A substation civil engineering quality intelligent detection and auxiliary control management system

Abstract

The application discloses a substation civil engineering quality intelligent detection and auxiliary control management system and belongs to the technical field of intelligent construction. The system comprises: an edge perception layer containing ground and aerial autonomous mobile collection equipment for continuously collecting three-dimensional point cloud data of a structure surface; a network transmission layer providing a communication link; a digital twin basement layer importing a design information model and constructing a digital twin; an intelligent analysis and calculation layer spatially registering the point cloud and the design model, calculating a normal distance set of an actual surface to a design model surface, and generating three-dimensional deviation distribution information; and an auxiliary business application layer superimposing the deviation information on the digital twin for visual display and alarming when the deviation is out of tolerance, and simultaneously comprising an auxiliary control execution module. The application changes traditional artificial discrete sampling detection into spatial continuous deviation distribution analysis, realizes full-coverage detection and three-dimensional visual evaluation of structure geometric size deviation, and links field equipment to complete automatic maintenance and safety control.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction and substation engineering management technology, specifically to an intelligent detection and auxiliary control management system for the quality of substation civil engineering projects. Background Technology

[0002] In substation civil engineering, the flatness and verticality deviations of structures such as concrete walls, equipment foundations, and structural columns directly determine the installation accuracy and operational safety of subsequent electrical equipment. Currently, in engineering practice, the detection of geometric dimensional deviations of these structures mainly relies on manual sampling measurements using tools such as total stations, levels, straightedges, and feeler gauges. Due to limitations in detection efficiency, on-site quality management personnel can only conduct spot checks on a limited number of measuring points on the structural surface after formwork installation acceptance and demolding, leaving a large number of areas with dimensional deviations unchecked. This discrete sampling detection method has two fundamental drawbacks: first, the spatial coverage is discontinuous, resulting in blind spots, and missed areas exceeding tolerances may affect equipment installation; second, discrete measuring point data cannot present a continuous spatial distribution of deviations on the structural surface, making it difficult to predict and warn of deviation trends, and also failing to provide spatial coordinate basis for subsequent precise rectification. For a long time, it has been generally believed in this field that manual discrete sampling detection, through reasonable layout of measuring points and increased sampling frequency, can meet the quality control needs of substation civil engineering, thus lacking the motivation to develop spatial continuous full-coverage detection technology. Therefore, how to achieve spatial continuity, full coverage detection, and three-dimensional visualization evaluation of geometric dimensional deviations in the civil engineering structure of substations is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0003] This invention provides an intelligent detection and auxiliary control management system for the quality of substation civil engineering projects, which solves the problems of spatial discontinuity, invisible deviation distribution, and lack of continuous evaluation capability caused by relying on manual discrete sampling detection in the prior art.

[0004] A substation civil engineering quality intelligent detection and auxiliary control management system includes:

[0005] The edge perception layer, configured at the substation civil construction site, includes ground-based autonomous mobile acquisition equipment and aerial autonomous mobile acquisition equipment, used to continuously acquire three-dimensional point cloud data of the structural surface at the construction site.

[0006] The network transport layer and the communication connection edge perception layer are used to provide a communication link for transmitting 3D point cloud data from the construction site to the upper-level node.

[0007] The digital twin base layer and the communication connection network transmission layer are used to import the design information model of the substation civil engineering project and receive the 3D point cloud data uploaded by the edge perception layer. The digital twin of the substation civil engineering project is constructed through virtual-real mapping.

[0008] The intelligent analysis and computing layer is connected to the digital twin base layer. It has a built-in 3D point cloud processing engine to spatially register 3D point cloud data with the design information model, calculate the set of normal distances from the actual structural surface to the design model surface, and generate 3D deviation distribution information.

[0009] The auxiliary business application layer connects the intelligent analysis and calculation layer and the digital twin base layer. It includes a deviation visualization display module and a deviation over-tolerance alarm module, which are used to overlay and render the three-dimensional deviation distribution information onto the digital twin for visualization display, and generate alarm information when the normal distance exceeds the preset allowable error threshold.

[0010] In the above technical solution, the edge perception layer achieves continuous and full spatial coverage acquisition of three-dimensional point clouds on the outer surface of the substation civil structure through the collaboration of ground and air mobile devices; the intelligent analysis and calculation layer transforms traditional discrete sampling detection into continuous distribution analysis of structural surface deviations by registering the point cloud with the design model and calculating the normal distance set; the auxiliary business application layer overlays the deviation information onto the digital twin in the form of a three-dimensional heat map, realizing the visualization and traceable assessment of the spatial distribution of deviations, and significantly improving the detection coverage and the spatiotemporal continuity of deviation assessment.

[0011] To more intuitively demonstrate the advantages of this invention over the prior art, the following table lists a comparison of key dimensions:

[0012] Detection coverage Sampling point coverage is typically less than 5%. 100% full coverage of structural surfaces Deviation visualization List of discrete measurement points Three-dimensional continuous deviation heatmap Detection efficiency 100㎡ requires 2 people for 4 hours One robot is needed for 100 square meters in 15 minutes. Data traceability Paper records are easily lost or altered. Blockchain-based evidence storage ensures immutability. Maintenance control Manual timed watering is uneven and delayed. Temperature difference linkage + feedforward control, fully automatic closed loop

[0013] Preferably, the ground-based autonomous mobile data acquisition equipment includes a quadruped robot swarm, each carrying a high-precision multi-line lidar, a high-definition panoramic gimbal camera, an infrared thermal imager, and a ground-penetrating radar probe. The aerial autonomous mobile data acquisition equipment includes a multi-rotor drone swarm, each carrying an oblique photography camera and a multispectral sensor. The edge perception layer also includes miniature temperature and humidity sensors and stress-strain fiber optic grating sensors embedded in concrete, as well as dust and noise environmental monitoring stations and intelligent high-definition monitoring PTZ cameras deployed at the construction site. The quadruped robot swarm and the multi-rotor drone swarm achieve integrated air-ground data acquisition through a preset collaborative inspection route. The quadruped robot swarm is responsible for scanning the structural surface from the ground to a height of 3 meters, while the multi-rotor drone swarm is responsible for oblique photography of structures above 3 meters and large-scale sites. The data collected by both are fused into a unified overall point cloud model in the digital twin base layer through coordinate registration. This preferred scheme enables the system to have multi-dimensional and multi-scale perception capabilities. The quadruped robot swarm and the drone formation form an air-ground collaborative data collection and coverage network, while fixed sensors provide continuous monitoring of the internal state of the structure, achieving full-range perception coverage from the macroscopic appearance to the internal hidden engineering.

[0014] Preferably, the network transport layer employs a customized slicing network based on 5G mobile communication technology combined with an industrial-grade passive optical network architecture. Specifically, enhanced mobile broadband network slices are allocated for uplink transmission of video streams and 3D point cloud data, while massive machine-type communication network slices are allocated for the miniature sensor nodes distributed within the concrete. This preferred solution, through a differentiated slice allocation strategy, balances the efficient transmission of high-bandwidth data with the stable access requirements of a massive number of low-power sensor nodes, ensuring the timely and reliable uploading of multi-source heterogeneous data in the field network environment.

[0015] Preferably, the intelligent analysis and computing layer receives 3D point cloud data from the high-precision multi-line lidar, applies a voxel filtering algorithm to remove outlier noise points, extracts line segments from the edges of building structural features, and applies an iterative nearest-point algorithm to register the preprocessed point cloud with the design information model. After registration, the set of normal distances from the actual structural surface to the design model surface is calculated. When the normal distance exceeds a preset tolerance threshold, an out-of-tolerance warning area is marked in the digital twin, and a 3D deviation heatmap is generated. This preferred scheme realizes a complete processing chain from raw point cloud acquisition to deviation visualization output. The combination of voxel filtering and the iterative nearest-point algorithm ensures the accuracy and robustness of point cloud registration, and the 3D deviation heatmap intuitively presents the spatial distribution pattern of the deviation.

[0016] Preferably, the intelligent analysis and calculation layer also invokes an improved instance segmentation neural network model with a built-in deformable convolution module to perform pixel-by-pixel classification and bounding box regression on the concrete surface image captured by the high-definition panoramic gimbal camera to segment defect areas. For identified crack defects, the crack skeleton centerline is extracted, the pixel distance is calculated along the normal direction of the centerline, and the actual physical width of the crack is converted by combining the camera calibration parameters and the depth information provided by the high-precision multi-line LiDAR. This preferred solution combines deep learning instance segmentation with LiDAR depth measurement, realizing fully automated processing of crack defects from image segmentation to physical width quantification, overcoming the problems of strong subjectivity and low efficiency in width estimation in traditional manual crack detection.

[0017] Preferably, during the rebar tying stage, the multi-rotor UAV formation acquires orthophotos and oblique images of the rebar skeleton. The intelligent analysis and calculation layer uses a motion reconstruction structure algorithm to reconstruct the 3D point cloud of the rebar skeleton, extracts the cylindrical geometric features to identify individual rebars, and calculates the actual horizontal spacing and row spacing. Simultaneously, the auxiliary business application layer is directly connected to augmented reality smart glasses worn by on-site personnel, overlaying and rendering the correct 3D rebar layout model from the design information model onto the real physical space. This preferred solution, through the combination of UAV aerial surveying and the motion reconstruction structure algorithm, achieves rapid 3D reconstruction and automated spacing detection of large-area rebar skeletons. The direct on-site overlay display through augmented reality glasses allows construction personnel to compare the differences between the design and the actual rebar positions in real time, enabling immediate detection and rectification of quality problems.

[0018] Preferably, the intelligent analysis and computing layer receives radar profile data echo images acquired by a ground-penetrating radar probe mounted on a tracked microrobot. It then applies a two-dimensional Hilbert transform to extract the instantaneous envelope attributes of the radar signal, automatically extracts the vertex coordinates of the hyperbola using a pattern recognition algorithm, and inverts the thickness of the reinforcing steel protective layer by combining time delay parameters and electromagnetic wave propagation speed. Furthermore, it extracts the cavity contour based on phase polarity reversal and amplitude amplification features, and performs three-dimensional perspective reconstruction within the digital twin. This preferred scheme applies ground-penetrating radar signal processing technology to the detection of concealed civil engineering works in substations. The two-dimensional Hilbert transform enhances the ability to identify weak signal characteristics of the radar echo, the automatic extraction of hyperbola vertices achieves high-precision inversion of the protective layer thickness, and the phase reversal and amplitude amplification features provide clear physical criteria for identifying cavity defects.

[0019] Preferably, the auxiliary business application layer also includes an auxiliary control execution module. This module communicates in real time with a miniature temperature and humidity sensor embedded in the concrete to calculate the internal and external temperature gradient. When the temperature difference approaches a set critical safety value, the auxiliary business application layer automatically sends an activation command to the automatic spray network controller and reduces the heating wire power of the intelligent temperature-controlled curing blanket. Simultaneously, based on wind speed and humidity data from an external weather station, a feedforward control algorithm dynamically adjusts the spraying duration and interval of the sprinkler system. This preferred solution achieves closed-loop automatic control for the curing of large-volume concrete. The temperature gradient monitoring and the linkage execution of spraying and heating temperature adjustment form a rapid response mechanism, while the meteorological feedforward control pre-adjusts the curing parameters according to environmental changes. Compared with the traditional manual timed watering curing scheme, this significantly improves the uniformity and adaptability of the curing process.

[0020] Preferably, the auxiliary business application layer constructs a dynamic personnel and equipment trajectory electronic fence system. This system determines boundary crossing behavior and triggers linked alarms and monitoring capture by calculating the three-dimensional coordinates of the smart safety helmet worn by personnel, which has a built-in ultra-wideband positioning chip. Regarding special equipment access control, the auxiliary business application layer uses a biometric terminal installed in the special equipment's cab and a controller local area network bus interception module to perform facial feature comparison and operator qualification verification. After successful verification, the bus interception module releases the engine ignition control signal. This preferred solution combines personnel positioning electronic fences with equipment activation biometrics. The electronic fence enables real-time alarms and evidence recording for boundary crossings in dangerous areas, while biometrics and controller local area network bus interception effectively physically block unlicensed operations at the hardware level, constructing a multi-level security protection system from behavior monitoring to operation access control.

[0021] Preferably, the system employs distributed ledger technology to extract the characteristic hash values ​​of front-end sensing data and back-end analysis results throughout the entire lifecycle of substation civil engineering projects, and uploads them to an immutable blockchain node network. During the project completion and acceptance phase, the auxiliary business application layer automatically organizes the corresponding data according to the hierarchical structure tree of the civil engineering projects, generating a three-dimensional digital twin acceptance body with embedded timestamps and spatial coordinates for digital acceptance file delivery. This preferred solution uses blockchain technology to solidify the hash values ​​of inspection data and auxiliary control records on the chain, ensuring the immutability and time-tested validity of quality data. The three-dimensional digital twin acceptance body integrates the quality data throughout the entire lifecycle in a spatially locateable and traceable manner, providing a highly credible data foundation for digital operation and maintenance after project completion and delivery.

[0022] Compared with existing technologies, this invention acquires three-dimensional point clouds of structural surfaces through autonomous mobile acquisition devices on the ground and in the air. By combining the point cloud with the design model registration and normal distance set calculation, it transforms traditional manual discrete sampling inspection into spatially continuous full-coverage deviation distribution analysis. Furthermore, it achieves three-dimensional visualization of deviations with the help of digital twins, thereby fundamentally solving the problems of discontinuous spatial coverage, invisible deviation distribution, and lack of continuous evaluation capability in the geometric dimension inspection of substation civil engineering structures. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the system architecture of the present invention. Detailed Implementation

[0024] Example 1: System Overall Architecture

[0025] See Figure 1This embodiment provides an intelligent detection and auxiliary control management system for the quality of substation civil engineering projects. The system includes an edge sensing layer, a network transmission layer, a digital twin base layer, an intelligent analysis and calculation layer, and an auxiliary business application layer.

[0026] The edge sensing layer is deployed at the substation construction site and includes ground-based autonomous mobile data acquisition equipment and aerial autonomous mobile data acquisition equipment to continuously acquire 3D point cloud data of the structural surfaces at the construction site. The ground-based autonomous mobile data acquisition equipment is preferably a swarm of quadruped robots, and the aerial autonomous mobile data acquisition equipment is preferably a formation of multi-rotor UAVs. Working together, they achieve blind-spot-free lidar scanning of the outer surfaces of structures such as substation firewalls, main transformer foundations, and structural columns.

[0027] The network transport layer connects to the edge sensing layer, providing a communication link for transmitting 3D point cloud data from the construction site to the upper-level node. In this embodiment, the network transport layer uses a customized slicing network based on 5G mobile communication technology in conjunction with an industrial-grade passive optical network architecture to ensure low-latency, high-bandwidth uplink transmission of massive point cloud data.

[0028] The digital twin's underlying communication layer connects to the network transmission layer, which imports design information models of the substation's civil engineering projects and receives 3D point cloud data uploaded by the edge perception layer. Through virtual-physical mapping, a digital twin of the substation's civil engineering project is constructed. This digital twin is dynamically updated as construction progresses, maintaining geometric and state consistency with the physical entity.

[0029] The intelligent analysis and computing layer communicates with the digital twin base layer and has a built-in 3D point cloud processing engine. This layer receives 3D point cloud data, spatially registers the point cloud with the design information model, calculates the set of normal distances from the actual structural surface to the design model surface, and generates 3D deviation distribution information.

[0030] The auxiliary business application layer connects the intelligent analysis and computing layer and the digital twin base layer, including a deviation visualization module and a deviation exceeding the tolerance alarm module. These modules overlay and render the 3D deviation distribution information onto the digital twin for visualization, and generate alarm information when the normal distance exceeds a preset tolerance threshold. This embodiment achieves a complete closed loop from data acquisition, transmission, modeling, analysis to visualization and alarm through the above five-layer architecture.

[0031] Example 2: Detailed Configuration of the Edge Sensing Layer

[0032] Building upon Example 1, the ground-based autonomous mobile data acquisition equipment of the edge perception layer includes a swarm of quadruped robots. Each quadruped robot carries a high-precision multi-line lidar, a high-definition panoramic gimbal camera, an infrared thermal imager, and a ground-penetrating radar probe. The aerial autonomous mobile data acquisition equipment includes a formation of multi-rotor drones, each equipped with an oblique photography camera and a multispectral sensor. Furthermore, the edge perception layer also includes miniature temperature and humidity sensors and stress-strain fiber optic grating sensors embedded within the concrete, as well as dust and noise monitoring stations and intelligent high-definition monitoring PTZ cameras deployed at the construction site. The quadruped robot swarm and the multi-rotor drone formation achieve integrated air-ground data acquisition through a pre-set collaborative inspection route. The quadruped robot swarm is responsible for scanning the structural surface from the ground to a height of 3 meters, while the multi-rotor drone formation is responsible for oblique photography of structures above 3 meters and large-scale sites. The data collected by both are fused into a unified overall point cloud model in the digital twin base layer through coordinate registration. This configuration enables the system to simultaneously acquire multi-dimensional data such as the external three-dimensional morphology of the structure, internal temperature and stress, and environmental parameters, providing sufficient data support for deviation analysis.

[0033] Example 3: Network Segmentation Configuration of the Network Transport Layer

[0034] Building upon Example 1, the network transport layer employs a customized slicing network based on 5G mobile communication technology, coupled with an industrial-grade passive optical network architecture. Enhanced mobile broadband network slices are allocated for uplink transmission of video streams and 3D point cloud data, ensuring a transmission rate of no less than 100Mbps. For numerous low-power, intermittently uploading sensor nodes distributed within concrete, such as miniature temperature and humidity sensors and stress-strain fiber Bragg grating sensors, massive machine-type communication network slices are allocated, supporting tens of thousands of terminals per cell, with individual sensor node transmit power below 20dBm. An on-site industrial-grade passive optical network serves as a supplement for highly reliable wired access to fixed equipment. This example addresses the network transmission bottleneck of multi-source heterogeneous data at substation civil construction sites through a differentiated slicing strategy.

[0035] Example 4: 3D Point Cloud Registration and Deviation Heatmap Generation

[0036] This embodiment details the processing flow of the intelligent analysis and computing layer. After the substation firewall is demolded, the system issues a detection task to the quadruped robot cluster. The quadruped robots move along a preset path, and a high-precision multi-line lidar collects point cloud data of the structural surface at a rate of hundreds of thousands of points per second. The raw point cloud data is transmitted back to the intelligent analysis and computing layer via the network transmission layer.

[0037] The intelligent analysis and computation layer first applies a voxel filtering algorithm to denoise and downsample the point cloud. The three-dimensional space is then divided into sections with sides of length... A voxel mesh is used, with only one representative point retained within each voxel, thus eliminating outlier noise points caused by dust or obstruction at the construction site. Then, line segments are extracted from the edges of building structural features, such as identifying geometric features like corners and column edges through normal estimation and curvature calculation.

[0038] Next, the iterative nearest point algorithm is applied to register the preprocessed point cloud with the design information model. Let the source point cloud be... The target point cloud is The iterative nearest-point algorithm iteratively solves for the rigid transformation matrix. Minimize the objective function:

[0039]

[0040] in yes after transformation The nearest point. Iterate until the transformation parameters converge.

[0041] After registration, calculate the set of normal distances from the actual structural surface to the design model surface. For each point in the point cloud... Find the nearest point on the surface of the design model. and its unit normal vector Then the signed normal distance Define when At that time, it was "positive excess tolerance". The timeframe is "negative out-of-tolerance". The system highlights the out-of-tolerance area in red or blue within the digital twin and generates a 3D deviation heatmap. The color mapping function is:

[0042]

[0043] This embodiment realizes automatic detection, quantitative calculation and intuitive visualization of deviation.

[0044] Example 5: AI Segmentation and Width Quantization of Cracks on Concrete Surface

[0045] This embodiment details the intelligent detection of cracks on concrete surfaces. A high-definition panoramic gimbal camera mounted on a quadruped robot simultaneously acquires images of the concrete surface during the detection process. The intelligent analysis and computation layer calls a pre-trained improved instance segmentation neural network model. This model is based on Mask R-CNN, and introduces a deformable convolution module into its backbone feature extraction network, allowing the sampling position of the convolution kernel to adaptively shift according to the crack direction, effectively improving the segmentation accuracy of irregular curved cracks.

[0046] The network output includes pixel-by-pixel classification masks and bounding box regression results. For the segmented crack / defect regions, the system further performs width quantization:

[0047] (1) Extract the center line of the crack skeleton. The crack region is peeled off layer by layer using a morphological thinning algorithm to obtain a curve with a width of one pixel.

[0048] (2) Calculate the pixel distance along the normal direction of the skeleton centerline. At each sampling point on the skeleton... Calculate the tangent direction at that point. normal direction Search for the left and right boundary points of the crack region along the normal direction. and Calculate the Euclidean distance between two points:

[0049]

[0050] (3) Convert to actual physical width. Utilize camera calibration parameters and the shooting distance provided by the LiDAR. Establish a mapping relationship between pixel coordinates and world coordinates. Let the physical size factor corresponding to the pixel size be... The actual width is:

[0051]

[0052] The average width of the crack is calculated by averaging multiple sampling points. If the average width exceeds the specification limit, the system automatically generates a defect work order. This embodiment achieves fully automated crack segmentation and millimeter-level precision quantization.

[0053] Example 6: Three-dimensional detection and AR overlay during the rebar binding stage

[0054] This embodiment describes the quality inspection of the reinforcing steel cage before concrete pouring. After the reinforcing steel is tied, a multi-rotor UAV formation takes off and hovers above the reinforcing steel cage. The onboard oblique photography cameras collect high-overlap orthophotos and oblique images along a planned flight path. All images are transmitted back to the intelligent analysis and calculation layer via the network transmission layer.

[0055] The intelligent analysis and computation layer employs a motion recovery structure algorithm for 3D reconstruction. This algorithm includes feature point extraction and matching, camera pose estimation, triangulation to generate sparse point clouds, and multi-view stereo matching to generate dense point clouds, ultimately obtaining a high-density 3D point cloud of a steel skeleton.

[0056] Next, cylindrical geometric features are extracted from the point cloud to identify individual rebars. A normal-based cylindrical fitting algorithm is used, employing the RANSAC method to fit the cylindrical parameters, thereby segmenting each rebar and calculating its central axis. The horizontal spacing between adjacent rebars is the projected distance between the two axes on the horizontal plane, and the row spacing is the vertical distance between the axes of the upper and lower layers of rebars. The calculated spacing values ​​are compared with the design values ​​in the design information model; deviations exceeding the allowable range are marked as unacceptable.

[0057] Meanwhile, the auxiliary business application layer is directly connected to the augmented reality smart glasses worn by on-site personnel via a 5G network. The intelligent analysis and computing layer sends the correct 3D rebar layout model from the design information model to the augmented reality glasses. The glasses utilize real-time positioning and mapping technology to identify feature points in the on-site environment, overlaying the virtual rebar model as semi-transparent colored lines onto the actual rebar skeleton in the operator's field of vision. Any deviation between the actual binding position and the virtual model is visually displayed with highlighted color differences, allowing operators to immediately identify problems such as missing rebar, misalignment, and non-compliant spacing, and provide on-site guidance for rectification. This embodiment achieves the immediate detection and rectification of rebar quality problems.

[0058] Example 7: Ground Penetrating Radar Detection and 3D Perspective Reconstruction of Concealed Engineering Projects

[0059] This embodiment describes the non-destructive testing of internal defects in hardened concrete. The system issues a testing task to a tracked microrobot. The robot, equipped with a ground-penetrating radar probe, moves at a constant speed along the concrete surface. The radar transmitting antenna emits high-frequency broadband electromagnetic waves into the structure, and the receiving antenna records the reflected echoes, forming a radar profile image.

[0060] After receiving the radar echo image, the intelligent analysis and computing layer first applies a two-dimensional Hilbert transform to extract the instantaneous envelope attributes. Let the original echo signal be... Its two-dimensional Hilbert transform is Then the instantaneous amplitude is:

[0061]

[0062] Envelope images enhance the visibility of weak signals, making it easier to identify rebar and void features.

[0063] The reinforcing bars appear as a single-branch hyperbola in radar images. The system automatically extracts the vertex coordinates of the hyperbola using a pattern recognition algorithm. Specifically, the method involves first extracting candidate hyperbola regions using gradient-based or morphological image segmentation, and then performing parabolic fitting on each region.

[0064]

[0065] in These are the coordinates of the hyperbola's vertices. The two-way travel time corresponding to each vertex is... And the known propagation speed of electromagnetic waves in concrete The thickness of the concrete cover for reinforcing bars can be determined by inverse calculation:

[0066]

[0067] For void defects, the radar echoes exhibit phase polarity reversal and abnormal amplitude amplification. The system extracts the void contour based on phase polarity detection and amplitude threshold segmentation. The extracted void contour is then reconstructed in three dimensions using translucent red voxels in a digital twin, visually displaying the void's depth, horizontal range, and estimated volume. This embodiment achieves high-precision non-destructive testing and three-dimensional visualization of the thickness of concealed engineering protective layers and internal defects.

[0068] Example 8: Closed-loop linkage control for temperature-controlled curing of large-volume concrete

[0069] This embodiment describes temperature control and automatic curing of large-volume concrete. Before pouring the large-volume foundation concrete for the main transformer, miniature temperature and humidity sensors are pre-embedded according to a spatial grid. After pouring, the sensors upload temperature data in real time.

[0070] The intelligent analysis and computing layer calculates the core temperature every 5 minutes. Surface temperature and ambient temperature And calculate the temperature difference:

[0071]

[0072] when or At the same time, the auxiliary control execution module of the auxiliary business application layer automatically issues control commands: sending an activation command to the automatic spray network controller, and simultaneously reducing the heating wire power of the intelligent temperature-controlled curing blanket pre-applied to the concrete surface. In addition, the system connects to an external weather station to obtain real-time wind speed. and relative humidity A feedforward control algorithm is used to predict the water evaporation rate. Dynamically adjust spray parameters: duration of a single spray cycle Interval period ,in This is an empirical coefficient. This embodiment achieves fully automated closed-loop control of concrete curing, effectively preventing temperature cracks.

[0073] Example 9: Personnel and Equipment Safety Auxiliary Control and Special Equipment Access Management

[0074] This embodiment describes construction safety management. An electronic fence system is constructed at the auxiliary business application layer: deep foundation pits and hoisting radius areas are designated as first-order high-risk restricted areas in the digital twin. On-site personnel wear smart safety helmets integrated with ultra-wideband positioning chips, and the positioning base station calculates the personnel's three-dimensional coordinates using a time-of-arrival algorithm. When an unauthorized person enters the restricted area, the system triggers a two-way linkage: the built-in buzzer in the safety helmet sounds and vibrates; simultaneously, the nearest intelligent high-definition monitoring PTZ camera automatically zooms to capture the face of the violator and pushes the violation record to the safety officer's mobile phone.

[0075] A biometric terminal and a controller area network (LAN) bus interception module are installed in the cab of the special equipment. Before startup, the operator undergoes facial recognition liveness detection. The system compares the operator's information with the background special operations personnel qualification database. If the verification is successful, the auxiliary business application layer sends an unlock command to the bus interception module, which closes the relay to release the engine ignition signal; if the verification fails, the relay remains open, and the vehicle cannot start. This embodiment achieves hardware-level security interception from behavior monitoring to operation authorization.

[0076] Example 10: Delivery of Digital Acceptance Documents Based on Blockchain

[0077] This embodiment describes the digital delivery during the final acceptance phase. Throughout the system's lifecycle, each time critical quality data is generated, its SHA-256 hash value is extracted and uploaded to the blockchain network. Hash value calculation: Here, 'data' represents the raw binary data stream. A blockchain transaction contains a data ID, timestamp, hash value, and the hash of the previous block, forming an immutable chain structure.

[0078] During final acceptance, the auxiliary business application layer automatically archives all data that has been blockchain-verified according to the sub-item project tree. The generated 3D digital twin acceptance body is an interactive, lightweight model, with each component linked to its corresponding quality data package and blockchain-verified transaction ID. Acceptance personnel can click on any component in the model to access its rebar spacing inspection report, concrete strength report, geometric dimension deviation heatmap, curing temperature record, etc. All displayed data is accompanied by a blockchain-verified mark and timestamp. If an attempt is made to modify historical data, and the recalculated hash value does not match the on-chain record, the system front-end immediately displays a red anti-counterfeiting alert. This embodiment provides a digital delivery archive with judicial-grade evidentiary value for substation civil engineering projects.

[0079] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart monitoring and auxiliary control management system for the quality of substation civil engineering projects, characterized in that, include: The edge perception layer, configured at the substation civil construction site, includes ground-based autonomous mobile acquisition equipment and aerial autonomous mobile acquisition equipment, used to continuously acquire three-dimensional point cloud data of the structural surface of the construction site; The network transmission layer, which is connected to the edge perception layer, is used to provide a communication link for transmitting the three-dimensional point cloud data from the construction site to the upper-level node. The digital twin base layer is connected to the network transmission layer for importing the design information model of the substation civil engineering project and receiving the three-dimensional point cloud data uploaded by the edge perception layer. The digital twin of the substation civil engineering project is constructed through virtual-real mapping. The intelligent analysis and computing layer is communicatively connected to the digital twin base layer. It has a built-in 3D point cloud processing engine, which is used to spatially register the 3D point cloud data with the design information model, calculate the set of normal distances from the actual structural surface to the surface of the design model, and generate 3D deviation distribution information. The auxiliary business application layer is communicatively connected to the intelligent analysis and calculation layer and the digital twin base layer. It includes a deviation visualization display module and a deviation overshoot alarm module, which are used to overlay and render the three-dimensional deviation distribution information onto the digital twin for visualization display, and generate alarm information when the normal distance exceeds a preset allowable error threshold.

2. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, The ground-based autonomous mobile data acquisition equipment includes a swarm of quadruped robots, each equipped with a high-precision multi-line lidar, a high-definition panoramic gimbal camera, an infrared thermal imager, and a ground-penetrating radar probe. The aerial autonomous mobile data acquisition equipment includes a formation of multi-rotor drones, each equipped with an oblique photography camera and a multispectral sensor. The edge perception layer also includes miniature temperature and humidity sensors and stress-strain fiber optic grating sensors embedded in concrete, as well as dust and noise monitoring stations and intelligent high-definition monitoring PTZ cameras deployed at the construction site. The quadruped robot swarm and the multi-rotor drone formation achieve integrated air-ground data acquisition through a pre-set collaborative inspection route. The quadruped robot swarm is responsible for scanning the structural surface from the ground to a height of 3 meters, while the multi-rotor drone formation is responsible for oblique photography of structures above 3 meters and large-scale sites. The data collected by both are fused into a unified overall point cloud model in the digital twin base layer through coordinate registration.

3. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, The network transmission layer adopts a customized slicing network based on fifth-generation mobile communication technology in conjunction with an industrial-grade passive optical network architecture. Specifically, enhanced mobile broadband network slices are allocated for uplink transmission of video streams and 3D point cloud data, while massive machine-type communication network slices are allocated for micro-sensor nodes distributed inside concrete.

4. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, The intelligent analysis and computing layer receives three-dimensional point cloud data from the high-precision multi-line lidar, applies a voxel filtering algorithm to remove outlier noise points, extracts line segments of the building structure feature edges, and applies an iterative nearest point algorithm to register the preprocessed point cloud with the design information model. After registration, the set of normal distances from the actual structural surface to the design model surface is calculated. When the normal distance exceeds the preset allowable error threshold, an out-of-tolerance warning area is marked in the digital twin and a three-dimensional deviation heat map is generated.

5. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, The intelligent analysis and calculation layer also calls an improved instance segmentation neural network model with a built-in deformable convolution module to perform pixel-by-pixel classification and bounding box regression on the concrete surface image captured by the high-definition panoramic gimbal camera to segment the defect area; for the identified crack defects, the center line of the crack skeleton is extracted, the pixel distance is calculated along the normal direction of the center line, and the actual physical width of the crack is converted by combining the camera calibration parameters and the depth information provided by the high-precision multi-line lidar.

6. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, During the rebar binding stage, the multi-rotor UAV formation acquires orthophotos and oblique images of the rebar skeleton. The intelligent analysis and calculation layer uses the motion recovery structure algorithm to reconstruct the three-dimensional point cloud of the rebar skeleton, extracts the cylindrical geometric features to identify individual rebars, and calculates the actual horizontal spacing and row spacing. At the same time, the auxiliary business application layer is directly connected to the augmented reality smart glasses worn by on-site personnel, and overlays and renders the correct three-dimensional rebar layout model in the design information model onto the real physical space.

7. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, The intelligent analysis and computing layer receives radar profile data echo images acquired by the ground-penetrating radar probe carried by the tracked microrobot, applies two-dimensional Hilbert transform to extract the instantaneous envelope attributes of the radar signal, automatically extracts the vertex coordinates of the hyperbola using a pattern recognition algorithm, and combines time delay parameters and electromagnetic wave propagation speed to inversely calculate the thickness of the steel reinforcement protective layer; and extracts the cavity contour based on phase polarity reversal and amplitude amplification features, and performs three-dimensional perspective reconstruction in the digital twin.

8. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, The auxiliary business application layer also includes an auxiliary control execution module. This module communicates in real time with a miniature temperature and humidity sensor embedded in the concrete to calculate the internal and external temperature gradient. When the temperature difference approaches the set critical safety value, the auxiliary business application layer automatically sends an activation command to the automatic spray network controller and reduces the heating wire power of the intelligent temperature-controlled curing blanket. At the same time, based on the wind speed and humidity data from the external weather station, the spraying duration and interval of the sprinkler system are dynamically adjusted using a feedforward control algorithm.

9. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, The auxiliary business application layer constructs a dynamic electronic fence system for personnel and equipment trajectories. It determines boundary crossing behavior and triggers linkage alarms and monitoring capture by calculating the three-dimensional coordinates of the smart safety helmet worn by personnel with built-in ultra-wideband positioning chips. In terms of special equipment access control, the auxiliary business application layer uses a biometric terminal installed in the special equipment cab and a controller local area network bus interception module to perform facial feature comparison and operator qualification verification. After the verification is passed, the bus interception module releases the engine ignition control signal.

10. The intelligent detection and auxiliary control management system for substation civil engineering quality according to claim 1, characterized in that, The intelligent inspection and auxiliary control management system for substation civil engineering quality adopts distributed ledger technology to extract the characteristic hash values ​​of the front-end perception data and the back-end analysis results of the entire life cycle of substation civil engineering, and upload them to an immutable blockchain node network. During the project completion and acceptance stage, the auxiliary business application layer automatically sorts out the corresponding data according to the hierarchical structure tree of the civil engineering components, and generates a three-dimensional digital twin acceptance body with embedded timestamps and spatial coordinates for digital acceptance file delivery.

Images