabilities. It reconstructs the 3D positions of reflective feature points by projecting a grating-coded pattern onto the edge of the wall panel and simultaneously imaging it with multiple cameras. This system should be pre-calibrated with the assembly platform coordinate system to ensure spatial consistency of the acquired 3D points.

[0085] Before scanning begins, the system uses an attitude synchronization module to identify the pre-assembly poses of the current wall panel and the target wall panel. Preferably, the reference coordinate system ∑ for the wall panel mounting surface is obtained using platform sensors or a laser total station. c With ∑ t And convert it into a unified site coordinate system ∑ g .

[0086] The image features of each identified reference point are used for sub-pixel-level precise localization via an image matching algorithm. Combining the known scanning baseline and triangulation relationships, its spatial position in the scanning device's coordinate system can be calculated. The resulting three-dimensional points... These represent the initial scan coordinates of the i-th reference point in the current wall panel and the target wall panel, respectively.

[0087] To achieve coordinate system unification, the original point data needs to be transferred from the local coordinate system ∑ c With Σ t Mapped to a unified geodetic coordinate system ∑ g The mapping relationship is as follows:

[0088]

[0089] Among them, T c ,T t ∈SE(3) is the homogeneous transformation matrix that transforms the scanning results from the device coordinate system to the site coordinate system; This is the input result for the point set in the final normalized coordinates. This transformation is obtained through a pre-completed calibration matrix and dynamically generated in combination with the actual workstation posture information.

[0090] After coordinate unification, the resulting point set may introduce undesirable deviations due to environmental occlusion, reflective interference, or sensor thermal noise, thus requiring error filtering. This embodiment employs a neighborhood consistency filtering method based on Euclidean distance thresholds, along with a fitting residual detection algorithm. For a given reference point p... i If the centroid distance between a point and other points in its local neighborhood exceeds the expected range, it is marked as a suspected outlier, and its coordinates are corrected using a weighted average of local point sets. The two point sets obtained here are:

[0091] P = {p1, p2, ..., p} n}, Q={q1,q2,…,q n};

[0092] This is the set of input points in the flexible rigid body registration frame of the present invention, which correspond to the same numbered spatial reference points in the current wall panel and the target wall panel respectively.

[0093] In the modeling process of this invention, this point set is not only used for initial attitude estimation, but also introduced into the modeling formula as a rigid matching error term in the flexible rigid body disturbance field.

[0094] It is important to emphasize that the acquisition and processing of point sets P and Q are the foundation of the accuracy of the formula model. Their completeness and consistency directly affect the accuracy of the attitude calculation and fine-tuning results. Therefore, this step occupies a key position in this invention.

[0095] In summary, this embodiment integrates operations such as structured light scanning, attitude unification, data noise filtering, and point set standardization to ultimately form a complete point set data structure that can be used for assembly modeling and control. It also ensures the accurate mapping effect of the data in rigid body matching and flexible vector disturbance modeling, providing reliable support for the engineering feasibility of the present invention.

[0096] S3. Calculate the initial rigid body transformation parameters between the wall panels based on the first point set and the second point set;

[0097] In this embodiment, regarding the attitude registration problem of the prefabricated firewall in the substation during assembly, step S3 is mainly used to calculate the initial rigid body transformation parameters of the current wall panel relative to the target wall panel, namely the rotation matrix R and the translation vector T, based on the obtained first and second point sets. This step provides an initial alignment reference before modeling the disturbance vector field and is the core foundation of the entire flexible assembly control process.

[0098] The first point set P = {p1, p2, ..., p} n} represents the spatial positioning reference point on the wall panel to be assembled, identified by the scanning system.

[0099] The second point set Q = {q1, q2, ..., q} n} represents the reference point on the target wall panel that corresponds one-to-one with the first point set.

[0100] in, n≥4 represents the size of the point set. All point set data have undergone coordinate system unification and filtering through the aforementioned steps, exhibiting good consistency and geometric stability.

[0101] To solve for rigid body transformations, we first need to center the two sets of points. Calculate the geometric centroids of the first and second point sets:

[0102]

[0103] By centralizing the point set using centroid information, a decentralized point set is obtained:

[0104] p′ i =p i -p mean ,q′i =q i -q mean ;

[0105] This process ensures that the point set is not affected by the overall translation during subsequent matching, which helps to separate the solution of rotation transformation from the translation component.

[0106] Next, construct the covariance matrix between the two sets of centered points:

[0107]

[0108] in, Let H represent the outer product tensor of the i-th point pair. The covariance matrix H essentially reflects the cooperative distribution structure of the two sets of points in corresponding directions in three-dimensional space.

[0109] Singular value decomposition (SVD) of the covariance matrix is ​​expressed as:

[0110] H=U·Σ·V T ;

[0111] in: It is an orthogonal matrix, representing the basis for direction transformation; It is a diagonal matrix containing the singular values ​​of the covariance matrix;

[0112] SVD decomposition can be implemented using a numerical linear algebra library, making it suitable for online execution in embedded solvers.

[0113] Calculate the initial rotation matrix from the singular value decomposition results:

[0114] R = V·U T ;

[0115] To ensure that R∈SO(3), which is a positive rotation matrix, if det(R)<0 is found, the third column of V needs to be negative and R needs to be recalculated.

[0116] After solving for the rotation matrix R, we further solve for the translation vector T, which represents the offset relationship between the current overall position of the wall panel and the target wall panel:

[0117] T = q mean -R·p mean ;

[0118] Thus, the complete rigid body transformation expression is obtained:

[0119]

[0120] This transformation can be applied to all edge reference points of the current wall panel to achieve initial correction of its spatial attitude.

[0121] From the perspective of optimization modeling, this rigid body matching process can also be expressed as the following least squares problem:

[0122]

[0123] in:

[0124] ||·|| denotes the Euclidean norm;

[0125] The objective function measures the spatial residual after rigid body transformation between each pair of reference points;

[0126] R and T are the target variables for optimization;

[0127] This problem has a closed-form solution, is computationally stable, requires no initial value iteration, and is suitable for fast solutions in assembly scenarios.

[0128] In summary, this embodiment fully discloses a method for solving the rigid body transformation parameters between the current wall panel and the target wall panel through processing procedures such as centering, covariance analysis, singular value decomposition, and rigid body least squares matching. This method offers stable accuracy and high computational efficiency, and the obtained parameters play a fundamental role in flexible error field modeling and fine-tuning control, constituting one of the core supporting components of the assembly correction technology solution of this invention.

[0129] S4. Construct a perturbation vector field based on the flexible rigid body transformation model, and establish an energy functional to describe the total error;

[0130] In this embodiment, the rigid body transformation parameters of the wall panel are obtained. Subsequently, in order to further improve the actual assembly alignment accuracy, a flexible rigid body disturbance model was constructed to compensate for the non-rigid errors and local geometric distortions of the structural components.

[0131] The core idea of ​​flexible disturbance modeling is to introduce a spatially continuously defined disturbance vector field ∈(x) on the basis of traditional rigid body transformation, which is used to describe the small offset behavior of the assembly reference point and its neighborhood under the action of force, deformation or manufacturing error.

[0132] The perturbation vector field is a mapping from a set of three-dimensional points in space to a three-dimensional vector:

[0133]

[0134] Wherein, Ω represents the spatial domain containing the reference point set, preferably the circumscribed cubic region of the point cloud of the wall panel or its expanded envelope.

[0135] To describe the combined effect of rigid body matching and flexible error correction, the total energy functional is defined in the following form:

[0136]

[0137] This energy functional consists of two main parts:

[0138] First, the smoothness constraint of the perturbation vector field in space:

[0139]

[0140] This term is derived from the first-order gradient tensor of the perturbation vector field. The Frobenius norm integral restricts drastic changes in the perturbation field throughout the entire domain, ensuring its physical rationality and continuity.

[0141] in, It is the Jacobian matrix of the perturbation vector field, representing the gradient distribution of the vector field in each spatial direction, ||·|| F The Frobenius norm is the square root of the sum of the squares of the matrix elements.

[0142]

[0143] This method preferably uses finite difference or finite element methods to calculate in discrete space to adapt to the discreteness of point cloud data.

[0144] Second, the reference point registration error term:

[0145]

[0146] in, This represents the i-th reference point on the current wall panel.

[0147] q i ∈Q is the target point on the target wall panel;

[0148] Rp i +T represents the displacement of the reference point under a rigid body configuration.

[0149] ∈(p i ) represents the vector correction of the point by the flexible disturbance.

[0150] This term measures the matching error of the perturbation field at key points, and is embedded in the energy functional to form a point constraint term.

[0151] Dirac function δ(xp) i In practical engineering implementation, it can be replaced by a discrete expression of the Gaussian kernel function or Kronecker delta, used for weighted error constraints at the reference point.

[0152] in:

[0153] The smoothness weight of the disturbance field is preferably determined by the structural stiffness and the accuracy of the application.

[0154] Registration accuracy weighting coefficient is used to adjust the balance between rigid body accuracy and flexible tolerance;

[0155] The two parameters are usually set by the prior structure or parameter tuning module during the system initialization phase and kept stable within a controllable range.

[0156] In practical implementation, this energy functional serves as the objective function for constructing the flexible perturbation field, and is minimized in subsequent steps using variational methods (or discretization optimization methods). Its optimal perturbation solution ∈ * (x) can be directly applied to the actual wall panel assembly path correction to assist the control mechanism in adjusting the end position of the wall panel.

[0157] Furthermore, to ensure the uniqueness and stability of the solution to the perturbation vector field, it is preferable to apply a Neumann zero boundary condition at the boundary of the solution domain, i.e.:

[0158]

[0159] Where n represents the normal vector outside the domain boundary, this condition is used to prevent the perturbation energy from growing abnormally at the boundary.

[0160] Overall, this step systematically introduces a flexible error field mechanism by constructing a disturbance energy model with spatial smoothness and point matching accuracy control capabilities, effectively connecting rigid body estimation results with flexible correction modules, and forming the basic constraint framework of the assembly error compensation system.

[0161] S5. Minimize the energy functional using the variational method to obtain the optimal perturbation vector field;

[0162] In this embodiment, based on the energy functional constructed in step S4 that depends on the perturbation vector field, the energy functional is further minimized using a variational method to obtain the optimal flexible perturbation vector field ∈ * (x) enables flexible correction and compensation of the wall panel assembly posture.

[0163] The energy functional form is as follows:

[0164]

[0165] in:

[0166] The domain of the perturbation vector field; For a perturbation field; Let ||·|| be the gradient tensor of the perturbation field; F Denotes the Frobenius norm; These are the weights for smoothing control and error control, respectively; δ(xp)i ) is the Dirac function, used to concentrate the error term at the i-th reference point p. i ;q i Rp is the reference point for the target wall panel. i +T represents the coordinates of the current reference point after rigid body transformation, ∈(p i ) is its flexible perturbation offset.

[0167] To minimize E[∈], its first variational derivative is set to zero. Specifically, the generalized Euler-Lagrange method is used, taking the variational derivative component by component, for each ∈ k Calculate the variation for (x)(k=1,2,3):

[0168]

[0169] Let's consider the variational processing of the first term:

[0170]

[0171] By integrating by parts, combined with natural boundary conditions (Neumann boundary), transformed into:

[0172] ∫ Ω -2αΔ∈ k ·δ∈ k dx;

[0173] Applying variational analysis to the second term, note that the Dirac function only applies when x = p i Non-zero:

[0174] δ(∫ Ω β∑ i δ(xp i )||r i -∈(p i )|| 2 dx)=∑ i 2β(∈ k (p i )-r i,k )δ∈ k (p i );

[0175] in:

[0176] r i =q i -Rp i -T represents the target deviation after rigid body transformation; r i,k It represents its k-th component.

[0177] Therefore, the Euler-Lagrange governing equations can be obtained as follows:

[0178]

[0179] Combining the three components k = 1, 2, 3, we obtain the partial differential governing equation in vector form:

[0180]

[0181] In this embodiment, the above-mentioned governing equations are numerically discretized within the spatial region Ω, preferably using the finite difference method or the finite element method: In the finite difference method, the region Ω is divided into a regular grid, and the node numbers are set as j∈{1,…,M}. Each disturbance component is denoted at node j as… The Laplace term is approximated using the central difference:

[0182]

[0183] Where h is the grid step size, and the offset number represents the adjacent grid node.

[0184] The Dirac function in the error term can be approximated in numerical implementation as:

[0185]

[0186] Where χ is the node containment function, in the node containing p i The value is 1 in the grid cells, otherwise it is 0.

[0187] Thus, the governing equations are constructed into the following discrete linear system:

[0188] A·∈=b;

[0189] in:

[0190] The system stiffness matrix is ​​constructed by discretizing the Laplace terms and relating them to the connections between nodes. This is the column vector of the expanded perturbation vector field at all nodes; This is the sum of the projections of the reference point error term onto the grid.

[0191] The error term for each reference point will be mapped to each node within its respective grid cell and distributed through interpolation coefficients.

[0192] For pseudolinear systems, iterative methods, such as the preconditional conjugate gradient method (PCG) or the multigrid method (MG), are preferred for solving to improve the efficiency of the solution.

[0193] After the solution is completed, the optimal perturbation vector field value ∈ at all spatial nodes can be obtained. * (x j This is further used to calculate each reference point p. iFinal attitude adjustment amount:

[0194] Δp i =∈ * (p i );

[0195] Preferably, at non-mesh nodes, trilinear interpolation or interpolation methods based on finite element shape functions can be used to calculate the perturbation value.

[0196] The disturbance vector is output to the control layer module as an attitude adjustment compensation amount, which is used to drive the fine-tuning control of the end effector of the wall panel, so as to realize flexible and high-precision assembly under the error compensation between the walls.

[0197] S6. Based on the optimal disturbance vector field, control multiple fine-tuning actuators of the wall panel to perform attitude fine-tuning and complete the physical assembly connection between the wall panel and the target wall panel.

[0198] In this embodiment, the optimal solution of the perturbation vector field is obtained through step S5. * After (x), the wall panel posture is further flexibly fine-tuned and controlled in combination with the wall panel structure configuration and assembly requirements, so as to finally realize the physical connection and positioning between the current wall panel and the target wall panel.

[0199] Preferably, the wall panel assembly structure of the present invention is provided with multiple fine-tuning actuators on its outer edge, structural surface or specific support frame, forming a multi-degree-of-freedom fine-tuning control unit. Each actuator is used to control the small-amplitude position and attitude adjustment of the wall panel in three-dimensional space along a specific direction (such as vertical, normal, tangential), and its movement is controlled by the calculation results of the flexible disturbance field.

[0200] In this embodiment, the perturbation vector field ∈ * The solution to (x) is defined in the wall panel region. Within, for any control reference point Its fine-tuning compensation amount is defined as:

[0201] Δp i =∈ * (p i );

[0202] in, This represents the required three-dimensional displacement correction at that point. The target motion can be obtained by extracting the disturbance field value at each actuator's point of action.

[0203] Preferably, to ensure control consistency and responsiveness, the center coordinates of the area controlled by each actuator j∈{1,2,...,m} are marked as x. j ∈Ω, its control input quantity Defined as:

[0204] u j=∈ * (x j );

[0205] Among them, u j This represents the spatial micro-displacement vector that the j-th actuator needs to output, the direction and magnitude of which are determined by the value of the disturbance vector field at that location.

[0206] When multiple actuators act on the same region, it is preferable to use a weighted average or local interpolation method to aggregate the disturbance field across regions, i.e.:

[0207] Where ∑ i w ij =1,w ij ≥0;

[0208] Among them, w ij The weighting coefficients can be calculated based on factors such as spatial distance, actuator coverage radius, and force-bearing area. This weighting strategy can achieve the fusion of redundant control constraints, improving system coordination and stability. Control commands for each fine-tuning actuator are output to the lower-level drive module via the system bus and refreshed in real time during the control cycle. After receiving the command, the actuator performs attitude fine-tuning operations through its internal position loop or incremental driver, causing the entire wall panel to generate a disturbance vector field ∈ * The micro-displacement behavior defined by (x).

[0209] In this embodiment, to prevent pose coupling errors generated during fine-tuning from affecting the overall registration, a control matrix can be preferably introduced. This represents the mapping relationship from the actuator array to the six-dimensional space of the rigid body. Based on the current actuator motion distribution and the target assembly residual, the system constructs a pseudo-inverse control strategy online.

[0210] u=J + ·Δξ;

[0211] in:

[0212] The correction amount for the desired rigid body transformation (including translation and rotation) at the end of the wall panel can be obtained by aggregating the disturbance field at multiple key points;

[0213] J + This represents the pseudo-inverse control matrix, used to allocate control resources under redundant control.

[0214] This represents a concatenated vector of control inputs from multiple actuators.

[0215] Through the above strategy, the actuators work together to accurately realize the physical representation of the flexible disturbance field on the structure.

[0216] After the actuator completes its fine-tuning, the wall panel as a whole will gradually tend to achieve a highly consistent attitude match with the target wall panel in space, and the remaining residual will be controlled by the flexible field micro-compensation.

[0217] In this embodiment, during the control adjustment phase, the system continuously monitors each reference point p. i The real-time pose is obtained, and its actual error is calculated:

[0218] e i =q i -(Rp i +T+Δp i );

[0219] If the preset error tolerance ||e is met i ||≤∈ th If so, the control determines that the current assembly can enter the final connection state.

[0220] After assembly and positioning, the wall panel will be physically connected to the target wall panel, preferably using plug-in, magnetic, bolt connection, or other mechanical coupling methods to achieve final fixation. The connection module is also triggered by the central control unit based on the error closed-loop results.

[0221] At this point, the flexible correction control of the wall panel posture and the actual physical assembly process have been integrated in a closed loop.

[0222] The assembly system for prefabricated firewalls in substations described below and the assembly method for prefabricated firewalls in substations described above can be referred to in correspondence.

[0223] Please see Figure 2 The present invention also provides an assembly system for prefabricated firewalls in substations, comprising:

[0224] The positioning and calibration module is used to set up spatial positioning reference points on the edge of the wall panel and store their three-dimensional coordinate information;

[0225] The three-dimensional point acquisition module is used to acquire the positions of reference points on the wall panel to be assembled and the target wall panel, forming a first point set and a second point set;

[0226] The initial attitude estimation module is used to calculate the rotation matrix and displacement vector based on the point set;

[0227] The perturbation modeling module is used to construct the energy functional of a flexible rigid body containing a perturbation vector field;

[0228] The numerical solution module is used to solve the optimal perturbation field based on the control partial differential equations.

[0229] The attitude fine-tuning module is used to control the electric push rods on the wall panel to make spatial fine-tuning according to the disturbance field;

[0230] The assembly connection module is used to position and close the wall panel after fine-tuning.

[0231] The closed-loop error verification module is used to rescan the reference point and determine whether the adjustment error is lower than the set threshold.

[0232] The device in this embodiment can be used to execute the above method embodiments, and its principle and technical effects are similar, so they will not be described again here.

[0233] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An assembly method for prefabricated firewalls in substations, characterized in that, Includes the following steps: S1. Set multiple spatial positioning reference points on the edges of the wall panel to be assembled and the target wall panel respectively; S2. Collect the three-dimensional coordinate data of spatial positioning reference points using a three-dimensional scanning device to form the first point set and the second point set; S3. Calculate the initial rigid body transformation parameters between the wall panels based on the first point set and the second point set; S4. Construct a perturbation vector field based on the flexible rigid body transformation model, and establish an energy functional to describe the total error; S5. Minimize the energy functional using the variational method to obtain the optimal perturbation vector field; S6. Based on the optimal disturbance vector field, control multiple fine-tuning actuators of the wall panel to perform attitude fine-tuning and complete the physical assembly connection between the wall panel and the target wall panel.

2. The assembly method for a prefabricated firewall in a substation according to claim 1, characterized in that, The steps for setting a spatial positioning reference point include the following sub-steps: At least four markers are evenly distributed along the edge of the wall panel, and the markers are reflective posts or metal inserts. During the factory manufacturing stage, the three-dimensional coordinates of each reference point are calibrated using high-precision positioning equipment; The reference point numbers are stored in the digital model for point mapping in the subsequent alignment process.

3. The assembly method for a prefabricated firewall in a substation according to claim 1, characterized in that, The process of acquiring 3D coordinate data includes the following sub-steps: Start the structured light scanning system and simultaneously locate the edges of the current wall panel and the target wall panel; Obtain the spatial coordinates of each reference point to form the first point set and the second point set respectively; The point set is subjected to coordinate system unification and noise filtering.

4. The assembly method for a prefabricated firewall in a substation according to claim 1, characterized in that, The steps for calculating the initial rigid body transformation parameters include: Calculate the centroids of the first and second point sets and center the point sets; Construct a covariance matrix based on a centralized point set and perform singular value decomposition on it; Based on the decomposition results, the rotation matrix R and the translation vector T are obtained, where T=q mean -R·p mean ; In the formula, U and V are matrices obtained from singular value decomposition, p mean q mean These are the centroids of the two point sets, respectively.

5. The assembly method for a prefabricated firewall in a substation according to claim 4, characterized in that, The initial rigid body transformation satisfies the following least squares optimization model: Among them, {p i } represents the current set of wall panel points, {q i Let} be the target wall panel point set, R be the rotation matrix, and T be the translation vector.

6. The assembly method for a prefabricated firewall in a substation according to claim 1, characterized in that, The steps to construct a flexible rigid body perturbation model include establishing the following form Energy functional: Where ∈(x) represents the perturbation vector field, α controls rigidity preservation, β controls matching accuracy, and δ is the Dirac function, ||·|| F This represents the Frobenius norm.

7. The assembly method for a prefabricated firewall in a substation according to claim 6, characterized in that, Solving for the optimal perturbation vector field includes: Variational differentiation of the energy functional E[∈] yields the perturbation control partial differential equation, which takes the form: αΔ∈(x)-β∑ i δ(xp i )(q i -Rp i -T-∈(p i ))=0; Where, ∈(x) is the perturbation vector field, Δ represents the Laplace operator, and q i and p i These represent the target and current reference point positions, respectively. i δ(xp i ) is the Dirac function; The governing equations are discretized in the spatial region Ω, and the equations are numerically discretized using the finite difference method or the finite element method. The optimal perturbation vector field ∈ is solved by numerical calculation. * (x), and output the adjustment amount Δx for each reference point. i This minimizes the system error.

8. The assembly method for a prefabricated firewall in a substation according to claim 7, characterized in that, The optimal solution of the perturbation vector field ∈ * (x) satisfies the following governing equations: Where E[∈] represents the energy functional.

9. The assembly method for a prefabricated firewall in a substation according to claim 1, characterized in that, The attitude fine-tuning and assembly connection process includes the following three sub-steps: The adjustment amount of each reference point of the disturbance vector field is converted into a control signal and sent to the electric micro actuator. Each push rod performs a corresponding displacement to correct the spatial attitude error of the wall panel; After the error converges, the wall panel connection and fixation are completed by structural limit buckles or sliding rail structure.

10. An assembly system for a prefabricated firewall in a substation, applied to the assembly method for a prefabricated firewall in a substation as described in any one of claims 1-9, characterized in that, include: The positioning and calibration module is used to set up spatial positioning reference points on the edge of the wall panel and store their three-dimensional coordinate information; The three-dimensional point acquisition module is used to acquire the positions of reference points on the wall panel to be assembled and the target wall panel, forming a first point set and a second point set; An initial attitude estimation module is used to calculate the rotation matrix and displacement vector based on the point set; The perturbation modeling module is used to construct the energy functional of a flexible rigid body containing a perturbation vector field; The numerical solution module is used to solve the optimal perturbation field based on the control partial differential equations. The attitude fine-tuning module is used to control the electric push rods on the wall panel to make spatial fine-tuning according to the disturbance field; The assembly connection module is used to position and close the wall panel after fine-tuning. The closed-loop error verification module is used to rescan the reference point and determine whether the adjustment error is lower than the set threshold.

Images