A construction project management method based on edge computing and hybrid large model adaptation

Abstract

The application discloses a kind of based on edge computing's construction engineering management method, system and storage medium, belong to construction engineering digital technology field.For mobile terminal in the service break problem under the environment without network, the application uses cloud edge collaborative architecture, and preposition lightweight model and business rules in smartphone APP, realize online and offline mode seamless switching.Material acceptance whole process controllable is achieved;Realize worker / mechanical operation whole process continuous management: before work, generate semi-structured form;Many nodes generate intermediate semi-structured form record state in the process of operation;After work, all node data are generated to complete structured form and carry out offline electronic signature.Network is recovered, and system automatically incremental synchronization data, and through digital thread, structured form is distributed to supplier, project account book, enterprise account book, cost early warning etc.system by key, realize whole process visualization and real-time early warning.The application realizes the service continuity under the environment without network, data authenticity, process visualization and digital collaboration.

Description

Technical Field

[0001] This invention relates to the field of digital and intelligent management technology for construction projects. Specifically, it relates to a construction project management method, system, and storage medium based on edge computing and hybrid large-scale model adaptation applications. It is particularly suitable for complex engineering site environments where network signals are unstable or completely absent, enabling digital and visualized scientific management of all elements and processes of construction projects, and ultimately producing data resources that can be quantified according to national standards. Background Technology

[0002] With the deepening development of digitalization in the construction industry, construction project management is rapidly evolving towards digitalization and intelligence. Existing technologies typically rely on cloud platforms for centralized data processing, but cloud-edge collaborative solutions depend on dedicated hardware, are costly, cannot adapt to linear engineering applications, lack offline environments to achieve closed-loop multi-node forms throughout the entire process, and lack hardware-level security and biometric-assisted verification for signatures. This makes it difficult to achieve refined management and control of the entire process of engineering operations, such as contract performance, cost accounting, progress control, and quality supervision.

[0003] However, many engineering project sites, such as those in mountainous areas, Gobi Desert, deserts, underground locations, tunnels, and islands, commonly experience a lack of or extremely unstable network signals. This prevents mobile applications (APPs) from interacting with the cloud in real time, forcing on-site operations (such as the acceptance of incoming materials and management records of labor / machinery usage) to be interrupted or revert to paper-based methods, resulting in gaps or even loss of control in enterprise management in critical areas. While solutions exist for deploying local servers or using satellite communications, these solutions have limited application scenarios, are complex to deploy, and involve bulky equipment, making them difficult to widely adopt in small and medium-sized enterprises and construction projects.

[0004] Furthermore, in traditional construction project management, the acceptance of incoming materials, labor usage, and machinery use often rely on paper-based records. This data is scattered, easily lost, and difficult to trace. It also fails to achieve full-process data structuring from the issuance of supply batches to on-site acceptance, task allocation, multi-node check-ins, and final settlement. This results in delayed cost accounting, lack of process control, and frequent occurrences of waste, leakage, and gray-area operations. How to achieve business continuity, data authenticity, process visualization, real-time early warning, intelligent management, and ultimately, quantifiable data resources that meet national standards in construction project site management has become an urgent technical challenge. Summary of the Invention

[0005] This invention aims to solve or improve the problems in the prior art mentioned above, and provides a construction project management method, system, and storage medium based on edge computing. This invention achieves closed-loop processing of business operations and secure data storage in the event of network interruption by deploying a lightweight edge inference model and local business logic on a mobile terminal (APP); it enables uninterrupted management of the entire process, including material procurement, labor employment, and machinery operations, even when offline by constructing a complete data chain from "multi-node semi-structured forms" to "structured forms"; and it achieves global cloud analysis and visual dashboard updates through automatic incremental synchronization and digital thread distribution after network recovery, forming a collaborative and intelligently evolving management system integrating cloud, edge, and terminal.

[0006] To achieve the above objectives, according to one aspect of the present invention, a construction project management method based on edge computing is provided. This method is executed by an app installed on a smartphone, the smartphone acting as an edge computing node and working collaboratively with a cloud platform, and is characterized by comprising the following steps:

[0007] S100: Deployment and initialization steps: The compressed and optimized lightweight edge inference model (model compression refers to removing redundant neural unit connections in the model through pruning, converting 32-bit floating-point parameters into 16-bit / 8-bit integers, compressing the parameters of the large model to 10% to 20% of the original model through knowledge distillation, and then transferring the compressed knowledge to the small model), as well as the project's basic data and local business rules are pre-installed in the APP of the smartphone;

[0008] S200: Mode adaptive switching step, the APP monitors the network connection status in real time. When the network connection is normal and stable, it runs in cloud online collaborative mode; when a network interruption or signal strength is detected to be lower than a preset threshold, it automatically and seamlessly switches to edge offline computing mode.

[0009] S300: Edge offline business closed-loop processing steps. In the edge offline computing mode, the APP utilizes local computing, storage, and sensing capabilities to independently complete the complete closed loop of construction project management business, specifically including:

[0010] S310: Task allocation and multi-node semi-structured form generation sub-step. Before the start of work in the morning, the site management personnel first conduct technical briefings, safety briefings and task allocation. Then, they fill out the form through the APP and take a photo for evidence. The system automatically loads the timestamp and location information to generate the associated semi-structured form. The form contains structured fields and unstructured data and is distributed to the project site management personnel's APP through near-field communication or Bluetooth to realize on-site team collaboration in offline state.

[0011] S320: Multi-node semi-structured form generation sub-step. During the operation, based on the operation rhythm and management needs, semi-structured forms for intermediate nodes are generated through the APP at multiple time nodes, including morning get off work hours, afternoon get off work hours, and evening overtime work hours. The operation status, operation time period, and evidence photos of this stage are recorded. The system automatically loads timestamps and location information to ensure the continuous traceability of the operation process.

[0012] S330: Sub-step of completion pricing and structured form generation. After the work is completed, the manager immediately conducts a post-shift summary through the APP, takes photos for evidence, and fills out a form to price the daily workload or duration. The system integrates the pre-shift task allocation content, intermediate node records, process photos, post-shift photos, pricing process and results to generate a complete structured form. It also uses the smartphone hardware security module to generate a legally valid offline electronic signature. The business documents containing the electronic signature are encrypted and stored in the local database to achieve a complete closed loop and uninterrupted record of worker / machine management in offline mode.

[0013] S340: Near-field collaboration sub-step. In edge offline computing mode, the APP can distribute the generated business documents to the mobile terminals of other relevant parties on site through near-field wireless communication or Bluetooth to achieve point-to-point business confirmation and collaboration.

[0014] S400: Data synchronization and digital thread distribution steps. When the network is restored, the APP automatically synchronizes the encrypted business data stored during the offline period to the cloud platform incrementally. The cloud platform performs global analysis on the merged data (including data conflicts), and then distributes the structured form to the team leader or supplier's APP, project ledger system, enterprise ledger system, cost early warning system and the sender's mailbox with one click through the digital thread, so as to realize the visual update of project progress, cost, etc.

[0015] S500: Model evolution steps: The cloud platform trains and optimizes the multi-constraint fusion scientific computing model based on the fused new data, and generates a lightweight model update package through model compression technology and sends it to the APP to optimize the local edge inference model; The data distribution and synchronization are achieved by real-time one-click distribution of structured forms through "digital threads".

[0016] Preferably, the offline electronic signature in step S330 is generated based on the hardware security chip of the smartphone, the device's unique identifier, and the operator's real-time biometric information to ensure the authenticity, integrity, and non-repudiation of the offline data.

[0017] Preferably, the multi-node semi-structured form generation in step S320 further includes: according to the granularity requirements of project management, form generation nodes can be customized and added, and the form of each node is independently associated with the main task sheet to form a complete work process data chain.

[0018] Preferably, the data conflict resolution priority rules in step S400 are as follows: ① Automatically collected data (such as weighbridge and sensor data) > Manually entered offline data; ② Valid data with later timestamps > Duplicate data with earlier timestamps; ③ Cloud-approved data > Edge-approved data. When a conflict occurs between the cloud and edge terminals in modifying the same business data, automatic coordination is performed according to the preset priority rules, and a conflict resolution log is recorded.

[0019] Preferably, the digital thread distribution in step S400 further includes: distributing a structured form to the cost early warning system, triggering the system to automatically compare the planned cost with the actual cost, and when the deviation exceeds a preset threshold, generating real-time early warning information and pushing it to relevant management personnel for source tracing and correction.

[0020] Preferably, the lightweight edge inference model is obtained by processing the large multi-constraint fusion scientific computing model in the cloud with model pruning, parameter quantization, or knowledge distillation techniques.

[0021] According to another aspect of the present invention, a construction project management system based on edge computing is provided for implementing the method described in any of the above claims, comprising:

[0022] The cloud service platform includes a multi-constraint fusion scientific computing hybrid large model module, an enterprise ledger module, a resource planning system module, a project ledger management module, a cost early warning module, and a model optimization and distribution service module;

[0023] The mobile edge computing terminal is a smartphone with a customized APP installed, the APP including:

[0024] Network status awareness and mode switching unit;

[0025] The multimodal data acquisition and preprocessing unit supports the acquisition of text, images, audio, location information, and timestamps.

[0026] Lightweight edge inference engine;

[0027] Local business workflow and rules engine, with pre-built business processes such as technical briefing, task allocation, attendance tracking, pricing, and settlement;

[0028] Semi-structured / structured form generation unit, supporting multi-node form generation and automatic association;

[0029] Hardware-secure electronic signature / seal unit;

[0030] Multi-node check-in and occurrence recording unit;

[0031] Local encrypted database;

[0032] Near-field cooperative communication unit;

[0033] Intelligent data synchronization client;

[0034] A data distribution and synchronization bus connects the cloud service platform with multiple mobile edge computing terminals, and is used to manage the real-time distribution of online data, the asynchronous uploading of offline data, the distribution of digital threads, and the distribution of model update packages.

[0035] According to another aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the construction project management method based on edge computing as described in any of the preceding claims.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1) Business continuity: Primarily using smartphones as edge terminals, the cloud-edge collaborative architecture enables seamless switching between networked and offline environments. In particular, material management and worker / machinery management can still achieve a complete closed loop from order placement to on-site acceptance of materials and from pre-shift briefing and process recording to post-shift settlement of worker / machinery management even when completely offline, effectively solving the problems of management gaps and loss of control in offline environments.

[0038] 2) Data structuring and traceability: A complete data chain from "semi-structured forms" to "structured forms" has been constructed. Through a multi-node semi-structured form generation mechanism, scattered paper documents, verbal instructions, and scattered scenarios are integrated into traceable and auditable structured data. Each operation link has form records, timestamps, location, and photo evidence, realizing a multi-node form closed loop, ensuring uninterrupted offline business processes and traceability.

[0039] 3) Data authenticity and security: Based on the hardware security and biometric module built into the smartphone, a legally valid electronic signature is generated offline. Combined with encrypted storage, automatic loading of timestamps and location information, the authenticity, integrity and immutability of business data are guaranteed from the source.

[0040] 4) Transparent and visible process: Through multi-node form generation, process photo evidence storage, and one-click distribution of digital threads, the operation process is made transparent, the cost structure is visualized, and the progress status is real-time. Management can keep track of the project dynamics in real time through a visual dashboard.

[0041] 5) Scientific and meticulous management: Based on real-time data, the system automatically triggers cost warnings. It can compare planned costs, actual costs, and completed output value in real time to achieve refined cost process control and avoid cost loss of control caused by post-event accounting.

[0042] 6) Efficient collaboration among multiple roles: Team leaders, workers, suppliers, finance, project managers and other roles can collaborate in real time on a unified data platform, greatly reducing communication costs and reconciliation disputes.

[0043] 7) Extremely low marginal cost deployment: Directly utilize the smartphones commonly used by on-site personnel as edge computing nodes, without the need for additional procurement and deployment of expensive dedicated hardware, greatly reducing the barriers to system promotion and use.

[0044] 8) Continuous intelligent evolution of the model: It forms an intelligent evolution closed loop of "edge acquisition - cloud training - edge update", which enables the on-site intelligence level to continuously improve with the accumulation of project data. Attached Figure Description

[0045] Figure 1 A schematic diagram of the overall architecture of a construction project management system based on edge computing provided in an embodiment of the present invention;

[0046] Figure 2 This is a flowchart illustrating the material acceptance management process in the edge offline mode in an embodiment of the present invention.

[0047] Figure 3 This is a schematic diagram of the offline, uninterrupted intelligent management process for the entire worker / machine operation in an embodiment of the present invention;

[0048] Figure 4 This is a sequence diagram illustrating the interaction between data synchronization, digital thread distribution, and model optimization and updating in an embodiment of the present invention. Detailed Implementation

[0049] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are only for explaining the present invention and do not constitute any limitation on the scope of protection of the present invention. Example 1: System Architecture

[0050] like Figure 1 As shown, the system of the present invention mainly includes a cloud service platform 100, a mobile edge computing terminal 200, and a data distribution and synchronization bus 300 connecting the two.

[0051] The cloud service platform 100 is deployed on a cloud server cluster and includes a core multi-constraint fusion scientific computing hybrid model 110. Model 110 employs a four-in-one constraint mechanism of "data pre-setting, boundary setting, threshold pre-setting, and data import," and integrates a recurrent neural network (RNN / LSTM / GRU) as a feature extractor and a decision tree algorithm model as a post-processing decision mechanism to construct a hybrid architecture model for intelligent analysis such as cost estimation, schedule prediction, and risk warning (the specific algorithm implementation will be described in detail in later embodiments). The management module includes an enterprise ledger system 120, an enterprise resource planning (ERP) system 130 (for managing information such as contracts, materials, personnel, and finances), a project ledger management module 131 (for recording real-time data such as project progress and completion), a cost warning module 132 (for real-time monitoring of cost deviations and triggering warnings), and a model optimization and distribution service module 140. This module is responsible for retraining or fine-tuning model 110 using newly gathered data and generating a lightweight version through pruning, quantization, and other techniques, ready for distribution.

[0052] The mobile edge computing terminal 200 is a smartphone used by on-site personnel (such as construction workers, material handlers, financial personnel, and other project and enterprise management personnel), on which a customized application (APP) 210 is installed. APP 210 integrates multiple functional units:

[0053] Network status awareness and mode switching unit 211: Real-time monitoring of network signals, and automatic seamless switching between online / offline modes according to preset strategies (such as three consecutive connection failures);

[0054] Multimodal data acquisition and preprocessing unit 212: calls hardware such as camera, GPS, and microphone to acquire image, positioning, and audio data, and automatically adds timestamps and watermarks to the acquired data;

[0055] Lightweight Edge Inference Engine 213: Loads and runs lightweight models downloaded from the cloud, providing on-site intelligent assistance, such as pre-approval of material acceptance compliance and estimation of engineering quantities;

[0056] Local Business Workflow and Rules Engine 214: It pre-configures standard business processes for enterprises, such as material acceptance process, labor task allocation management process, and machine shift management process, to guide users to operate step by step;

[0057] Semi-structured / structured form generation unit 215: Automatically generates semi-structured forms (such as pre-shift handover forms, intermediate node record forms) or complete structured forms (such as completion settlement forms) according to the business stage, and supports automatic association of multi-node forms;

[0058] Hardware-secure electronic signature / seal unit 216: calls the TEE or security chip of the mobile phone, and combines fingerprint or facial recognition to generate an offline signature and electronic seal that comply with the Electronic Signature Law.

[0059] Multi-node check-in and time recording unit 217: Supports check-in at multiple nodes such as morning work, lunch break, afternoon get off work, and overtime work, automatically loads check-in timestamps and location information, and associates them with the corresponding semi-structured forms.

[0060] Local encrypted database 218: Uses technologies such as SQLCipher to encrypt and store business documents to ensure data security;

[0061] Near-field cooperative communication unit 219: Based on Bluetooth or Wi-Fi Aware technology, it enables point-to-point data exchange with other terminals in offline mode, such as distributing task orders or verifying documents;

[0062] Intelligent data synchronization client (including team leaders, suppliers, etc.) 220: responsible for automatically synchronizing local encrypted data incrementally to the cloud after the network is restored, and receiving model update packages sent from the cloud.

[0063] The data distribution and synchronization bus 300 is responsible for all data flow. In online mode, it enables real-time message distribution; in offline mode, it acts as a synchronization agent, managing encrypted data packets uploaded by APP 210 and securely injecting them into the cloud platform. It is also responsible for pushing model update packages generated in the cloud and structured forms distributed through digital threads to designated APPs and system modules. Example 2: Material Acceptance Process in Edge Offline Mode

[0064] Combination Figure 2 This section details the process of material acceptance in a mountainous area project without network coverage, corresponding to steps S200-S330.

[0065] S201 Mode Switching: Material clerk Xiao Wang enters a remote mountainous area. His mobile app 210's network status sensing unit 211 detects a complete network loss and, according to a preset strategy, automatically and seamlessly switches the app from "online mode" to "edge offline mode." The app interface displays a message indicating that it is currently offline, and the data will be saved locally.

[0066] S202 Business Initiation: After Xiao Wang inspected the appearance, quality, quantity, and specifications of the incoming steel and confirmed that they met the requirements for this supply, he collected the relevant paper documents, then opened the APP and selected the "Material Acceptance" module. The local business workflow engine 214 loaded the pre-set acceptance template and automatically associated it with the current project's contract and list.

[0067] S203 Data Acquisition: Xiao Wang selects the "Steel - Rebar" purchase order and enters the actual quantity received as 18 tons. Acquisition unit 212 calls the camera to take pictures of the steel nameplate and on-site stacking evidence, and automatically records the current timestamp (2025-05-20 14:30:23) and the approximate location information within the work area calculated by base station or inertial navigation.

[0068] S204 Edge Intelligent Assistance: Before submission, the APP calls the lightweight edge inference engine 213 to conduct a compliance pre-review of the acceptance data. Based on the planned consumption of this construction section (15 tons planned for this week) and historical acceptance data, the pre-review model determines that the 18 tons accepted this time is within a reasonable range (considering material preparation) and gives a "normal acceptance quantity" prompt. If it exceeds the budget, it displays "abnormal usage quantity".

[0069] S205 Offline Electronic Signature: After Xiao Wang confirms that everything is correct, the signing process begins. Unit 216 prompts Xiao Wang to perform fingerprint verification. After successful fingerprint verification, Unit 216 uses the project department's electronic seal private key stored in the phone's security chip to sign the document's hash value, which contains all collected data, timestamps, and location information, generating an encrypted, tamper-proof offline electronic signature and attaching Xiao Wang's digital certificate.

[0070] S206 Local Encrypted Storage and Near-Field Distribution: The generated complete structured acceptance form (including signatures) is encrypted and stored in the local database 218. Simultaneously, Xiao Wang sends the acceptance form to the mobile apps of supplier representatives and related personnel within the same area via near-field communication or Bluetooth, completing the offline material acceptance process. Example 3: Offline, uninterrupted, intelligent management of the entire worker / machine operation process

[0071] This embodiment aims to illustrate how to achieve uninterrupted management of worker and machinery operations in a completely offline state through edge computing terminals. By generating semi-structured forms across multiple nodes and closing the loop to the final structured form, it ensures that labor and machinery management operations can continue even in a network-free environment. Once the network is restored, it automatically integrates into the global management system, achieving intelligent and visualized scientific management of engineering projects. The specific process is as follows: Figure 3 As shown.

[0072] Application Scenario: In a highway tunnel project, a labor team is responsible for pouring the secondary lining concrete, while a machinery team is responsible for the wet spraying robotic operation. The project site is located deep in the mountains, with no network signal available 24 / 7, and all operations are completed in an offline edge computing mode.

[0073] S1: Work Preparation: Technical / Safety briefing, task assignment, and semi-structured form generation

[0074] Timeframe: 7:00 AM on the same day, pre-shift meeting.

[0075] Operator: Construction worker Zhang.

[0076] Operating procedures:

[0077] After completing the pre-shift task assignment, safety briefing, and technical briefing for S311, Engineer Zhang opened the mobile app, which automatically entered offline mode: He went to "Manual Management" → "Task Assignment" and selected "Second Lining Team - Li Qiang Team". He filled in the task details, such as "Secondary Lining Concrete Pouring for Section ZK12+345~ZK12+365".

[0078] S312 Photo Evidence: Engineer Zhang took photos of the pre-shift meeting on-site, including: photos of the handover documents and group photos of the team members. The photos are automatically overlaid with timestamps (2025-3-25 7:02:32) and location information (relative position within the area calculated based on inertial navigation).

[0079] S313 generates a semi-structured form: The system packages the above structured fields (team, station number, volume, and handover content) with unstructured data (photos) to generate an initial semi-structured form: This form is encrypted and stored in Zhang Gong's mobile phone local database, and its status is "in progress".

[0080] S314 Near-Field Distribution: Zhang Gong uses Bluetooth to send a summary of the pre-shift handover form to team leader Li Qiang's mobile app with one click. After receiving the message, the team leader can view today's tasks offline and complete the paperless handover confirmation.

[0081] S315 Mechanical Team Parallel Operation: Mechanical Administrator Wang simultaneously assigns mechanical tasks and provides technical / safety briefings. He enters "Mechanical Management" → "Work Location", selects "Wet Spraying Robot - Operator Zhao", fills in the task content and planned number of shifts, takes photos of the equipment's arrival status, and generates a semi-structured form for mechanical operation. This form is encrypted and stored in Wang's local database on his mobile phone, with the status "in progress", and is then distributed to the operator's APP to complete the paperless briefing.

[0082] S2: During the operation: Manual management of multi-node semi-structured form generation

[0083] Time points: 12:00 (morning get off work end point), 14:00 (afternoon work start point); 18:00 (afternoon get off work end point), 19:00 (evening overtime work start point).

[0084] Operator: Construction worker Zhang.

[0085] Operating procedures:

[0086] S321 Midday Off-get off work Node Continues Semi-Structured Form Generation: At 12:00, Li Qiang's team completed their morning work and prepared to leave. Engineer Zhang opened the app and clicked "Continue Generating Intermediate Node Records" under the corresponding task sheet. The system automatically imported the previous semi-structured form. Engineer Zhang took photos of the morning work area and the team's departure, and filled in a brief summary of the morning's work (e.g., "Completed 16 cubic meters of concrete pouring"). The system generated an intermediate node semi-structured form, recording the work status, attendance information, and evidence photos for this stage, and stored it locally with encryption.

[0087] S322 Evening Overtime Node Semi-structured Form Generation: Due to work requirements, Li Qiang's team needs to work overtime until 23:00 (afternoon shift ends the same as S321). At 19:00, workers arrive on time. Zhang opens the APP, clicks "Overtime Check-in," fills in relevant information, takes photos for evidence, etc., and generates a semi-structured form that is saved locally. The system automatically switches to overtime mode. At 23:00, the work is completed, and Zhang clicks "Generate Intermediate Node Record." The system automatically records the overtime period (19:00-23:00), and Zhang takes photos of the nighttime work and the finished surface. The system generates an evening overtime node semi-structured form, recording the work status and evidence photos during the overtime period.

[0088] Parallel operation of the S323 mechanical team: At the end of each work phase, mechanical manager Wang also generates a corresponding intermediate node semi-structured form to record equipment operating status, spraying effect photos, oil gauge readings, etc.

[0089] S3: After get off work: Conduct post-shift summary, pricing, and generate structured forms.

[0090] Timeframe: 23:30 on the same day, post-shift summary.

[0091] Operator: Construction worker Zhang.

[0092] Operating procedures:

[0093] S331 Post-shift summary and data entry: Engineer Zhang and the team leader conduct a post-shift summary. Engineer Zhang opens the initial task sheet assigned in the morning in the APP, and the system automatically displays all related forms generated today: semi-structured form, lunchtime node record sheet, and evening overtime node record sheet.

[0094] S332 Completion Quantity Confirmation: Zhang Gong calculates and fills in the actual completion quantity or the number of working days for the day according to the form requirements.

[0095] S333 Automatic Pricing: The system automatically calculates the total output value of the work team for the day based on the preset unit price of the labor contract, combined with the clock-in time and completed volume at each node. The pricing process is completed offline locally.

[0096] S334 generates a complete structured form: The system integrates the semi-structured form, intermediate node record sheets, post-shift summary information, automatic pricing results, and all evidence photos to generate a complete structured form: the daily labor settlement sheet. This form fully records the data of the entire work group from before the start of the shift to after the end of the shift.

[0097] S335 Offline Electronic Signature: Engineer Zhang uses fingerprint verification to generate an offline electronic signature via a hardware security chip, which is then attached to the settlement document. The settlement document is encrypted and stored in a local database.

[0098] S336 Mechanical Team Parallel Operation: Engineer Wang also integrated the pre-shift structured forms and intermediate node record sheets of mechanical operations, combined with shift pricing, to generate a complete structured form: the daily mechanical settlement sheet, and performed offline electronic signature.

[0099] Overall Effectiveness: Despite the lack of internet access throughout the day, all aspects of worker and machinery management—from technical / safety briefings, task assignments, process recording to final settlement—were completed continuously and completely on the app, ensuring uninterrupted management even in offline conditions. All data was stored in layers and interconnected, forming a traceable, end-to-end data chain.

[0100] S400: Data Distribution: Digital Thread-Driven Visualization and Early Warning

[0101] Timeframe: After the project site personnel returned to their accommodation that evening and the network was restored.

[0102] Operator: The system will execute automatically.

[0103] S401 Incremental Synchronization: After Zhang's mobile phone connects to the Internet, the APP automatically starts the background synchronization service, and uploads all structured forms generated during the offline period (including the pre-shift structured forms and intermediate record sheets) to the cloud bus 300 in the order of transactions.

[0104] S402 Digital Thread One-Click Distribution: After receiving the data, the cloud platform decrypts, verifies, and reconstructs the data chain to restore the complete operational process. Then, through the digital thread, the final structured settlement statement is distributed in real-time to:

[0105] Team leader / supplier APP: Li Qiang received a push notification on his mobile phone and can view today's complete work record and his output value or salary details;

[0106] Project ledger system: Automatically adds a completed entry for "16 cubic meters of concrete for the secondary lining team", and automatically updates the project progress chart;

[0107] Enterprise accounting system: Automatically generates an accounts payable, and the financial system can view the project's labor cost expenditure in real time;

[0108] Cost early warning system: The system compares planned costs with actual costs to analyze whether the cost of that part is normal;

[0109] The person in charge's outbox: Mr. Zhang's APP "Sent Items" permanently saves this settlement form and all related intermediate node forms for future reference.

[0110] S403 Visual Dashboard Update: APP Dashboard: "Today's Labor Completion" bar chart added in real time; "Cumulative Cost Curve" rising; "Personnel Operation Heat Map" showing the activity trajectory of the work team in the area and its distribution at various time points, and the project progress Gantt chart progressing.

[0111] The beneficial effects of this embodiment:

[0112] Uninterrupted offline operation: Through a multi-node semi-structured form generation mechanism, it ensures that the daily management of workers / machinery is uninterrupted, undelayed, and undegraded even in a completely offline environment.

[0113] Full process traceability: From pre-shift briefing to intermediate recording and post-shift settlement, a complete data chain is formed, with each key node having forms, photos, timestamps, and location information as evidence.

[0114] Data structuring: The scattered process records of project construction are eventually compiled into a complete structured settlement statement, providing a high-quality data foundation for subsequent progress display, cost analysis, payment, efficiency assessment and risk warning.

[0115] Management visualization: After the network is restored, all process data is automatically integrated into the APP's visual dashboard, allowing management to grasp the complete dynamics of on-site operations in real time. Example 4: Data Synchronization, Digital Thread Distribution, and Model Evolution

[0116] Combination Figure 4 The detailed process of each step is explained in detail.

[0117] S401 Triggers Synchronization: That evening, when Engineer Zhang and Engineer Wang returned to their residence and their mobile phones connected to Wi-Fi or mobile signal, APP210's background service was automatically awakened and began executing the synchronization task.

[0118] S402 Incremental Data Upload: The synchronization client unit 220 first initiates a synchronization request to the cloud bus 300 and uploads all new business data packets (encrypted) since the last synchronization. These data packets are packaged in transaction order to ensure that the cloud can restore the business process in the correct order.

[0119] S403 Data Verification and Conflict Resolution: After receiving data, the cloud bus 300 decrypts it and sends it to the platform 100. Module 130 first verifies the validity of the offline electronic signature on each document. If it finds that the same material acceptance form already has a record automatically generated by the weighbridge system in the cloud, and the quantities are inconsistent (a conflict occurs), then automatic coordination is performed according to the preset rules (the same data classification and priority rules described in S400, with offline data as a backup, but both records are retained), and the coordination result is recorded in the log.

[0120] S404 Digital Thread Distribution: After successful verification, the system triggers the digital thread distribution mechanism to automatically push the structured form to various business systems (project ledger, enterprise ledger, cost warning, etc.) to achieve real-time data synchronization across multiple systems.

[0121] S405 Cloud-based Global Computation and Model Update: After merging all offline data, the cloud-based large model 110 is triggered to perform a global recalculation, updating the project's cost ledger and risk status. Simultaneously, the model optimization and distribution module 130 uses the newly added data to incrementally train model 110. After training, through knowledge distillation technology, a portion of the large model's "knowledge" is compressed into a lightweight student model, generating a differential update package.

[0122] S406 Model Distribution: The cloud pushes this lightweight differential update package to Zhang and Wang's APP 210 via bus 300. The APP loads the update package silently on the next launch or in the background, updating the local lightweight inference engine 213.

[0123] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art can make various improvements and modifications without departing from the spirit and principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A construction project management method based on edge computing and hybrid large-scale model adaptation, characterized in that, Collaborating with a cloud platform using mobile edge computing terminals (including smartphones, tablets, industrial handheld terminals, etc.) includes the following steps: S100: Pre-loads a lightweight edge inference model and initial project data into a smartphone app; S200: The APP automatically switches between cloud-based online collaboration mode and edge offline computing mode based on network conditions; S300: In edge offline computing mode, the APP independently executes the following business processing and local collaboration for construction site management; S310: Offline acceptance management, structured form generation, electronic signature and local storage for material arrival batch fulfillment orders; S320: Offline management records of labor operation batch performance, generation of multi-node semi-structured and structured settlement sheets, electronic signatures and local storage; S330: Offline management records of mechanical construction batch performance, generation of multi-node semi-structured and structured settlement sheets, electronic signatures and local storage; S340: In addition to the structured forms mentioned above, all multimodal data for on-site construction management during the construction process shall be recorded offline and temporarily stored locally. S400: When the network is restored, the APP automatically synchronizes offline business data incrementally to the cloud platform; the cloud platform performs data fusion and global analysis, and optimizes the edge model.

2. The method according to claim 1, characterized in that, In steps S310, S320, and S330, the generation of the electronic signature relies on the smartphone's hardware security chip, the device's unique identifier, and the operator's biometric information.

3. The method according to claim 1, characterized in that, In step S310, before generating the structured acceptance form, the lightweight edge inference model is invoked to perform a compliance pre-review of the acceptance data.

4. The method according to claim 1, characterized in that, In the edge offline computing mode, the acceptance form generated in step S310 or the settlement form generated in steps S320 and S330 can be distributed to the smartphones of other relevant parties on site via near-field wireless communication or Bluetooth.

5. The method according to claim 1, characterized in that, In step S400, incremental synchronization refers to uploading only new data generated during network interruption and performing transmission and integrity verification according to the order of service occurrence.

6. The method according to claim 1, characterized in that, In step S400, the optimized edge model refers to the cloud platform training and optimizing the multi-constraint fusion scientific computing hybrid model based on the fused data, and generating an update package by compressing part of its capabilities through model compression technology, which is then sent to a smartphone APP.

7. A construction project management system based on edge computing and hybrid large model adaptation for implementing the method of any one of claims 1-6, characterized in that: • The cloud service platform includes a multi-constraint fusion scientific computing hybrid large model module, an enterprise ledger system module, and a cost control early warning system module; • The mobile edge computing terminal is a smartphone with a customized APP installed. The APP includes a network status awareness module, an offline business processing module, a lightweight edge inference model engine, a local encrypted database, and a data synchronization module. • Data distribution and synchronization service, used to manage real-time online data distribution and asynchronous offline data upload; The offline business processing module supports on-site business management such as material acceptance and labor / machinery control, and can generate structured business forms with offline electronic signatures in offline mode; The mobile edge computing terminal's APP also integrates a near-field communication module, used for point-to-point data exchange with other terminals in offline mode.

8. A computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the construction project management method based on edge computing and multi-constraint fusion scientific computing hybrid large model adaptation as described in any one of claims 1-6.

Images