Municipal communication pipeline laying construction management system

Abstract

The invention relates to the field of municipal communication pipelines, and discloses a municipal communication pipeline-based laying construction management system, which comprises the following steps of: acquiring images, position information and voice data of a construction site, preprocessing the acquired content by utilizing a lightweight edge calculation model, and judging whether data exception exists or not; on the basis of spatio-temporal information matching and an image content recognition algorithm, construction data collected on site are compared with the engineering plan model, and whether areas which are not constructed according to specifications are recognized or not is judged; uniformly converting the unstructured data into construction record items in a standard format by using a multi-modal feature fusion method, and judging whether the unstructured data is successfully converted or not; dynamically updating a construction progress state in combination with a project construction plan and real-time acquired data, and generating a visual Gantt chart display; and training the historical engineering data based on a machine learning model, and automatically pushing early warning information to a management terminal. The method has the advantage of improving the communication pipeline construction management capability.

Description

Technical Field

[0001] The present invention relates to the field of municipal communication pipelines, and specifically to a construction management system for municipal communication pipeline laying. Background Art

[0002] With the advancement of urban informatization construction, municipal communication pipelines, as important carriers for communication, power, signal and other lines, their laying construction management has been increasingly emphasized. Existing construction management systems for municipal communication pipeline laying mostly rely on the comprehensive management of GIS maps, engineering node information, progress tracking and drawing materials to achieve construction visualization, standardization and processization. However, in the prior art, there is a lag problem in the synchronization of on-site data collection and the system during the construction process. Especially during the pipeline burial process, construction workers usually record the construction situation through paper forms or unstructured pictures, and these data need to be manually sorted out and uploaded to the management system later, resulting in problems such as low data entry efficiency, high error rate and poor real-time performance, which seriously affect the construction quality traceability and the accuracy of on-site management. Therefore, it is necessary to design a construction management system for municipal communication pipeline laying that can improve the construction management of communication pipelines. Summary of the Invention

[0003] Aiming at the deficiencies of the prior art, the present invention provides a construction management system for municipal communication pipeline laying, which has the advantages of improving the construction management of communication pipelines and solves the problems in the above background art.

[0004] To achieve the above purpose of improving the construction management of municipal communication pipelines, the present invention provides the following technical solutions: A construction management system for communication pipeline laying, including: Data acquisition module: Collect images, location information and voice data at the construction site, preprocess the collected content using a lightweight edge computing model, and judge whether there is data abnormality. If so, enter the data verification module; Data verification module: Compare the construction data collected on-site with the engineering plan model based on the spatio-temporal information matching and image content recognition algorithms, and judge whether there is an area where construction is not carried out according to the specifications. If so, enter the structured conversion module; Structured conversion module: Use the multi-modal feature fusion method to uniformly convert the unstructured data into construction record items in a standard format, and judge whether the unstructured data is successfully converted. If so, enter the progress update module; Progress update module: Combine the project construction plan and the real-time collected data to dynamically update the construction progress status, and generate a visual Gantt chart for display, and judge whether construction delay is found. If so, enter the intelligent warning module; Intelligent warning module: Train the historical engineering data based on a machine learning model, automatically push warning information to the management terminal, and suggest rectification measures.

[0005] Preferably, the process of preprocessing the collected content using a lightweight edge computing model is as follows: Compress and encode the images, voices, and location information collected on-site and perform synchronous processing, and extract visual and voice features through the convolutional neural network and voice recognition model embedded in the edge device; Use the rule matching and probability judgment method to classify and compare the preliminary feature results and generate preliminary judgment labels; Use the preprocessed data with labels as the input for subsequent data verification and mark the indexes of potential abnormal data.

[0006] Preferably, the process of judging whether there is data abnormality is as follows: Extract the timestamp and GPS coordinates of the preprocessed data, and calculate the spatial and temporal deviation values from the set operation nodes in the predetermined project plan; If the deviation value exceeds the set tolerance threshold, or the image recognition result does not match the engineering operation template, it is determined that the data is abnormal.

[0007] Preferably, the process of comparing the construction data collected on-site with the engineering plan model based on the spatio-temporal information matching and image content recognition algorithm is as follows: Based on the time window and location points set by the operation plan node, perform spatial projection and chronological arrangement on the uploaded data; Combine the semantic segmentation algorithm to label the construction status of the pipeline in the image and perform semantic alignment with the standard drawings; Output the operation compliance score according to the comparison result and identify the areas with abnormally low scores.

[0008] Preferably, the process of judging whether an area where construction is not carried out according to the specification is identified is as follows: Statistically analyze the area information with comparison scores lower than the specification threshold, and perform correlation analysis in combination with the missing items of the construction content in the image; If it is detected that a key step is missing or the construction sequence is incorrect, it is determined that the construction area is not constructed according to the specification, and the construction area information is encoded as a unit to be rectified.

[0009] Preferably, the process of uniformly converting unstructured data into construction record items in a standard format using the multi-modal feature fusion method is as follows: Extract features from the images, voices, and location information respectively to generate image feature vectors, semantic text descriptions, and spatio-temporal labels; Adopt the attention mechanism to fuse the features of each modality, establish the semantic mapping relationship between the unstructured data and the engineering information template, and construct structured form items according to the mapping relationship.

[0010] Preferably, the process of judging whether the conversion of unstructured data is successful is as follows: Perform integrity and consistency checks on the keyword fields in the structured results. If a field is missing or the content is ambiguous, mark the conversion as failed; Use the conversion accuracy evaluation model to compare the conversion samples with the manually annotated results and generate an automated conversion quality score; If the score is greater than or equal to the set standard, it is determined that the conversion is successful; If the score is less than the set standard, it is determined that the conversion is unsuccessful.

[0011] Preferably, the process of dynamically updating the construction progress status and generating a visual Gantt chart for display is as follows: Parse the timestamps and operation types in the structured construction records and map them to the corresponding process nodes in the project plan; Calculate the actual progress ratio of each node and generate a comparison curve in combination with the original planned progress; Use the visualization engine to generate a Gantt chart and update it to the management terminal's large screen or mobile terminal.

[0012] Preferably, the process of determining whether construction delays are found is as follows: Perform a difference analysis on the task progress line and the actual collected data progress line in the Gantt chart to determine the offset time of each key process; If the node delay on the critical path exceeds the set tolerance days, or the cumulative delay of multiple non-critical nodes causes the critical path to be blocked, it is determined that there is a construction delay.

[0013] Preferably, the process of automatically pushing warning information to the management terminal is as follows: Based on the trained classification model, perform classification and labeling on the problems; Combine historical data with current project parameters to predict the duration of the delay and the impact on the overall project duration; According to the prediction results, generate a rectification suggestion report and synchronize the reminder to the management terminal through the message push system.

[0014] Compared with the prior art, the present invention provides a construction management system based on municipal communication pipeline laying, which has the following beneficial effects: By integrating a variety of key technologies such as image recognition, speech processing, edge computing, multi-modal feature fusion, structured transformation, and intelligent warning, the present invention realizes the closed-loop management of data collection, abnormal determination, normative comparison, structured archiving, progress dynamic update, and delay warning throughout the construction process. The system has the advantages of strong real-time performance, high processing efficiency, and high degree of automation, which can significantly improve the informatization and intelligent management level of the municipal communication pipeline laying project, reduce the errors and lags caused by manual inspections and paper records, and effectively prevent the occurrence of construction delays and illegal operations. At the same time, the visual Gantt chart and responsibility attribution analysis function generated by the system facilitate managers to comprehensively grasp the construction progress and potential risks, realize refined, visual, and intelligent project scheduling and decision-making support, and have good application and promotion value. Brief Description of the Drawings

[0015] Figure 1 It is a schematic structural diagram of the present invention; Figure 2 It is a schematic diagram of the method of the present invention. Detailed Embodiments

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment 1: Please refer to Figure 1 As shown, a construction management system based on municipal communication pipeline laying described in an embodiment of the present invention includes: Data acquisition module: Collect images, location information, and voice data at the construction site, preprocess the collected content using a lightweight edge computing model, and determine whether there is data abnormality. If so, enter the data verification module.

[0018] The process of preprocessing the collected content using the lightweight edge computing model in the data acquisition module is: Compress and encode the images, voices, and location information collected on-site and perform synchronous processing. Use the convolutional neural network and speech recognition model embedded in the edge device to extract visual and speech features respectively; Use the high-resolution cameras deployed at the construction site to take images at a set frequency, and compress the original images using the JPEG2000 or H.265 encoding algorithm, retaining the clarity of the key areas to ensure subsequent recognition accuracy; Use a pick-up microphone to collect voice data such as construction workers' conversations and instructions, perform front-end noise reduction processing using MFCC feature extraction and endpoint detection, and then compress and store it through AAC or SPEEX encoding; Combine the GNSS module to obtain the longitude, latitude, altitude, and timestamp data of the acquisition device, uniformly package it in the standard GeoJSON format, bind it with the image and voice data, and form a traceable data packet; Use the scheduling module built into the edge computing device to perform multi-source alignment processing on the image, voice, and location data according to the timestamp, ensuring the temporal consistency of each modality data and providing a basis for subsequent feature extraction; Run a lightweight CNN in the edge device to perform forward inference on the image, and extract features including object categories, position boxes, construction status, etc.; Perform scene segmentation on the image through a semantic segmentation network, identify the position relationships of the construction area, equipment, personnel, and pipelines, and output a semantic label map; Use the optimized RNN-T model at the edge to perform real-time recognition on the voice data, convert the construction conversation into structured text, and extract the key verbs and material terms therein; Unify all visual and speech features and write them into the local cache database, and align them with the acquisition timestamp as the index to form a complete feature record.

[0019] Use the rule matching and probability judgment method to classify and compare the preliminary feature results and generate preliminary judgment labels; The system has a built-in set of standard construction action templates to perform combined matching on the image and speech recognition results; If a certain rule is satisfied, the corresponding operation label is marked. If multiple rules may be matched simultaneously, use a probability model based on a Bayesian network to evaluate the confidence of the candidate labels, output the most likely operation type, and assign a confidence score; If there are conflicts between the labels, the system makes a judgment through the context judgment mechanism and only retains the logically valid labels; All generated labels are converted into the standard encoding format and are accompanied by time, location, and source data pointers; Use the preprocessed data with labels as the input for subsequent data verification and mark the indexes of potential abnormal data.

[0020] The process of judging whether there is data abnormality in the data acquisition module is as follows: Extract the timestamps and GPS coordinates of the preprocessed data, and calculate the spatial and temporal deviation values from the set operation nodes in the predetermined project plan; the system imports the operation node schedule table containing information such as operation types, predetermined time periods, and geographical locations from the project management platform; parse the acquisition timestamps and corresponding GPS location information of each record from the preprocessed data packets to ensure one-to-one correspondence with modal data such as images and voices; use the Haversine formula calculation interface to calculate the spherical distance between the acquisition location and the set location of the target operation node as the spatial deviation value; compare the difference between the acquisition time and the predetermined operation time range. If the acquisition time is earlier or later than the set operation time period, take the absolute time difference as the temporal deviation value; input the spatial deviation and temporal deviation into the operation matching engine simultaneously to evaluate whether this data record can match a specific operation node; if both deviations are less than the preset threshold, it is determined that the match is successful; If the deviation value exceeds the set tolerance threshold, or the image recognition result does not match the engineering operation template, it is determined that the data is abnormal; After the abnormal data is marked, it is transmitted by the system to the data verification module to trigger further comparison and analysis operations.

[0021] Data verification module: Based on the spatio-temporal information matching and image content recognition algorithms, compare the construction data collected on-site with the engineering plan model to determine whether areas where construction is not carried out according to specifications are identified. If so, enter the structured conversion module.

[0022] The process of comparing the construction data collected on-site with the engineering plan model based on the spatio-temporal information matching and image content recognition algorithms in the data verification module is as follows: Based on the time window and location points set by the operation plan nodes, perform spatial projection and chronological arrangement on the uploaded data; use the GIS spatial matching algorithm to map each piece of data into the corresponding construction area or node section to form a data-operation node pairing relationship; compare the timestamp of each record with the time window of the node, and only retain the data within the valid time range, and eliminate the out-of-bounds records; sort the valid data within each operation node according to the timestamp to form a continuous chronological data stream, providing a basis for subsequent state change recognition and trend analysis; Combine semantic segmentation algorithms to finely annotate the construction states such as pipeline laying, excavation, and backfilling in the image, and perform semantic alignment with the standard drawings; uniformly adjust the uploaded images to a fixed size and color gamut space, and perform enhancement processing such as denoising and histogram equalization to improve the model recognition effect; use a pre-trained or self-developed lightweight semantic segmentation network to annotate the pixel-level regions in the image, and output the mask map and semantic labels; through semantic mask analysis, quantify the area, position, and contour features of each construction state region in the image, and construct a structured state description; import the construction standard drawings or reference operation drawings corresponding to the current node, and extract their semantic annotations and spatial structure features; use image feature matching and semantic relationship graphs to perform semantic alignment between the construction state map and the standard drawings, and calculate the structural similarity and state matching degree; Output the operation compliance score according to the comparison result, and identify the abnormally low score regions for subsequent processing.

[0023] The process of judging whether an area where construction is not carried out according to the specifications is identified in the data verification module is as follows: Statistically analyze the area information with a comparison score lower than the specification threshold, and perform correlation analysis in combination with the missing items of the construction content in the image; the system screens the image records or image regions with scores lower than the set threshold from the compliance score results, extracts their corresponding spatial positions, timestamps, and operation node information, and marks them as suspected abnormal regions; perform in-depth semantic analysis on the suspected abnormal region images, extract the construction state labels, and construct a construction state vector or missing item matrix; use the built-in construction specification rule engine to compare the identified state missing items with the standard process requirements that the node should have, and judge whether there are structural omissions or step omissions; record the comparison analysis results in a structured form, including: abnormal type, list of missing labels, image and semantic mask position indexes; If a key step is missing or the construction sequence is incorrect, it is determined that the area is not constructed according to the specifications; the system sets key construction steps, such as the foundation cushion layer, sand and gravel leveling layer, pipeline laying, and protective layer backfilling, etc., each corresponding to necessary state labels and their sequence logics; perform sequence verification operations on the identified areas, including judging whether the semantic labels are complete, whether the order of appearance of the semantic labels conforms to the construction logic, and whether the spatial hierarchy between the states meets the overlay logic; once it is determined that there is a key step missing or the sequence is violated, the system marks the area as not constructed according to the specifications, and attaches the violation type code and description text.

[0024] Encode the area information into a unit to be rectified and transfer it to the structured conversion module.

[0025] Structured conversion module: Use the multi-modal feature fusion method to uniformly convert the unstructured data into construction record items in a standard format, and judge whether the unstructured data is successfully converted. If so, enter the progress update module.

[0026] In the structured conversion module, the process of uniformly converting unstructured data into construction record items in a standard format using the multimodal feature fusion method is as follows: Feature extraction is performed on the image, voice, and location information respectively to generate an image feature vector, a semantic text description, and a spatio-temporal label; a lightweight convolutional neural network is used to process the construction image, extract the key elements and their location information in the construction scene, and output a visual feature vector; the voice data collected on-site is input into a pre-trained speech recognition model, converted into readable text, and then the BERT model is called for semantic extraction to output a semantic description vector; The attention mechanism is used to fuse the features of each modality, and a semantic mapping relationship between the unstructured data and the engineering information template is established; dimensional normalization and temporal alignment processing are performed on the feature vectors of different modalities to ensure that they can interact in a unified semantic space; a multi-head attention fusion module is constructed to calculate the feature vectors of the two modalities, representing the contribution degrees of different features to the final task, and the formula for calculating the attention weight matrix between them is: ; In the formula, Q, K, and V are the Q, K, and K matrices respectively, is the feature dimension, is the modality similarity matrix; the information of each modality is weighted and combined according to the weights to generate a fused global representation vector, and the fused features are semantically compared with the standard process table in the engineering information template, and process inference is performed through nearest neighbor matching, vector similarity calculation, or a rule-based decoder; Structured form items are constructed according to the mapping relationship, including fields such as construction location, operation time, process category, and personnel information.

[0027] In the structured conversion module, the process of judging whether the unstructured data is successfully converted is as follows: Integrity and consistency detection are performed on the key fields in the structured result. If a field is missing or the content is ambiguous, the conversion is marked as failed; the core fields are extracted from the structured data template, such as construction location, operation time, process category, construction personnel, equipment number, etc.; the effective format standard and the allowable default rule for each field are defined; the structured data records are traversed to check whether there are null values, NULL, or illegal placeholders; if a missing field exists, the current record is marked as a conversion failure; the content of the field is verified to meet semantic or logical consistency, for example: whether the time field conforms to the process time sequence constraint, whether the location information falls within the construction area boundary, and whether the process category is consistent with the semantic segmentation image recognition label; keyword matching or a semantic recognition model is used to identify fuzzy descriptions and compare them with the standard template; if there is a fuzzy expression, the field is recorded as having ambiguous content and the conversion is marked as failed; Using the conversion accuracy evaluation model, compare the conversion samples with the manually annotated results to generate an automated conversion quality score; If the score is greater than or equal to the set standard, it is determined that the conversion is successful; If the score is less than the set standard, it is determined that the conversion is unsuccessful.

[0028] The technical solution of this implementation is as follows: A multi-modal feature fusion method is used to perform standardized conversion on unstructured construction data, specifically including: extracting the visual feature vector of the image, the semantic text description of the voice, and the spatio-temporal tags of the location information, using the attention mechanism to fuse the feature information of different modalities, establishing a semantic mapping relationship between the unstructured data and the engineering operation template, and then generating structured record items including fields such as construction location, operation time, process category, and personnel information, and judging whether the conversion is successful through integrity and consistency verification. It realizes the automatic parsing of complex and multi-source unstructured data and the generation of standardized forms, improves the accuracy and efficiency of data processing, and effectively reduces the burden of manual recording and manual entry; through the attention mechanism, it enhances the information synergy between modalities, ensures accurate keyword field recognition, further improves the integrity and consistency of construction records, provides a high-quality data basis for subsequent progress update and anomaly warning, and thus significantly improves the informatization management level and automated operation and maintenance ability of municipal construction projects.

[0029] Example 2: As Figure 1 shown, a construction management system based on municipal communication pipeline laying further includes the following modules: Progress update module: Combining the project construction plan and the real-time collected data, dynamically update the construction progress status, generate a visual Gantt chart for display, and judge whether construction delays are found. If so, enter the intelligent warning module.

[0030] The process of dynamically updating the construction progress status and generating a visual Gantt chart display in the progress update module is as follows: Parse the timestamps and operation types in the structured construction records, and map them to the corresponding process nodes in the engineering plan; Extract the key fields from the structured data: construction timestamp, operation type, construction location, operation personnel, etc.; According to the operation type and timestamp, map the construction records to the corresponding process nodes in the engineering plan; Use the time interval and spatial position of the process nodes for precise matching to ensure the accurate attribution of the construction records; Calculate the actual progress ratio of each node, and generate a comparison curve in combination with the original planned progress; For each process node, calculate the quantitative indicators of the completed operations according to the corresponding construction records; Calculate the actual completion ratio of the node: ; Obtain the planned completion ratio and time progress point corresponding to this node in the project plan, generate the time series comparison data of the actual progress and the planned progress, sort the actual progress ratio and the planned progress ratio by time to form a hyperbolic comparison data sequence; Use the visualization engine to generate a Gantt chart containing task progress, node status, and estimated completion time, and update it to the management terminal large screen or mobile terminal.

[0031] In the progress update module, the process of determining whether construction delay is found is as follows: Perform a difference analysis on the task progress line and the actual collected data progress line in the Gantt chart to determine the offset time of each key process; extract the planned start time, planned completion time, and actual completion time (calculated based on real-time collected data) of each key process node from the Gantt chart data structure; for each key node, calculate the difference between the actual completion time and the planned completion time: ; If is greater than or equal to 0, it means that the process node is delayed; If is less than 0, it means that the extraction is completed; If the delay of a node on a critical path exceeds the set tolerance days, or the cumulative delay of multiple non-critical nodes causes the critical path to be blocked, it is determined that there is construction delay; Associate the delay information with the responsible work team and operation content for the intelligent warning module to generate reminders.

[0032] Intelligent warning module: Train the historical project data based on the machine learning model, automatically push warning information to the management terminal, and recommend rectification measures.

[0033] In the intelligent warning module, the process of automatically pushing warning information to the management terminal is as follows: Based on the trained classification model, classify and label problems such as construction delay, abnormal construction records, and data loss; Combine historical data with current project parameters to predict the delay duration and the impact on the overall project duration; According to the prediction results, generate a rectification suggestion report, and synchronize the reminder to the management terminal through the message push system, supporting forms such as voice broadcast and mobile terminal notification.

[0034] Embodiment 3: Please refer to Figure 2 As shown, a construction management method based on municipal communication pipeline laying includes the following steps: S1: Collect the image, location information, and voice data at the construction site, preprocess the collected content using a lightweight edge computing model, and determine whether there is data abnormality. If so, enter the data verification module; S2: Compare the construction data collected on-site with the engineering plan model based on the spatio-temporal information matching and image content recognition algorithm to determine whether areas where construction is not carried out according to specifications are identified. If so, enter the structured conversion module; S3: Use the multi-modal feature fusion method to uniformly convert the unstructured data into construction record items in a standard format, and determine whether the conversion of the unstructured data is successful. If so, enter the progress update module; S4: Combine the project construction plan and the real-time collected data, dynamically update the construction progress status, and generate a visual Gantt chart for display. Determine whether construction delays are found. If so, enter the intelligent warning module; S5: Train the historical engineering data based on the machine learning model, automatically push warning information to the management terminal, and recommend rectification measures.

[0035] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0036] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A construction management system based on the laying of municipal communication pipelines, characterized in that, include: Data collection module: collects images, location information and voice data of the construction site, and uses a lightweight edge computing model to pre-process the collected content to determine whether there are data anomalies. If so, it enters the data verification module; Data verification module: Based on the spatiotemporal information matching and image content recognition algorithm, the construction data collected on site are compared with the engineering plan model to determine whether areas that are not constructed according to the specifications are identified. If so, the module enters the structured conversion module; Structured conversion module: Use multimodal feature fusion method to uniformly convert unstructured data into construction record items in a standard format, and determine whether the unstructured data is successfully converted. If so, enter the progress update module; Progress update module: Combines the project construction plan with real-time data collection to dynamically update the construction progress status and generate a visual Gantt chart to determine whether construction delays are found. If so, enter the intelligent early warning module; Intelligent early warning module: Based on the machine learning model, historical engineering data is trained to automatically push early warning information to the management terminal and recommend corrective measures.

2. The construction management system based on municipal communication pipeline laying according to claim 1, characterized in that, The process of preprocessing the collected content using the lightweight edge computing model is as follows: The images, voices and location information collected on site are compressed and encoded and processed synchronously, and the visual and voice features are extracted respectively through the convolutional neural network and voice recognition model embedded in the edge device; Using rule matching and probability judgment methods, the preliminary feature results are classified and compared, and preliminary judgment labels are generated; The labeled preprocessed data is used as input for subsequent data verification and potential abnormal data indexes are marked.

3. The construction management system based on municipal communication pipeline laying according to claim 2, characterized in that, The process of judging whether there is data anomaly is as follows: Extract the timestamp and GPS coordinates of the pre-processed data, and calculate the spatial and temporal deviation values ​​of the operation nodes set in the predetermined engineering plan; If the deviation value exceeds the set tolerance threshold, or the image recognition result does not match the engineering operation template, the data is judged to be abnormal.

4. A construction management system for municipal communication pipeline laying according to claim 3, characterized in that, The process of comparing the construction data collected on site with the engineering plan model based on spatiotemporal information matching and image content recognition algorithm is as follows: Based on the time window and location point set by the job plan node, the uploaded data is spatially projected and time-series arranged; Combined with the semantic segmentation algorithm, the construction status of the pipeline in the image is annotated and semantically aligned with the standard drawing; Output the job compliance score based on the comparison results and identify abnormally low-scoring areas.

5. The construction management system based on municipal communication pipeline laying according to claim 4, characterized in that, The process of determining whether an area that has not been constructed in accordance with the specifications is as follows: Statistically compare the area information with scores below the standard threshold, and conduct correlation analysis based on the missing items of construction content in the image; If it is detected that a key step is missing or the construction sequence is wrong, it is determined that the construction area has not been constructed according to the specifications, and the construction area information is coded as a unit to be rectified.

6. The construction management system based on municipal communication pipeline laying according to claim 5, wherein, The process of converting unstructured data into construction record items in a standard format using the multimodal feature fusion method is as follows: Perform feature extraction on images, speech and location information to generate image feature vectors, semantic text descriptions and spatiotemporal labels; The attention mechanism is used to fuse the features of each modality, establish the semantic mapping relationship between unstructured data and engineering information templates, and construct structured form items based on the mapping relationship.

7. The construction management system based on municipal communication pipeline laying according to claim 6, characterized in that, The process of judging whether the unstructured data conversion is successful is as follows: Perform integrity and consistency checks on the keyword fields in the structured results. If a field is missing or the content is ambiguous, mark the conversion as failed; Use the conversion accuracy evaluation model to compare the conversion samples with the manually annotated results and generate an automated conversion quality score; If the score is greater than or equal to the set standard, the conversion is determined to be successful; If the score is less than the set standard, the conversion is determined to be unsuccessful.

8. A construction management system for municipal communication pipeline laying according to claim 7, characterized in that, The process of dynamically updating the construction progress status and generating a visual Gantt chart is as follows: Parse the timestamps and operation types in the structured construction records and map them to the corresponding process nodes in the project plan; Calculate the actual progress ratio of each node and generate a comparison curve in combination with the original planned progress; Use the visualization engine to generate a Gantt chart and update it to the management terminal large screen or mobile terminal.

9. The construction management system for laying municipal communication pipelines according to claim 8, characterized in that, The process of determining whether construction delays are found is as follows: Perform a difference analysis on the task progress line and the actual collected data progress line in the Gantt chart to determine the offset time of each key process; If the node delay on the critical path exceeds the set tolerance days, or the cumulative delay of multiple non-critical nodes causes the critical path to be blocked, it is determined that there is a construction delay.

10. A construction management system for municipal communication pipeline laying according to claim 9, characterized in that, The process of automatically pushing warning information to the management terminal is as follows: Based on the trained classification model, classify and label the problems; Combine historical data with current project parameters to predict the duration of the delay and the impact on the overall project duration; According to the prediction results, generate a rectification suggestion report and synchronize the reminder to the management terminal through the message push system.

Images