Method and device for positioning and mounting side wall of assembly trolley of laminated assembly type subway station

Abstract

The invention provides a side wall positioning and mounting method and device for an assembly trolley of an overlapped assembly type subway station, and belongs to the technical field of underground engineering construction. The method comprises the following steps: an environment perception and initial positioning step: scanning a construction environment, and determining spatial positions of a side wall and an H-shaped steel groove relative to a preset BIM model through a target identification and point cloud registration technology; a transverse moving and coarse positioning step: planning a path based on a positioning result, and controlling the assembly trolley through real-time pose estimation and visual tracking to enable the side wall to move to a position above the target; an accurate positioning and leveling step: obtaining deviation and real-time attitude through visual detection and multi-sensor fusion technology, and driving an adjusting mechanism to carry out accurate alignment and leveling; a gap detecting and filling step: after the side wall falls in place, measuring the shape of the gap and guiding filling construction; according to the invention, global absolute position calibration is realized, error accumulation is avoided, construction safety is ensured through non-contact measurement, and positioning precision and reliability are ensured by using real-time feedback control.

Description

Technical Field

[0001] This invention belongs to the field of prefabricated building construction technology, specifically relating to a method and device for positioning and installing the side wall of a composite prefabricated subway station assembly trolley. Background Technology

[0002] Prefabricated subway stations are becoming a development trend in subway construction. By disassembling the station structure into various components for factory prefabrication and realizing prefabricated construction on site, the level of mechanized operation can be significantly improved, thereby improving quality, accelerating construction progress, and reducing labor input and material usage.

[0003] Currently, prefabricated metro stations are widely constructed using assembly trolleys. After the prefabricated components reach the required strength, they are transported to the site by specialized vehicles. A gantry crane is used to lift each component for posture adjustment and alignment. The assembly trolley is then used to temporarily fix and fine-tune the components, completing the assembly of components such as side walls, central slabs, columns, and longitudinal beams.

[0004] During assembly, positioning typically relies on discrete point surveying equipment such as total stations, supplemented by manual string lines. This method struggles to obtain comprehensive information about the construction environment, resulting in incomplete positioning data. Furthermore, it is highly susceptible to error accumulation, leading to significant deviations in the final alignment of the sidewalls and H-shaped steel channels in the global coordinate system, severely impacting the overall linearity and connection quality of the station structure. Moreover, frequent entry of construction personnel under the assembly trolley or close proximity to heavy sidewall components for close-range measurement, observation, and adjustment exposes them to the risks of moving and potential falls of heavy components, making it difficult to effectively guarantee construction safety.

[0005] During the precise alignment and leveling stage of the sidewall, existing methods rely heavily on the visual observation and experience of the operators. The human eye cannot accurately identify and quantify alignment deviations, and the adjustment process requires time to adjust the angle, which affects construction efficiency. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method and device for positioning and installing the side wall of a prefabricated metro station assembly trolley, ensuring installation accuracy and improving construction safety and reliability.

[0007] The technical solution of this invention is: a method for positioning and installing the side wall of a prefabricated metro station assembly trolley, comprising the following steps:

[0008] Environmental perception and initial positioning: The construction environment is scanned using non-contact 3D measurement sensors to obtain point cloud and image data. Through target recognition and point cloud registration technology, the spatial position of the side wall and H-shaped steel channel mounted on the assembly trolley relative to the preset BIM model is determined.

[0009] Lateral movement and coarse positioning: Based on the initial positioning results, the lateral movement path of the assembly trolley is planned, and during the lateral movement, the assembly trolley is controlled in a closed loop through real-time pose estimation and visual tracking, so that the side wall moves above the H-shaped steel channel.

[0010] Precise positioning and leveling: Before the side wall is lowered, visual inspection and multi-sensor fusion technology are used to obtain the sub-pixel level alignment deviation between the H-beam and the slot and the real-time attitude of the side wall. Based on the alignment deviation and real-time attitude data, the adjustment mechanism on the assembly trolley is driven to perform precise alignment and height leveling.

[0011] Gap detection and filling: After the side wall is in place, the three-dimensional shape of the gap between its lower surface and the H-shaped steel channel is measured by three-dimensional scanning technology. Based on this three-dimensional gap shape data, the filling construction is planned and guided.

[0012] Furthermore, in the environmental perception and initial localization step, the target recognition employs a deep learning-based target detection algorithm, which includes the following steps:

[0013] The point cloud and image data acquired by the non-contact three-dimensional measurement sensor are used as input;

[0014] Output the identification results and location information of the side wall contour features and H-shaped steel channel in the BIM model coordinate system.

[0015] Furthermore, the point cloud registration employs an iterative nearest-point algorithm, which includes the following steps:

[0016] The real-time collected site point cloud data is registered with the theoretical point cloud data of the preset BIM model.

[0017] By minimizing the spatial distance between the field point cloud and the theoretical point cloud, the pose transformation matrix of the side wall mounted on the assembly trolley relative to the target installation position is calculated.

[0018] Furthermore, in the lateral movement and coarse localization steps, the real-time pose estimation employs an adaptive Monte Carlo localization algorithm, which includes the following steps:

[0019] Based on the motion control commands of the assembly trolley and the data from the inertial measurement unit, the pose of the sidewall is predicted;

[0020] The point cloud data collected in real time, which includes the side wall and environmental features, is matched with the preset BIM model to update the confidence level of the side wall pose.

[0021] The algorithm parameters are dynamically adjusted based on the confidence level, and the high-precision real-time pose of the side wall in the BIM model coordinate system is output as the estimation result.

[0022] Furthermore, in the lateral movement and coarse localization steps, the visual tracking employs a feature point matching method based on ORB features. This ORB feature point matching method includes the following steps:

[0023] ORB feature points are extracted and feature descriptors are generated in a preset area on the side wall surface and the adjacent H-shaped steel channel surface mounted on the assembly trolley.

[0024] During the lateral movement of the assembly trolley, the ORB feature points are continuously matched and tracked between adjacent frames of the video sequence;

[0025] Based on the successfully matched feature point pairs, the relative positional change between the side wall and the H-shaped steel channel is calculated, and this change is fed back to the closed-loop motion control system of the assembly trolley.

[0026] Furthermore, in the precise positioning and leveling step, the visual detection and multi-sensor fusion technology includes the following steps:

[0027] Detailed images of the lower H-beam and groove of the side wall are obtained using a vision camera;

[0028] Edge and line features are extracted from the detailed image, and the distance deviation between them is calculated with sub-pixel precision;

[0029] Simultaneously acquire the linear acceleration and angular velocity data of the sidewalls measured by the inertial measurement unit;

[0030] The distance deviation and the data from the inertial measurement unit are fused together by an extended Kalman filter to estimate the real-time attitude angle of the sidewall.

[0031] Real-time acquisition of 3D point cloud data of the upper surface of the sidewalls already in place and those to be installed;

[0032] The three-dimensional point cloud data is processed, and the spatial equations of the two wall surfaces are calculated using a plane fitting algorithm, and the height difference is solved.

[0033] The height difference is used as a control signal to drive the adjustment mechanism on the assembly trolley, so that the surface height of the side wall to be installed is leveled with the side wall that is already in place.

[0034] Furthermore, in the gap detection and filling step, the three-dimensional scanning technology is structured light three-dimensional scanning, which measures the gap by projecting a light spot pattern and analyzing its deformation, and generates a three-dimensional model of the gap; based on the three-dimensional model of the gap, a deep learning model is used to analyze the gap features to automatically plan a filling scheme.

[0035] Furthermore, in the gap detection and filling step, digital twin technology is used to simulate the filling effect to guide the precise fabrication of the pads on site.

[0036] An electronic device, including a processor and a memory, the memory storing a computer program, characterized in that: when the processor executes the computer program, it implements the steps of the method described in any of the preceding claims.

[0037] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0038] The beneficial effects of this invention are:

[0039] (1) This invention achieves absolute position calibration of the side wall and steel channel in the global coordinate system by comparing point cloud registration with BIM model, thus avoiding error accumulation;

[0040] (2) By using visual inspection and multi-sensor fusion technology, the sub-pixel level alignment deviation between the H-beam and the slot and the real-time attitude of the side wall are obtained, so as to achieve precise control of the tilting, rotation and other attitudes of the side wall.

[0041] (3) By adopting non-contact three-dimensional measurement sensors, visual inspection and other technologies, the risks of personnel conducting close-range manual measurements under or near heavy components are avoided, thus improving construction safety;

[0042] (4) Uncertainty can be dynamically estimated through real-time pose estimation and visual tracking, and the deviation can be continuously corrected during the lateral movement to prevent the side wall from colliding with the surrounding structure, thus ensuring the safety and reliability of construction. Attached Figure Description

[0043] Figure 1 This is a flowchart of the side wall positioning and installation method of the composite prefabricated subway station assembly trolley in this invention. Detailed Implementation

[0044] Various exemplary embodiments of the invention will now be described in detail with reference to the accompanying drawings. The descriptions of the exemplary embodiments are merely illustrative and are in no way intended to limit the invention or its application or use. The invention can be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to make the invention thorough and complete, and to fully express the scope of the invention to those skilled in the art. It should be noted that, unless otherwise specifically stated, the relative arrangement of components and steps, the composition of materials, numerical expressions, and values ​​set forth in these embodiments should be interpreted as merely exemplary and not as limiting.

[0045] The terms "first," "second," and similar words used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different parts. Words such as "including" or "comprising" mean that the element preceding the word encompasses the element listed after it, without excluding the possibility of encompassing other elements. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0046] like Figure 1 As shown, a method for positioning and installing the side wall of a prefabricated assembly trolley for a composite subway station is disclosed, including the following steps:

[0047] S1, Environmental perception and initial positioning: The construction environment is scanned using non-contact 3D measurement sensors to obtain point cloud and image data. Through target recognition and point cloud registration technology, the spatial position of the side wall and H-shaped steel channel mounted on the assembly trolley relative to the preset BIM model is determined.

[0048] S2, Lateral movement and coarse positioning: Based on the initial positioning results, the lateral movement path of the assembly trolley is planned, and during the lateral movement, the assembly trolley is controlled in a closed loop through real-time pose estimation and visual tracking, so that the side wall moves above the H-shaped steel channel.

[0049] S3, Precise Positioning and Leveling: Before the side wall is lowered, visual inspection and multi-sensor fusion technology are used to obtain the sub-pixel level alignment deviation between the H-beam and the slot and the real-time attitude of the side wall. Based on the alignment deviation and real-time attitude data, the adjustment mechanism on the assembly trolley is driven to perform precise alignment and height leveling.

[0050] S4, Gap Detection and Filling: After the side wall is in place, the three-dimensional shape of the gap between its lower surface and the H-shaped steel channel is measured by three-dimensional scanning technology. Based on this three-dimensional gap shape data, the filling construction is planned and guided.

[0051] Specifically, by comparing point cloud registration with the BIM model, the absolute position calibration of the sidewall and steel channel in the global coordinate system was achieved, avoiding error accumulation. Through visual inspection and multi-sensor fusion technology, the sub-pixel alignment deviation of the H-beam and the channel opening and the real-time posture of the sidewall were obtained, enabling precise control of the sidewall's tilt, rotation, and other postures. The use of non-contact 3D measurement sensors and visual inspection technologies avoided the risk of personnel manually measuring at close range under or near heavy components, improving construction safety. Real-time pose estimation and visual tracking enabled dynamic estimation of uncertainties and continuous correction during lateral movement, preventing collisions between the sidewall and surrounding structures and ensuring the safety and reliability of construction.

[0052] In some embodiments, in the environmental perception and initial localization step, target recognition employs a deep learning-based target detection algorithm, which includes the following steps:

[0053] Point cloud and image data acquired by a non-contact 3D measurement sensor are used as input;

[0054] Output the side wall outline features and the identification results and location information of the H-shaped steel channel in the BIM model coordinate system.

[0055] Specifically, deep learning-based object detection algorithms also include the following steps:

[0056] The real-time acquired point cloud is preprocessed, downsampled using a voxel grid filter, and noise points are removed using a statistical outlier removal algorithm. The preprocessed point cloud and image data are then used as input.

[0057] Theoretical point cloud data is extracted from the BIM model, and theoretical point clouds of H-shaped steel channels and adjacent structures are generated based on the design coordinates. The point cloud density is matched with the point cloud collected on site.

[0058] In some embodiments, point cloud registration employs an iterative nearest point algorithm, which includes the following steps:

[0059] The real-time collected site point cloud data is registered with the theoretical point cloud data of the preset BIM model.

[0060] By minimizing the spatial distance between the on-site point cloud and the theoretical point cloud, the pose transformation matrix of the side wall mounted on the assembly trolley relative to the target installation position is calculated.

[0061] Specifically, the iterative nearest point algorithm includes the following steps:

[0062] Nearest point correspondence establishment: For each point in the field point cloud, find the nearest neighbor in the theoretical point cloud. Use the KD-tree data structure to accelerate the search process and set the maximum correspondence distance threshold.

[0063] Pose transformation matrix calculation: Constructing the objective function:

[0064] min T Σ||P site -T·P bim || 2

[0065] Among them, P site Point cloud for the site; P bim The point cloud of the BIM model is obtained; the optimal rigid body transformation matrix T is solved by SVD decomposition.

[0066] Iterative optimization: Repeat the above two steps until the convergence condition is met. For example, continue until the number of iterations exceeds 100.

[0067] In some embodiments, during the lateral movement and coarse localization steps, the real-time pose estimation employs the Adaptive Monte Carlo Localization (AMCL) algorithm, which includes the following steps:

[0068] Based on the motion control commands of the assembly trolley and the data from the inertial measurement unit, the pose of the sidewall is predicted;

[0069] The point cloud data, which includes side wall and environmental features, is collected in real time and matched with the preset BIM model to update the confidence level of the side wall pose.

[0070] The algorithm parameters are dynamically adjusted based on the confidence level, and the high-precision real-time pose of the side wall in the BIM model coordinate system is output as the estimation result.

[0071] Specifically, the adaptive Monte Carlo localization algorithm includes the following steps:

[0072] Initialization: Generate a swarm of particles around the target location, with each particle containing a pose assumption and weights;

[0073] Prediction Update: Based on the trolley motion commands and IMU data, predict the pose of each particle at the next moment, and add motion noise that conforms to a Gaussian distribution to simulate uncertainty;

[0074] Observation update: Match real-time point cloud data with the BIM model, calculate the observation likelihood under the pose assumption of each particle, and reallocate particle weights according to the matching degree.

[0075] Resampling: The timing of resampling is dynamically determined based on the number of effective particles. Low-weight particles are eliminated, high-weight particles are replicated, and random perturbations are added to maintain particle diversity.

[0076] Pose output: Calculate the weighted average pose of the particle swarm, output the best estimated pose of the sidewall in the BIM coordinate system, and evaluate the positional reliability based on the particle distribution covariance.

[0077] The inertial measurement unit (IMU) data is acquired through IMU sensor modules directly mounted on the side wall to ensure that the acquired inertial measurement unit data can reflect the movement of the side wall.

[0078] In some embodiments, during the lateral movement and coarse localization steps, visual tracking employs an ORB-based feature point matching method, which includes the following steps:

[0079] ORB feature points are extracted and feature descriptors are generated in a preset area on the side wall surface and the adjacent H-shaped steel channel surface mounted on the assembly trolley.

[0080] During the lateral movement of the assembly trolley, ORB feature points are continuously matched and tracked between adjacent frames of the video sequence;

[0081] Based on the successfully matched feature point pairs, the relative positional change between the side wall and the H-shaped steel channel is calculated, and this change is fed back to the closed-loop motion control system of the assembly trolley.

[0082] Specifically, the feature point matching method based on ORB features includes the following steps:

[0083] Feature extraction and initialization: Pre-define textured areas on the side wall surface and H-shaped steel channel surface, use ORB algorithm to extract feature points and generate 256-bit binary descriptors, set the number of image pyramid layers to 8, and the maximum number of feature points to 1000.

[0084] Real-time tracking process: For adjacent frames of the video sequence, Hamming distance is used for feature descriptor matching. A ratio test is applied to eliminate mismatches, with a ratio threshold set to 0.7. The RANSAC algorithm is used to further optimize the matching pairs and remove outliers. Based on the matched feature point pairs, the homography matrix or fundamental matrix between adjacent frames is calculated. The matrix is ​​decomposed to obtain the relative translation and rotation changes of the sidewall and the H-shaped steel channel. When the number of matching points is insufficient, optical flow-based tracking is used as a supplement.

[0085] Control feedback: The calculated relative position change (Δx, Δy, Δθ) is converted into control commands, and the assembly trolley speed adjustment signal is generated by the PID controller. A position change threshold is set, and emergency braking is triggered immediately when the threshold is exceeded. As an example of the position change threshold, emergency braking is triggered when Δx or Δy exceeds 50mm, or Δθ exceeds 2°.

[0086] In some embodiments, the visual inspection and multi-sensor fusion technology in the precise positioning and leveling step includes the following steps:

[0087] Detailed images of the lower H-beam and groove of the side wall are obtained using a vision camera;

[0088] Edge and line features are extracted from the detailed image, and the distance deviation between them is calculated with sub-pixel precision; the linear acceleration and angular velocity data of the sidewall measured by the inertial measurement unit are acquired simultaneously.

[0089] The distance deviation and inertial measurement unit data are input into an extended Kalman filter for fusion processing to predict the real-time attitude angle of the sidewall.

[0090] Real-time acquisition of 3D point cloud data of the upper surface of the sidewalls already in place and those to be installed;

[0091] The three-dimensional point cloud data is processed, and the spatial equations of the two wall surfaces are calculated using a plane fitting algorithm, and the height difference is solved.

[0092] The height difference is used as a control signal to drive the adjustment mechanism on the assembly trolley, so that the surface height of the side wall to be installed is leveled with the side wall that is already in place.

[0093] Specifically, attitude adjustment involves acquiring detailed images of the H-beam and the slot using an industrial camera, calculating the distance deviation between them using edge detection and sub-pixel technology, simultaneously acquiring angular velocity and acceleration data from the IMU, fusing the visual and IMU data using an extended Kalman filter, and outputting the real-time attitude angle of the sidewall. This drives the adjustment mechanism to perform pose correction, achieving precise alignment between the H-beam and the slot. Height leveling, after attitude adjustment, involves acquiring point clouds of the upper surfaces of adjacent sidewalls through 3D scanning, calculating the height difference between the positioned sidewall and the sidewall to be installed using a plane fitting algorithm, converting the height difference into a control signal, and driving the leveling mechanism to adjust the height, making the surfaces of both sidewalls flush. The preferred visual camera is a high-resolution industrial camera with at least 50 megapixels.

[0094] In some embodiments, in the gap detection and filling step, the three-dimensional scanning technology is structured light three-dimensional scanning, which measures the gap by projecting a light spot pattern and analyzing its deformation, and generates a three-dimensional model of the gap.

[0095] Based on the 3D model of the gap, a deep learning model is used to analyze the gap features in order to automatically plan the filling scheme.

[0096] Specifically, a structured light projector projects a coded light spot pattern onto the gap area, and an industrial camera collects the deformed light spot modulated by the gap shape. Based on the principle of triangulation, a three-dimensional point cloud of the gap is reconstructed, generating an accurate three-dimensional model containing the gap size, depth, and shape. The three-dimensional gap model is then input into a pre-trained deep learning network, which uses a 3D convolution-based neural network to analyze the spatial characteristics of the gap, automatically identify the gap type, calculate the volume and shape of the required filling material, and output an optimized filling scheme, including the amount of material and the construction path.

[0097] In some embodiments, digital twin technology is used to simulate the filling effect in the gap detection and filling steps to guide the precise fabrication of the on-site pads.

[0098] In some embodiments, the construction of the digital twin model involves importing the three-dimensional model of the gap obtained by structured light scanning into the digital twin system to establish a virtual model that includes the material properties and contact mechanical characteristics of the pad. The filling effect of pads with different thicknesses is simulated in the virtual environment. The filling effect is simulated and analyzed by running filling simulations of multiple pad schemes, analyzing the contact stress distribution between the pad and the gap, evaluating the load-bearing performance and stability of different pad schemes, and selecting the optimal pad parameters (thickness, shape, material).

[0099] In the above embodiments, the non-contact three-dimensional measurement sensor is a binocular vision sensor or a three-dimensional lidar.

[0100] The side wall positioning and installation device for the modular metro station assembly trolley includes electronic equipment and computer-readable storage media.

[0101] In some embodiments, an electronic device is disclosed, including a processor and a memory, the memory storing a computer program, the processor executing the computer program to implement the steps of the method in any of the above embodiments.

[0102] In some embodiments, a computer-readable storage medium is disclosed having a computer program stored thereon, which, when executed by a processor, implements the steps of the method in any of the above embodiments.

[0103] The various embodiments of the present invention have now been described in detail. To avoid obscuring the concept of the invention, some details known in the art have not been described. Those skilled in the art will fully understand how to implement the technical solutions disclosed herein based on the above description.

[0104] The embodiments described above only illustrate some implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for positioning and installing the side wall of a prefabricated assembly trolley for a modular subway station, characterized in that, Includes the following steps: Environmental perception and initial positioning: The construction environment is scanned using non-contact 3D measurement sensors to obtain point cloud and image data. Through target recognition and point cloud registration technology, the spatial position of the side wall and H-shaped steel channel mounted on the assembly trolley relative to the preset BIM model is determined. Lateral movement and coarse positioning: Based on the initial positioning results, the lateral movement path of the assembly trolley is planned, and during the lateral movement, the assembly trolley is controlled in a closed loop through real-time pose estimation and visual tracking, so that the side wall moves above the H-shaped steel channel. Precise positioning and leveling: Before the side wall is lowered, visual inspection and multi-sensor fusion technology are used to obtain the sub-pixel level alignment deviation between the H-beam and the slot and the real-time attitude of the side wall. Based on the alignment deviation and real-time attitude data, the adjustment mechanism on the assembly trolley is driven to perform precise alignment and height leveling. Gap detection and filling: After the side wall is in place, the three-dimensional shape of the gap between its lower surface and the H-shaped steel channel is measured by three-dimensional scanning technology. Based on this three-dimensional gap shape data, the filling construction is planned and guided.

2. The method for positioning and installing the side wall of the prefabricated metro station assembly trolley according to claim 1, characterized in that: In the environmental perception and initial localization step, the target recognition employs a deep learning-based target detection algorithm, which includes the following steps: The point cloud and image data acquired by the non-contact three-dimensional measurement sensor are used as input; Output the identification results and location information of the side wall contour features and H-shaped steel channel in the BIM model coordinate system.

3. The method for positioning and installing the side wall of the prefabricated metro station assembly trolley according to claim 2, characterized in that: The point cloud registration employs an iterative nearest point algorithm, which includes the following steps: The real-time collected site point cloud data is registered with the theoretical point cloud data of the preset BIM model. By minimizing the spatial distance between the field point cloud and the theoretical point cloud, the pose transformation matrix of the side wall mounted on the assembly trolley relative to the target installation position is calculated.

4. The method for positioning and installing the side wall of the prefabricated metro station assembly trolley according to claim 1, characterized in that: In the lateral movement and coarse localization steps, the real-time pose estimation employs an adaptive Monte Carlo localization algorithm, which includes the following steps: Based on the motion control commands of the assembly trolley and the data from the inertial measurement unit, the pose of the sidewall is predicted; The point cloud data collected in real time, which includes the side wall and environmental features, is matched with the preset BIM model to update the confidence level of the side wall pose. The algorithm parameters are dynamically adjusted based on the confidence level, and the high-precision real-time pose of the side wall in the BIM model coordinate system is output as the estimation result.

5. The method for positioning and installing the side wall of the prefabricated metro station assembly trolley according to claim 4, characterized in that: In the lateral movement and coarse localization steps, the visual tracking employs an ORB-based feature point matching method, which includes the following steps: ORB feature points are extracted and feature descriptors are generated in a preset area on the side wall surface and the adjacent H-shaped steel channel surface mounted on the assembly trolley. During the lateral movement of the assembly trolley, the ORB feature points are continuously matched and tracked between adjacent frames of the video sequence; Based on the successfully matched feature point pairs, the relative positional change between the side wall and the H-shaped steel channel is calculated, and this change is fed back to the closed-loop motion control system of the assembly trolley.

6. The method for positioning and installing the side wall of the prefabricated metro station assembly trolley according to claim 1, characterized in that, In the precise positioning and leveling step, the visual detection and multi-sensor fusion technology includes the following steps: Detailed images of the lower H-beam and groove of the side wall are obtained using a vision camera; Edge and line features are extracted from the detailed image, and the distance deviation between them is calculated with sub-pixel precision; Simultaneously acquire the linear acceleration and angular velocity data of the sidewalls measured by the inertial measurement unit; The distance deviation and the data from the inertial measurement unit are fused together by an extended Kalman filter to estimate the real-time attitude angle of the sidewall. Real-time acquisition of 3D point cloud data of the upper surface of the sidewalls already in place and those to be installed; The three-dimensional point cloud data is processed, and the spatial equations of the two wall surfaces are calculated using a plane fitting algorithm, and the height difference is solved. The height difference is used as a control signal to drive the adjustment mechanism on the assembly trolley, so that the surface height of the side wall to be installed is leveled with the side wall that is already in place.

7. The method for positioning and installing the side wall of the prefabricated metro station assembly trolley according to claim 1, characterized in that: In the gap detection and filling step, the three-dimensional scanning technology is structured light three-dimensional scanning, which measures the gap by projecting a light spot pattern and analyzing its deformation, and generates a three-dimensional model of the gap. Based on the 3D model of the gap, a deep learning model is used to analyze the gap features in order to automatically plan the filling scheme.

8. The method for positioning and installing the side wall of the prefabricated metro station assembly trolley according to claim 1, characterized in that: In the gap detection and filling step, digital twin technology is used to simulate the filling effect to guide the precise fabrication of the pads on site.

9. An electronic device, characterized in that, The method includes a processor and a memory, the memory storing a computer program, characterized in that: when the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images