Engineering construction whole process supervision method and system based on multi-technology fusion

Abstract

The invention discloses an engineering construction whole process supervision method and system based on multi-technology fusion, and relates to the technical field of engineering supervision, and the supervision steps are as follows: S1, collecting construction site data through a mobile terminal multi-source channel, carrying out the space-time alignment and association fusion of data preprocessing, and carrying out the panoramic visual monitoring of a construction site; abnormal events and potential risks are identified through a preset rule and a machine learning algorithm; and S2, analyzing the engineering data by using a natural language processing technology. According to the invention, on the aspect of data and risk management and control, panoramic visual monitoring is realized through multi-source data fusion, and risks are accurately identified through an intelligent algorithm; the NLP technology analyzes the engineering data, automatically generates the check list, ensures compliance, realizes conversion from passive supervision to active prevention and from artificial experience to data driving, has remarkable effects in the aspects of improving management efficiency, reducing risks, saving cost and enhancing transparency, and is suitable for various large complex engineering scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of engineering supervision, in particular to a multi-technology fusion-based whole-process engineering construction supervision method and system. BACKGROUND

[0002] As an important basic project for guaranteeing national food security and agricultural modernization, high-standard farmland construction involves multiple parties, multiple processes and multiple links, has the characteristics of long time span, complex process, high quality and safety requirements, and the traditional farmland construction supervision relies on manual site inspection and paper file management, which has the problems of single supervision means, incomplete information collection, serious data island, low supervision efficiency and difficulty in realizing whole-process closed-loop management, and cannot meet the comprehensive supervision needs of modern high-standard farmland construction projects in quality, safety, progress and compliance;

[0003] With the rapid development of technologies such as Internet of Things, big data and artificial intelligence, intelligent engineering supervision has gradually become a key way to improve farmland construction quality and supervision efficiency. However, the existing supervision systems are mostly limited to single data type or management of a certain link, lack of fusion processing and intelligent analysis capability of multi-source heterogeneous data, and cannot realize dynamic perception, real-time monitoring and intelligent early warning of the whole construction process, making it difficult to form a whole life cycle supervision closed loop throughout the construction, supervision and audit links. In view of this, a multi-technology fusion-based whole-process engineering construction supervision method and system are proposed. SUMMARY

[0004] To solve the above technical problems, a multi-technology fusion-based whole-process engineering construction supervision method and system are provided, which solve the above problems.

[0005] To achieve the above purposes, the technical scheme adopted by the present application is as follows: a multi-technology fusion-based whole-process engineering construction supervision method, the supervision steps are:

[0006] S1, collecting construction site data through mobile terminal multi-source channels, pre-processing the data for spatio-temporal alignment and correlation fusion, visualizing monitoring of the construction site, and identifying abnormal events and potential risks through preset rules and machine learning algorithms;

[0007] S2, using natural language processing technology to analyze engineering data, constructing a knowledge graph to match regulations, technical standards and historical cases, and customizing inspection checklists through natural language generation;

[0008] S3, real-time recording of construction logs, labor personnel attendance, qualification and whole-process data of start-up and resumption of work, archiving according to time sequence and supporting multi-dimensional retrieval;

[0009] S4. Verify the rectification effect through image comparison, retain the chain of evidence, link the rectification results with the assessment, and connect with the compliance module to ensure that the rectification is compliant;

[0010] S5: Configure the concealed works acceptance and design change approval process based on the process engine, automatically verify material compliance to assist decision-making; electronically archive approval documents and visually display project progress and quality;

[0011] S6. Intelligently allocate resources according to project progress; quickly initiate emergency response to unexpected situations; and assist management decision-making by visualizing key indicators on a large screen.

[0012] S7. Regularly assess the effectiveness of supervision and analyze the occurrence rate of problems and the efficiency of rectification.

[0013] Preferably, in step S1, the mobile terminal includes a fixed camera device, a mobile sentry system, a smart helmet, a drone device, and an environmental quality sensor;

[0014] All heterogeneous data collected from multiple sources by mobile terminals are uniformly aggregated into the platform's "data access layer." Through highly automated ETL, including extraction, transformation, and loading processes, the data is standardized, cleaned, format-aligned, and stored in a structured manner.

[0015] All collected data is labeled with timestamps and spatial location information, and is linked to specific project stages and construction task milestones.

[0016] Preferably, the step of identifying abnormal events and potential risks in step S1 is as follows:

[0017] Feature extraction is performed on the fused construction site data. Based on wavelet analysis, time series features and spatial distribution patterns in the data are mined to construct feature vectors of personnel behavior trajectories, equipment operating parameters and environmental monitoring indicators.

[0018] A basic judgment model is established using preset rules, setting equipment operating temperature thresholds and restricted areas for personnel activities, judging the data, and triggering warnings for data that clearly violates the rules;

[0019] Machine learning algorithms are used for deep analysis. Historical anomalous event data is trained using random forests to build a classification model to identify known risk patterns.

[0020] Cluster analysis is used to uncover anomalous clusters in data and identify new or potential risks.

[0021] Establish a dynamic feedback mechanism to optimize model parameters by combining the identification results with manually reviewed data, and identify and provide early warnings for abnormal events and potential risks at the construction site.

[0022] Preferably, the wavelet analysis mining steps are as follows:

[0023] Based on the characteristics of construction site data, wavelet basis functions are selected. After selection, discrete wavelet transform is performed on the fused construction site original data to decompose the data into subsequences of different frequencies, namely low-frequency approximate components and high-frequency detail components.

[0024] The subsequences obtained from the decomposition are reconstructed, and the approximate components and detail components at different levels are combined by wavelet inverse transform to obtain data features at different scales, observe the changing trends of data at different scales, and extract key information reflecting the characteristics of the time series.

[0025] The construction site was divided into regions, and wavelet analysis was performed on the data in each region. The analysis results of different regions at the same scale were compared to find the spatial distribution differences and patterns of the data.

[0026] The mined features are constructed into feature vectors containing multi-dimensional information.

[0027] Preferably, the specific steps for identifying known risk patterns are as follows:

[0028] After preprocessing the historical anomalous event data, the processed data is divided into a training set and a test set, allocated in a 7:3 ratio;

[0029] Construct a random forest model, set parameters including the number of decision trees and the number of features considered when splitting nodes, take the feature vectors in the training set as input, and output the corresponding abnormal event categories to learn the mapping relationship between data features and risk patterns;

[0030] After training is complete, use the test set for evaluation, calculate the accuracy and recall metrics, and determine the model's ability to identify known risk patterns. If the evaluation results are not ideal, adjust the model parameters and retrain until the evaluation results reach the ideal state.

[0031] The model is applied to real-time data from the construction site, and timely warnings are issued when anomalies are detected.

[0032] Preferably, the engineering data in step S2 includes electronic documents, construction drawings, meeting minutes, audio recordings, and video recordings.

[0033] The knowledge graph construction uses word segmentation technology to break down engineering data into individual words and phrases, uses part-of-speech tagging to classify the grammatical categories of each word, and uses named entity recognition to extract key entities from the data, including project name, equipment model, personnel name and time and location. It also uses dependency parsing and semantic role tagging to mine the relationships between entities, and stores the entities and relationships in a structured way to build a knowledge graph for the engineering domain.

[0034] The customized checklist generation process involves regulatory personnel describing their inspection needs in both spoken and written language based on the matching results. The checklist is then customized according to the different inspection requirements and concerns of each project. Through template filling, regulatory requirements and technical standards are transformed into clear, easy-to-understand, and operable inspection items.

[0035] Preferably, the construction log in step S3 includes the daily work content, construction progress, personnel arrangement, use of machinery and equipment, material consumption, safety inspection and abnormal situations at the construction site;

[0036] Implement real-name management for laborers to achieve attendance registration, work group distribution statistics, and attendance rate analysis;

[0037] For the construction commencement and resumption stages, it provides functions for construction plan formulation, task breakdown and node control. Construction units can submit commencement / resumption plans in the system, allocate time nodes for construction tasks, and generate commencement / resumption records.

[0038] The entire process data is generated into standardized electronic archives, which are automatically summarized into monthly construction reports.

[0039] Preferably, in step S4, the rectification effect is compared and verified by acquiring images before and after rectification using high-definition equipment, and using the SIFT feature point matching algorithm to compare and analyze the changes in the images before and after rectification to determine the rectification effect.

[0040] The chain of evidence is stored on the blockchain;

[0041] The assessment of rectification results is conducted by establishing tiered standards, incorporating scoring into multi-party performance evaluations, rewards and penalties, and qualifications, building a dynamically updated compliance module database, automatically comparing rectification plans with laws and regulations, and reviewing documents.

[0042] Preferably, in step S6, a collaborative platform is built based on cloud computing and microservice architecture to push messages in real time;

[0043] Intelligent resource allocation is driven by project progress, dynamically analyzes resource supply and demand based on genetic algorithms, and automatically generates allocation plans.

[0044] A multi-technology integrated engineering construction process monitoring system includes:

[0045] The perception data fusion module is configured to collect on-site data, integrate and preprocess the data, perform collaborative analysis of data fusion, provide panoramic monitoring and perception of the construction site, and detect abnormal events and potential risks.

[0046] The AI-driven engineering supervision module analyzes uploaded data based on knowledge graphs, performs semantic analysis, extracts keywords, and quickly locates relevant legal provisions, technical standards, and historical cases; it also supports supervisors in generating customized inspection checklists through verbal expression.

[0047] The process recording module is configured to record the entire process of construction logs, labor logs, attendance management, and commencement and resumption of work in real time.

[0048] The supervision and management module is configured to retain data records, operation traces and responsibility chains through a multi-unit collaborative problem reporting, rectification plan formulation, execution process tracking, image comparison and evidence collection and assessment results linkage mechanism, and automatically connect to the assessment system and compliance module to carry out project supervision and management.

[0049] The engineering quality assessment module includes: a process-driven approval unit, a problem feedback and rectification unit, an archiving unit, a display unit, and a scheduling unit;

[0050] The process-driven approval unit manages the approvals and processes in the project based on the process-driven engine and approval management mechanism;

[0051] The problem feedback and rectification unit, based on the data monitored by the supervision and management module, provides feedback on problems from multiple channels, generates a unique problem number for each reported problem, and provides a rectification plan.

[0052] The archiving unit organizes all relevant documents for all stages and aspects of the project, including drawings, design documents, construction notices, meeting minutes, material certificates, and key documents related to quality acceptance records. These documents are stored through a standardized electronic process.

[0053] The display and scheduling units provide visual displays, allowing for an intuitive presentation of the project progress.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0055] This invention proposes a multi-source data fusion approach for data and risk management, enabling panoramic visual monitoring and intelligent algorithms for accurate risk identification. NLP technology analyzes engineering data, automatically generating checklists to ensure compliance. In terms of process management, full-cycle data recording ensures traceability, closed-loop rectification and intelligent approval improve quality control accuracy and decision-making efficiency, intelligent resource allocation optimizes progress, a unified platform breaks down information barriers, an emergency response mechanism facilitates rapid response to emergencies, and regular evaluation drives continuous improvement of regulatory strategies. This method achieves a shift from passive supervision to proactive prevention, and from manual experience to data-driven approaches, demonstrating significant effectiveness in improving management efficiency, reducing risks, saving costs, and enhancing transparency. It is applicable to various large-scale and complex engineering scenarios. Attached Figure Description

[0056] Figure 1 This is a flowchart of the regulatory steps of the present invention;

[0057] Figure 2 This is a framework diagram of the monitoring system of the present invention. Detailed Implementation

[0058] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0059] Reference Figure 1 As shown, a method for supervising the entire construction process based on the integration of multiple technologies is proposed, and the supervision steps are as follows:

[0060] S1. Collect construction site data through multiple channels via mobile terminals, perform spatiotemporal alignment and correlation fusion on the preprocessed data, conduct panoramic visual monitoring of the construction site, and identify abnormal events and potential risks through preset rules and machine learning algorithms.

[0061] S2. Utilize natural language processing technology to analyze engineering data, construct a knowledge graph to match legal provisions, technical standards, and historical cases; generate customized checklists through natural language processing.

[0062] S3: Real-time recording of construction logs, labor attendance and qualification data, and data on the entire process of work commencement and resumption approvals, archived in time sequence and supporting multi-dimensional retrieval;

[0063] S4. Verify the rectification effect through image comparison, retain the chain of evidence, link the rectification results with the assessment, and connect with the compliance module to ensure that the rectification is compliant;

[0064] S5: Configure the concealed works acceptance and design change approval process based on the process engine, automatically verify material compliance to assist decision-making; electronically archive approval documents and visually display project progress and quality;

[0065] S6. Intelligently allocate resources according to project progress; quickly initiate emergency response to unexpected situations; and assist management decision-making by visualizing key indicators on a large screen.

[0066] S7. Regularly assess the effectiveness of supervision and analyze the occurrence rate of problems and the efficiency of rectification.

[0067] This invention system addresses the multi-participant, multi-stage, and multi-standard management characteristics of high-standard farmland construction. It integrates advanced information technology to achieve a fully intelligent closed-loop system from data collection to decision support. Compared to traditional monitoring methods and existing systems, this invention possesses the following significant technical and application advantages:

[0068] Based on multiple types of intelligent terminals, an integrated perception system covering the sky, ground, and people is constructed, covering key locations and operation scenarios at the construction site, forming a comprehensive, real-time, and multi-dimensional dynamic monitoring capability for the project. All perception data is automatically associated with project nodes according to time and space, ensuring that the supervision information has a high degree of contextual traceability, laying the foundation for the visualization and digital supervision of the project process.

[0069] By building a unified data access and processing platform, the system supports the cleaning, tagging, and structured storage of various data types, including video, images, audio, sensor data, and documents. The data is standardized during the access process and uniformly incorporated into the main engineering data domain, breaking down data silos and greatly improving the consistency, accessibility, and automatic processing capabilities of regulatory information.

[0070] The system is equipped with an AI semantic analysis and knowledge graph engine, which can not only automatically understand and extract information from unstructured content such as documents, drawings, and voice, but also automatically generate scenario-based checklists, early warning rules, and rectification suggestions based on policies, regulations, quality standards, and historical issues. This truly embeds AI capabilities into the core workflow of supervision and improves the depth and accuracy of intelligent assisted decision-making.

[0071] The platform supports multi-unit collaborative problem reporting, rectification plan formulation, execution process tracking, image comparison and evidence collection, and assessment result linkage mechanism. Data records, operation traces, and responsibility chains are retained for each problem from discovery to handling. It automatically connects to the assessment system and compliance module to form a closed-loop control and accountability system for engineering governance.

[0072] The system has a built-in multi-level quality standard library, covering national, local, industry and project-defined standards. It supports automatic scoring of process quality, trend analysis of historical quality problems and visual acceptance of hidden works. Through a mandatory traceability mechanism (images, videos, test reports), it ensures that the quality of each process is verifiable and the responsibility is traceable.

[0073] The system has the capability of full-process electronic document archiving, supports the automatic generation of document summaries, keywords, tags, version comparisons and full-text search, and has an embedded policy matching and compliance verification engine. All approval records, uploaded documents and operation traces can be automatically generated into an "audit preparation package" to meet the compliance requirements of internal and external inspections, accountability checks and government supervision.

[0074] In step S1, the mobile terminal includes fixed camera equipment, mobile sentry system, smart helmet, drone equipment, and environmental quality sensor;

[0075] All heterogeneous data collected from multiple sources by mobile terminals are uniformly aggregated into the platform's "data access layer." Through highly automated ETL, including extraction, transformation, and loading processes, the data is standardized, cleaned, format-aligned, and stored in a structured manner.

[0076] All collected data is labeled with timestamps and spatial location information, and is linked to specific project stages and construction task milestones.

[0077] Fixed video equipment is deployed in the main construction passage, water conservancy hub, key construction nodes and key safety areas. It has high-resolution monitoring and night infrared functions to capture construction dynamics, personnel flow and abnormal events in real time.

[0078] The mobile sentry system is equipped with a 360° panoramic camera, high-definition video recording and real-time data transmission module, supports remote control and automatic patrol functions, flexibly covers complex on-site environments and blind spots, and realizes dynamic monitoring and instant response.

[0079] The smart safety helmet, worn by on-site staff, is equipped with a built-in high-definition camera, microphone, GPS positioning, and multiple environmental sensors to upload audio and video, geographical location, and personnel status data in real time, thereby improving on-site safety management and emergency response capabilities.

[0080] The drone equipment performs aerial photography missions on a regular basis to acquire high-altitude panoramic images and automatically generate construction terrain change maps, engineering coverage maps and 3D models, providing intuitive data support for terrain analysis, progress monitoring and quality assessment.

[0081] Environmental quality sensors are deployed in key locations to monitor environmental parameters (such as wind speed, temperature, humidity, and gas concentration) and engineering quality indicators (such as concrete strength and soil compaction) in real time, ensuring the safety of the construction environment and the control of engineering quality.

[0082] The steps for identifying anomalous events and potential risks in step S1 are as follows:

[0083] Feature extraction is performed on the fused construction site data. Based on wavelet analysis, time series features and spatial distribution patterns in the data are mined to construct feature vectors of personnel behavior trajectories, equipment operating parameters and environmental monitoring indicators.

[0084] A basic judgment model is established using preset rules, setting equipment operating temperature thresholds and restricted areas for personnel activities, judging the data, and triggering warnings for data that clearly violates the rules;

[0085] Machine learning algorithms are used for deep analysis. Historical anomalous event data is trained using random forests to build a classification model to identify known risk patterns.

[0086] Cluster analysis is used to uncover anomalous clusters in data and identify new or potential risks.

[0087] Establish a dynamic feedback mechanism to optimize model parameters by combining the identification results with manually reviewed data, and identify and provide early warnings for abnormal events and potential risks at the construction site.

[0088] This application utilizes wavelet analysis to effectively extract the spatiotemporal features of data, constructing feature vectors encompassing personnel, equipment, and the environment. This comprehensively captures construction dynamics, providing rich information for risk identification. The pre-set rule model can quickly respond to obvious violations, such as equipment overheating or personnel crossing boundaries, enabling immediate early warnings. The machine learning algorithm trains the model using historical data to accurately identify known risk patterns. The combination of these two approaches balances efficiency and accuracy. Cluster analysis can uncover anomalous clusters in the data, helping to identify new risks that are not yet clearly defined, filling the gaps in rules and historical experience, and improving risk prediction capabilities. The dynamic feedback mechanism combines the identification results with manually reviewed data, continuously optimizing model parameters, enabling the risk identification system to adapt to the complex changes at the construction site, and continuously improving the accuracy and reliability of early warnings.

[0089] The steps for mining using wavelet analysis are as follows:

[0090] Based on the characteristics of construction site data, wavelet basis functions are selected. After selection, discrete wavelet transform is performed on the fused construction site original data to decompose the data into subsequences of different frequencies, namely low-frequency approximate components and high-frequency detail components.

[0091] The subsequences obtained from the decomposition are reconstructed, and the approximate components and detail components at different levels are combined by wavelet inverse transform to obtain data features at different scales, observe the changing trends of data at different scales, and extract key information reflecting the characteristics of the time series.

[0092] The construction site was divided into regions, and wavelet analysis was performed on the data in each region. The analysis results of different regions at the same scale were compared to find the spatial distribution differences and patterns of the data.

[0093] The mined features are constructed into feature vectors containing multi-dimensional information.

[0094] By using discrete wavelet transform, the original data is decomposed into low-frequency approximate components (reflecting the overall trend) and high-frequency detail components (highlighting local abrupt changes). Then, the subsequences of each layer are reconstructed by inverse transform. The data change trend can be observed from different scales. The low-frequency components can identify long-term abnormal trends in equipment operation (such as continuous power fluctuations), while the high-frequency components can capture sudden interference (such as instantaneous voltage drops). This enables multi-level analysis of time series characteristics and avoids the one-sidedness of single-scale analysis.

[0095] Data is divided by construction area and wavelet analysis is performed. By comparing the characteristic differences of different areas at the same scale (such as the distribution of equipment vibration frequency in different work areas), spatial distribution anomalies can be accurately located.

[0096] By integrating time series features (such as equipment operating cycle patterns) with spatial distribution patterns (such as regional environmental parameter differences) into a multi-dimensional feature vector, richer input dimensions are provided for subsequent machine learning models.

[0097] The specific steps for identifying known risk patterns are as follows:

[0098] After preprocessing the historical anomalous event data, the processed data is divided into a training set and a test set, allocated in a 7:3 ratio;

[0099] Construct a random forest model, set parameters including the number of decision trees and the number of features considered when splitting nodes, take the feature vectors in the training set as input, and output the corresponding abnormal event categories to learn the mapping relationship between data features and risk patterns;

[0100] After training is complete, use the test set for evaluation, calculate the accuracy and recall metrics, and determine the model's ability to identify known risk patterns. If the evaluation results are not ideal, adjust the model parameters and retrain until the evaluation results reach the ideal state.

[0101] The model is applied to real-time data from the construction site, and timely warnings are issued when anomalies are detected.

[0102] This application focuses on data processing and model optimization to significantly improve the efficiency of engineering risk management. In terms of data utilization, the training and test sets are divided in a 7:3 ratio, allowing the model to fully learn from historical data features while independently testing and verifying generalization ability to avoid overfitting. Random forest modeling, through the ensemble of multiple decision trees, can capture multi-dimensional information interaction relationships, has greater tolerance for noisy data, and ensures recognition stability. In the evaluation and optimization phase, accuracy and recall are used to quantify model performance, allowing for targeted parameter adjustments to ensure recognition accuracy meets engineering requirements. In practical applications, the model is linked with real-time data from the construction site, enabling rapid identification of risks conforming to historical patterns and immediate warnings, significantly shortening risk response time. This method transforms expert experience into model feature mapping relationships, achieving the digitization and standardization of experience, reducing reliance on human experience, and minimizing errors caused by subjective judgment. It has significant effects in improving risk management efficiency and reducing enterprise management costs.

[0103] The engineering documents in step S2 include electronic documents, construction drawings, meeting minutes, audio recordings, and video recordings.

[0104] The knowledge graph construction uses word segmentation technology to break down engineering data into individual words and phrases, uses part-of-speech tagging to classify the grammatical categories of each word, and uses named entity recognition to extract key entities from the data, including project name, equipment model, personnel name and time and location. It also uses dependency parsing and semantic role tagging to mine the relationships between entities, and stores the entities and relationships in a structured way to build a knowledge graph for the engineering domain.

[0105] The customized checklist generation process involves regulatory personnel describing their inspection needs in both spoken and written language based on the matching results. The checklist is then customized according to the different inspection requirements and concerns of each project. Through template filling, regulatory requirements and technical standards are transformed into clear, easy-to-understand, and operable inspection items.

[0106] This application transforms complex engineering information into efficient management tools through multi-type data processing and intelligent construction, significantly improving the scientific nature and convenience of supervision. In terms of knowledge graph construction, it performs full-type analysis of diverse data such as electronic documents, drawings, and audio-visual materials. Using word segmentation and part-of-speech tagging NLP technology, it accurately extracts key entities such as project names and equipment models. Through dependency parsing, it mines the logical relationships between entities, stores them in a structured manner to form a knowledge graph, breaks down information silos, realizes the systematic integration of engineering knowledge, and provides a clear knowledge network for subsequent retrieval and decision-making.

[0107] The generation of customized checklists further enhances the practicality of supervision. Supervisors only need to describe their needs in natural language, and the system can match legal provisions and technical standards based on a knowledge graph. By filling in templates, professional requirements are transformed into visual and actionable checklist items. This avoids the omissions and time-consuming process of manually compiling checklists, while also meeting the personalized inspection needs of different projects. This makes quality and safety inspections more targeted, promotes the transformation of project management from experience-driven to data-driven, and effectively improves the efficiency and standardization of supervision.

[0108] The construction log in step S3 includes the daily work content, construction progress, personnel arrangement, use of machinery and equipment, material consumption, safety inspection and abnormal situations at the construction site.

[0109] Implement real-name management for laborers to achieve attendance registration, work group distribution statistics, and attendance rate analysis;

[0110] For the construction commencement and resumption stages, it provides functions for construction plan formulation, task breakdown and node control. Construction units can submit commencement / resumption plans in the system, allocate time nodes for construction tasks, and generate commencement / resumption records.

[0111] The entire process data is generated into standardized electronic archives, which are automatically summarized into monthly construction reports.

[0112] In terms of data recording, this application includes construction logs that cover multi-dimensional information such as work content, progress, personnel, and equipment; real-name management of laborers enables attendance and dynamic monitoring of work teams; and detailed task node control in the start-up and resumption of work processes. Together, these three elements construct a precise data profile of the entire construction process, enabling managers to grasp the on-site dynamics in real time and avoid information lag or omission. At the data application level, standardized electronic archives and automatically generated monthly construction reports achieve orderly data storage and efficient retrieval, facilitating traceability and auditing.

[0113] In step S4, the rectification effect is compared and verified by collecting images before and after rectification using high-definition equipment, and using the SIFT feature point matching algorithm to compare and analyze the changes in the images before and after rectification to determine the rectification effect.

[0114] The chain of evidence is stored on the blockchain;

[0115] The assessment of rectification results is conducted by establishing tiered standards, incorporating scoring into multi-party performance evaluations, rewards and penalties, and qualifications, building a dynamically updated compliance module database, automatically comparing rectification plans with laws and regulations, and reviewing documents.

[0116] This application utilizes the SIFT feature point matching algorithm to perform pixel-level comparison of high-definition rectification images before and after rectification. This allows for intuitive identification of subtle changes, avoiding subjective biases and omissions in manual verification, and providing objective and quantitative evaluation basis for rectification acceptance, ensuring problem resolution. By storing key data such as images and approval records during the rectification process on the blockchain, and leveraging the decentralized and timestamped characteristics of blockchain, an immutable and traceable electronic evidence chain is formed. This not only enhances regulatory credibility but also provides legal effect in dispute resolution, reducing the cost of post-event liability determination.

[0117] In step S6, a collaborative platform is built based on cloud computing and microservice architecture to enable real-time message push.

[0118] Intelligent resource allocation is driven by project progress, dynamically analyzes resource supply and demand based on genetic algorithms, and automatically generates allocation plans.

[0119] Reference Figure 2 As shown, a multi-technology integrated engineering construction process monitoring system includes:

[0120] The perception data fusion module is configured to collect on-site data, integrate and preprocess the data, perform collaborative analysis of data fusion, provide panoramic monitoring and perception of the construction site, and detect abnormal events and potential risks.

[0121] The AI-driven engineering supervision module analyzes uploaded data based on knowledge graphs, performs semantic analysis, extracts keywords, and quickly locates relevant legal provisions, technical standards, and historical cases; it also supports supervisors in generating customized inspection checklists through verbal expression.

[0122] The process recording module is configured to record the entire process of construction logs, labor logs, attendance management, and commencement and resumption of work in real time.

[0123] The supervision and management module is configured to retain data records, operation traces and responsibility chains through a multi-unit collaborative problem reporting, rectification plan formulation, execution process tracking, image comparison and evidence collection and assessment results linkage mechanism, and automatically connect to the assessment system and compliance module to carry out project supervision and management.

[0124] The engineering quality assessment module includes: a process-driven approval unit, a problem feedback and rectification unit, an archiving unit, a display unit, and a scheduling unit;

[0125] The process-driven approval unit manages the approvals and processes in the project based on the process-driven engine and approval management mechanism;

[0126] The problem feedback and rectification unit, based on the data monitored by the supervision and management module, provides feedback on problems from multiple channels, generates a unique problem number for each reported problem, and provides a rectification plan.

[0127] The archiving unit organizes all relevant documents for all stages and aspects of the project, including drawings, design documents, construction notices, meeting minutes, material certificates, and key documents related to quality acceptance records. These documents are stored through a standardized electronic process.

[0128] The display and scheduling units provide visual displays, allowing for an intuitive presentation of the project progress.

[0129] This application utilizes fixed cameras, mobile terminals, smart safety helmets, drone inspections, and environmental / engineering sensors to achieve multi-dimensional real-time perception of construction sites and construct a data model that binds "time and space + process". Based on uploaded drawings, documents, videos, and audio unstructured data, it automatically parses key information and constructs a multimodal knowledge graph in the field of farmland engineering, supporting semantic indexing, intelligent reasoning, and compliance verification. It realizes a complete document processing workflow from uploading, identification, classification, approval, comparison to archiving, and embeds an automatic standard specification matching engine to support policy auditing and regulatory closed loop.

[0130] Based on a quality standard library and scoring model, the system automatically scores and evaluates construction data. If the test fails, the system automatically suspends the relevant approval process and triggers the rectification process, forming a closed-loop quality control. All issues support classification and numbering, assignment of responsible units, uploading of comparative data, rectification plan tracking, overdue warnings and responsibility tracing, building a traceable full-process issue management chain.

[0131] The system has a built-in process modeler that supports custom nodes, approver configuration, timeout rules, and audit trails to ensure standardization and timeliness during project execution. Based on logs, approval records, image data, and timelines, the system automatically compiles a complete audit data package for a specific stage to meet the needs of internal and external verification and accountability.

[0132] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A method for supervising the entire construction process based on the integration of multiple technologies, characterized in that: The regulatory steps are as follows: S1. Collect construction site data through multiple channels via mobile terminals, perform spatiotemporal alignment and correlation fusion on the preprocessed data, conduct panoramic visual monitoring of the construction site, and identify abnormal events and potential risks through preset rules and machine learning algorithms. S2. Utilize natural language processing technology to analyze engineering data, construct a knowledge graph to match legal provisions, technical standards, and historical cases; generate customized checklists through natural language processing. S3: Real-time recording of construction logs, labor attendance and qualification data, and data on the entire process of work commencement and resumption approvals, archived in time sequence and supporting multi-dimensional retrieval; S4. Verify the rectification effect through image comparison, preserve the evidence chain, link the rectification results with the assessment, and connect with the compliance module to ensure that the rectification is compliant; S5: Configure the concealed works acceptance and design change approval process based on the process engine, automatically verify material compliance to assist decision-making; electronically archive approval documents and visually display project progress and quality; S6. Intelligently allocate resources according to project progress; quickly initiate emergency response to unexpected situations; and assist management decision-making by visualizing key indicators on a large screen. S7. Regularly assess the effectiveness of supervision and analyze the occurrence rate of problems and the efficiency of rectification.

2. The method for monitoring the entire construction process based on multi-technology integration as described in claim 1, characterized in that, In step S1, the mobile terminal includes fixed camera equipment, mobile sentry system, smart helmet, drone equipment, and environmental quality sensor; All heterogeneous data collected from multiple sources by mobile terminals are uniformly aggregated into the platform's "data access layer." Through highly automated ETL processes, including extraction, transformation, and loading, the data is standardized, cleaned, format-aligned, and stored in a structured manner. All collected data is labeled with timestamps and spatial location information, and is linked to specific project stages and construction task milestones.

3. The method for monitoring the entire construction process based on multi-technology integration as described in claim 1, characterized in that, The steps for identifying anomalous events and potential risks in step S1 are as follows: Feature extraction is performed on the fused construction site data. Based on wavelet analysis, time series features and spatial distribution patterns in the data are mined to construct feature vectors of personnel behavior trajectories, equipment operating parameters and environmental monitoring indicators. A basic judgment model is established using preset rules, setting equipment operating temperature thresholds and restricted areas for personnel activities, judging the data, and triggering warnings for data that clearly violates the rules; Machine learning algorithms are used for deep analysis. Historical anomalous event data is trained using random forests to build a classification model to identify known risk patterns. Cluster analysis can be used to uncover anomalous clusters in data and identify new or potential risks. Establish a dynamic feedback mechanism to optimize model parameters by combining the identification results with manually reviewed data, and identify and provide early warnings for abnormal events and potential risks at the construction site.

4. The method for supervising the entire construction process based on multi-technology integration as described in claim 3, characterized in that, The steps for mining using wavelet analysis are as follows: Based on the characteristics of construction site data, wavelet basis functions are selected. After selection, discrete wavelet transform is performed on the fused construction site original data to decompose the data into subsequences of different frequencies, namely low-frequency approximate components and high-frequency detail components. The subsequences obtained from the decomposition are reconstructed, and the approximate components and detail components at different levels are combined by wavelet inverse transform to obtain data features at different scales, observe the changing trends of data at different scales, and extract key information reflecting the characteristics of the time series. The construction site was divided into regions, and wavelet analysis was performed on the data in each region. The analysis results of different regions at the same scale were compared to find the spatial distribution differences and patterns of the data. The mined features are constructed into feature vectors containing multi-dimensional information.

5. The method for monitoring the entire construction process based on multi-technology integration as described in claim 3, characterized in that, The specific steps for identifying known risk patterns are as follows: After preprocessing the historical anomalous event data, the processed data is divided into a training set and a test set, allocated in a 7:3 ratio; Construct a random forest model, set parameters including the number of decision trees and the number of features considered when splitting nodes, take the feature vectors in the training set as input, and output the corresponding abnormal event categories to learn the mapping relationship between data features and risk patterns; After training is complete, use the test set for evaluation, calculate the accuracy and recall metrics, and determine the model's ability to identify known risk patterns. If the evaluation results are not ideal, adjust the model parameters and retrain until the evaluation results reach the ideal state. The model is applied to real-time data from the construction site, and timely warnings are issued when anomalies are detected.

6. The method for supervising the entire construction process based on multi-technology integration as described in claim 1, characterized in that, The engineering documents in step S2 include electronic documents, construction drawings, meeting minutes, audio recordings, and video recordings. The knowledge graph construction uses word segmentation technology to break down engineering data into individual words and phrases, uses part-of-speech tagging to classify the grammatical categories of each word, and uses named entity recognition to extract key entities from the data, including project name, equipment model, personnel name and time and location. It also uses dependency parsing and semantic role tagging to mine the relationships between entities, and stores the entities and relationships in a structured way to build a knowledge graph for the engineering domain. The customized checklist generation process involves regulatory personnel describing their inspection needs in both spoken and written language based on the matching results. The checklist is then customized according to the different inspection requirements and concerns of each project. Through template filling, regulatory requirements and technical standards are transformed into clear, easy-to-understand, and operable inspection items.

7. The method for supervising the entire construction process based on multi-technology integration as described in claim 1, characterized in that, The construction log in step S3 includes the daily work content, construction progress, personnel arrangement, use of machinery and equipment, material consumption, safety inspection and abnormal situations at the construction site. Implement real-name management for laborers to achieve attendance registration, work group distribution statistics, and attendance rate analysis; For the construction commencement and resumption stages, it provides functions for construction plan formulation, task breakdown and node control. Construction units can submit commencement / resumption plans in the system, allocate time nodes for construction tasks, and generate commencement / resumption records. The entire process data is generated into standardized electronic archives, which are automatically summarized into monthly construction reports.

8. The method for supervising the entire construction process based on multi-technology integration as described in claim 1, characterized in that, In step S4, the rectification effect is compared and verified by collecting images before and after rectification using high-definition equipment, and using the SIFT feature point matching algorithm to compare and analyze the changes in the images before and after rectification to determine the rectification effect. The chain of evidence is stored on the blockchain; The assessment of rectification results is conducted by establishing tiered standards, incorporating scoring into multi-party performance evaluations, rewards and penalties, and qualifications, building a dynamically updated compliance module database, automatically comparing rectification plans with laws and regulations, and reviewing documents.

9. A method for supervising the entire construction process based on multi-technology integration as described in claim 1, characterized in that, In step S6, a collaborative platform is built based on cloud computing and microservice architecture to enable real-time message push. Intelligent resource allocation is driven by project progress, dynamically analyzes resource supply and demand based on genetic algorithms, and automatically generates allocation plans.

10. A multi-technology integrated engineering construction process monitoring system, characterized in that, include: The perception data fusion module is configured to collect on-site data, integrate and preprocess the data, perform collaborative analysis of data fusion, provide panoramic monitoring and perception of the construction site, and detect abnormal events and potential risks. The AI-driven engineering supervision module analyzes uploaded data based on knowledge graphs, performs semantic analysis, extracts keywords, and quickly locates relevant legal provisions, technical standards, and historical cases; it also supports supervisors in generating customized inspection checklists through verbal expression. The process recording module is configured to record the entire process of construction logs, labor logs, attendance management, and commencement and resumption of work in real time. The supervision and management module is configured to retain data records, operation traces and responsibility chains through a multi-unit collaborative problem reporting, rectification plan formulation, execution process tracking, image comparison and evidence collection and assessment results linkage mechanism, and automatically connect to the assessment system and compliance module to carry out project supervision and management. The engineering quality assessment module includes: a process-driven approval unit, a problem feedback and rectification unit, an archiving unit, a display unit, and a scheduling unit; The process-driven approval unit manages the approvals and processes in the project based on the process-driven engine and approval management mechanism; The problem feedback and rectification unit, based on the data monitored by the supervision and management module, provides feedback on problems from multiple channels, generates a unique problem number for each reported problem, and provides a rectification plan. The archiving unit organizes all relevant documents for all stages and aspects of the project, including drawings, design documents, construction notices, meeting minutes, material certificates, and key documents related to quality acceptance records. These documents are stored through a standardized electronic process. The display and scheduling units provide visual displays, allowing for an intuitive presentation of the project progress.

Images