Power construction acceptance method and system based on image processing technology

Abstract

The invention relates to the technical field of electric power engineering, in particular to an electric power construction acceptance method and system based on an image processing technology, and the method comprises the steps: receiving an electric power construction drawing, recognizing an electric power equipment mark and an association mark through a deep neural network, and generating a design parameter set; a multi-view depth image of a construction site is collected, point cloud data and topology rules are fused to construct a live-action three-dimensional model, and an actual measurement parameter set is extracted; carrying out space coordinate system alignment on the design space coordinates and the actually measured space coordinates, calculating size and angle deviations and forming a comprehensive deviation data set; and performing compliance judgment according to a safety threshold value of the power equipment, marking a standard exceeding position, generating a deviation thermodynamic diagram, generating a dynamic rectification scheme in combination with a historical acceptance database and construction environment information, and finally outputting a power acceptance report containing the thermodynamic diagram and rectification suggestions. According to the invention, automatic identification and intelligent rectification of the power construction deviation can be realized, and the acceptance efficiency and the quality traceability are improved.

Description

Technical Field

[0001] This invention relates to the field of power engineering technology, and in particular to a power construction acceptance method and system based on image processing technology. Background Technology

[0002] In the construction of power engineering projects, the installation accuracy of power equipment has a critical impact on the stability and safety of system operation. Especially in scenarios such as high-voltage and ultra-high-voltage transmission and transformation stations, switching stations, and distribution stations, the spatial layout, installation angles, structural connections, and geometric dimensions of power equipment must strictly comply with the requirements of construction drawings and national standards. With the rise of intelligent construction and digital acceptance methods, utilizing drawing analysis, 3D modeling, and digital acceptance technologies for closed-loop control of construction accuracy has become an important trend in promoting intelligent acceptance and quality traceability management of power engineering projects.

[0003] However, current power equipment construction and acceptance mainly rely on manual measurement and on-site experience, which presents several technical bottlenecks. On the one hand, the traditional acceptance process relies on surveyors manually reading drawings and using tape measures or laser rangefinders to measure the size and position of each piece of equipment on-site. This is inefficient, susceptible to human error, and makes it difficult to accurately obtain key parameters such as equipment attitude and angle. On the other hand, while some existing acceptance methods using BIM-assisted modeling can improve the level of digitalization to a certain extent, they generally lack automatic alignment algorithms based on point cloud registration and intelligent drawing parsing. This makes it impossible to accurately identify dimensional and angular deviations, and there is a lack of visualized deviation location and a historical case-driven rectification decision-making mechanism. Summary of the Invention

[0004] This invention provides a power construction acceptance method and system based on image processing technology, and provides an integrated acceptance technology solution that integrates depth drawing recognition, three-dimensional reconstruction, spatial alignment, deviation analysis and dynamic rectification generation, so as to achieve efficient, accurate and intelligent control of power equipment construction quality.

[0005] A power construction acceptance method based on image processing technology includes the following steps: S1: Receives the input power construction drawings, identifies the power equipment markings and associated annotations in the drawings through a deep neural network, and outputs a set of design parameters; S2: Collect multi-view depth images of the construction site, integrate point cloud data and equipment topology to construct a real-scene 3D model, extract the measured size data, measured spatial coordinates and measured angle vectors of the power equipment in the real-scene 3D model, and generate a set of measured parameters. S3: Align the design spatial coordinates with the measured spatial coordinates in a spatial coordinate system, and compare the deviations based on the aligned coordinate system to obtain a comprehensive deviation dataset; S4: Call the power equipment safety threshold library to determine the compliance of the comprehensive deviation dataset, mark the location of the deviation exceeding the standard in the real scene 3D model, optimize the rectification suggestions by combining historical acceptance data, and output a power acceptance report including a deviation heat map and dynamic rectification plan.

[0006] Optionally, the set of design parameters includes the design dimensions of the equipment, design spatial coordinates, and design angle vectors.

[0007] Optionally, the deviation comparison in S3 includes: Compare the design dimension data with the measured dimension data to generate a dimension deviation map; By comparing the designed angle vector with the measured angle vector, an angle offset matrix is ​​generated. The dimensional deviation map and the angular offset matrix are combined to form a comprehensive deviation dataset.

[0008] Optionally, the received power construction drawings are used to identify power equipment markings and associated annotations in the drawings via a deep neural network, including: S11: Receive the input power construction drawings, perform preprocessing operations on the power construction drawings, and generate standardized power construction drawings; S12: Input the standardized power construction drawings into a deep neural network, and use the deep neural network to identify the power equipment markings in the standardized power construction drawings to generate power equipment marking recognition results; S13: Based on the power equipment mark recognition results, locate the associated annotations in the standardized power construction drawings that are associated with each power equipment mark, and extract the text information of the associated annotations; S14: Construct a set of design parameters based on the power equipment marking and identification results and the associated annotation text information.

[0009] Optionally, the step of constructing a set of design parameters based on the power equipment tag identification results and associated annotation text information includes: S141: Parse the size parameters in the text information to generate the design size data of the device; S142: Calculate the geometric center coordinates of the power equipment marked on the standardized power construction drawings and map them to design space coordinates; S143: Identify the direction identifier of the power equipment markings and convert it into a design angle vector; S144: Aggregate the design dimension data, design spatial coordinates, and design angle vector of the device to construct the design parameter set.

[0010] Optionally, the acquisition of multi-view depth images of the construction site, and the fusion of point cloud data with equipment topology to construct a realistic 3D model, includes: S21: Collect multi-view depth images of the construction site, perform distortion removal and spatiotemporal synchronization processing on the multi-view depth images, and generate synchronized multi-view depth images; S22: Convert the synchronous multi-view depth image into point cloud data, and fuse the point cloud data using a point cloud registration algorithm to generate fused point cloud data; S23: Based on the fused point cloud data and combined with the preset power equipment topology relationship rule library, construct a real-scene 3D model including equipment connection relationships.

[0011] Optionally, the step of extracting the measured dimensions, measured spatial coordinates, and measured angle vectors of the power equipment in the real-world 3D model to generate a set of measured parameters specifically involves: S24: Extract the geometric parameters of the power equipment from the real-world 3D model and generate measured dimension data; S25: Perform calculations and detections in the unified coordinate system of the real-world 3D model to generate measured spatial coordinates and measured angle vectors. The calculations and detections are specifically as follows: Calculate the spatial position of the centroid of each power device and generate the measured spatial coordinates; The main axis direction of each power device is detected and converted into a measured angle vector; S26: Aggregate the measured size data, measured spatial coordinates, and measured angle vectors to construct the measured parameter set.

[0012] Optionally, S3 includes: S31: Input the design space coordinates and the measured space coordinates into the space coordinate system alignment module, calculate the coordinate transformation matrix through the best fitting transformation algorithm, and generate the aligned design space coordinates and the aligned measured space coordinates. S32: Perform parallel deviation calculation based on the aligned design space coordinates and the aligned measured space coordinates, specifically: The design dimension data and the measured dimension data of each device are compared one by one, the absolute difference is calculated and mapped to the topological position of the device, and a dimension deviation map is generated. Compare the design angle vector with the measured angle vector for each device, calculate the angle offset, and arrange them according to the spatial distribution of the devices to generate an angle offset matrix; S33: Merge the size deviation map and the angle offset matrix according to the device identification code to construct the comprehensive deviation dataset.

[0013] Optionally, S4 includes: S41: Input the comprehensive deviation dataset into the power equipment safety threshold library, compare the deviation value with the safety threshold for each device, and generate a compliance judgment result and a list of devices with excessive deviations; S42: Highlight the locations of the devices in the list of devices with excessive deviations in the real-world 3D model, and generate a deviation heatmap by binding the deviation values; S43: Call the historical acceptance database, match similar cases according to the equipment type and deviation value in the list of equipment with excessive deviations, and generate a dynamic rectification plan framework; S44: Optimize the dynamic rectification scheme framework to generate the final dynamic rectification scheme. The optimization of the dynamic rectification scheme framework is as follows: The construction path is corrected based on the equipment topology in the real-world 3D model. Adjust the tool list based on the current construction site environmental parameters; S45: Combine the deviation heatmap and the final dynamic rectification plan to generate a power acceptance report.

[0014] A power construction acceptance system based on image processing technology, used to implement the aforementioned power construction acceptance method based on image processing technology, includes the following modules: Drawing recognition module: Used to receive input power construction drawings, and to identify power equipment markings and their associated annotations in the power construction drawings through a deep neural network, extract design dimension data, design spatial coordinates and design angle vectors, and generate a set of design parameters; Real-world modeling module: used to collect multi-view depth images of the construction site, integrate point cloud data with the topology of power equipment, construct a real-world 3D model and extract measured size data, measured spatial coordinates and measured angle vectors, and generate a set of measured parameters; Spatial coordinate system alignment module: used to perform best-fit transformation on the design spatial coordinates and the measured spatial coordinates, and output the aligned parameter set; Deviation analysis module: used to calculate the size deviation and angular offset of the device based on the aligned parameter set, and generate size deviation map and angular offset matrix to construct a comprehensive deviation dataset; Compliance assessment and rectification module: This module compares the comprehensive deviation dataset with the power equipment safety threshold library, outputs compliance assessment results, deviation heatmaps and dynamic rectification plans, and generates a power acceptance report.

[0015] The beneficial effects of this invention are: This invention automatically identifies equipment markings, dimensional information, and spatial coordinates in power construction drawings using a deep neural network, forming a standardized set of design parameters. Combined with the real-scene modeling process in S2, which integrates multi-view depth images and point cloud registration, a high-fidelity on-site 3D model is constructed, and the set of measured parameters is extracted. In S3, rigid transformation coordinate alignment and parallel deviation calculation effectively identify subtle errors in dimensions and angles, forming a refined comprehensive deviation dataset. Compared to traditional methods of manual measurement and comparison, the recognition accuracy is significantly improved.

[0016] This invention introduces a safety threshold database for power equipment to automatically determine compliance with deviations. It highlights the locations of equipment exceeding standards in a 3D model and uses a deviation heatmap to visually present the degree and distribution of deviations, allowing technicians to easily identify key rectification targets. Simultaneously, it matches similar deviation cases with a historical acceptance database and dynamically optimizes rectification paths and tool lists based on equipment topology and site environment. This results in targeted and rapid-response dynamic rectification plans, avoiding traditional "experience-based" judgments and repetitive work, and improving the efficiency and feasibility of on-site construction decisions.

[0017] This invention encapsulates all process data in a structured manner during the acceptance process, ultimately outputting a power acceptance report that includes deviation heatmaps, compliance analysis, and dynamic rectification suggestions. The report supports PDF export, system integration, and platform archiving. It can serve as third-party acceptance documentation, construction handover documents, and a basis for subsequent maintenance, effectively avoiding engineering management challenges such as "fragmented acceptance data," "vague problem identification," and "lack of rectification guidelines." This helps construction units, contractors, and operation and maintenance units achieve high-quality collaborative delivery and clear accountability in closed-loop engineering management. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the system flow according to an embodiment of the present invention. Detailed Implementation

[0020] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0021] It should be noted that the use of terms such as "an embodiment," "an embodiment," "an exemplary embodiment," and "some embodiments" in the specification indicates that the described embodiment may include a specific feature, structure, or characteristic, but not every embodiment necessarily includes that specific feature, structure, or characteristic. Furthermore, when a specific feature, structure, or characteristic is described in connection with an embodiment, implementing such a feature, structure, or characteristic in conjunction with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the art.

[0022] Generally, terms can be understood at least partly from their use in context. For example, depending at least partly on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in a singular sense, or a combination of features, structures, or characteristics in a plural sense. Additionally, the term "based on" can be understood not necessarily to convey an exclusive set of factors, but rather, alternatively, depending at least partly on the context, to allow for the presence of other factors that are not necessarily explicitly described.

[0023] like Figure 1 As shown, a power construction acceptance method based on image processing technology includes the following steps: S1: Receives the input power construction drawings, identifies the power equipment markings and associated annotations in the drawings through a deep neural network, and outputs a set of design parameters; S2: Collect multi-view depth images of the construction site, integrate point cloud data and equipment topology to construct a real-scene 3D model, extract the measured size data, measured spatial coordinates and measured angle vectors of the power equipment in the real-scene 3D model, and generate a set of measured parameters. S3: Align the design spatial coordinates with the measured spatial coordinates in a spatial coordinate system, and compare the deviations based on the aligned coordinate system to obtain a comprehensive deviation dataset; S4: Call the power equipment safety threshold library to determine the compliance of the comprehensive deviation dataset, mark the location of the deviation exceeding the standard in the real scene 3D model, optimize the rectification suggestions by combining historical acceptance data, and output a power acceptance report including a deviation heat map and dynamic rectification plan.

[0024] S1 includes: S11: Receive the input power construction drawings, perform preprocessing operations on the power construction drawings, and generate standardized power construction drawings, specifically: Power construction drawings originate from different design units, resulting in inconsistencies in drawing size, resolution, format (e.g., JPG, PNG, PDF), and line style. To ensure the accuracy of subsequent neural network recognition, standardized preprocessing of the drawings is necessary, including: Grayscale and binarization of drawings; Reset the drawing resolution to a uniform DPI (e.g., 300 DPI). Remove drawing borders and drawing frame information; Rotate and correct to align the main direction of the drawing with the coordinate system; Size normalization processing ensures that different image sizes are uniformly mapped to a set scale (e.g., 1 unit pixel = 0.1m). Standardized power construction drawings are output as input for subsequent identification.

[0025] S12: Input standardized power construction drawings into a deep neural network, and use the deep neural network to identify power equipment markings in the standardized power construction drawings, generating power equipment marking recognition results, specifically: A deep neural network is used to accurately identify electrical equipment markings in standardized power construction drawings, outputting the marking results including bounding boxes, category labels, and confidence scores. This neural network adopts the YOLOv5 object detection architecture, possessing end-to-end drawing object recognition capabilities, balancing high accuracy with fast inference efficiency. The specific network structure and recognition process are as follows: Network structure composition: Deep neural networks mainly consist of the following four modules: Input module: Receives standardized drawing images with a size of 640×640 and performs normalization processing (pixel values ​​divided by 255).

[0026] Backbone network: CSPDarknet53 is used as the backbone network structure to extract multi-level features of the drawing image and output feature maps of different scales.

[0027] This includes cross-stage residual structures (CSP) to enhance semantic feature representation; The multi-scale feature pyramid (FPN) mechanism is used to preserve the details of small equipment markings in drawings.

[0028] Neck Network: Introducing the Path Aggregation Network (PANet) structure to fuse features from different levels, improving the representation of device edges and morphology; Downsampling paths enhance global semantics; Upsampling paths enhance local details.

[0029] Output header: Outputs three predicted feature maps at different resolutions (corresponding to small, medium, and large-sized devices, respectively), with each location generating: : Coordinates of the center point of the bounding box, width and height; Classification probability (e.g., circuit breakers, instrument transformers, transformers, etc.); Confidence score.

[0030] Model training and inference process: Training dataset construction: A large number of labeled power construction drawing images were used as the training set. The bounding boxes and equipment types of each power device were manually labeled to construct the labeled dataset. The format is as follows: ; This represents a labeled dataset.

[0031] Training process: The loss function is a multi-task loss, including bounding box regression loss, category classification loss, and confidence loss. The optimizer uses SGD or Adam, and the learning rate uses a Cosine Annealing strategy; During training, the network automatically enhances the drawing images (e.g., by rotation, affine transformation, blurring) to improve its robustness to different drawing formats.

[0032] Reasoning process: Input the standardized power construction drawings into the network and output a set of candidate power equipment bounding boxes; Perform non-maximum suppression (NMS) on the output to remove duplicate or overlapping predictions; The final result set of power equipment identification results is obtained. , is represented as: ; in, Indicates the first Confidence score for each device tag.

[0033] S13: Based on the power equipment marker recognition results, locate the associated annotations in the standardized power construction drawings that are associated with each power equipment marker, and extract the text information of the associated annotations, specifically: Equipment markings are usually accompanied by text annotations (such as specifications, model, angle, etc.) on their left, top, or upper right. Using the Tesseract OCR engine's text recognition method, text boxes are extracted from a search area defined near the equipment marking's bounding box. This is then combined with rule matching and natural language keyword templates (such as "model," "mm," etc.) to complete the associated annotation extraction.

[0034] For example, if the text "Model: ACB-2000 Width 800mm Height 1000mm Angle 45°" appears in the upper right corner of a circuit breaker label, this information will be parsed into size data and angle information.

[0035] S141: Parse the dimensional parameters in the text information to generate the design dimensional data of the device, specifically: The text content recognized by OCR is structured and parsed using regular expressions, such as extracting keywords like "width," "height," and "diameter," and unifying the units (mm → m) to generate equipment design dimensions. ,For example: ; S142: Calculate the geometric center coordinates of the electrical equipment markings on standardized electrical construction drawings and map them to design space coordinates. (Using the drawing scale...) Based on (unit: meters / pixel), center the bounding box Mapping to two-dimensional coordinates in the design coordinate system , is represented as: ; The scale of the drawing is as follows: Given m / px, and the center coordinates of a certain device are (500, 300), then the mapped design space coordinates are: ; S143: Identify the directional markings on electrical equipment and convert them into design angle vectors. Directional markings may be presented as arrows, rotation symbols, or angle numbers. If a specific angle is specified (e.g., "angle 45°"), convert it into a two-dimensional angle vector. , is represented as: ; in ,but: ; If the drawing does not show the angle, it can be estimated by combining the orientation of the elements (such as the direction of the edge of the equipment frame).

[0036] S15: Aggregate the design dimensions, spatial coordinates, and angle measurements of the equipment to construct a set of design parameters. Finally, for each piece of electrical equipment, construct its corresponding set of design parameters. ; The combination of multiple devices forms the overall design parameter set for the system: ; Table 1 shows an example of the set of design parameters after analyzing a certain section of drawings: Table 1 S21: Acquire multi-view depth images of the construction site, perform distortion correction and spatiotemporal synchronization processing on the multi-view depth images, and generate synchronized multi-view depth images. This is done by deploying structured light or TOF depth camera arrays (such as Intel RealSense D455, Azure Kinect, etc.) to simultaneously capture images of the construction site from multiple fixed angles, obtaining a sequence of original multi-view depth images. ,in Indicates the number of cameras.

[0037] To improve image quality and multi-view registration accuracy, the following steps need to be performed on the depth map of each frame: Distortion correction: based on intrinsic parameter matrix and distortion coefficient vector Perform radial and tangential distortion correction; Spatiotemporal alignment: utilizing a unified timestamp Inter-frame alignment is performed on all depth image streams to ensure that all depth frames are sampled within the same time window.

[0038] The output is a set of synchronously corrected multi-view depth images, denoted as: ; S22: Convert the synchronized multi-view depth images into point cloud data, fuse the point cloud data using a point cloud registration algorithm, and generate fused point cloud data, specifically: First, each depth map Projecting the image into a 3D point cloud based on camera intrinsic and extrinsic parameters. The conversion formula is as follows: ; in For pixel coordinates, For depth value, This is the inverse matrix of the camera's intrinsic parameters.

[0039] Then, the Iterative Closest Point (ICP) algorithm is used to analyze all point cloud sets. Align them to the same coordinate system to obtain fused point cloud data. , is represented as: ; in Indicates the first The rigid transformation matrix for point cloud registration with a camera includes rotation and translation parameters.

[0040] S23: Based on fused point cloud data and combined with a pre-defined power equipment topology relationship rule base, construct a real-world 3D model including equipment connection relationships, including: Power equipment typically has standardized and relatively fixed spatial connection patterns, such as transformer-switch cabinet and busbar-circuit breaker connection methods.

[0041] In this invention, the preset power equipment topology relationship rule library is stored in the form of a graph structure, where nodes represent equipment categories and edges represent typical connection relationships, as shown in the following example: Combining the spatial distribution of point clouds, an algorithm for constructing spatial adjacency graphs is used to... The dense point clusters in the image are classified, matched, and connected to construct a realistic 3D model that includes geometric and connectivity relationships. This model is denoted as... .

[0042] S24: Extract the geometric parameters of the power equipment from the real-world 3D model and generate measured dimensional data, including: For each cluster of power devices in the model, perform the following processing steps: Boundary fitting: The minimum bounding rectangle or bounding box algorithm is used to extract the boundary of the point cluster; Dimension calculation: Calculate the projected range in the length, width, and height directions respectively, and generate the measured dimension data. .

[0043] As shown in the diagram (illustrative): S25: Calculate the spatial position of the centroid of each power device and generate the measured spatial coordinates, including: For each device cluster Its centroid coordinates The calculation formula is: ; The obtained three-dimensional coordinates represent the position of the device in the unified coordinate system of the real-world three-dimensional model.

[0044] The principal axis direction of each power device is detected and converted into a measured angle vector, including: Principal Component Analysis (PCA) algorithm is used to extract the principal axis direction vectors of point clusters. That is, the eigenvector corresponding to the largest eigenvalue of the point cluster covariance matrix is ​​taken as the principal direction: ; After normalizing the vector, it is projected onto a two-dimensional plane as the measured angle vector: ; S26: Aggregate measured dimensional data, measured spatial coordinates, and measured angle vectors to construct a set of measured parameters. Finally, construct the set of measured parameters for each device. ; The collection of multiple devices constitutes the overall set of measured parameters for the system: ; The generated set of measured parameters is shown in the table below: S3 includes: S31: Input the design spatial coordinates and measured spatial coordinates into the spatial coordinate system alignment module, calculate the coordinate transformation matrix through the best-fit transformation algorithm, and generate the aligned design spatial coordinates and aligned measured spatial coordinates. Specifically: Because there are rigid rotation and translation differences between the point cloud coordinate system at the construction site and the design coordinate system in the construction drawings, the best-fit transformation algorithm needs to be executed through the spatial coordinate system alignment module to complete the coordinate alignment.

[0045] The rigid transformation estimation method using the Kabsch algorithm has the following input: Design spatial coordinate set ; Measured spatial coordinate set ; Its goal is to solve for the optimal rotation matrix. With translation vector To minimize the mean square error between the transformed design coordinates and the measured coordinates: ; The aligned coordinates are output as follows: (Already used as the target coordinate system); The alignment results are used to ensure spatial consistency in subsequent dimensional and angular deviation calculations.

[0046] S32: Perform parallel deviation calculations based on the aligned design space coordinates and the aligned measured space coordinates, including parallel comparison of the design parameters and measured parameters for each power device, and calculation of its dimensional deviation map and angular offset matrix, specifically: 1. Dimensional Deviation Chart Calculation: For each equipment number Extract its design dimension data , is represented as: long ,Width ,high , and its measured dimensions long ,high Calculate the absolute difference in each dimension direction: ; Mapping the above differences to the topological spatial location of the device (visualized in a 3D model) generates a dimensional deviation map. Its structure is as follows: 2. Angle offset matrix calculation: Extract the design angle vector for each device. With the measured angle vector ; The offset angle is calculated using the vector angle formula. , is represented as: ; Combine all device numbers with their angular offsets and spatial locations to construct an angular offset matrix. The structure is as follows: S33: Merge the dimensional deviation map and the angular offset matrix according to the equipment identification code to construct a comprehensive deviation dataset, specifically: Using the equipment identification code as a unique primary key, the dimensional deviation and angular offset data are combined at the field level to generate a unified comprehensive deviation dataset.

[0047] S4 includes: S41: Input the comprehensive deviation dataset into the power equipment safety threshold library, compare the deviation value with the safety threshold for each device, and generate compliance judgment results and a list of devices with excessive deviations.

[0048] This invention pre-defines a safety threshold library for power equipment. This library is based on national standards, industry specifications, and equipment manuals, and sets allowable deviation ranges for dimensions and angles for different equipment categories, as shown in Table 2. Table 2: Examples of Safety Deviation Thresholds for Power Equipment For each device number in the comprehensive deviation dataset Extract its deviation term: Size deviation vector ; Angle deviation Compare the upper limit value found in the security threshold database for the corresponding device type. Determine if any dimension exceeds the limit: ; Output a list of compliance assessment results for the devices, and a set of all devices with out-of-limit deviations, denoted as the list of devices with out-of-limit deviations. .

[0049] S42: Highlight the locations of devices in the list of devices exceeding the deviation limit in the realistic 3D model, bind the deviation values ​​to generate a deviation heatmap, specifically: In real-world 3D model Mid-positioning Spatial coordinates of each device And bind it to the corresponding deviation value vector. .

[0050] Using a visual rendering method: The deviation level is represented by a color gradient (blue→red), with a larger deviation resulting in a darker color. Reliability is indicated by transparency; Label the specific deviation values ​​with text labels on the equipment model.

[0051] Generate a deviation heatmap The deviation distribution is color-coded using spatial grid units, making it easier for on-site technicians to intuitively identify key problem areas.

[0052] S43: Retrieve the historical acceptance database, match similar cases based on the equipment type and deviation value in the list of equipment exceeding the standard deviation, and generate a dynamic rectification plan framework. Specifically: This system uses its built-in historical acceptance database, which includes: Equipment category, deviation type, deviation value; Corresponding corrective measures (such as reinstallation, adjusting bolts, and correcting rotation angles); Rectification time, required tools and construction sequence.

[0053] Using a similarity matching strategy, search the database for devices that match the current device. Historical cases that meet the following conditions: Same or similar equipment types; Similar size or angle deviations (threshold set to) : The similarity function is defined as follows: ; in As a historical case, This is for adjusting the coefficient.

[0054] Before output The rectification strategies of several similar cases are integrated to form a dynamic rectification plan framework, including a draft rectification strategy, standard process numbers, and suggestions for optional tool configurations.

[0055] S44: Optimize the framework of the dynamic rectification plan and generate the final dynamic rectification plan, specifically as follows: By combining the results of the real-scene modeling with the current on-site environment, the framework of the rectification plan is customized to output a final, feasible, and dynamic rectification plan, which includes: Topology path correction: Based on the topology diagram of the equipment rooms in the realistic 3D model, the construction path is replanned. For example, crossing adjacent high-voltage conductors is avoided, auxiliary support points are set up, and the work sequence is adjusted to reduce repeated disassembly and assembly. The optimized logic diagram is as follows: Original path: D01 → D02 → D03 (conflict); After optimization: D02 → D03 → D01 (avoidance); Tool list adjustment: Based on site environmental parameters (such as temperature, humidity, and space accessibility) and site working condition data (collected from BIM or smart sensor nodes), select appropriate operating tools and methods, for example: Confined space → miniature socket tool; High-temperature environment → Insulating and cooling gloves; High humidity → Prioritize electrical drying operations; The final output is a structured dynamic rectification plan document, including equipment number, rectification method, required tools, work steps sequence, and risk warnings.

[0056] S45: Aggregate the deviation heatmap and final dynamic rectification plan, and generate a power acceptance report, specifically: By combining the deviation heat map with the final dynamic rectification plan according to the equipment number, a standardized power acceptance report is generated. The report can be exported as PDF, BIM attachment or system audit data, and connected to the quality inspection platform and the construction unit's collaborative platform to form a closed-loop traceability record.

[0057] like Figure 2 As shown, a power construction acceptance system based on image processing technology is used to implement the aforementioned power construction acceptance method based on image processing technology, and includes the following modules: Drawing recognition module: Used to receive input power construction drawings, and to identify power equipment markings and their associated annotations in the power construction drawings through a deep neural network, extract design dimension data, design spatial coordinates and design angle vectors, and generate a set of design parameters; Real-world modeling module: used to collect multi-view depth images of the construction site, integrate point cloud data with the topology of power equipment, construct a real-world 3D model and extract measured size data, measured spatial coordinates and measured angle vectors, and generate a set of measured parameters; Spatial coordinate system alignment module: used to perform best-fit transformation on the design spatial coordinates and the measured spatial coordinates, and output the aligned parameter set; Deviation analysis module: used to calculate the size deviation and angular offset of the device based on the aligned parameter set, and generate size deviation map and angular offset matrix to construct a comprehensive deviation dataset; Compliance assessment and rectification module: This module compares the comprehensive deviation dataset with the power equipment safety threshold library, outputs compliance assessment results, deviation heatmaps and dynamic rectification plans, and generates a power acceptance report.

[0058] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0059] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for power construction acceptance based on image processing technology, characterized in that, The method comprises the following steps: S1: receiving an input power construction drawing, identifying power equipment marks and associated annotations in the drawing through a deep neural network, and outputting a design parameter set; S2: collecting multi-view depth images of the construction site, fusing point cloud data and equipment topology relationship to construct a real scene three-dimensional model, extracting measured size data, measured spatial coordinates and measured angle vectors of the power equipment in the real scene three-dimensional model, and generating a measured parameter set; S3: aligning the design spatial coordinates with the measured spatial coordinates in the spatial coordinate system, comparing the deviations based on the aligned coordinate system, and obtaining a comprehensive deviation data set; S4: calling a power equipment safety threshold library to determine the compliance of the comprehensive deviation data set, marking the deviation positions that exceed the standard in the real scene three-dimensional model, optimizing rectification suggestions combined with historical acceptance data, and outputting a power acceptance report including a deviation heat map and a dynamic rectification scheme.

2. The power construction acceptance method based on image processing technology according to claim 1, characterized in that, The design parameter set includes design size data, design spatial coordinates and design angle vectors of the equipment.

3. The power construction acceptance method based on image processing technology according to claim 2, characterized in that, The deviation comparison in S3 includes: comparing the design size data with the measured size data to generate a size deviation map; comparing the design angle vectors with the measured angle vectors to generate an angle deviation matrix; combining the size deviation map and the angle deviation matrix to form a comprehensive deviation data set.

4. The power construction acceptance method based on image processing technology according to claim 3, characterized in that, The method of receiving an input power construction drawing and identifying power equipment marks and associated annotations in the drawing through a deep neural network comprises: S11: receiving an input power construction drawing, performing a preprocessing operation on the power construction drawing to generate a standardized power construction drawing; S12: inputting the standardized power construction drawing into a deep neural network, identifying power equipment marks in the standardized power construction drawing through the deep neural network, and generating a power equipment mark identification result; S13: based on the power equipment mark identification result, locating the associated annotations in the standardized power construction drawing associated with each power equipment mark, and extracting the text information of the associated annotations; S14: constructing a design parameter set according to the power equipment mark identification result and the text information of the associated annotations.

5. The method of claim 4, wherein the method is based on image processing techniques. The method of constructing a design parameter set according to the power equipment mark identification result and the text information of the associated annotations comprises: S141: analyzing the size parameters in the text information to generate design size data of the equipment; S142: calculating the geometric center coordinates of the power equipment mark in the standardized power construction drawing and mapping them as design spatial coordinates; S143: identifying the direction identifier of the power equipment mark and converting it into a design angle vector; S144: aggregating the design size data, design spatial coordinates and design angle vectors of the equipment to construct the design parameter set.

6. The power construction acceptance method based on image processing technology according to claim 5, characterized in that, The method of collecting multi-view depth images of the construction site, fusing point cloud data and equipment topology relationship to construct a real scene three-dimensional model comprises: S21: collecting multi-view depth images of the construction site, performing distortion removal and space-time synchronization processing on the multi-view depth images to generate synchronized multi-view depth images; S22: converting the synchronized multi-view depth images into point cloud data, fusing the point cloud data through a point cloud registration algorithm, and generating fused point cloud data; S23: Based on the fusion point cloud data, a real scene three-dimensional model including device connection relationship is constructed in combination with a preset power equipment topology relationship rule library.

7. The method of claim 6, wherein the method is characterized by: The measured size data, measured spatial coordinates and measured angle vector of the power equipment in the real scene three-dimensional model are extracted to generate a measured parameter set, specifically: S24: Geometric parameters of the power equipment are extracted from the real scene three-dimensional model to generate measured size data; S25: Calculation and detection are performed in the unified coordinate system of the real scene three-dimensional model to generate measured spatial coordinates and measured angle vector, and the calculation and detection are specifically: The centroid spatial position of each power equipment is calculated to generate measured spatial coordinates; The main shaft direction of each power equipment is detected and converted into a measured angle vector; S26: The measured size data, measured spatial coordinates and measured angle vector are aggregated to construct the measured parameter set.

8. The power construction acceptance method based on image processing technology according to claim 7, characterized in that, The S3 includes: S31: The design spatial coordinates and the measured spatial coordinates are input into a spatial coordinate system alignment module, a coordinate conversion matrix is calculated by a best fitting transformation algorithm to generate aligned design spatial coordinates and aligned measured spatial coordinates; S32: Parallel deviation calculation is performed based on the aligned design spatial coordinates and the aligned measured spatial coordinates, specifically: The design size data of the equipment is compared with the measured size data device by device to calculate absolute difference values and map them to the equipment topology position to generate a size deviation atlas; The design angle vector is compared with the measured angle vector device by device to calculate angle offset values and arrange them according to the spatial distribution of the equipment to generate an angle offset matrix; S33: The size deviation atlas and the angle offset matrix are merged according to the equipment identification code to construct the comprehensive deviation data set.

9. The power construction acceptance method based on image processing technology according to claim 8, characterized in that, The S4 includes: S41: The comprehensive deviation data set is input into a power equipment safety threshold library, and the deviation values are compared with the safety threshold device by device to generate compliance judgment results and an out-of-tolerance deviation equipment list; S42: The equipment positions in the out-of-tolerance deviation equipment list are highlighted and labeled in the real scene three-dimensional model, and a deviation heat map is generated by binding the deviation values; S43: A historical acceptance database is called, similar cases are matched according to the equipment types and deviation values in the out-of-tolerance deviation equipment list to generate a dynamic rectification scheme framework; S44: The dynamic rectification scheme framework is optimized to generate a final dynamic rectification scheme, and the optimization of the dynamic rectification scheme framework is specifically: The construction path is corrected based on the device topology relationship in the real scene three-dimensional model; The tool list is adjusted in combination with the current construction site environment parameters; S45: The deviation heat map and the final dynamic rectification scheme are aggregated to generate a power acceptance report.

10. A method and system for power construction acceptance based on image processing technology, used to implement the method for power construction acceptance based on image processing technology according to any one of claims 1-9, characterized in that, The following modules are included: A drawing recognition module is used to receive an input power construction drawing, identify power equipment marks and their associated annotations in the power construction drawing through a deep neural network, extract design size data, design spatial coordinates and design angle vector, and generate a design parameter set; The real scene modeling module is configured to collect multi-view depth images of the construction site, fuse point cloud data and a power equipment topology relationship, construct a real scene three-dimensional model, and extract measured size data, measured spatial coordinates and measured angle vectors to generate a measured parameter set. The spatial coordinate system alignment module is configured to perform a best fitting transformation on the design spatial coordinates and the measured spatial coordinates, and output an aligned parameter set. The deviation analysis module is configured to calculate size deviations and angle deviations of the equipment based on the aligned parameter set, generate a size deviation atlas and an angle deviation matrix, and construct a comprehensive deviation data set. The compliance determination and rectification module is configured to compare the comprehensive deviation data set with a power equipment safety threshold library, output a compliance determination result, a deviation heat map and a dynamic rectification scheme, and generate a power acceptance report.

Images