Intelligent Management Methods and Systems for the Entire Lifecycle of Labor Employment in Engineering Projects
By constructing a digital twin foundation based on BIM and GIS, combining VR simulation and smart wearable devices, configuring differentiated attendance strategies, and introducing a labor-specific earned value model, the problems of safety risks, data disconnect, and system scalability in labor management have been solved. This has enabled full-cycle data fusion and visualized control, improving security and decision support capabilities.
Patent Information
- Application Number
- CN202611147871.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-30
- Publication Date
- 2026-11-17
AI Technical Summary
The existing labor management system suffers from problems such as insufficient pre-entry control of laborers, significant safety risks, poor universality of attendance collection schemes, disconnect between salary settlement and project progress, superficial application of digital twins, and poor system scalability, resulting in insufficient data dimensions and inadequate decision support capabilities.
By integrating BIM and GIS technologies to build a digital twin foundation, combining WBS decomposition to generate an S-curve for labor cost input, introducing VR high-risk scenario simulation and smart wearable devices, configuring differentiated attendance strategies, introducing a labor-specific earned value model, realizing full-cycle data fusion and visual management and control, and adopting a decoupled architecture data platform for data governance and early warning.
It has achieved full-cycle safety risk prevention and control, closed-loop data storage, full-scenario attendance adaptation, dynamic cost early warning and credit asset accumulation, and improved the ability to prevent safety accidents, legal effect and decision support.
Smart Images

Figure CN123117592A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of smart construction site and digital management technology of construction engineering labor, and more specifically, relates to a smart management method and system for the entire life cycle of labor employment in engineering projects. Background Technology
[0002] The construction industry is a typical labor-intensive industry, and the management of laborers is a core aspect of on-site management, safety control, and cost control for engineering projects. Currently, the mainstream labor management model in the industry remains relatively rudimentary, generally relying on traditional methods such as paper-based file registration, manual offline sign-in, single attendance tracking via turnstiles and facial recognition, offline signing of paper labor contracts, and monthly manual salary calculation. This approach suffers from multiple industry pain points, including data fragmentation, lagging control, frequent disputes, and insufficient precision in cost control.
[0003] With the advancement of building information systems (BIS) and smart construction site technologies, information systems for real-name labor management have gradually emerged in the market. Existing related technical solutions mainly fall into three categories: The first is a basic real-name registration system, integrating only personnel identity entry, gate facial recognition attendance, and simple salary calculation functions; the second is a real-name platform with multiple subsystems, adding independent modules for project filing, team management, and salary settlement; the third is a personnel positioning solution integrating digital twins, relying on a BIM twin base and positioning hardware to achieve simple on-site personnel location display. While these existing technologies can achieve basic personnel information registration, simple on-site positioning, and basic salary calculation, they still have inherent technical defects that cannot be avoided in actual project implementation, specifically manifested in the following five aspects: (1) The pre-entry management of laborers is too simplistic and lacks prevention and control at the source of safety risks. The existing real-name system only collects basic identity information such as ID cards and facial recognition data. It does not connect to the enterprise's historical labor database or the government's labor supervision platform to retrieve personnel's historical records of violations, poor performance, and illegal employment. There is no automatic screening and early warning mechanism. At the same time, there is no intelligent matching and assessment method for personnel's physiological condition, psychological stress resistance and the job they are applying for. For high-risk positions such as high-altitude, deep foundation pits and confined spaces, it is impossible to identify personnel with insufficient physical endurance and poor psychological stress resistance in advance. This can easily lead to personnel with potential safety hazards "entering the site with illness", creating potential safety hazards on site.
[0004] (2) The safety education, job evaluation, and employment contract management chain is fragmented, and there is no complete and traceable electronic evidence chain. Current safety training often uses offline centralized lectures or simple VR video demonstrations, only recording training attendance information. It cannot collect and quantify physiological and psychological feedback data of personnel under high-risk scenario simulations in real time, and cannot objectively determine the training absorption effect and the suitability of personnel for their positions. Labor contracts are mainly signed on paper, and training records, personnel evaluation results, and employment contracts are stored independently with no data linkage. Once labor disputes or safety accident liability determination occur, a complete and tamper-proof integrated electronic evidence chain cannot be provided, making it difficult to provide evidence.
[0005] (3) The attendance collection scheme has poor universality and insufficient ability to integrate and analyze multi-source data. Existing attendance solutions heavily rely on facial recognition at gates at closed construction site entrances and exits, and are only suitable for point-like closed projects such as building construction and factories. For linear projects without fixed closed boundaries, such as highways, railways, and pipelines, they cannot achieve full coverage of personnel attendance. Furthermore, attendance data is only used for the single purpose of payroll calculation and is not linked and integrated with data on personnel's on-site work trajectories, work progress, and work efficiency for analysis, resulting in low data value mining.
[0006] (4) Salary settlement is completely disconnected from project progress and cost control system, and there is a lack of dynamic employment risk prediction capability. The existing system's payroll calculation relies solely on attendance hours and does not link to project work breakdown structure (WBS), enterprise construction quotas, or planned labor cost input curves. This makes it impossible to dynamically compare actual labor costs with the project's planned labor costs. Furthermore, it does not introduce earned value management methods to quantitatively assess the rationality of labor cost inputs. It can only calculate labor costs after the fact and cannot provide early warnings when labor costs exceed the budget or when labor resources are misallocated. This makes it difficult to support refined cost control and dynamic resource allocation for projects.
[0007] (5) The application of digital twins is superficial and cannot achieve closed-loop visual traceability of the entire life cycle of laborers. Existing digital twin construction site solutions only focus on three-dimensional display of construction progress and single-point positioning of personnel on site. They do not integrate the business data of the entire process of labor personnel entry verification, on-site operation, salary settlement, and exit archiving. They cannot fully map the entire life cycle status of a single person from entry, on-the-job operation, salary payment to exit archiving in the twin model, nor do they support cross-project retrieval of personnel's performance and violation history files over many years, resulting in a lack of digital traceability capabilities for the entire life cycle of personnel.
[0008] In addition, existing labor management systems generally adopt an integrated architecture that is highly coupled with business functions, underlying data processing, AI computing, and early warning push. Data cleaning, model training, alarm distribution and front-end business modules are deeply bound together, resulting in poor system scalability, high iteration and maintenance costs, and an inability to flexibly adjust early warning rules, AI evaluation models and business processes according to different project types and management needs.
[0009] Therefore, there is an urgent need for a technical solution that can integrate multi-source data, adapt to various engineering scenarios, and realize digital closed-loop management of the entire labor employment cycle, in order to solve the technical problems of the above-mentioned non-closed control chain, insufficient data dimensions, and insufficient decision support capabilities. Summary of the Invention
[0010] To address the aforementioned problems in existing technologies, this invention provides a method and system for intelligent management of the entire labor employment cycle in engineering projects. It integrates BIM and GIS technologies to construct a digital twin foundation, generating an S-curve of labor cost input based on WBS decomposition and quota data as a cost control benchmark. In the on-site entry phase, it connects with enterprise and government regulatory platforms to screen resumes, and utilizes VR high-risk scenario simulation combined with smart wearable devices to collect physiological parameters. An AI model quantifies job matching, and electronic signatures are used to construct traceable electronic files. In the operational phase, differentiated attendance strategies are configured for point-like and linear projects, respectively, using gate eagle-eye fusion and drone mobile terminal splicing. This invention generates standardized time-series attendance datasets and maps them to digital twin trajectories. In the settlement phase, it links attendance data with an S-curve benchmark, introduces a labor-specific earned value model, and combines actual project output with deviation analysis and tiered early warning. In the exit phase, it fully verifies performance data and archives it to form labor credit assets that can be accessed across projects. Through a decoupled architecture between the business application layer and the underlying data platform, this invention achieves real-time fusion, intelligent analysis, and visualized management of labor employment data throughout the entire lifecycle, from entry, operation, settlement to exit. This results in proactive prevention of employment safety risks, closed-loop evidence preservation, legal validity, adaptation to all attendance scenarios, dynamic early warning of cost deviations, and the accumulation of corporate credit assets.
[0011] To achieve the above objectives, one aspect of the present invention provides a method for intelligent management of the entire lifecycle of labor use in engineering projects, comprising the following steps: S1: Acquire multi-dimensional basic data of the project, integrate BIM 3D geometric data and GIS geospatial data to build a full-domain digital twin foundation; generate the S-curve of labor cost input throughout the project cycle based on WBS work breakdown structure, enterprise and market labor quota data, as a benchmark for dynamic management and control of labor costs. S2: Receive applications from workers to enter the site, and perform cross-project resume verification and government platform real-name verification in parallel. Combine VR high-risk scene physiological time sequence collection and AI stress adaptation assessment to build an integrated entry access mechanism that includes resume screening, quantitative determination of personnel and positions, and electronic signing, forming an unalterable full-link electronic file for personnel. S3: Automatically identify project types, distinguish between point-like closed projects and linear open projects, and configure differentiated multi-source attendance collection strategies; integrate multi-source attendance data to generate a standardized time-series attendance dataset; map the spatial coordinates contained in the dataset to the digital twin base to generate dynamic work trajectories; S4: Based on standardized time-series attendance data, calculate the actual labor cost of WBS work packages, link the full-cycle labor cost input S-curve benchmark to calculate cost deviation, introduce a labor-specific earned value model, combine the actual output value of sub-projects to calculate the cost performance index (CPI), configure independent two-level risk thresholds for point-like closed projects and linear open projects respectively, and push labor risk warnings in a hierarchical manner based on the joint judgment results of deviation rate and labor cost input performance index. S5: Respond to personnel departure requests by fully verifying attendance, salary, and material performance data. After salary settlement, generate encrypted electronic departure vouchers, mark personnel departure status on the digital twin base, and archive all project performance data to the enterprise labor credit database to generate personnel credit profiles that can be retrieved across projects. S6: Map the full-cycle business data flow from steps S1 to S5 to the digital twin platform in real time to build an integrated interactive interface that supports full-cycle traceability for individual soldiers, cross-project credit queries, and visualization of multi-dimensional cost indicators. S7: Set up an independent data platform that is decoupled from all upper-level business steps to complete unified governance of all heterogeneous data, incremental iteration of AI models, and precise push of risk stratification and classification across the entire chain.
[0012] Furthermore, the multidimensional basic data of the project in step S1 includes: geometric space dimension data, construction organization dimension data, economic quota dimension data, and schedule plan dimension data; The geometric spatial dimension data includes at least the project's BIM 3D geometric data and GIS geospatial data; The construction organization dimension data includes at least construction drawings and construction organization design documents; The economic quota dimension data should include at least the company's internal construction quotas and the latest labor unit prices for different trades in the project location. The schedule dimension data includes at least a list of work packages generated based on the WBS hierarchical breakdown, and the time milestone nodes bound to each work package.
[0013] Furthermore, the generation of the S-curve for full-cycle labor cost input in step S1 specifically includes: Acquire BIM 3D geometric data and GIS geospatial data of the target project, perform coordinate registration and fusion of the two, and construct a high-precision digital twin base covering the entire project area; Import the construction drawings, construction organization design documents, internal construction quotas, and the latest labor unit prices for different trades in the project location into the digital twin base; perform WBS hierarchical decomposition on the project scope, decompose the project into independent work packages at each level, and bind the planned start time, planned end time, and key milestone nodes to each work package; By combining construction quotas, labor unit prices, and the schedule milestones of each work package, the labor costs required for each milestone are calculated cumulatively over time, generating a cumulative S-curve of labor cost input throughout the project lifecycle.
[0014] Furthermore, step S2 specifically includes: The system acquires the facial biometrics of workers through facial recognition devices and obtains their real-name information through OCR recognition or NFC reading devices. It also connects to the company's historical labor database and the government-level labor real-name supervision platform to cross-verify the authenticity of the workers' identities and automatically screens whether the workers have any cross-project historical violations, poor performance records, or industry bans. If high-risk records are found, an entry interception warning is immediately generated and pushed to the security management terminal. Those who fail the review are prohibited from entering the site. Workers who pass the initial screening are guided into the VR safety education experience hall, where high-risk work scenarios are simulated based on the types of jobs they have applied for. During the immersive experience, the smart wearable devices worn by the workers are connected via Bluetooth to collect their physiological parameters such as heart rate and blood pressure fluctuations in different stress scenarios in real time. The collected time-series physiological parameters are input into a pre-trained AI analysis model to assess the psychological adaptability, stress response ability, and physical endurance of workers in a simulated high-risk environment, and output a quantitative job matching assessment report. For personnel who pass the matching assessment, their identity data, violation screening records, VR training records, physiological parameter logs, AI assessment reports, and signed electronic contracts are linked and packaged together, and blockchain technology or timestamp technology is used for evidence preservation to form an immutable and traceable electronic entry file.
[0015] Furthermore, step S3 specifically includes: Identify the type of engineering scenario and configure targeted attendance collection strategies; if it is a point-like closed project, enable the dual verification mode of turnstile + eagle eye; if it is determined to be a linear open project, enable the wide-area stitching mode of drone + mobile terminal + along-line video; through multi-sensor data fusion algorithm, stitch together to generate continuous position information of personnel on long-distance linear work surface. Each standardized attendance record generated is spatially correlated with the BIM work surface division data of the construction area to parse out the specific work position or construction section where the person is currently located and extract their real-time spatial coordinates; the output is a standardized time-series attendance dataset containing the fields of person ID, timestamp, spatial coordinates and work package. Based on the standardized time-series attendance dataset, the spatial coordinates contained in the dataset are mapped to the BIM+GIS digital twin base constructed in step S1 through a real-time data interface, driving the virtual personnel model in the twin space to move synchronously and generating a dynamic work trajectory consistent with the physical world.
[0016] 2. Further, step S4 specifically includes: Retrieve standardized time-series attendance datasets, associate them with the salary calculation rules in electronic labor contracts to automatically calculate and summarize individual salary details; according to the WBS work package affiliation, collect and summarize the salary details of the personnel to which they belong, calculate the total actual labor cost of each work package within a specific time period, and form a dataset of actual labor cost expenditure for the project. Retrieve the S-curve of labor cost input for the entire project cycle and extract the planned labor cost value corresponding to the current statistical period; compare the actual labor cost expenditure dataset AC with the planned labor cost value PV, and calculate the absolute deviation value SV and the relative deviation rate SV%; SV = PV - AC; relative deviation rate SV% = (PV - AC) / PV × 100%; The earned value management approach is adopted, and the actual completed output value (EV) of each sub-project is introduced as a key variable; a labor-specific earned value cost assessment model is constructed to calculate the labor cost input performance index (CPI); CPI = EV / AC, where AC is the actual labor cost; Two independent early warning thresholds are set according to the type of project: For point-based closure projects, the preset first-level mild warning threshold range is: And the relative deviation rate of labor costs For linear open engineering, a preset first-level mild early warning threshold range is defined as follows: And the relative deviation rate of labor costs When the deviation rate Labor cost input performance index If the cost falls within the corresponding threshold range, it is determined to be a slight fluctuation in labor costs. Cost reminder information is automatically generated, highlighting the current abnormal work packages and personnel attendance density. Lightweight messages are pushed to the mobile terminals of on-site construction workers and labor managers to remind them to pay attention to on-site labor dynamics and to conduct on-site review or fine-tuning when necessary. For point-based closure projects, a pre-set Level II severe warning trigger condition is established: Or relative deviation rate of labor costs For linear open engineering projects, a pre-defined level-two severe early warning trigger condition is established: Or relative deviation rate of labor costs When the deviation rate Labor cost input performance index If any one of the threshold conditions is met, it is determined as a project-level overspending alarm; an in-depth analysis report containing a preliminary diagnosis of the cause of the deviation is automatically generated and pushed to the project manager and the company's cost control center.
[0017] Furthermore, step S5 specifically includes: Automatically correlate and query the full-dimensional business data of workers to be discharged in this project, and construct a discharge verification checklist; The system automatically calculates and confirms that all payments have been settled based on the verification checklist; after confirmation, it generates an encrypted electronic exit certificate containing a unique hash value or digital signature. The virtual twin status of the corresponding worker to be discharged in the digital twin base is permanently marked from "on duty" to "discharged", the real-time mapping update of the worker's work trajectory is terminated, and the worker to be discharged is grayscaled or moved into the historical personnel layer in the visualization interface. All business data generated by the workers to be discharged from the project are packaged and archived, and synchronously written into the enterprise's global labor credit database to form credit archive data entries corresponding to the workers to be discharged. The credit archive data entries are strongly linked to the unique identifiers of the workers to be discharged, forming a dynamically updated labor credit profile.
[0018] Furthermore, step S6 specifically includes: A high-throughput message middleware is built as a real-time data synchronization channel; the multi-source heterogeneous data collected in steps S1-S5 is processed by streaming ETL and then synchronously written into the digital twin base. On top of the digital twin platform, a multi-dimensional visual interactive interface is generated, rendering a heat map of personnel at the construction site, the density of job distribution, the concentration of personnel in key areas, and the real-time employment status of each WBS work package. The interface allows managers to initiate a full-cycle traceability query for a single person by clicking on the virtual personnel model in the twin interface or entering the personnel's identity information. The interface also supports cross-project and cross-cycle data fusion queries.
[0019] Furthermore, step S7 specifically includes: A layered and decoupled architecture is adopted, encapsulating all business logic in steps S1 to S6 into a business application layer, while data governance, model calculation, and early warning distribution are separated into an independent data platform; the two layers interact with each other through a standardized API gateway. The data platform uniformly receives raw perception data and derived business data from all stages; for multi-source heterogeneous data, it performs standardized cleaning processes and data anonymization; the cleaned data is classified and stored according to subject domains, and a distributed data warehouse is built to form a unique data asset pool for the project. Through the built-in algorithm engine of the data platform, and by utilizing the massive amount of historical data accumulated, the AI stress assessment model in step S2 and the labor-specific earned value cost model in step S4 are continuously incrementally learned and their parameters are optimized. By using a unified rules engine built into the data platform, the data flow of the entire business chain can be monitored in real time to identify multiple types of risk events; differentiate between point-like and linear engineering risk control matrices, classify the identified risk events into levels, and push them to the corresponding terminals in layers according to operational, management, and strategic levels.
[0020] A second aspect of the present invention provides a smart management system for the entire lifecycle of labor employment in engineering projects, used to implement the aforementioned smart management method for the entire lifecycle of labor employment in engineering projects, comprising: The project basic data acquisition module is used to acquire multi-dimensional basic data of the project, integrate BIM 3D geometric data and GIS geospatial data to build a full-domain digital twin foundation; and generate the S-curve of labor cost input throughout the project cycle based on WBS work breakdown structure, enterprise and market labor quota data, as a benchmark for dynamic management and control of labor costs. The multi-dimensional personnel entry control module is used to receive labor personnel entry applications, and to perform cross-project resume verification and government platform real-name verification in parallel. It combines VR high-risk scenario physiological time sequence collection and AI stress adaptation assessment to build an integrated entry access mechanism that includes resume screening, quantitative determination of personnel and positions, and electronic signing, forming an unalterable full-link electronic file for personnel. The operation data acquisition and trajectory construction module is used to automatically identify the project type, distinguish between point-like closed projects and linear open projects, configure differentiated multi-source attendance acquisition strategies, integrate multi-source attendance data to generate a standardized time-series attendance dataset, and map the spatial coordinates contained in the dataset to the digital twin base to generate a dynamic operation trajectory. The payroll calculation and cost early warning module is used to calculate the actual labor cost of WBS work packages based on standardized time-series attendance data, link the full-cycle labor cost input S-curve benchmark to calculate cost deviation, introduce a labor-specific earned value model, and calculate the cost performance index (CPI) based on the actual output value of sub-projects. It configures independent two-level risk thresholds for point-like closed projects and linear open projects respectively, and pushes labor risk early warnings in a tiered manner based on the joint judgment results of deviation rate and labor cost input performance index. The exit closed-loop archiving module is used to fully verify attendance, salary, and material performance data in response to personnel exit requests. After salary is settled, an encrypted electronic exit voucher is generated, the personnel exit status is marked on the digital twin base, and the entire project performance data is archived to the enterprise labor credit database to generate personnel credit profiles that can be retrieved across projects. The digital twin visualization building module is used to map the full-cycle business data flow to the digital twin base in real time, and build an integrated interactive interface that supports full-cycle traceability for individual soldiers, cross-project credit query and visualization of multi-dimensional cost indicators. The data platform processing and early warning module is used to set up an independent data platform that is decoupled from all upper-level business steps, and to complete the unified governance of all heterogeneous data, incremental iteration of AI models, and precise push of risk stratification and classification across the entire chain.
[0021] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: (1) This invention breaks through the limitation of existing real-name systems that only collect basic identity information. By using facial recognition technology to connect the enterprise's historical database with the government's regulatory platform, it can automatically screen cross-project historical performance and violation records. In particular, it introduces VR high-risk scenario simulation in conjunction with smart wearable devices to collect physiological parameters. It uses pre-trained AI models to quantitatively assess the psychological adaptability and stress response capabilities of personnel, realizing the essential safety transformation from "identity compliance" to "person-job fit". It effectively solves the safety hazard of "entering the site with illness" and provides a quantitative basis for the prevention of safety accidents.
[0022] (2) This invention links VR safety training, physiological parameter collection, AI job suitability assessment, and electronic contract signing into an irreversible closed-loop process, and uses blockchain or timestamp technology to associate, encapsulate, and solidify identity data, training records, assessment reports, and contract texts. This mechanism not only achieves objective quantification of training effectiveness, but also constructs a judicial-grade electronic evidence chain covering the entire process of "entry-operation-exit," greatly enhancing the legal effect of labor dispute resolution and safety accident liability determination.
[0023] (3) For point-like closed projects, this invention adopts a dual verification mode of "gate face recognition + Eagle Eye visual tracking (ReID)" to eliminate proxy attendance data through spatiotemporal alignment; for linear open projects, it pioneers a multi-source heterogeneous data splicing and fusion algorithm of "inspection drone + mobile terminal positioning + along-line video recognition" to solve the problem of continuous position perception in scenarios without fixed entrances and exits. This differentiated adaptation strategy breaks the dependence of traditional attendance on physical fences and achieves full coverage and high-precision attendance for all types of projects.
[0024] (4) This invention extends labor management down to the dimension of project cost control, creatively comparing standardized attendance data with the "labor cost input S-curve" generated based on WBS decomposition; further, it introduces a labor-specific earned value assessment model (CPI) to identify deviations such as "inefficient overspending" and "benign overspending" based on the actual output value of the project. This mechanism enables managers to grasp the labor input-output ratio in real time, realizing the transformation from a simple "payroll tool" to a "cost risk control center," providing data-driven decision support for resource allocation and process improvement.
[0025] (5) In the exit process, this invention verifies all performance data and archives the full-cycle business data of personnel in this project (including entry assessment, trajectory, salary, violations, etc.) into the enterprise's global labor credit database with a unique identity. By constructing a cross-project callable labor credit profile, a credit joint prevention mechanism of "one breach of trust, network-wide early warning" is realized, and the leap from single-project personnel management to enterprise-level refined management of human resources assets is completed.
[0026] (6) This invention adopts a system architecture that separates the business application layer from the underlying support layer. Through an independent data middle platform, it uniformly processes the data across the entire chain, which not only ensures the consistency and security of data standards, but also realizes independent incremental learning and parameter tuning of the AI stress assessment model and the earned value cost model. This architecture solves the stubborn problems of high coupling and poor scalability between traditional system modules, and gives the system the ability to continuously iterate and evolve, adapting to the long-term operation and maintenance needs of large-scale engineering projects. Attached Figure Description
[0027] Figure 1 This is a schematic diagram of the structure of a smart management system for the entire lifecycle of labor employment in an engineering project, according to an embodiment of the present invention. Figure 2 This is a flowchart illustrating a method for intelligent management of the entire labor lifecycle of an engineering project, as described in an embodiment of the present invention. Detailed Implementation
[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.
[0029] To address the technical shortcomings of existing engineering labor management systems, such as lagging on-site safety control, narrow applicability to attendance scenarios, disconnect between wages and costs, shallow twin visualization, and difficulty in iterating on a tightly coupled system, etc. Figure 1 As shown, the first aspect of the present invention provides a smart management system for the entire lifecycle of labor employment in engineering projects, comprising: The project basic data acquisition module is used to acquire multi-dimensional basic data of the project, integrate BIM 3D geometric data and GIS geospatial data to build a full-domain digital twin foundation; and generate the S-curve of labor cost input throughout the project cycle based on WBS work breakdown structure, enterprise and market labor quota data, as a benchmark for dynamic management and control of labor costs. The multi-dimensional personnel entry control module is used to receive labor personnel entry applications, and to perform cross-project resume verification and government platform real-name verification in parallel. It combines VR high-risk scenario physiological time sequence collection and AI stress adaptation assessment to build an integrated entry access mechanism that includes resume screening, quantitative determination of personnel and positions, and electronic signing, forming an unalterable full-link electronic file for personnel. The operation data acquisition and trajectory construction module is used to automatically identify the project type, distinguish between point-like closed projects and linear open projects, configure differentiated multi-source attendance acquisition strategies, integrate multi-source attendance data to generate a standardized time-series attendance dataset, and map the spatial coordinates contained in the dataset to the digital twin base to generate a dynamic operation trajectory. The payroll calculation and cost early warning module is used to calculate the actual labor cost of WBS work packages based on standardized time-series attendance data, link the full-cycle labor cost input S-curve benchmark to calculate cost deviation, introduce a labor-specific earned value model, and calculate the cost performance index (CPI) based on the actual output value of sub-projects. It configures independent two-level risk thresholds for point-like closed projects and linear open projects respectively, and pushes labor risk early warnings in a tiered manner based on the joint judgment results of deviation rate and labor cost input performance index. The exit closed-loop archiving module is used to fully verify attendance, salary, and material performance data in response to personnel exit requests. After salary is settled, an encrypted electronic exit voucher is generated, the personnel exit status is marked on the digital twin base, and the entire project performance data is archived to the enterprise labor credit database to generate personnel credit profiles that can be retrieved across projects. The digital twin visualization building module is used to map the full-cycle business data flow to the digital twin base in real time, and build an integrated interactive interface that supports full-cycle traceability for individual soldiers, cross-project credit query and visualization of multi-dimensional cost indicators. The data platform processing and early warning module is used to set up an independent data platform that is decoupled from all upper-level business steps, and to complete the unified governance of all heterogeneous data, incremental iteration of AI models, and precise push of risk stratification and classification across the entire chain.
[0030] Furthermore, the project basic data acquisition module includes: a digital twin base construction submodule, a construction plan generation submodule, an economic quota submodule, a labor unit price submodule, a WBS decomposition submodule, and an S-curve generation submodule; The digital twin base construction submodule is used to acquire the BIM 3D geometric data and GIS geospatial data of the target project, and to perform coordinate registration and fusion of the two to construct a BIM+GIS digital twin base covering the entire project area, providing a spatial carrier for personnel trajectory mapping and visualization. The construction plan generation submodule is used to import construction drawings and construction organization design documents (including at least the schedule and construction process plan) to define the physical construction logic and on-site operation specifications of the project. The economic quota submodule is used to obtain internal construction quota data of enterprises (covering standard labor efficiency, work type ratio parameters and machine shift quotas for each type of work). The labor unit price submodule is used to obtain the latest labor market unit price data for different types of work in the project location, and to store and manage it according to the type of work. The WBS decomposition submodule is used to break down the project scope into independent work packages based on the WBS (Work Breakdown Structure), and to bind the planned start time, planned end time and key milestone nodes to each work package; The S-curve generation submodule is used to calculate the cumulative labor cost input at each time node according to the time dimension based on the WBS decomposition results, construction quota data, labor unit price and each work package schedule node, and generate the cumulative S-curve of labor cost input for the whole project cycle (as a benchmark for dynamic management and control of labor costs). Before construction begins and after the completion of the construction drawings, acquire and store multi-dimensional basic data of the project to build a data benchmark for the full life cycle management of labor services: construct a digital twin foundation by integrating the project's BIM 3D model and GIS geospatial data; import construction drawings and construction organization design documents containing schedule plans and construction processes; retrieve internal construction quotas covering standard labor efficiency and ratio parameters for each type of work, as well as local labor market unit price data categorized by type of work; based on the WBS work breakdown structure, break down the project into work packages that are associated with key milestone nodes, and combine construction quotas, labor unit prices, and the planned schedule nodes of each work package to calculate the cumulative labor cost input at each node according to the time dimension, generating a cumulative S-curve of full life cycle labor cost input as a dynamic comparison benchmark; Furthermore, the multi-dimensional personnel entry control module includes: a face verification and resume screening subunit, a safety education VR subunit, a physiological parameter collection subunit, an AI job suitability assessment subunit, and an electronic signing and file packaging subunit; The facial verification and resume screening subunit is used to collect the facial biometric information of workers and compare it with the enterprise's historical labor database and the national or provincial labor real-name supervision platform to automatically retrieve and update the personnel's historical performance data and violation records, and generate an entry interception warning when high-risk criminal records or age limits are identified. The VR safety education sub-unit is used to guide workers who have passed the initial screening into the VR safety education experience hall to simulate at least one high-risk work scenario among high-altitude work, deep foundation pit work, and confined space work, so as to carry out visual safety training. The physiological parameter acquisition subunit is used during VR training to connect via Bluetooth to the smart wearable devices worn by workers, and to collect real-time physiological parameters such as heart rate and blood pressure in corresponding high-risk work scenarios. The physiological parameters are then spatiotemporally correlated and recorded with specific simulated scenarios. Differentiated physiological safety thresholds are preset for different high-risk jobs: the upper limit of resting heart rate for high-altitude work is 110 beats / minute, and diastolic blood pressure is ≤95mmHg; the upper limit of heart rate for deep foundation pit work is 120 beats / minute. The AI-powered job suitability assessment subunit inputs collected time-series physiological parameters into a pre-trained AI analysis model. Based on massive amounts of historical data, it assesses workers' psychological adaptability, stress response, and physical endurance in a simulated environment, outputs a quantitative job matching assessment report, and determines eligibility for entry or recommends job reassignment based on the assessment results. The AI analysis model is initially pre-trained based on over 100,000 cross-project personnel time-series physiological data points combined with job suitability tags. The system automatically triggers incremental model learning when more than 500 new personnel physiological data points are added monthly, synchronously updating the physiological thresholds for each job type. When a worker's physiological indicators exceed the preset threshold for their corresponding job type three consecutive times within the VR scenario, the system directly determines job incompatibility and blocks their entry. The electronic signing and file encapsulation subunit is used to guide workers to complete the online signing of electronic labor contracts on integrated terminal devices or mobile devices after the assessment is deemed qualified. It also associates, encapsulates, and secures personnel identity data, violation screening records, VR training records, physiological parameter logs, AI assessment reports, and signed electronic contracts to form an unalterable and traceable electronic file for entry. The multi-dimensional personnel entry control module implements full-chain management of laborers entering the site, integrating multi-source identity verification and job suitability quantitative assessment functions: Facial biometrics are obtained through a face capture and comparison unit, and cross-connected with the enterprise's labor database and the government's real-name supervision platform to cross-screen historical performance records, violation records, and age compliance, automatically generating entry interception warnings for high-risk personnel; after initial screening by the job matching assessment unit, personnel are guided into a VR safety education unit, where, in simulated high-risk work scenarios such as high-altitude, deep foundation pits, and confined spaces, the physiological parameter collection unit collects stress data such as heart rate and blood pressure through smart wearable devices; the AI analysis and assessment unit inputs the aforementioned physiological data into a pre-trained model to assess the personnel's psychological adaptability and stress response capabilities, outputting a job suitability report; finally, the electronic contract signing unit guides qualified personnel to complete the online signing of electronic labor contracts on an integrated terminal or mobile device, and encapsulates identity data, violation screening records, training physiological parameters, assessment reports, and contract text together to form an unalterable and traceable electronic file.
[0031] Furthermore, the operation data acquisition and trajectory construction module includes: a camera acquisition subunit, an attendance strategy configuration subunit, a personnel positioning and association subunit, an operation trajectory mapping subunit, and an attendance data governance subunit; The video acquisition subunit is used to deploy eagle-eye camera equipment in the construction site and use computer vision algorithms to analyze the video stream in real time to identify the facial features, behavior and movement trajectory of the workers. The attendance strategy configuration subunit is used to identify the type of engineering scenario and configure differentiated attendance logic. For point-like closed projects, the gate facial recognition data is used as the core data source, and the Eagle Eye ReID visual tracking trajectory is used as an auxiliary verification. The gate facial data and Eagle Eye visual tracking data are fused, and spatiotemporal alignment and cross-validation are performed to generate high-confidence attendance records. If the gate record and the Eagle Eye trajectory within the site do not match within 15 minutes for the same person, it is judged as a proxy attendance and the attendance data is removed. For linear open projects, the inspection drone imagery, mobile smart terminal RTK positioning, and along-line video stream data are fused, and continuous location information is generated by stitching together through a multi-sensor data fusion algorithm. The drone only operates in sunny and windless environments. In rainy or snowy weather, the along-line video + RTK mobile terminal positioning is used as the main method. If the positioning coordinates drift by more than 5m, the preceding and following time-series coordinate interpolation correction is automatically adopted. When fusing multi-source data, the RTK positioning weight is 0.6, the drone visual weight is 0.3, and the along-line video weight is 0.1. The weighted calculation generates standardized spatial coordinate position information. The personnel location and association sub-unit is used to spatially associate standardized attendance records with BIM work surface division data of the construction area, analyze the specific work position or construction section where the personnel are located, and extract real-time spatial coordinates. ; The work trajectory mapping sub-unit is used to map real-time spatial coordinates to the BIM+GIS digital twin base through digital twin technology, driving the synchronous movement of personnel models in the virtual space to dynamically display the spatial distribution and movement trajectory of workers. The attendance data governance subunit is used to clean, deduplicate and format multi-source heterogeneous attendance data, and output a standardized time-series attendance dataset containing fields such as "personnel ID-timestamp-spatial coordinates-work package". This dataset is then associated with the work trajectory data and stored for use by the payroll calculation module. The operation data acquisition and trajectory construction module performs full-scene adaptive management of on-site operations and attendance, configuring differentiated perception strategies for different engineering forms: the camera acquisition unit uses eagle-eye equipment and computer vision algorithms to identify personnel characteristics and behaviors in real time; for point-like closed projects such as building construction, the gate attendance unit integrates gate face recognition and eagle-eye visual tracking dual data sources, performing spatiotemporal alignment and cross-validation to generate high-confidence attendance; for linear open projects such as highways, the drone inspection unit integrates inspection drone images, mobile smart terminal positioning, and roadside video stream data, and generates high-confidence attendance through multi-sensor stitching. The system generates continuous location information to address blind spots in attendance tracking; the personnel identification and positioning unit associates the acquired location information with the BIM work surface division data to analyze the specific workstations of personnel; the work trajectory generation unit uses digital twin technology to map spatiotemporal data onto the BIM+GIS base, driving the virtual model to move synchronously to dynamically display personnel distribution and movement trajectories; the attendance management unit finally cleans and merges multi-source heterogeneous data, outputting a standardized time-series attendance dataset containing fields such as "personnel ID-timestamp-spatial coordinates-work package", and associates and stores the work trajectory data as a unified benchmark for payroll calculation and cost analysis.
[0032] Furthermore, the payroll calculation and cost early warning module includes: an automatic payroll calculation subunit, an actual labor cost statistics subunit, an S-curve comparison subunit, an earned value management analysis subunit, and an employment risk early warning subunit; The automatic pay calculation subunit is used to retrieve the standardized time-series attendance dataset, associate it with the pay calculation rules agreed upon by the electronic contract signing unit, automatically calculate the basic salary, overtime pay and various allowances and subsidies of the workers, and summarize and generate individual pay details; The actual labor cost statistics subunit is used to collect and summarize the individual salary details according to the WBS work package attribution and time dimension, and generate a dataset of actual labor cost expenditures that reflects the actual investment of the project. The S-curve comparison subunit is used to retrieve the full-cycle labor cost input S-curve generated by the project basic data construction module, extract the planned labor cost value for the current statistical period, dynamically compare the actual labor cost expenditure dataset with the planned value, and calculate the absolute deviation value and relative deviation rate. The Earned Value Management (EVM) analysis sub-unit is used to introduce the actual completed output value of sub-projects, construct a labor-specific EVM assessment model, calculate the labor cost input performance index, and comprehensively quantify the labor matching degree in combination with the deviation direction to identify the nature of deviations such as "inefficient overspending" or "efficient cost saving." It also adds three types of abnormal labor judgment rules: if the actual number of employees present exceeds the corresponding WBS work package staffing quota by 15%, it is judged as having redundant personnel; if the actual working hours consumed per unit of work exceed the quota working hours by 20%, it is judged as low efficiency; and if the actual settlement average unit price is 10% higher than the local market labor unit price during the same period, it is judged as an abnormal unit price. The labor risk early warning subunit is used to preset two sets of independent two-level risk thresholds according to engineering scenarios. Based on the joint judgment result of the relative deviation rate and the labor cost input performance index, it performs hierarchical early warning distribution. Differentiated early warning judgment standards are configured for point-like closed projects and linear open projects respectively. When the indicator falls into the first-level mild early warning range, a cost reminder is pushed to the site administrator. When the indicator meets any triggering condition of the second-level severe early warning, an in-depth analysis report containing root cause diagnosis of personnel redundancy, low efficiency, and abnormal unit price is generated and pushed to the project manager and the company's cost control center. The aforementioned payroll calculation and cost early warning module implements dynamic quantitative control over payroll settlement and labor costs, constructing a closed-loop management chain of "calculation-comparison-analysis-early warning": The automatic payroll calculation unit links standardized time-series attendance datasets with payroll calculation rules in electronic contracts to automatically calculate basic wages, overtime pay, and various allowances and subsidies for laborers, generating detailed payroll data; the actual labor cost statistics unit aggregates and summarizes the payroll data according to WBS work packages and time dimensions, forming an actual labor cost expenditure dataset; the S-curve comparison unit retrieves the full-cycle labor cost input S-curve benchmark generated in the early stages of the project, extracting the same... The system dynamically compares the planned value with the actual output value of each project to calculate the absolute deviation value and relative deviation rate. The earned value management analysis unit incorporates the actual output value of each sub-project to construct a labor-specific earned value assessment model. It comprehensively quantifies the employment matching degree by calculating the labor cost input performance index (CPI) and identifies deviations such as "inefficient overspending" or "efficient cost saving". The employment risk warning unit distinguishes between two types of project scenarios and configures independent threshold rules. Based on the joint judgment results of the deviation rate and CPI, when the indicators exceed the preset threshold of the corresponding scenario, it automatically generates an in-depth analysis report containing root cause diagnoses such as personnel redundancy and low efficiency, and pushes it to the corresponding management terminal in a tiered manner according to risk level.
[0033] Furthermore, the exit closed-loop archiving module includes: an exit application acceptance subunit, an exit confirmation subunit, and a data archiving and credit building subunit; The exit application acceptance sub-unit is used to respond to the workers' voluntary exit application or contract expiration instructions, and automatically link and query the workers to be exited in this project's full-cycle attendance records, salary payment records, safety violation records, and equipment and material return records, so as to build a full performance verification list; The exit confirmation subunit is used to perform automatic accounting based on the verification list, generate an encrypted electronic exit certificate containing a unique hash value or digital signature after confirming that all payments have been settled, permanently mark the status of the virtual twin of the corresponding worker to be exited in the digital twin base from "on duty" to "exited", terminate the real-time mapping update of its work trajectory, and perform grayscale processing or move the virtual twin into the historical personnel layer in the visualization interface. The data archiving and credit building sub-unit is used to package and archive the entry assessment data, full-cycle attendance data, salary settlement data, violation records and exit evaluation data generated by the worker to be discharged in this project. After strongly associating the archived data items with the unique identity of the worker to be discharged (such as ID card number or biometric ID), it is written into the enterprise's global labor credit database to form a dynamically updated labor credit profile, which can be retrieved and queried in the subsequent project entry screening stage. The exit closed-loop archiving module implements closed-loop management of the entire process of worker departure, achieving zero economic material losses and credit asset accumulation: The exit application acceptance unit responds to workers' voluntary exit applications or contract expiration instructions, automatically retrieving the worker's full-cycle attendance records, payroll records, safety violation records, and equipment and material return records for the project, constructing a comprehensive performance verification list; the exit confirmation unit performs payroll settlement based on the verification list, generating an encrypted electronic exit certificate with a unique hash value or digital signature, and storing the worker's virtual twin in the digital twin system. The status of the individual is permanently marked from "on duty" to "left the site," and the real-time mapping update of their work trajectory and the migration of the visualization layer are simultaneously terminated. Finally, the data archiving unit packages and archives all business data generated by the individual in this project, including entry assessment, attendance trajectory, salary settlement, violation records, and exit evaluation. Based on the strong correlation between the data and the unique identity identifier (such as ID card number or biometric ID), the data is written into the enterprise's global labor credit database to form a dynamically updated cross-project labor credit profile, which can be retrieved during subsequent project entry screening to achieve "one breach of trust, network-wide early warning" credit joint prevention.
[0034] Furthermore, the digital twin visualization construction module includes: a real-time data injection subunit, a real-time employment status twin scene rendering subunit, a micro-level traceability query subunit, and a multi-dimensional analysis subunit; The real-time data injection subunit is used to build a high-throughput message middleware based on the publish-subscribe model, establish a real-time data synchronization pipeline from the business front end to the digital twin base, perform streaming ETL processing on all business data, and inject project basic data, dynamic business data and exit status data into the BIM+GIS fusion base in real time according to a unified data schema to drive the virtual model to respond in milliseconds. The real-time employment status twin scene rendering subunit is used to render and generate an integrated twin visualization interactive interface. On the basis of reproducing the physical construction site topography, building model and temporary facilities, it overlays and displays the labor management theme layer, intuitively presenting the construction site personnel heat map, job distribution density and each WBS work package. The micro-traceability query subunit is used to respond to the triggered operation of the virtual personnel model in the twin interface, retrieve and display the complete digital file of the corresponding personnel, including entry identity verification record, AI job matching degree assessment report, full cycle operation trajectory playback, salary settlement details and exit settlement voucher, to achieve full cycle penetration traceability of a single personnel; The multidimensional analysis subunit is used to provide a multidimensional data fusion query interface, supporting queries on real-time personnel density and safety compliance based on the work surface dimension, comparison of the deviation trend of phased labor costs from the planned S-curve based on the time dimension, and retrieval of the performance credit rating of specific personnel across projects, providing data support for management decisions; The digital twin visualization construction module maps the full volume of business data streams generated from S1 to S5 to the BIM+GIS digital twin base in real time, constructing a high-fidelity integrated twin model of the entire labor employment cycle, breaking down data silos and achieving penetrating management: by building a high-throughput message middleware to establish a real-time data synchronization pipeline, the project's basic data, dynamic business data, and exit status data are streamed and injected according to a unified schema, driving the virtual personnel model and physical entity status in the twin space to synchronize at millisecond levels; based on the integrated visualization interactive interface, in addition to intuitively displaying the construction site personnel heat map, work... In addition to displaying the distribution density and real-time employment status of each WBS work package, it also supports responding to triggered operations on virtual personnel models to perform full-cycle traceability queries for single personnel, and to access the corresponding personnel's entry verification records, AI job matching assessment reports, full-time work trajectory playback, salary settlement details, and exit settlement vouchers. It also provides multi-dimensional data fusion query interfaces, supporting queries on real-time personnel density based on the work surface dimension, comparison of the deviation trend of phased labor costs from the planned S-curve based on the time dimension, and retrieval of the performance credit rating of specific personnel across projects, transforming dry business data into intuitive decision support.
[0035] Furthermore, the data platform processing and early warning module includes: a data governance and storage subunit, an AI model iteration subunit, a risk identification and judgment subunit, and a hierarchical early warning distribution subunit; The data governance and storage subunit is used to uniformly receive the raw perception data and derived business data collected by various business modules through a standardized API gateway, perform a cleaning process to remove redundant fields, repair missing values, unify data formats and desensitize data, and classify and store data according to personnel, cost and spatial subject domains to build a unique distributed data asset pool for the project. The AI model iteration subunit is used to independently support the operating environment of the AI stress assessment model in step S2 and the labor-specific earned value cost assessment model in step S4. It utilizes accumulated cross-project historical data (including personnel physiological response database and work efficiency cost database) to continuously incrementally learn and dynamically optimize the model parameters, thereby achieving autonomous evolution of the algorithm model. The risk identification and judgment subunit is used to build a unified rule engine to monitor the data flow of the entire business chain in real time, identify multiple risk events such as entry interception, job mismatch, attendance loss, and labor cost overrun, and judge the level of the identified risks (such as prompt, warning, and serious alarm) based on the differentiated risk control matrix that distinguishes between point-like closed and linear open projects. The tiered early warning distribution subunit is used to execute a tiered push strategy based on the risk level. Operational-level early warnings (such as attendance record reminders) are pushed to the mobile terminals of front-line team leaders, management-level early warnings (such as cost overrun analysis) are pushed to the PC terminals of project managers, and strategic-level early warnings (such as collective labor risks) are pushed to the enterprise-level monitoring dashboard, so as to achieve accurate delivery and closed-loop handling of risk information.
[0036] The data platform processing and early warning module, as the underlying support hub decoupled from the business application layer, uniformly receives multi-source heterogeneous data from the entire chain through a standardized API gateway. It performs cross-module data cleaning, format standardization, spatiotemporal coordinate transformation, and distributed storage to build a unique data asset pool for the project. It independently hosts the operating environment of the AI stress assessment model and the labor-specific earned value cost model. It uses accumulated cross-project historical data to continuously incrementally learn and dynamically optimize the model parameters, realizing the autonomous evolution of algorithm capabilities. The built-in unified rule engine monitors the business data flow in real time. When it identifies various risk events such as entry interception, job mismatch, attendance loss, and labor cost overrun, it intelligently judges the risk level based on the differentiated risk control matrix that distinguishes between point-like closed and linear open projects. It then pushes early warning information to the mobile terminals of front-line team leaders, the PC terminals of project managers, or the enterprise-level monitoring screen according to the risk level, realizing accurate reach and closed-loop handling of risk events.
[0037] The modules of this invention work collaboratively to form a complete closed loop: "physical entity → twin model construction → real-time perception and data acquisition → AI training and twin data fusion → hazard prediction → closed-loop verification → proactive handling → twin model iteration." Through virtual-real interaction and data feedback mechanisms, the system is endowed with continuously evolving intelligent management capabilities. Engineering projects applicable to this invention include, but are not limited to, water conservancy and water affairs, hydropower, thermal power, municipal engineering, building construction, integrated transportation, ecological and environmental protection, and mining engineering.
[0038] Figure 1The system showcases the overall layered architecture of a smart management system for the entire lifecycle of labor employment in engineering projects. It adopts a bottom-up, multi-layered collaborative architecture of "external interface - hardware perception - data collection - middle platform governance - business application - digital twin visualization." Each layer achieves data interoperability and business linkage through standardized interfaces, forming a closed-loop management system covering the entire lifecycle of laborers from entry, operation, salary to departure. The entire architecture relies on the decoupled design of the business middle platform and data middle platform, forming a complete technical link from multi-source hardware data collection and intelligent early warning analysis to closed-loop business control and digital twin visualization, enabling intelligent management of the entire labor employment lifecycle. The system is divided from bottom to top into an external system and hardware perception layer, a data collection layer, a data middle platform and early warning layer, a business application layer, and a visualization layer. Each layer is interconnected through data interfaces and business flows.
[0039] The system's underlying layer interfaces with three types of external systems and multiple IoT sensing devices, forming the source for collecting all business data. External systems include national or provincial / municipal government regulatory platforms (for identity verification and retrieval of historical violation records), the company's existing labor database (for cross-project historical performance data retrieval), and bank payment systems (for salary payment and settlement confirmation). Hardware sensing devices include ID card readers (for reading real-name information), facial recognition terminals (for collecting facial biometrics), smart bracelets / wearable devices (for real-time collection of physiological parameters such as heart rate and blood pressure during VR training), turnstiles and facial recognition attendance machines (for attendance at point-based closed project entrances and exits), high-definition cameras / eagle-eye devices (for personnel behavior and trajectory recognition within the site), inspection drones (for wide-area personnel positioning and video acquisition in linear open projects), and mobile smart terminals (such as worker mobile apps or RTK positioning devices for reporting GPS / BeiDou coordinates). The above devices respectively collect raw multi-dimensional data such as personnel identity data, physiological parameters, location trajectory, attendance records, and video streams.
[0040] The data acquisition layer is responsible for unified access, protocol conversion, and preliminary verification of data from the underlying hardware and external systems, providing standardized data services to the upper layer. The data acquisition layer receives raw data reports from hardware devices (including facial images, physiological waveforms, GPS coordinates, video frames, etc.), and simultaneously connects with external systems to obtain standard data services (such as real-name information returned by government platforms and historical performance tags returned by enterprise databases). Through external data interface, the layer cross-compares ID card information, facial features, and enterprise databases and government platforms to output preliminary screening results such as the legality of personnel identity, age compliance, and historical violation records, providing data support for subsequent entry access.
[0041] The data platform and early warning layer are the core of the system, adopting an independent architecture decoupled from the business application layer. They are responsible for the unified governance of all data, the operation of AI models, and risk warning functions across business processes; specifically including: Data cleaning and fusion: Perform redundancy removal, missing data repair, format standardization, spatiotemporal coordinate transformation, and data anonymization on multi-source heterogeneous data (structured forms, unstructured videos / images, time-series physiological signals) from the acquisition layer to form a unified subject domain data warehouse; Early warning analysis engine: With a built-in rule engine and machine learning model, combined with reference data such as personnel physiological response baselines and job quotas / efficiencies from the historical database, it performs multi-dimensional risk identification. Specifically, this includes: age-related / violation-based warnings based on age thresholds and past records; warnings about mismatches between personnel and positions based on AI analysis models; and warnings about labor cost overruns based on deviations from actual labor costs and the S-curve. Warning results are pushed to the corresponding management terminals in tiers according to risk level (prompt, warning, severe alarm). AI Model Iteration: The data platform independently hosts the physiological stress assessment model and the labor-specific earned value cost model, using continuously flowing project data to incrementally learn and dynamically optimize model parameters. The business application layer, based on the data services and early warning capabilities provided by the platform, executes the specific business logic throughout the entire labor employment lifecycle, divided into three main functional domains: Project basic data collection: responsible for building the BIM+GIS digital twin foundation, importing construction drawings, construction organization design, enterprise construction quotas and labor unit prices, executing WBS work package decomposition, and generating the full-cycle labor cost input S-curve based on quotas and unit prices as a benchmark for subsequent dynamic cost comparison; Multi-dimensional personnel entry control: This involves sequentially verifying identity (facial and ID card information), screening resumes (connecting with enterprise and government platforms), VR safety education and physiological parameter collection, and AI job matching assessment, culminating in the signing of an electronic labor contract on an integrated terminal. All process data (identity, violation records, physiological parameters, assessment reports, and electronic contracts) are linked and encapsulated using blockchain or timestamp technology to form a traceable, judicial-grade electronic file. Full-cycle management of personnel operations includes: generating work trajectories (mapping attendance locations to BIM work surfaces onto a twin base), summarizing attendance data (integrating multi-source data from turnstiles, surveillance cameras, drones, and mobile terminals), calculating salaries (linking attendance to contract rules), and confirming departure (full performance verification, salary settlement, generating encrypted departure vouchers and marking "departed" in the twin base). After departure, the personnel's full-cycle data is archived to the enterprise's global labor credit database, forming a labor credit profile that can be accessed across projects.
[0042] The presentation layer, based on a BIM+GIS digital twin foundation, maps all data generated by the business application layer into a real-time visual interactive interface, providing managers with panoramic decision support; specifically including: Personnel files and dynamic profiles: Displaying workers' identity information, job type, contract status, real-time location, historical trajectory, and credit score; Labor Cost and Risk Warning Dashboard: Displays a dynamic comparison of actual labor costs with the S-curve, deviation rate, CPI index, and tiered warning information in chart form; Personnel entry and exit closed-loop display: intuitively presents personnel presence status (on duty / already left), exit and settlement progress, and cross-project flow trajectory; Integrated visual interactive interface: Supports full-cycle traceability of a single person (one file per person from entry to exit), heat map of personnel density on the work surface, phased labor cost trend analysis and other multi-dimensional data fusion queries.
[0043] Through the collaborative operation of the aforementioned five-layer architecture, the entire system constructs a complete technical closed loop: "physical entity perception → multi-source data acquisition → middle platform cleaning and AI analysis → business closed-loop processing (entry → operation → salary → exit) → twin visualization display → early warning feedback to the business end." In particular, the decoupled design of the data middle platform enables business modules to be upgraded independently, AI models to be continuously iterated, and early warning rules to be dynamically configured, thereby realizing an intelligent leap in labor employment management from "passive response" to "proactive quantitative prevention and control, dynamic cost early warning, and enterprise-level credit accumulation."
[0044] It should be noted that the intelligent management system for the entire lifecycle of labor employment in the engineering project provided in this embodiment can be a computer program (including program code) running on a computer device. For example, the intelligent management system for the entire lifecycle of labor employment in the engineering project is an application software. The intelligent management system for the entire lifecycle of labor employment in the engineering project can be used to execute the corresponding steps in the methods provided in the embodiments of this application.
[0045] like Figure 2 As shown, a second aspect of the present invention provides a method for intelligent management of the entire lifecycle of labor employment in engineering projects, implemented using the aforementioned intelligent management system for the entire lifecycle of labor employment in engineering projects, comprising the following steps: S1: Acquire multi-dimensional basic data of the project, integrate BIM 3D geometric data and GIS geospatial data to build a full-domain digital twin foundation; generate the S-curve of labor cost input throughout the project cycle based on WBS work breakdown structure, enterprise and market labor quota data, as a benchmark for dynamic management and control of labor costs. S2: Receive applications from workers to enter the site, and perform cross-project resume verification and government platform real-name verification in parallel. Combine VR high-risk scene physiological time sequence collection and AI stress adaptation assessment to build an integrated entry access mechanism that includes resume screening, quantitative determination of personnel and positions, and electronic signing, forming an unalterable full-link electronic file for personnel. S3: Automatically identify project types, distinguish between point-like closed projects and linear open projects, and configure differentiated multi-source attendance collection strategies; integrate multi-source attendance data to generate a standardized time-series attendance dataset; map the spatial coordinates contained in the dataset to the digital twin base to generate dynamic work trajectories; S4: Based on standardized time-series attendance data, calculate the actual labor cost of WBS work packages, link the full-cycle labor cost input S-curve benchmark to calculate cost deviation, introduce a labor-specific earned value model, combine the actual output value of sub-projects to calculate the cost performance index (CPI), configure independent two-level risk thresholds for point-like closed projects and linear open projects respectively, and push labor risk warnings in a hierarchical manner based on the joint judgment results of deviation rate and labor cost input performance index. S5: Respond to personnel departure requests by fully verifying attendance, salary, and material performance data. After salary settlement, generate encrypted electronic departure vouchers, mark personnel departure status on the digital twin base, and archive all project performance data to the enterprise labor credit database to generate personnel credit profiles that can be retrieved across projects. S6: Map the full-cycle business data flow from steps S1 to S5 to the digital twin platform in real time to build an integrated interactive interface that supports full-cycle traceability for individual soldiers, cross-project credit queries, and visualization of multi-dimensional cost indicators. S7: Set up an independent data platform that is decoupled from all upper-level business steps to complete unified governance of all heterogeneous data, incremental iteration of AI models, and precise push of risk stratification and classification across the entire chain.
[0046] The steps of the method of the present invention will be described in detail below.
[0047] (1) Project basic data collection Step S1 provides a data foundation and cost control benchmark for subsequent full-cycle management.
[0048] Specifically, step S1 includes: First, acquire the BIM 3D geometric data and GIS geospatial data of the target project, perform coordinate registration and fusion of the two, and construct a high-precision digital twin base covering the entire project area. Subsequently, the construction drawings, construction organization design documents, internal construction quotas (including standard work efficiency and ratio parameters for each type of work) and the latest labor unit prices for the project location were imported. Based on the above data, the project scope was hierarchically decomposed using WBS (Work Breakdown Structure), and the project was broken down into independent work packages at each level. The planned start time, planned end time and key milestone nodes were then bound to each work package. Finally, by combining the construction quota, labor unit price and the schedule nodes of each work package, the labor cost required for each node is calculated cumulatively according to the time dimension, and the cumulative S-curve of labor cost input throughout the project cycle is generated; this S-curve serves as the benchmark for dynamic comparison of labor costs. The dynamic correction mechanism of the cumulative S-curve of labor cost input is as follows: when a design change occurs, the corresponding WBS work package quantity is automatically updated based on the engineering visa, the full-cycle planned labor cost of the work package is recalculated, and the global S-curve is updated synchronously; if the change exceeds ±10%, the current period S-curve deviation warning threshold is temporarily relaxed by 2% to avoid short-term invalid alarms caused by the change. The formula for calculating the phased planned labor cost of a single work package within the planned duration is as follows: in, For work package In the Planned labor costs for the period; For work package The Middle The planned number of personnel for each job type; The location of the project The unit price of labor for a particular job; Standard working hours within a single statistical period; The first Each work package is bound to a planned start time and a planned end time; The formula for calculating the cumulative S-curve of total labor cost input over the entire lifecycle is: in, Indicates as of The total cumulative labor costs planned for the project. This represents the total number of WBS work packages; For accumulating intermediate variables, representing the period from the 1st period to the 2nd period. The index for traversing the period; For the first The WBS work package is in the first Planned labor costs during the period; Furthermore, the multi-dimensional basic data of the project mentioned in step S1 includes geometric spatial dimension data, construction organization dimension data, economic quota dimension data, and schedule dimension data; the geometric spatial dimension data includes at least the project's BIM (Building Information Modeling) three-dimensional geometric data (including geometric topology information of building structure, temporary facilities, and pipeline lines) and GIS (Geographic Information System) geospatial data (including topography, coordinate reference, and surrounding environment information); The construction organization dimension data includes at least construction drawings (design deliverables) and construction organization design documents (including overall deployment, construction technology, schedule and construction plan). The economic quota dimension data includes at least the company's internal construction quota data (covering the standard labor efficiency and work allocation parameters for each type of work) and the latest labor unit price for each type of work in the project location.
[0049] The schedule data includes at least a list of work packages generated from the hierarchical breakdown of the WBS (Work Breakdown Structure), and the project milestones (planned start time and planned end time) associated with each work package.
[0050] (2) Multi-dimensional management and control of personnel entry Step S2 aims to move the safety control checkpoint forward, achieving pre-entry screening and quantitative suitability assessment of personnel through multi-dimensional data fusion. Specifically, it includes: First, multi-source identity verification is implemented: facial biometric features of workers are acquired through facial recognition devices, while real-name information of ID cards is obtained through OCR recognition or NFC reading devices; the system cross-verifies the authenticity of personnel identities by connecting to the company's historical labor database and the government-level labor real-name supervision platform, and automatically screens for cross-project violations, poor performance records, or industry-restricted records. If high-risk records are found, an entry interception warning is immediately generated and pushed to the security management terminal, and those who fail the verification are prohibited from entering the site. Secondly, conduct physiological stress assessments in VR scenarios: guide workers who pass the initial screening to enter the VR safety education experience center, and simulate corresponding high-risk work scenarios (including but not limited to high-altitude work, deep foundation pit work, and confined space work) based on their declared job types; during the immersive experience, connect to the smart wearable devices (such as smart bracelets and heart rate belts) worn by the workers via Bluetooth to collect their physiological parameters such as heart rate and blood pressure fluctuations in real time under different stress scenarios; among them, preset differentiated physiological safety thresholds for different high-risk jobs: the upper limit of resting heart rate for high-altitude work is 110 beats / minute, and diastolic blood pressure is ≤95mmHg; the upper limit of heart rate for deep foundation pit work is 120 beats / minute. Next, AI job matching analysis is performed: the collected time-series physiological parameters are input into a pre-trained AI analysis model. This model is trained based on massive historical personnel physiological data and job suitability labels, and is used to assess the psychological adaptability, stress response ability, and physical endurance of workers in simulated high-risk environments, outputting a quantitative job matching assessment report. If the assessment results show that the personnel's physiological indicators exceed the safety threshold for the job, it is judged as "person-job mismatch," and it is recommended to transfer to another job or not to hire them. The AI analysis model is initially pre-trained based on more than 100,000 cross-project personnel time-series physiological data and job suitability labels. The system automatically triggers incremental learning of the model when more than 500 new personnel physiological data are added each month, and updates the physiological critical thresholds for each job synchronously. When a person's physiological indicators exceed the preset threshold for the corresponding job three times in a VR scene, it is directly judged as person-job mismatch and the person is blocked from entering the site. Finally, a full-chain electronic file is constructed: For personnel who pass the matching assessment, standardized electronic labor contracts are retrieved on integrated terminals or mobile devices, guiding them to complete online signing; an index relationship is automatically established to associate and encapsulate personnel identity data, violation screening records, VR training records, physiological parameter logs, AI assessment reports, and signed electronic contracts, and solidify the evidence using blockchain technology or timestamp technology to form an immutable and traceable electronic file for entry, providing a judicial-grade electronic evidence chain for the determination of liability in the event of subsequent labor disputes or safety accidents.
[0051] (3) Multi-source attendance and trajectory collection for personnel operations Step S3 adaptively configures multi-source perception strategies based on the physical boundary characteristics of different engineering scenarios, solving the problem of traditional attendance systems failing in non-closed environments, and achieving real-time mapping of physical space work trajectories to digital twin spaces. Specifically, this includes: First, identify the type of engineering scenario and configure the collection strategy: automatically read the project attribute information. If it is determined to be a point-like closed project such as a building or factory, then the dual verification mode of "gate + eagle eye" is activated; if it is determined to be a linear open project such as a highway, railway, or pipeline, then the wide-area splicing mode of "drone + mobile terminal + video along the line" is activated.
[0052] Secondly, perform multi-source data fusion and standardization processing: For point-based closed-loop projects: entrance and exit turnstiles obtain the first attendance data source through facial recognition, while surveillance cameras deployed within the site capture personnel movement trajectories to obtain the second attendance data source through computer vision algorithms (such as ReID technology). Using the turnstile location as a spatial anchor point, the facial ID is bound and associated with the trajectory ID captured by the surveillance cameras the moment a person passes through the turnstile. Subsequently, the two data sources are spatiotemporally aligned and cross-validated to eliminate abnormal data such as proxy attendance or recognition errors, generating standardized attendance records with high confidence. If the turnstile record and the surveillance camera trajectory within the site do not match within 15 minutes for the same person, it is determined to be proxy attendance, and this attendance data is removed. For linear open engineering projects: High-definition gimbals and visual recognition modules mounted on inspection drones are used, combined with GPS / BeiDou coordinates uploaded by workers' mobile smart terminals (such as mobile phones with positioning apps or RTK devices), and video stream data from fixed-point cameras along the route. A multi-sensor data fusion algorithm is employed to stitch together continuous position information of personnel on long-distance linear work surfaces, solving the attendance blind spot problem caused by the lack of fixed entrances and exits. The drones operate only in sunny, windless conditions; in rainy or snowy weather, the system switches to primarily use video along the route and RTK mobile terminal positioning. If the positioning coordinates drift by more than 5 meters, automatic interpolation of preceding and following time-series coordinates is used for correction. During multi-source data fusion, RTK positioning has a weight of 0.6, drone vision has a weight of 0.3, and video along the route has a weight of 0.1, and standardized spatial coordinate position information is generated through weighted calculation. The merged attendance data is uniformly cleaned, deduplicated, and formatted. Each standardized attendance record is then spatially correlated with the BIM work surface division data of the construction area to determine the specific workstation or construction section where the personnel are currently located and extract their real-time spatial coordinates. The final output is a standardized time-series attendance dataset containing the fields "Personnel ID-Timestamp-Spatial Coordinates-Work Package".
[0053] Next, construct a dynamic work trajectory: Based on the standardized time-series attendance dataset, the spatial coordinates contained in the dataset are mapped to the BIM+GIS digital twin base constructed in step S1 through a real-time data interface, driving the virtual personnel model in the twin space to move synchronously and generating a high-fidelity dynamic work trajectory consistent with the physical world.
[0054] Finally, data distribution and application: The standardized time-series attendance dataset is synchronously distributed to the payroll calculation module in step S4 and the visualization module in step S6, serving as a unified data benchmark for payroll calculation and multi-dimensional display, respectively, to ensure the consistency between business data and the twin display.
[0055] (4) Dynamic salary calculation and early warning of labor cost risks Step S4 breaks away from the limitations of traditional payroll management that relies solely on attendance records, deeply integrating labor data into the project cost control system. It achieves dynamic quantitative monitoring and early warning of labor costs by introducing earned value analysis. Specifically, this includes: First, based on the standardized attendance dataset, the actual labor costs are automatically calculated: the standardized time-series attendance dataset output by step S3 is retrieved, and the salary calculation rules in the electronic labor contract stored in step S2 (including but not limited to hourly unit price, piece-rate unit price, overtime rate, job type subsidy and reward and punishment clauses) are associated to automatically calculate the current basic wage, overtime wage and various allowances and subsidies for each worker, and generate individual salary details; then, according to the WBS work package classification defined in step S1, the salary details of the personnel to which they belong are collected and summarized, and the total actual labor cost of each work package in a specific time period is calculated to form the project actual labor cost expenditure dataset AC.
[0056] Secondly, the deviation analysis of the linkage labor cost input benchmark is performed: the S-curve of the project's full-cycle labor cost input (planned benchmark) pre-generated in step S1 is retrieved, and the planned labor cost value PV corresponding to the current statistical period is extracted; the actual labor cost expenditure dataset AC is compared with the planned labor cost value PV of the same period, and the absolute deviation value (SV=PV-AC) and relative deviation rate (SV%=(PV-AC) / PV×100%) are calculated to quantify the current overspending or surplus status of labor costs.
[0057] Secondly, a labor-specific earned value cost assessment model is introduced for multi-dimensional analysis: To further identify the nature of cost deviations, the earned value management (EVM) method is adopted, and the actual completed output value (EV) of each sub-project is introduced as a key variable; a labor-specific earned value cost assessment model is constructed to calculate the labor cost input performance index (CPI=EV / AC, where AC is the actual labor cost); if CPI<1, it indicates that the labor cost input-output ratio is low, and there is inefficient labor use or idle labor; combining the deviation direction (positive or negative SV) and the performance index (the magnitude of CPI), the current labor matching degree is comprehensively quantified. For example, if the actual labor cost exceeds the budget but the output value does not meet the target (SV is negative and CPI<1), it is judged as "inefficient overspending"; if the actual labor cost is saved and the output value meets the target (SV is positive and CPI≥1), it is judged as "efficient cost saving". In addition, three types of abnormal employment judgment rules are added: if the actual number of employees exceeds the quota of the corresponding WBS work package by 15%, it is judged that there is a redundancy of personnel; if the actual working hours consumed per unit of project exceeds the quota working hours by 20%, it is judged that the work efficiency is low; if the actual settlement average unit price is 10% higher than the local market labor unit price in the same period, it is judged that the unit price is abnormal. Finally, implement tiered and graded labor risk warnings: pre-set two sets of independent two-level risk thresholds and corresponding risk control strategies according to the different types of engineering scenarios, and implement tiered and graded warning distribution based on the joint judgment results of deviation rate (SV%) and labor cost input performance index (CPI); 1) For point-based enclosure projects (buildings, factories, etc.): The preset threshold range for Level 1 mild warning is: And the relative deviation rate of labor costs When the indicators fall within the above range simultaneously, it is judged as a slight fluctuation in labor costs; cost reminder information is automatically generated, highlighting the current abnormal work packages and personnel attendance density, and is pushed to the mobile terminals of on-site construction workers and labor managers through lightweight messages, prompting them to pay attention to the dynamics of on-site labor, and to conduct on-site review or fine-tuning when necessary, so as to achieve early detection and early intervention of risks. Level II Severe Warning Triggering Conditions: Or relative deviation rate of labor costs If any condition is met, it will be judged as a project-level overspending alarm; an in-depth analysis report containing a preliminary diagnosis of the cause of the deviation (such as personnel redundancy, low efficiency, abnormal unit price, etc.) will be automatically generated and pushed to the project manager and the company's cost control center. 2) For linear open engineering projects (highways, railways, pipelines): Preset Level 1 Mild Warning Threshold Range: And the relative deviation rate of labor costs When the indicators fall within the above range simultaneously, it is judged as a slight fluctuation in labor costs; cost reminder information is automatically generated, highlighting the current abnormal work packages and personnel attendance density, and is pushed to the mobile terminals of on-site construction workers and labor managers through lightweight messages, prompting them to pay attention to the dynamics of on-site labor, and to conduct on-site review or fine-tuning when necessary, so as to achieve early detection and early intervention of risks. Level II Severe Warning Triggering Conditions: Or relative deviation rate of labor costs If any condition is met, it will be judged as a project-level overspending alarm; an in-depth analysis report containing a preliminary diagnosis of the cause of the deviation (such as personnel redundancy, low efficiency, abnormal unit price, etc.) will be automatically generated and pushed to the project manager and the company's cost control center to provide managers with a basis for decision-making on resource allocation, personnel optimization or process improvement. The analysis report should include at least the following diagnostic dimensions: Staffing diagnosis: Compare and analyze the deviation between the actual number of employees present and the planned number of employees for the work package to identify whether there is "redundancy" or invalid attendance during non-working hours; Work efficiency diagnosis: By combining the BIM engineering quantity model with the actual working hours, the ratio of the actual working hours per unit of engineering quantity to the quota working hours is calculated to identify whether there is "low efficiency" or slack work. Cost unit price diagnosis: Compare the deviation between the actual settlement average unit price and the market labor unit price imported in step S1 to identify whether there is "abnormal unit price" or unreasonable remuneration outside the contract; Schedule matching diagnosis: Combine EV (actual output value) analysis to distinguish between "benign overspending" (reasonable increase in input due to schedule overrun) and "malicious overspending" (cost waste caused by schedule delay).
[0058] (5) Closed-loop archiving of personnel departure Step S5 aims to achieve closed-loop management of the entire process of worker departure, ensuring that all economic and material transactions are cleared, and that personnel performance data is accumulated as core credit assets for the enterprise, supporting cross-project personnel selection and risk control. Specifically, it includes: First, in response to workers' voluntary withdrawal requests or contract expiration instructions, a full-scale performance data retrieval mechanism is initiated: automatically linking and querying the full-dimensional business data of workers to be withdrawn in this project to construct a withdrawal verification list; the list includes at least: full-cycle attendance records generated in step S3 (verifying whether there are any absence deductions or unpaid overtime), salary payment records generated in step S4 (verifying whether there are any unpaid wages or undeducted advances), violation penalty records entered by the on-site safety management department, and records of labor protection supplies requisition and construction machinery return registered by the materials management department (verifying whether there are any material loss or damage deductions).
[0059] Secondly, the process involves payroll settlement and electronic voucher generation: Automatic calculations are performed based on the verification checklist to confirm all payments are settled (including the net amount after deducting various fines and compensation payments from wages due). After verification, an encrypted electronic exit voucher containing a unique hash value or digital signature is generated. This encrypted electronic exit voucher contains basic employee information, employment period, final settlement amount, and material return status. It allows employees to download and save it via mobile devices and can also serve as digital evidence in labor disputes.
[0060] Secondly, the status of personnel in the digital twin base is updated synchronously: the status of the virtual twin of the corresponding worker in the digital twin base is permanently marked from "on duty" to "left the site", the real-time mapping update of the work trajectory of the worker to be left the site is terminated, and the worker is grayscaled or moved into the historical personnel layer in the visualization interface to ensure that the status of the twin space is consistent with the physical site.
[0061] Finally, a cross-project labor credit file is constructed: all business data generated by the workers to be leaving the project (including entry assessment data, full-cycle attendance data, payroll settlement data, violation records, and exit evaluation data) are packaged and archived, and synchronously written into the enterprise's global labor credit database, forming a credit archive data entry corresponding to the workers to be leaving; the credit archive data entry is strongly correlated with the unique identifier of the workers to be leaving (such as ID card number or biometric ID), forming a dynamically updated labor credit profile. When the workers to be leaving subsequently participate in other projects of the enterprise, the system can directly retrieve the cross-project credit file in the entry screening stage of step S2, realizing a credit joint prevention mechanism of "one breach of trust, network-wide early warning", and completing the leap from single-project management to enterprise-level human resource asset management.
[0062] (6) Full lifecycle visualization of digital twins Step S6 aims to break down data silos by mapping the full volume of business data generated from S1 to S5 to a digital twin platform in real time, constructing a high-fidelity panoramic view of labor employment, and achieving penetrating management from a macro-level perspective to a micro-level understanding of personnel. Specifically, this includes: First, establish a real-time data synchronization pipeline: build a high-throughput message middleware (such as Kafka, RabbitMQ) to establish a real-time data synchronization channel from the business front-end (attendance equipment, mobile terminals, VR devices, etc.) to the digital twin base; perform streaming ETL processing (Extract-Transform-Load) on the collected multi-source heterogeneous data: standardize the format of the raw data, transform the spatiotemporal coordinates and clean redundant fields according to the predefined unified data schema to generate a structured data stream; through the real-time data injection interface, synchronously write the project's static basic data (WBS structure, S-curve), dynamic business data (entry files, attendance trajectory, salary changes, risk warnings) and exit status data into the digital twin base built by BIM+GIS integration, and use in-memory computing technology to ensure that every subtle change in the physical world (such as personnel location movement, job type change, cost fluctuation) can achieve millisecond-level response in the twin space.
[0063] Secondly, an integrated digital twin visualization interface is constructed: On top of the digital twin platform, a multi-dimensional visualization interface is generated, rendering the construction site's personnel heat map, job distribution density, personnel gathering in key areas, and the real-time labor status of each WBS work package. This interface not only reproduces the physical site's topography, building model, and temporary facilities, but also overlays a labor management thematic layer. Through this virtual-real mapping, managers can intuitively view the construction site's personnel heat map, job distribution density, personnel gathering in key areas, and the real-time labor status of each WBS work package within the same view.
[0064] Secondly, it enables full-cycle traceability for individual personnel: managers can initiate full-cycle traceability queries for a single person by clicking on the virtual personnel model in the twin interface or entering the personnel's identity information. The system responds to the query request, retrieves and displays the person's complete digital profile, including: identity verification records upon entry and AI job matching assessment reports, complete work trajectory playback and attendance anomaly records during their time on site, detailed salary settlements for each period, historical violation penalty records, and finally, exit settlement vouchers. This "one person, one file" panoramic view provides intuitive visual support for tracing safety incidents and collecting evidence in labor disputes.
[0065] Finally, the system supports multi-dimensional data fusion queries and decision support: it provides flexible data interaction interfaces, supporting cross-project and cross-cycle data fusion queries. Managers can retrieve the movement trajectory and performance credit rating of specific personnel across multiple historical projects with a single click (cross-project credit file retrieval); they can query real-time personnel density and safety compliance of specific areas based on the work surface dimension; and they can compare and analyze the deviation trend of phased labor costs from the planned S-curve based on the time dimension. Through these multi-dimensional visualization query functions, dry business data is transformed into intuitive decision-making basis, significantly improving the precision of project management and the speed of emergency response.
[0066] (7) Data platform for unified processing and early warning distribution Step S7 aims to address the problems of high coupling between modules, poor data consistency, and difficulty in algorithm iteration in traditional labor service systems by building an independent data platform decoupled from upper-layer business logic. This provides a unified data governance foundation and intelligent computing power support for full-cycle intelligent management. Specifically, it includes: First, a decoupled layered data architecture is constructed: a "business application layer - underlying support layer" separation architecture design is adopted; all business logic in S1 to S6 is encapsulated into the business application layer, while data governance, model calculation, and early warning distribution are separated into independent data middle platforms; the two layers interact with each other through a standardized API gateway to ensure that the functional upgrades or additions and subtractions of business modules do not affect the consistency and stability of the underlying data.
[0067] Secondly, unified governance and storage of all data are implemented: the data platform uniformly receives raw sensor data (such as facial images, physiological waveforms, GPS coordinates, and video streams) and derived business data (such as attendance records, pay slips, and contract texts) from all stages of the process. For multi-source heterogeneous data, a standardized cleaning process is executed, including removing redundant fields, repairing missing values, standardizing data formats (such as spatiotemporal coordinate standardization), and data anonymization. The cleaned data is then categorized and stored according to subject areas (such as personnel, cost, and spatial themes), constructing a distributed data warehouse to form a unique "data asset pool" for the project.
[0068] Secondly, it supports independent incremental iterative computation of the AI model: The data platform has a built-in algorithm engine that independently supports the operation environment of the AI stress assessment model in step S2 and the labor-specific earned value cost assessment model in step S4. Unlike traditional embedded algorithms, this platform utilizes massive amounts of accumulated historical data (a cross-project database of personnel physiological responses and work efficiency costs) to continuously perform incremental learning and parameter tuning on the model. For example, as new project data flows in, the model automatically updates the "physiological threshold for high-risk jobs" and "work efficiency deviation coefficient," allowing the assessment results and scenario-specific differentiated warning thresholds to dynamically evolve with actual working conditions without requiring modifications to the front-end business code.
[0069] Finally, a cross-process, tiered risk warning distribution system is implemented: the data platform has a built-in unified rule engine that monitors the data flow of the entire business chain in real time, identifying multiple risk events, including but not limited to "entry interception, mismatch between personnel and positions, attendance omissions, and labor cost overruns." Based on a differentiated preset risk control matrix that distinguishes between point-like closed and linear open projects, the identified risk events are graded (e.g., alerts, warnings, critical alarms), and a tiered push strategy is executed: operational-level warnings (e.g., attendance record reminders) are pushed to the mobile devices of front-line team leaders, management-level warnings (e.g., cost overrun analysis) are pushed to the PCs of project managers, and strategic-level warnings (e.g., collective labor risks) are pushed to the enterprise-level monitoring dashboard, achieving accurate delivery and closed-loop handling of risk information.
[0070] This invention can solve the problems of existing technologies, such as single dimension of entry control, lagging safety assessment, poor adaptability to attendance scenarios, disconnect between salary and progress, and difficulty in full-cycle traceability. It achieves the beneficial effects of pre-emptive prevention of inherent safety in labor employment, closed-loop electronic evidence across the entire chain, high-precision attendance in all scenarios, dynamic cost early warning, and enterprise-level credit asset accumulation.
[0071] This invention achieves proactive prevention of employment safety through multi-dimensional entry verification and AI-powered job matching. Existing technologies only collect basic identity information of workers, lacking effective screening of historical performance records and violation records, and failing to address the matching assessment of personnel's physiological conditions with job positions, leading to unsafe personnel entering the workforce. This invention automatically retrieves personnel's cross-project historical performance data and violation records through facial scanning and comparison with the enterprise database, generating early warning pushes. Simultaneously, the ID card information reading unit connects to national or provincial / municipal labor real-name supervision platforms, automatically completing age compliance verification and marking historical violation records. Furthermore, during VR safety training, this invention uses smart wearable devices to collect personnel's physiological parameters such as heart rate and blood pressure in real time, associating physiological responses with high-risk work scenarios (high-altitude work, deep foundation pit work, confined space work), and inputting this data into a pre-trained AI analysis model to assess whether the personnel's psychological adaptability, stress response ability, and physical condition meet the requirements of the declared job position, generating a job matching assessment report. This shifts the focus of workplace safety management from "whether training has been completed" to preventative assessment of "whether the personnel are suitable for the position," thus eliminating the risk of safety accidents caused by unqualified personnel entering the workplace or mismatch between personnel and positions.
[0072] This invention constructs a traceable electronic evidence chain through an integrated closed-loop system of education, assessment, and contract signing. In existing technologies, safety training often employs offline centralized lectures or simple VR demonstrations, lacking real-time quantitative analysis of the physiological and psychological states of trainees during the training process. Furthermore, contract signing remains primarily paper-based, disconnected from training records and assessment results, making it difficult to form a complete legal evidence chain. This invention links VR safety education, smart wearable physiological parameter collection, AI job matching assessment, and online electronic contract signing into a closed-loop process. Electronic labor contracts are signed using an integrated terminal in conjunction with a mobile device, and contract information is linked and stored with personnel information, training records, and assessment results, forming a traceable electronic archive. This mechanism not only enables quantifiable assessment of training effectiveness but also ensures complete archiving of all safety-related data (identity verification records, violation screening results, physiological reaction data, job matching conclusions, and contract signing vouchers) from personnel entry to pre-employment, providing an immutable electronic evidence chain to support labor dispute resolution and safety accident liability determination.
[0073] This invention overcomes the adaptability bottlenecks of different engineering types by integrating differentiated attendance systems with multi-source data. Existing attendance systems mainly rely on access control gates and simple facial recognition, which are poorly adapted to construction scenarios without closed conditions, such as linear engineering projects (highways, railways, pipelines, etc.). Furthermore, attendance data is only used for salary calculation and fails to be integrated with dimensions such as personnel trajectory and work progress for analysis. This invention designs differentiated attendance schemes for point-like and linear engineering projects: For point-like projects, gates are installed at key entrances and exits, and facial recognition attendance data is integrated and verified with eagle-eye camera visual recognition data to generate attendance records; for linear projects, personnel location information is collected by combining drone inspections and mobile smart terminals, and spatiotemporal correlation analysis is performed with visual recognition data to generate attendance records. Simultaneously, personnel identification and positioning units associate location information with work area division data to generate work trajectory data. Therefore, this invention not only meets the efficient and accurate attendance requirements of closed point-like projects but also solves the attendance problems of open linear projects with high personnel mobility and no fixed entrances and exits, achieving universal adaptability across engineering types.
[0074] This invention achieves dynamic quantitative monitoring of labor costs by linking payroll management with earned value analysis. Existing systems rely solely on attendance data for payroll calculation, lacking integration with Work Breakdown Structure (WBS) and labor cost planning, thus failing to dynamically assess the rationality of labor input and the risk status of project labor costs. This invention automatically generates payroll data by linking attendance data with electronic contract payroll rules, and summarizes actual labor cost expenditure data by work package and time period. More importantly, this invention applies the Earned Value Management (EVM) methodology from project management to the specific field of labor employment, dynamically comparing actual labor cost expenditure with the pre-generated labor cost S-curve based on WBS, construction quotas, and labor unit prices, calculating deviation values and deviation rates, and comprehensively evaluating the rationality of labor cost input in conjunction with the actual completed output value of the project. For point-like closed and linear open projects, two-level early warning thresholds are configured to adapt to different working conditions; when indicators exceed the preset thresholds for the corresponding scenarios, an automatic labor risk warning is generated. This technological feature enables project managers to monitor in real time whether labor costs are over budget or in surplus, and whether labor allocation matches the project schedule. It provides a quantitative basis for decision-making on cost correction and resource optimization, filling the technological gap between existing labor management systems and project schedule and cost management.
[0075] This invention achieves end-to-end traceability of personnel trajectories through full-cycle data fusion and digital twin visualization. Existing applications of digital twin technology in construction site management primarily focus on displaying construction progress and limited personnel location, failing to achieve twin mapping and dynamic traceability across the entire lifecycle of personnel, from entry, on-site operations, payroll settlement to exit management. This invention constructs a digital twin foundation based on BIM models and GIS data from the initial project data collection phase. In subsequent stages such as entry data collection, operation data collection, payroll management, and exit management, entry data, operation trajectory data, attendance data, job matching evaluation data, labor cost data, employment risk warning data, and exit status data are integrated into the twin foundation in real time, forming a full-cycle digital twin model of labor employment. Through a visual interactive interface, managers can intuitively view the entry time, operation trajectory at various times, job change records, payroll settlement status, and exit time of any worker throughout the project lifecycle. Furthermore, the cross-project file traceability unit displays the movement trajectory of workers across multiple projects and their performance credit rating information within the digital twin base, enabling end-to-end full-cycle traceability management of workers from joining one project to completing multiple projects.
[0076] This invention decouples business modules from underlying support, ensuring efficient system operation and continuous iteration. Existing labor management systems often mix underlying functions such as data processing, storage, and early warning push with business modules, resulting in high system coupling, poor scalability, and high maintenance costs. This invention decouples the business application layer (project basic data collection, personnel entry data collection, personnel operation data collection, salary management, exit management, and digital twin visualization) from the underlying support layer (data platform and early warning management module). The data platform and early warning management module uniformly receive personnel-related data collected by various business modules, perform data cleaning, fusion, standardization processing, and storage, and provide an operating environment for AI analysis models. Simultaneously, this module centrally monitors the evaluation results of each business module, and generates early warning notifications and pushes them to designated management terminals when preset early warning conditions are triggered, based on differentiated risk control rules for different project types. This architecture enables business modules to be upgraded and expanded independently without affecting the underlying data. AI models can continuously perform incremental learning and parameter optimization using accumulated historical data. Early warning rules can be dynamically adjusted according to the actual operation of the project. From the system architecture level, this invention ensures high scalability, high maintainability, and continuous iterative evolution capabilities.
[0077] This application also provides a computer-readable storage medium storing a computer program that is executed by a processor to implement... Figure 2 The methods provided in each step are detailed in the implementation methods provided in the above steps, and will not be repeated here.
[0078] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A whole-cycle intelligent management method for labor employment of an engineering project, characterized in that, Includes the following steps: S1: Acquire multi-dimensional basic data of the project, integrate BIM 3D geometric data and GIS geospatial data to build a full-domain digital twin foundation; generate the S-curve of labor cost input throughout the project cycle based on WBS work breakdown structure, enterprise and market labor quota data, as a benchmark for dynamic management and control of labor costs. S2: Receive applications from workers to enter the site, and perform cross-project resume verification and government platform real-name verification in parallel. Combine VR high-risk scene physiological time sequence collection and AI stress adaptation assessment to build an integrated entry access mechanism that includes resume screening, quantitative determination of personnel and positions, and electronic signing, forming an unalterable full-link electronic file for personnel. S3: Automatically identify project types, distinguish between point-like closed projects and linear open projects, and configure differentiated multi-source attendance collection strategies; integrate multi-source attendance data to generate a standardized time-series attendance dataset; map the spatial coordinates contained in the dataset to the digital twin base to generate dynamic work trajectories; S4: Based on standardized time-series attendance data, calculate the actual labor cost of WBS work packages, link the full-cycle labor cost input S-curve benchmark to calculate cost deviation, introduce a labor-specific earned value model, combine the actual output value of sub-projects to calculate the cost performance index (CPI), configure independent two-level risk thresholds for point-like closed projects and linear open projects respectively, and push labor risk warnings in a hierarchical manner based on the joint judgment results of deviation rate and labor cost input performance index. S5: Respond to personnel departure requests by fully verifying attendance, salary, and material performance data. After salary settlement, generate encrypted electronic departure vouchers, mark personnel departure status on the digital twin base, and archive all project performance data to the enterprise labor credit database to generate personnel credit profiles that can be retrieved across projects. S6: Map the full-cycle business data flow from steps S1 to S5 to the digital twin platform in real time to build an integrated interactive interface that supports full-cycle traceability for individual soldiers, cross-project credit queries, and visualization of multi-dimensional cost indicators. S7: Set up an independent data platform that is decoupled from all upper-level business steps to complete unified governance of all heterogeneous data, incremental iteration of AI models, and precise push of risk stratification and classification across the entire chain.
2. The intelligent management method for the entire lifecycle of labor employment in engineering projects according to claim 1, characterized in that, The multidimensional basic data of the project in step S1 includes: geometric space dimension data, construction organization dimension data, economic quota dimension data, and schedule plan dimension data; The geometric spatial dimension data includes at least the project's BIM 3D geometric data and GIS geospatial data; The construction organization dimension data includes at least construction drawings and construction organization design documents; The economic quota dimension data should include at least the company's internal construction quotas and the latest labor unit prices for different trades in the project location. The schedule dimension data includes at least a list of work packages generated based on the WBS hierarchical breakdown, and the time milestone nodes bound to each work package.
3. The intelligent management method for the entire lifecycle of labor employment in engineering projects according to claim 2, characterized in that, The specific process for generating the S-curve of total lifecycle labor cost input in step S1 is as follows: Acquire BIM 3D geometric data and GIS geospatial data of the target project, perform coordinate registration and fusion of the two, and construct a high-precision digital twin base covering the entire project area; Import the construction drawings, construction organization design documents, internal construction quotas, and the latest labor unit prices for different trades in the project location into the digital twin base; perform WBS hierarchical decomposition on the project scope, decompose the project into independent work packages at each level, and bind the planned start time, planned end time, and key milestone nodes to each work package; By combining construction quotas, labor unit prices, and the schedule milestones of each work package, the labor costs required for each milestone are calculated cumulatively over time, generating a cumulative S-curve of labor cost input throughout the project lifecycle.
4. A method for intelligent management of the entire lifecycle of labor employment in engineering projects according to any one of claims 1-3, characterized in that, Step S2 specifically includes: The system acquires the facial biometrics of workers through facial recognition devices and obtains their real-name information through OCR recognition or NFC reading devices. It also connects to the company's historical labor database and the government-level labor real-name supervision platform to cross-verify the authenticity of the workers' identities and automatically screens whether the workers have any cross-project historical violations, poor performance records, or industry bans. If high-risk records are found, an entry interception warning is immediately generated and pushed to the security management terminal. Those who fail the review are prohibited from entering the site. Workers who pass the initial screening are guided into the VR safety education experience hall, where high-risk work scenarios are simulated based on the types of jobs they have applied for. During the immersive experience, the smart wearable devices worn by the workers are connected via Bluetooth to collect their physiological parameters such as heart rate and blood pressure fluctuations in different stress scenarios in real time. The collected time-series physiological parameters are input into a pre-trained AI analysis model to assess the psychological adaptability, stress response ability, and physical endurance of workers in a simulated high-risk environment, and output a quantitative job matching assessment report. For personnel who pass the matching assessment, their identity data, violation screening records, VR training records, physiological parameter logs, AI assessment reports, and signed electronic contracts are linked and packaged together, and blockchain technology or timestamp technology is used for evidence preservation to form an immutable and traceable electronic entry file.
5. The intelligent management method for the entire lifecycle of labor employment in engineering projects according to claim 4, characterized in that, Step S3 specifically includes: Identify the type of engineering scenario and configure targeted attendance collection strategies; if it is a point-like closed project, enable the dual verification mode of turnstile + eagle eye; if it is determined to be a linear open project, enable the wide-area stitching mode of drone + mobile terminal + along-line video; through multi-sensor data fusion algorithm, stitch together to generate continuous position information of personnel on long-distance linear work surface. Each standardized attendance record generated is spatially correlated with the BIM work surface division data of the construction area to parse out the specific work position or construction section where the person is currently located and extract their real-time spatial coordinates; the output is a standardized time-series attendance dataset containing the fields of person ID, timestamp, spatial coordinates and work package. Based on the standardized time-series attendance dataset, the spatial coordinates contained in the dataset are mapped to the BIM+GIS digital twin base constructed in step S1 through a real-time data interface, driving the virtual personnel model in the twin space to move synchronously and generating a dynamic work trajectory consistent with the physical world.
6. The intelligent management method for the entire lifecycle of labor employment in engineering projects according to claim 5, characterized in that, Step S4 specifically includes: Retrieve standardized time-series attendance datasets, associate them with the salary calculation rules in electronic labor contracts to automatically calculate and summarize individual salary details; according to the WBS work package affiliation, collect and summarize the salary details of the personnel to which they belong, calculate the total actual labor cost of each work package within a specific time period, and form a dataset of actual labor cost expenditure for the project. Retrieve the S-curve of labor cost input for the entire project cycle and extract the planned labor cost value corresponding to the current statistical period; compare the actual labor cost expenditure dataset AC with the planned labor cost value PV, and calculate the absolute deviation value SV and the relative deviation rate SV%; SV = PV - AC; relative deviation rate SV% = (PV - AC) / PV × 100%; The earned value management approach is adopted, and the actual completed output value (EV) of each sub-project is introduced as a key variable; a labor-specific earned value cost assessment model is constructed to calculate the labor cost input performance index (CPI); CPI = EV / AC, where AC is the actual labor cost; Two independent early warning thresholds are set according to the type of project: For point-based closure projects, the preset first-level mild warning threshold range is: And the relative deviation rate of labor costs For linear open engineering, a preset first-level mild early warning threshold range is defined as follows: And the relative deviation rate of labor costs When the deviation rate Labor cost input performance index If the cost falls within the corresponding threshold range, it is determined to be a slight fluctuation in labor costs. Cost reminder information is automatically generated, highlighting the current abnormal work packages and personnel attendance density. Lightweight messages are pushed to the mobile terminals of on-site construction workers and labor managers to remind them to pay attention to on-site labor dynamics and to conduct on-site review or fine-tuning when necessary. For point-based closure projects, a pre-set Level II severe warning trigger condition is established: Or relative deviation rate of labor costs For linear open engineering projects, a pre-defined level-two severe early warning trigger condition is established: Or relative deviation rate of labor costs When the deviation rate Labor cost input performance index If any one of the threshold conditions is met, it is determined as a project-level overspending alarm; an in-depth analysis report containing a preliminary diagnosis of the cause of the deviation is automatically generated and pushed to the project manager and the company's cost control center.
7. The intelligent management method for the entire lifecycle of labor employment in engineering projects according to claim 6, characterized in that, Step S5 specifically includes: Automatically correlate and query the full-dimensional business data of workers to be discharged in this project, and construct a discharge verification checklist; The system automatically calculates and confirms that all payments have been settled based on the verification checklist; after confirmation, it generates an encrypted electronic exit certificate containing a unique hash value or digital signature. The virtual twin status of the corresponding worker to be discharged in the digital twin base is permanently marked from "on duty" to "discharged", the real-time mapping update of the worker's work trajectory is terminated, and the worker to be discharged is grayscaled or moved into the historical personnel layer in the visualization interface. All business data generated by the workers to be discharged from the project are packaged and archived, and synchronously written into the enterprise's global labor credit database to form credit archive data entries corresponding to the workers to be discharged. The credit archive data entries are strongly linked to the unique identifiers of the workers to be discharged, forming a dynamically updated labor credit profile.
8. A method for intelligent management of the entire lifecycle of labor employment in engineering projects according to any one of claims 1-3 and 5-7, characterized in that, Step S6 specifically includes: A high-throughput message middleware is built as a real-time data synchronization channel; the multi-source heterogeneous data collected in steps S1-S5 is processed by streaming ETL and then synchronously written into the digital twin base. On top of the digital twin platform, a multi-dimensional visual interactive interface is generated, which renders the heat map of personnel at the construction site, the density of job distribution, the concentration of personnel in key areas, and the real-time employment status of each WBS work package. The interface allows managers to initiate a full-cycle traceability query for a single person by clicking on the virtual personnel model in the twin interface or entering the personnel's identity information. The interface also supports cross-project and cross-cycle data fusion queries.
9. A method for intelligent management of the entire lifecycle of labor use in engineering projects according to claim 8, characterized in that, Step S7 specifically includes: A layered and decoupled architecture is adopted, encapsulating all business logic in steps S1 to S6 into a business application layer, while data governance, model calculation, and early warning distribution are separated into an independent data platform; the two layers interact with each other through a standardized API gateway. The data platform uniformly receives raw perception data and derived business data from all stages; for multi-source heterogeneous data, it performs standardized cleaning processes and data anonymization; the cleaned data is classified and stored according to subject domains, and a distributed data warehouse is built to form a unique data asset pool for the project. Through the built-in algorithm engine of the data platform, and by utilizing the massive amount of historical data accumulated, the AI stress assessment model in step S2 and the labor-specific earned value cost model in step S4 are continuously incrementally learned and their parameters are optimized. By using a unified rules engine built into the data platform, the data flow of the entire business chain can be monitored in real time to identify multiple types of risk events; differentiate between point-like and linear engineering risk control matrices, classify the identified risk events into levels, and push them to the corresponding terminals in layers according to operational, management, and strategic levels.
10. A smart management system for the entire lifecycle of labor employment in engineering projects, characterized in that, A method for implementing intelligent management of the entire labor lifecycle of engineering projects as described in any one of claims 1-9, comprising: The project basic data acquisition module is used to acquire multi-dimensional basic data of the project, integrate BIM 3D geometric data and GIS geospatial data to build a full-domain digital twin foundation; and generate the S-curve of labor cost input throughout the project cycle based on WBS work breakdown structure, enterprise and market labor quota data, as a benchmark for dynamic management and control of labor costs. The multi-dimensional personnel entry control module is used to receive labor personnel entry applications, and to perform cross-project resume verification and government platform real-name verification in parallel. It combines VR high-risk scenario physiological time sequence collection and AI stress adaptation assessment to build an integrated entry access mechanism that includes resume screening, quantitative determination of personnel and positions, and electronic signing, forming an unalterable full-link electronic file for personnel. The operation data acquisition and trajectory construction module is used to automatically identify the project type, distinguish between point-like closed projects and linear open projects, configure differentiated multi-source attendance acquisition strategies, integrate multi-source attendance data to generate a standardized time-series attendance dataset, and map the spatial coordinates contained in the dataset to the digital twin base to generate a dynamic operation trajectory. The payroll calculation and cost early warning module is used to calculate the actual labor cost of WBS work packages based on standardized time-series attendance data, link the full-cycle labor cost input S-curve benchmark to calculate cost deviation, introduce a labor-specific earned value model, and calculate the cost performance index (CPI) based on the actual output value of sub-projects. Independent two-level risk thresholds are configured for point-like closed projects and linear open projects respectively. Based on the joint judgment results of deviation rate and labor cost input performance index, layered labor risk early warning is pushed. The exit closed-loop archiving module is used to fully verify attendance, salary, and material performance data in response to personnel exit requests. After salary is settled, an encrypted electronic exit voucher is generated, the personnel exit status is marked on the digital twin base, and the entire project performance data is archived to the enterprise labor credit database to generate personnel credit profiles that can be retrieved across projects. The digital twin visualization building module is used to map the full-cycle business data flow to the digital twin base in real time, and build an integrated interactive interface that supports full-cycle traceability for individual soldiers, cross-project credit query and visualization of multi-dimensional cost indicators. The data platform processing and early warning module is used to set up an independent data platform that is decoupled from all upper-level business steps, and to complete the unified governance of all heterogeneous data, incremental iteration of AI models, and precise push of risk stratification and classification across the entire chain.
