A building space light quantitative measurement and construction working hour data statistical analysis system

Abstract

This invention discloses a system for quantitative measurement of building space illumination and statistical analysis of construction time data, relating to the field of smart construction site management technology. It includes a comprehensive management center, which is communicatively connected to the following modules: an environmental dynamic sensing module, used to deploy an intelligent sensing system at the construction site, employing a particle swarm optimization algorithm to collect and analyze illumination data from multiple points in real time, generating a real-time illuminance distribution map and dynamically identifying abnormal illumination intervals. This invention, by deploying an intelligent illumination sensing network covering the entire construction space, can collect multi-point, multi-dimensional illumination data, and automatically clean and correct the data using a particle swarm optimization algorithm, identifying abnormal illumination intervals and generating a high-precision real-time illuminance distribution map. This changes the traditional extensive mode of construction management, which lacks quantitative monitoring of environmental factors, elevating illumination conditions from vague perception to precise measurement and visual control, providing accurate and reliable environmental status input for construction decisions.

Description

Technical Field

[0001] This invention relates to the field of smart construction site management technology, specifically a system for quantitative measurement of building space illumination and statistical analysis of construction time data. Background Technology

[0002] With the acceleration of urbanization, the requirements for the functionality and comfort of building spaces are increasing. Reasonable lighting design can significantly improve the user experience of the space, while quantitative measurement can objectively assess the quality of lighting conditions and provide a scientific basis for building design. At the same time, lighting measurement provides data support for subsequent energy consumption analysis, environmental protection and sustainable development, promoting the development of green buildings. In terms of construction management, the statistics of working hours data help optimize the construction process and resource allocation, and improve construction efficiency.

[0003] For example, a refined engineering construction management system, published in Chinese Patent Publication No. CN111898853A, records in detail the material consumption, man-hours, and process time of each stage of construction through the application of material flow management data collection. It also records the details of the construction process in real time. The system can quickly, efficiently, and accurately generate statistical analysis reports on various information, which are traceable and queryable, and at the same time realize the management of quality, progress, and personnel.

[0004] While existing construction management systems have achieved automated management of materials, schedules, and personnel processes, they still lack real-time quantitative perception and dynamic impact analysis of environmental factors at the construction site, especially lighting conditions. This makes it difficult to identify the immediate interference of different lighting intensities on the work efficiency of construction workers, resulting in construction time statistics remaining at the static recording level. Consequently, they cannot reflect the actual fluctuations in working hours caused by environmental changes. Furthermore, due to the lack of a dynamic correlation model between lighting parameters and construction task execution, it is difficult to adaptively adjust the construction plan under abnormal lighting conditions, leading to the continued use of fixed schedules, resulting in reduced construction efficiency and potential safety risks. Simultaneously, the impact of long-term lighting changes on the cumulative fatigue of construction workers and construction quality has not been systematically modeled and analyzed. This results in a lack of environmental factor-based quality risk early warning and optimization mechanisms throughout the construction process, limiting the level of refinement and intelligence in construction management. Therefore, this paper proposes a quantitative measurement system for building space lighting and a statistical analysis system for construction time data to address the aforementioned problems. Summary of the Invention

[0005] To solve the above-mentioned technical problems, the present invention is implemented through the following technical solution: a system for quantitative measurement of building space illumination and statistical analysis of construction time data, including a comprehensive management center, wherein the comprehensive management center is communicatively connected to the following modules: The environmental dynamic perception module is used to deploy an intelligent sensing system at the construction site, use the particle swarm algorithm to collect and analyze illumination data from multiple points in real time, generate a real-time illuminance distribution map and dynamically identify abnormal illumination intervals, so as to realize quantitative monitoring and impact assessment of illumination conditions. The construction time impact prediction module is used to integrate digital twin technology to build a virtual model of the construction process. Combined with reinforcement learning algorithms, it predicts the fluctuation impact of real-time lighting data on construction time, realizing the leap from static recording to dynamic simulation. The intelligent task sorting module receives prediction data on the impact of work hour fluctuations, and actively intervenes in the construction plan in conjunction with the construction task network diagram. It dynamically adjusts the task order on non-critical paths when there are abnormal lighting conditions, actively avoids environmental interference, and improves the adaptability and robustness of the plan to dynamic environmental changes. The resource elastic configuration module is used to dynamically adjust the configuration scheme of personnel and equipment based on the lighting environment prediction and task rescheduling results, under the premise of meeting the total project duration and process logic constraints, generate an adaptive scheduling scheme that is resistant to lighting interference, and push it to the integrated management center to optimize resource allocation and improve resource utilization and overall construction efficiency. The quality risk early warning module is used to establish a three-dimensional risk assessment model of light-fatigue-quality by fusing convolutional neural networks and time series prediction models. It provides intelligent early warning of fatigue accumulation and quality risks caused by light environment during long-term construction, and ensures construction quality and personnel safety from the source.

[0006] Preferably, the environmental dynamic perception module includes a light network perception unit and an abnormal light recognition unit; The light network sensing unit is used to deploy an intelligent sensing system composed of intelligent light sensors in key work areas and personnel activity areas of the construction site, collect light data from multiple points in real time, and build a multi-source, continuous, and high-precision light sensing network covering the entire construction space. The abnormal lighting identification unit is used to clean, correct, and fuse the collected multi-point lighting data. It uses the particle swarm optimization algorithm to dynamically identify and remove outliers, generate real-time global and local illuminance distribution maps, delineate and mark the normal operating illuminance range and abnormal lighting range, realize intelligent diagnosis of environmental conditions, quickly locate lighting anomalies in the working environment, and achieve automatic diagnosis from data to risk.

[0007] Preferably, the execution steps of the illumination network sensing unit include: Based on the construction site layout and operational requirements, the lighting monitoring points for key work areas and personnel activity areas are determined, and intelligent lighting sensor nodes with wireless communication capabilities are deployed to build an intelligent sensing system covering the entire construction space, which can accurately capture environmental changes in key areas. The system collects real-time light data, including light intensity, spectral characteristics and timestamps, from each monitoring point through an intelligent sensing system. Combined with the Internet of Things protocol, the data is synchronously uploaded to the edge computing gateway to form a multi-source, multi-dimensional raw light dataset, realizing fully automatic, uninterrupted and low-latency aggregation of on-site light environment data. The original illumination dataset is processed for temporal alignment and spatial registration to generate real-time environmental illumination field data based on time series and spatial coordinates, providing a high-quality, structured, and spatiotemporally unified data foundation for high-level analysis and visualization.

[0008] Preferably, the execution steps of the abnormal lighting recognition unit include: Noise filtering and data correction were performed on the received multi-point illumination data. The sliding window mean method and sensor calibration parameters were used for data cleaning to eliminate equipment errors and transient interference. The signal-to-noise ratio of the processed data was significantly improved, laying a high-quality foundation for subsequent analysis. The particle swarm optimization algorithm is used to perform cluster analysis and outlier detection on the cleaned multi-point illumination data, dynamically identify outlier areas in the illumination distribution, and delineate normal illumination ranges and abnormal illumination ranges based on preset illumination operation thresholds. This can accurately locate abnormal illumination points and improve the accuracy of early warning. Based on the spatiotemporal distribution characteristics of abnormal illumination intervals, global and local real-time illuminance distribution maps are generated, and abnormal areas and their deviation degrees are marked in a visual form on the maps, thereby realizing intelligent diagnosis and abnormal location of environmental conditions.

[0009] Preferably, the time impact deduction module includes a virtual-real mapping construction unit and a time impact deduction unit; The virtual-real mapping construction unit, based on BIM, schedule planning and personnel information, integrates digital twin technology to construct a virtual model of the construction process that is synchronized with the physical space, and maps multi-point lighting data to the corresponding spatiotemporal locations in the virtual model of the construction process in real time, providing a visualized environmental background for construction activities, realizing precise synchronization between the physical world and the virtual model, and providing a visualized analysis and decision-making platform. The time impact estimation unit is used to run a simulation proxy model based on reinforcement learning algorithm in a digital twin environment. Based on the mapped illumination data, it dynamically simulates and predicts the changes in the work efficiency of each type of work under different illumination conditions, calculates the actual fluctuation impact of illumination on standard working hours, and dynamically predicts the fluctuation of work efficiency of each type of work, so as to achieve accurate quantification and forward-looking assessment of the time impact.

[0010] Preferably, the execution steps of the virtual-real mapping construction unit include: Based on the construction BIM model, schedule data and personnel organization structure, digital twin technology is integrated to build a virtual model of the construction process that includes geometric structure, construction procedures and resource attributes, so as to realize the structural synchronization between physical space and virtual model, and support multi-terminal collaboration and remote monitoring. Real-time multi-point illumination data is mapped to the corresponding spatial location and construction time node in the construction process virtual model through coordinate transformation and time synchronization mechanism, forming a virtual illumination field layer consistent with the real environment. The spatiotemporal superposition effect of the virtual illumination field layer and construction activities is dynamically displayed in the construction process virtual model, providing a real-time and visualized environmental background and data association interface for the construction process. This enables precise integration and visualization of environmental data and construction activities, helping managers to intuitively grasp the impact of the environment on construction.

[0011] Preferably, the execution steps of the time impact simulation unit include: In the virtual model of the construction process, a simulation agent model based on reinforcement learning algorithm is constructed. The operation activities of each type of work are used as intelligent agents, and real-time lighting data is used as environmental state input. The lighting-efficiency mapping relationship is established through reinforcement learning training, so that the system can automatically learn and establish response models of the operation efficiency of each type of work under different lighting conditions, thereby improving the objectivity of the analysis. Based on the real-time mapped virtual layer data of the illumination field, the simulation agent model is driven to dynamically simulate the change process of the work efficiency of each type of work under different illumination conditions, output the actual work efficiency coefficient under the influence of illumination, realize the real-time dynamic prediction of the work efficiency of each type of work, and provide an immediate and quantitative decision basis for adjusting working hours. By combining the standard time quota for construction tasks and calculating the impact of time fluctuations caused by sunlight based on the work efficiency coefficient, a time impact report categorized by job type and process is generated. This quantifies the environmental impact into specific time fluctuation data, forming a traceable and analyzable special report to support management optimization.

[0012] Preferably, the execution steps of the task intelligent sorting module include: By receiving data on the impact of work hour fluctuations and combining the process dependencies and critical path analysis in the construction task network diagram, the set of non-critical path tasks that are significantly affected by lighting can be identified, thereby achieving efficient location of non-critical tasks and centralized identification of interference. Based on the temporal prediction and spatial distribution of abnormal lighting intervals, non-critical path tasks are dynamically prioritized and rearranged. The execution order or time period is adjusted under the premise of meeting the process logic to avoid or mitigate lighting interference, effectively reduce environmental interference, and ensure stable construction rhythm. The adjusted construction task sequence is generated and verified through simulation in a digital twin environment to confirm the feasibility of the adjusted task sequence under the constraints of schedule and resources. The optimized construction schedule plan is then output to ensure the feasibility and executability of the plan adjustment.

[0013] Preferably, the execution steps of the resource elastic configuration module include: Based on the predicted data of the lighting environment and the optimized construction schedule, we analyze the changes in personnel and equipment demand in different time periods and regions, identify resource allocation conflicts and redundancies, effectively avoid resource idleness and over-concentration, and improve the overall resource utilization rate. Under the condition of meeting the overall project period and construction process constraints, a resource scheduling optimization model is established to dynamically adjust personnel shifts, equipment allocation plans and material supply rhythm, generate an adaptive resource allocation scheme that resists light interference, realize the dynamic matching of resources and light environment, and reduce efficiency loss caused by environmental interference. The optimized resource allocation scheme is pushed to the integrated management center and synchronized with the resource entities in the digital twin environment to achieve real-time visualization and dynamic control of resource scheduling, enhance scheduling transparency and response speed, and support managers to monitor and make rapid decisions in real time.

[0014] Preferably, the execution steps of the quality risk early warning module include: Based on long-term collected illumination data, personnel fatigue monitoring data, and construction quality inspection records, a three-element risk assessment model of illumination-fatigue-quality is constructed, which integrates convolutional neural network and time series prediction model. The correlation between illumination fluctuations, fatigue accumulation, and quality deviation is established, effectively identifying the hidden hazards of long-term poor illumination to personnel health and construction quality. Real-time input of current and predicted illumination data into the risk assessment model dynamically calculates the fatigue risk index and quality deviation probability caused by illumination environment in each region and process, and compares them with their respective preset thresholds to achieve real-time quantitative assessment and advanced perception of potential risks. When the fatigue risk index or the probability of quality deviation exceeds its respective preset threshold, a graded early warning mechanism is automatically triggered, pushing early warning information to managers and recommending targeted intervention measures based on the risk type, including adjusting lighting, arranging intermittent rest, or strengthening process inspection, which significantly improves the speed of risk response and the accuracy of handling, and curbs quality accidents from the source.

[0015] This invention provides a system for quantitative measurement of building space illumination and statistical analysis of construction time data. It has the following beneficial effects: (I) This system for quantitative measurement of building space illumination and statistical analysis of construction time data, by deploying an intelligent illumination sensor network covering the entire construction space, can collect illumination data from multiple points and dimensions, and automatically clean and correct the data and identify abnormal illumination intervals by combining the particle swarm algorithm, generating a high-precision real-time illuminance distribution map. This changes the extensive mode of traditional construction management that lacks quantitative monitoring of environmental factors, and improves the illumination conditions from vague perception to precise measurement and visual control, providing accurate and reliable environmental status input for construction decision-making.

[0016] (II) This system for quantitative measurement of building space illumination and statistical analysis of construction time data accurately maps real-time illumination data to a virtual model and constructs simulation agents based on deep reinforcement learning for each key trade. It can dynamically simulate and predict changes in work efficiency under different illumination conditions, scientifically calculate the actual fluctuation impact of illumination on standard working hours, and achieve a fundamental leap in construction time management from static recording and post-event statistics to dynamic simulation and pre-event prediction. It provides an intelligent analysis tool for quantitatively assessing environmental interference and formulating construction plans.

[0017] (III) This quantitative measurement system for building space illumination and statistical analysis of construction time data can automatically and dynamically intervene in the construction plan based on the illumination impact prediction results through intelligent task sorting and flexible resource allocation. Under the premise of ensuring the total construction period and process logic, it can intelligently adjust the sequence of non-critical path tasks and optimize the allocation of personnel, equipment and other resources to generate an adaptive scheduling scheme that resists illumination interference. This significantly improves the resilience and flexibility of the construction plan in response to environmental changes, realizes the transformation from fixed scheduling to dynamic response, and from passive response to active avoidance, and effectively reduces efficiency loss and safety risks caused by abnormal illumination. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the workflow of a system for quantitative measurement of building space illumination and statistical analysis of construction time data according to the present invention; Figure 2 This is a data flow diagram of a system for quantitative measurement of building space illumination and statistical analysis of construction time data according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1, please refer to Figure 1 , Figure 2 This invention provides a technical solution: a system for quantitative measurement of building space illumination and statistical analysis of construction time data, including a comprehensive management center, which is connected to the following modules for communication: The environmental dynamic perception module is used to deploy an intelligent sensing system at the construction site, use the particle swarm algorithm to collect and analyze light data from multiple points in real time, generate a real-time illuminance distribution map and dynamically identify abnormal light intervals, realize quantitative monitoring and impact assessment of light conditions, and provide accurate and real-time environmental data support for subsequent work time simulation and risk warning. The environmental dynamic perception module includes a light network sensing unit and an abnormal light identification unit. The lighting network sensing unit is used to deploy an intelligent sensing system composed of intelligent lighting sensors in key work areas and personnel activity areas of the construction site. It collects lighting data from multiple points in real time, and builds a multi-source, continuous, and high-precision lighting sensing network covering the entire construction space. Based on the construction layout and work process requirements, it determines the lighting monitoring points in key work areas and personnel activity areas, and deploys intelligent lighting sensor nodes with wireless communication capabilities to build an intelligent sensing system covering the entire construction space. This system can accurately capture environmental changes in key areas. The intelligent sensing system collects lighting data, including light intensity, spectral characteristics, and timestamps, from each monitoring point in real time. Combined with the Internet of Things protocol, the data is synchronously uploaded to the edge computing gateway to form a multi-source, multi-dimensional raw lighting dataset. This enables fully automatic, uninterrupted, and low-latency aggregation of on-site lighting environment data. The raw lighting dataset is then processed for time alignment and spatial registration to generate real-time environmental lighting field data based on time series and spatial coordinates. This provides a high-quality, structured, and spatiotemporally unified data foundation for high-level analysis and visualization. It should be noted that, based on the construction site layout and work process analysis, key work surfaces (rebar tying areas, formwork support surfaces, concrete pouring surfaces, decoration work surfaces, etc.) and areas with high-frequency personnel activity (material storage areas, passageways, temporary rest areas, etc.) were identified as core locations for light monitoring. Intelligent light sensor nodes equipped with high-precision photoelectric sensing elements (silicon photodiode sensors) were selected. Each node has a built-in LoRa communication module, supporting low-power wide-area network transmission. Sensor nodes are deployed evenly at 30-50 meter intervals on construction floors or work areas according to a gridded or critical path deployment principle. Additional monitoring points are added at high-altitude work surfaces, dark corners, and areas with significant changes in lighting, forming a multi-layered, redundant intelligent sensor network covering the entire construction space. All nodes are networked via a wireless self-organizing network protocol to ensure continuous and reliable data transmission. The deployed intelligent sensor system collects real-time data on light intensity, spectral characteristics, and high-precision timestamps at each monitoring point at a sampling frequency of once per second. The collected data is transmitted via the LoRaWAN protocol through the construction... The LoRa gateway deployed on-site synchronously uploads data to the edge computing gateway in JSON format. The edge computing gateway performs preliminary verification and timestamp alignment on the received raw data, removes obvious outliers, and stores it in a structured manner according to sensor ID, coordinate location, and time series, forming a raw illumination dataset containing spatial location attributes, time attributes, and multi-dimensional illumination indicators. The edge computing gateway performs time synchronization and alignment on the data uploaded by each sensor node based on the unified Coordinated Universal Time (UTC). At the same time, based on the predefined spatial coordinates of each sensor in the construction BIM model, it performs spatial registration on the raw dataset. Through time series interpolation algorithm and spatial kriging interpolation, it fuses the asynchronous illumination data of discrete points to generate continuous and complete real-time environmental illumination field data. This data is expressed in a gridded form, with the grid resolution set to 1 meter × 1 meter as needed. Each grid cell contains the illumination intensity, color temperature, and confidence index of the corresponding spatial location at a specific time point. The final generated illumination field data is pushed to the integrated management center in real time through a RESTful API interface. The abnormal lighting identification unit is used to clean, correct, and fuse multi-point lighting data. It uses a particle swarm optimization algorithm to dynamically identify and remove outliers, generating real-time global and local illuminance distribution maps. It delineates and marks normal operating illuminance ranges and abnormal lighting ranges, enabling intelligent diagnosis of environmental conditions and rapid location of lighting anomalies in the working environment. This achieves automatic diagnosis from data to risk. The unit performs noise filtering and data correction on the received multi-point lighting data, using a sliding window averaging method and sensor calibration parameters for data cleaning to eliminate equipment errors and transient interference. The processed data has a significantly improved signal-to-noise ratio, providing a foundation for subsequent processing. Further analysis lays a high-quality foundation. Particle swarm optimization is used to perform cluster analysis and outlier detection on the cleaned multi-point illumination data. Outlier areas in the illumination distribution are dynamically identified, and normal and abnormal illumination intervals are defined according to preset illumination operation thresholds. This can accurately locate abnormal illumination points and improve the accuracy of early warning. Based on the spatiotemporal distribution characteristics of abnormal illumination intervals, global and local real-time illumination distribution maps are generated, and abnormal areas and their deviations are marked in a visual form on the maps. This enables intelligent diagnosis and anomaly location of the environmental status. The visualization is intuitive and clear, helping managers to quickly locate and respond to environmental anomalies. It should be noted that after acquiring the raw data of light intensity, spectral characteristics, and timestamps from multiple points, the data is preprocessed to improve its quality. A sliding window with a fixed duration of 5 seconds is used to calculate a moving average of the data stream uploaded by each sensor node. This sliding window averaging method can effectively smooth reading spikes caused by momentary occlusion, electronic noise, or short-term interference. Simultaneously, based on the independent calibration parameters of each smart light sensor at the factory (including zero-point drift correction coefficients)... and sensitivity correction coefficient The original voltage signal is linearly corrected and converted into an illuminance value in standard units (lux, lx). The calibration formula is as follows: ,in, The sensor output voltage, For dark voltage, after cleaning and correction, the inherent equipment errors and instantaneous environmental interference in the original dataset are significantly suppressed. For the cleaned and corrected illuminance dataset, a particle swarm optimization (PSO) clustering analysis method is applied for anomaly detection. The particle swarm size is set to 50, and the maximum number of iterations is set to 200. A multi-dimensional feature space is constructed using the illuminance values ​​and spatiotemporal coordinates between data points. The algorithm automatically searches for and delineates core areas with dense data point distribution within the feature space, and identifies outliers far from these core areas as potential anomalies. Simultaneously, based on the "Standard for Lighting Design of Buildings" GB50034-2013 and specific construction work safety regulations, illuminance threshold ranges for different work areas are preset. Specifically, the normal illuminance range for fine decoration work surfaces is set to 300 lx to 750 lx, and for material stacking channels, it is set to 150 lx to 300 lx. The set of outliers identified by the PSO algorithm is compared and fused with the preset work threshold ranges to dynamically define and mark abnormal lighting ranges in the current construction environment that are below the safety lower limit or above the glare upper limit. Record its spatial coordinates, time range, and degree of deviation; based on the identified abnormal lighting intervals and their spatiotemporal distribution characteristics, automatically generate real-time illuminance distribution maps at two levels: global and local. The global map uses the entire construction plane as the base map, with 1m×1m grids as units, and continuously renders the real-time illuminance values ​​of each grid unit using a color gradient (from dark blue to bright yellow). The local map focuses on the identified abnormal intervals and is visualized at a resolution of 0.5m×0.5m. In the distribution map, abnormal lighting intervals are marked with a bright outline (red flashing boundary), and their specific illuminance values, deviations from the safety threshold, and percentages are overlaid. All distribution maps are updated in real time through the digital dashboard of the integrated management center and linked to the BIM model for three-dimensional positioning display. A structured anomaly diagnosis report is generated simultaneously, including the anomaly area number, location description (linked to BIM component ID), anomaly start and end time, duration, maximum deviation value, and suggested inspection priority (high / medium / low), realizing intelligent diagnosis and precise positioning of environmental conditions. The construction time impact prediction module is used to integrate digital twin technology to build a virtual model of the construction process. Combined with reinforcement learning algorithms, it predicts the impact of real-time illumination data on the fluctuation of construction time, realizing the leap from static recording to dynamic simulation. It quantifies the environmental impact at the time level and provides a scientific basis for dynamic adjustment of the plan. The construction time impact prediction module includes a virtual-real mapping construction unit and a construction time impact prediction unit. The virtual-real mapping construction unit, based on BIM, schedule planning, and personnel information, integrates digital twin technology to construct a virtual model of the construction process synchronized with the physical space. It maps multi-point illumination data to the corresponding spatiotemporal locations in the virtual model in real time, providing a visualized environmental background for construction activities. This achieves precise synchronization between the physical world and the virtual model, providing a visual analysis and decision-making platform. Based on the construction BIM model, schedule planning data, and personnel organizational structure, it integrates digital twin technology to construct a virtual model of the construction process, including geometric structure, construction procedures, and resource attributes. This achieves structural synchronization between the physical space and the virtual model, supports multi-terminal collaboration and remote monitoring, and maps real-time collected multi-point illumination data to the corresponding spatial locations and construction time nodes in the virtual model through coordinate transformation and time synchronization mechanisms. This forms a virtual illumination field layer consistent with the real environment, and dynamically displays the spatiotemporal overlay effect of the virtual illumination field layer and construction activities in the virtual model. This provides a real-time, visualized environmental background and data association interface for the construction process, achieving precise integration and visualization of environmental data and construction activities, and assisting managers in intuitively understanding the impact of the environment on construction. It should be noted that, based on the construction BIM model and schedule data, a virtual construction process model with geometric structure, work procedures, and resource attribute associations is constructed. The virtual construction process model is imported using a lightweight BIM format and, combined with the work procedure logic and time parameters clearly defined in the construction organization design, achieves a 3D dynamic display of each construction stage through a timeline-driven approach. The virtual construction process model is divided into twin units based on the construction area. Each twin unit is associated with the start and end times of the corresponding work procedure, the required trades, and the type of equipment, and supports automatic verification and updates with actual on-site progress data. The virtual construction process model is deployed through the WebGL platform, enabling real-time access and interactive operation across multiple terminals. Through coordinate transformation and time synchronization mechanisms, multi-point illumination data collected by on-site intelligent sensors is accurately mapped to the virtual construction process model. Coordinate transformation is performed based on the transformation matrix between the construction control network coordinate system and the BIM model coordinate system to ensure that the physical location of each sensor node corresponds to its coordinate system in the virtual construction process model. To ensure consistency, the system synchronizes timestamps from all data sources using the Network Time Protocol (NTP), with errors controlled within milliseconds. The mapped illumination data is then overlaid on the surface of the virtual model of the construction process as a dynamic layer, forming a spatially continuous virtual illumination field. Within the digital twin platform, the virtual illumination field layer and construction activities are spatiotemporally overlaid and visualized. Real-time illuminance data is linked to process time nodes, and the distribution of illumination intensity in each area is dynamically displayed on the surface of the virtual model of the construction process using a coloring algorithm. The impact of illumination conditions on operations is simulated in conjunction with the construction progress. The virtual illumination field data is stored in a grid format with a grid resolution of 0.5m × 0.5m. Each grid cell contains illuminance value, color temperature, and update time attributes. The digital twin platform provides layered display control functionality, allowing users to independently access illumination layers, process models, or resource distribution views. Users can also interactively query illumination data and related process information at any location by clicking, forming a visual analysis interface that integrates environmental background and construction process. The time-related impact simulation unit is used in a digital twin environment to run a simulation agent model based on reinforcement learning algorithms. Based on mapped illumination data, it dynamically simulates and predicts changes in the work efficiency of various trades under different illumination conditions, calculates the actual fluctuation impact of illumination on standard working hours, and dynamically predicts the fluctuation of work efficiency for each trade. This achieves accurate quantification and forward-looking assessment of time-related impacts. A simulation agent model based on reinforcement learning algorithms is constructed within the virtual construction process model, with each trade's work activities as intelligent agents and real-time illumination data as environmental state input. Through reinforcement learning training, an illumination-efficiency mapping relationship is established, enabling the system to automatically learn and establish different illumination environments. The system employs response models to assess the operational efficiency of various trades, enhancing the objectivity of the analysis. Based on real-time mapped virtual layer data of the illumination field, it drives a simulation proxy model to dynamically simulate the changes in operational efficiency of each trade under different illumination conditions. The model outputs the actual operational efficiency coefficients under the influence of illumination, enabling real-time dynamic prediction of operational efficiency for each trade. This provides an immediate and quantitative basis for adjusting work hours. Combined with the standard work hour quotas for construction tasks, the system calculates the impact of work hour fluctuations caused by illumination based on the operational efficiency coefficients and generates work hour impact reports categorized by trade and process. This quantifies the environmental impact into specific work hour fluctuation data, forming a traceable and analyzable special report to support management optimization. It should be noted that in the virtual construction model, independent simulation agents are constructed for each key construction trade (steel reinforcement, carpentry, concrete work, decoration, etc.). Each agent is designed based on a deep deterministic policy gradient algorithm. Its state space includes real-time mapped light intensity, color temperature, and spatial coordinates. The action space is defined as the adjustable work rate and operational precision of the trade within a unit work cycle. Model training adopts a combination of offline and online modes. In the initial training phase, historical construction datasets are used, containing at least 200 workdays covering different seasons and weather conditions with corresponding light-time records. Strategy optimization is performed by maximizing the objective through a cumulative reward function, which is set as the ratio of actual work efficiency to standard efficiency. After training, the simulation agent model can receive real-time light field virtual layer data in the digital twin environment, with 0.5-meter grid units and an update frequency of 1 Hz, and dynamically adjust the work efficiency output of each trade. In the system operation phase, the simulation agent model drives the agents of each trade to perform dynamic efficiency deduction based on the real-time incoming light field virtual layer data. Each agent calculates the efficiency based on the illuminance and color temperature values ​​of its grid unit. Based on the characteristics of different work types and the work stage, the system calculates the work efficiency coefficient under the current environment through a trained strategy network. This coefficient is defined as the ratio of the actual work efficiency to the efficiency under standard interference-free conditions. The system updates the efficiency coefficient of each work type every 5 minutes and displays it as a heat map overlaid on the corresponding construction area in the digital twin platform. At the same time, it automatically records the efficiency coefficient change curve of each process throughout the entire work period and stores it in a structured database. Based on the dynamically calculated work efficiency coefficient, combined with the entered standard work hour quota database for construction tasks, the system automatically calculates the impact of work hour fluctuations caused by light. During the calculation, the standard work hour quota and the real-time efficiency coefficient are weighted and accumulated on a process-by-process basis to obtain the estimated actual work hour required. The difference between the estimated work hour and the standard work hour is the fluctuating work hour caused by light. A work hour impact analysis report is automatically generated every 30 minutes. The report is classified by work type and process, and lists in detail the affected work area, fluctuating work hour, average efficiency coefficient and deviation percentage. The report output format conforms to engineering management specifications, supports PDF and Excel format export, and is displayed in real time through the digital dashboard of the integrated management center. The formula for calculating the impact of working hour fluctuations is as follows: ; ; In the formula: Indicates in job type Process The impact of fluctuations in working hours caused by lighting conditions; Indicate job type Process The standard time quota is the standard time required to complete the process under undisturbed conditions. Indicates time Job type Process The efficiency coefficient of the operation at the location; Indicates time Job type Process The actual operational efficiency of a unit is measured by the amount of work completed per unit of time. Indicate job type Process The baseline operating efficiency under standard, undisturbed conditions is determined by industry quotas. Indicate process The total number of work periods is calculated cumulatively by dividing the periods into fixed time granularities; if This indicates that the lighting environment leads to increased working hours. This indicates that the lighting environment is conducive to operation, reducing working hours; The intelligent task sorting module receives prediction data on the impact of work hour fluctuations, and actively intervenes in the construction plan in conjunction with the construction task network diagram. It dynamically adjusts the task order on non-critical paths when there are abnormal lighting conditions, actively avoids environmental interference, and improves the adaptability and robustness of the plan to dynamic environmental changes. The resource elastic configuration module is used to dynamically adjust the configuration scheme of personnel and equipment based on the lighting environment prediction and task rescheduling results, under the premise of meeting the total project duration and process logic constraints, generate an adaptive scheduling scheme that is resistant to lighting interference, and push it to the integrated management center to optimize resource allocation and improve resource utilization and overall construction efficiency. The quality risk early warning module is used to establish a three-dimensional risk assessment model of light-fatigue-quality by fusing convolutional neural networks and time series prediction models. It provides intelligent early warning of fatigue accumulation and quality risks caused by light environment during long-term construction, and ensures construction quality and personnel safety from the source.

[0021] Example 2, as Figure 1 , Figure 2As shown, based on Embodiment 1, the present invention provides a technical solution: the execution steps of the intelligent task sorting module include: receiving data on the impact of work hour fluctuations, combining the process dependency relationship and critical path analysis in the construction task network diagram, identifying the set of non-critical path tasks that are greatly affected by light, realizing efficient location of non-critical tasks and centralized identification of interference, dynamically prioritizing non-critical path tasks according to the time prediction and spatial distribution of abnormal light intervals, adjusting the execution order or time period under the premise of satisfying the process logic, so as to avoid or alleviate light interference, effectively reduce environmental interference, ensure stable construction rhythm, generate an adjusted construction task sequence, and verify it through simulation in a digital twin environment to confirm the feasibility of the adjusted task sequence under the constraints of schedule and resources, and output an optimized construction schedule plan to ensure the feasibility and executability of the plan adjustment; It should be noted that by analyzing the construction task network diagram, the dependencies and critical path information of all processes are obtained. In the time fluctuation impact report generated every 30 minutes, process data where the fluctuation time caused by light exceeds a preset threshold are extracted. The process number, construction area, and fluctuation amount are automatically matched with non-critical path tasks in the network diagram to filter out the set of non-critical path tasks significantly affected by light. During the matching process, only tasks with a total float time greater than the light impact fluctuation time of the process are selected to ensure that the tasks have room for adjustment. The identification results are output as a structured list, including task ID, original planned time period, expected fluctuation time, and available float time. Based on the spatiotemporal prediction data of abnormal light intervals, the selected non-critical path tasks are dynamically rearranged. The rearrangement process follows the process logic constraints and safety operation interval requirements, aiming to minimize the impact of light. A heuristic scheduling algorithm is used to re-arrange the tasks within the available float time window. Allocate work time slots or adjust execution order, ensuring that all adjustments do not affect the logical relationship between subsequent preceding and succeeding processes, and that the time slot offset of a single adjustment does not exceed 50% of the total floating time of the task. The optimized task sequence includes an explanation of the adjustment reasons and an assessment of the benefits of light avoidance. The adjusted task sequence will be automatically imported into the digital twin platform for simulation verification. Based on the updated task time parameters, resource requirements, and real-time light field virtual layer data, the virtual model of the construction process will be driven to perform full-cycle dynamic simulation. During the simulation, the adjusted plan will be monitored and verified in real time to ensure that it meets the requirements of total project duration constraints, zero floating of the critical path, and continuous balance of resources (personnel, equipment). If the simulation results identify resource conflicts or project delay risks, iterative optimization will be automatically initiated until a feasible solution is generated. Finally, the optimized construction schedule plan that has passed verification will be output in the form of standardized Gantt charts and network diagrams and simultaneously updated to the baseline plan in the project management system as the basis for on-site execution. The execution steps of the resource elastic allocation module include: based on the predicted data of the lighting environment and the optimized construction schedule, analyzing the changes in personnel and equipment demand in different time periods and regions, identifying resource allocation conflicts and redundancies, effectively avoiding resource idleness and over-concentration, improving the overall resource utilization rate, and establishing a resource scheduling optimization model under the condition of meeting the constraints of the total project period and construction process logic. This model dynamically adjusts personnel shift arrangements, equipment allocation plans and material supply rhythms, generates an adaptive resource allocation scheme that resists lighting interference, realizes dynamic matching between resources and the lighting environment, reduces efficiency losses caused by environmental interference, pushes the optimized resource allocation scheme to the integrated management center, and synchronizes updates with the resource entities in the digital twin environment to achieve real-time visualization and dynamic control of resource scheduling, enhance scheduling transparency and response speed, and support real-time monitoring and rapid decision-making by management personnel. It should be noted that after generating the optimized construction schedule, a refined resource demand analysis is conducted based on the adjusted start and end times of each process, the required trades and equipment types, and combined with lighting environment prediction data. Using 15-minute intervals as the basic time unit and individual construction areas as spatial units, the required number of personnel for each trade and the number of shifts for various equipment (such as tower cranes, pump trucks, and welding machines) are calculated for each time period. During the analysis, the total demand for the same type of resources in different areas within the same time period is automatically compared with the actual available resources on site. When the total demand exceeds 90% of the preset safety factor for available resources, it is identified as a resource allocation conflict point. Simultaneously, for predicted... During periods of good lighting conditions (illuminance consistently above 300 lx) and an expected work efficiency coefficient higher than 1.05, if resource demand is less than 70% of available resources, these periods are marked as potential resource redundancy points, providing a basis for dynamic allocation. Based on the identified conflicts and redundancy points, and under the hard constraints of total project duration, zero fluctuation on the critical path, and process logic, a mixed-integer programming scheduling optimization model is established with the dual objectives of maximizing resource balance and minimizing lighting interference costs. Model decision variables include shift adjustments for various work types, usage time offsets for critical equipment, and fine-tuning of auxiliary material arrival times. An improved genetic algorithm is used for optimization, with a population size set to [missing information]. The model employs an iteration count of 100 and 500 to ensure rapid convergence to a satisfactory solution under complex constraints. The model outputs an adaptive resource allocation scheme that explicitly provides detailed scheduling instructions for each resource entity over the next 24 to 72 hours, along with the reasons for the adjustments (avoiding the predicted strong glare period from 14:00 to 15:30) and an assessment of the expected efficiency improvements. After the optimization scheme is generated, a structured list of scheduling instructions is pushed to the integrated management center in real time via a standardized data interface. The integrated management center then triggers two synchronous operations: first, it distributes personnel shifts and equipment scheduling information to the project management system and mobile terminal applications to guide on-site supervisors and resource administrators in execution. Second, the updated resource spatiotemporal distribution data will be synchronized to the digital twin environment. In the digital twin platform, the status, location, and planned trajectory of each resource entity will be updated in real time according to the new scheme. Managers can use the visualization dashboard of the integrated management center or directly operate the digital twin model to dynamically monitor the resource load rate, conflict warning status, and deviation between the plan and actual execution at any time and in any region in the future by pulling the timeline or filtering by region. If any resource cannot be in place according to the scheduling instructions due to any sudden situation on site, the closed-loop process from demand analysis to optimized scheduling will be automatically detected and restarted to achieve full-process, visualized, adaptive, and dynamic control of resource scheduling. The execution steps of the quality risk early warning module include: based on long-term collected lighting data, personnel fatigue monitoring data, and construction quality inspection records, constructing a three-element risk assessment model of lighting-fatigue-quality that integrates convolutional neural networks and time-series prediction models; establishing the correlation between lighting fluctuations, fatigue accumulation, and quality deviations; effectively identifying the hidden hazards of long-term poor lighting to personnel health and construction quality; inputting current and predicted lighting data into the risk assessment model in real time; dynamically calculating the fatigue risk index and quality deviation probability caused by lighting environment in each area and process; comparing them with their respective preset thresholds; realizing real-time quantitative assessment and proactive perception of potential risks; when the fatigue risk index or quality deviation probability exceeds their respective preset thresholds, automatically triggering a graded early warning mechanism; pushing early warning information to management personnel; and recommending targeted intervention measures based on the risk type, including adjusting lighting, arranging intermittent rest, or strengthening process inspection; significantly improving the speed of risk response and the accuracy of handling; and curbing quality accidents from the source. It should be noted that an integrated historical database is used, containing at least six consecutive months of time-series data on light intensity covering different seasons and day / night times. Simultaneously collected data includes fatigue monitoring indicators such as heart rate variability and body movement frequency obtained through smart safety helmets or wearable devices, as well as construction quality inspection records for the corresponding time periods and regions. Based on this multi-source heterogeneous data, a hybrid risk assessment model integrating a one-dimensional convolutional neural network (CNN) and a long short-term memory network (LSTM) is constructed. The CNN has three layers with kernel sizes of 3, 5, and 3, used to extract local spatiotemporal features of light intensity fluctuations. Its output feature vector is then input into an LSTM layer with 128 hidden units to learn the long-range temporal dependencies between light intensity fluctuation patterns and subsequent fatigue accumulation trends and quality deviations. The model is trained through supervised learning using light intensity sequences from the past 24 hours. As input, the model uses the fatigue risk level of personnel (divided into low, medium, and high levels) and the probability of quality deviation in key processes within the next 4 hours as joint prediction objectives. A weighted sum of mean squared error and cross-entropy is used as the loss function, and the model is optimized through backpropagation until its prediction accuracy on the validation set stabilizes at over 85%. In the operational phase, the risk assessment model deployed on the on-site edge computing server receives real-time illuminance grid data from the environmental dynamic sensing module and light intensity prediction data for the next 2 hours from meteorological services or autoregressive models via a standard data interface. A 3-hour sliding window of data, consisting of the measured sequence of the previous hour and the predicted sequence of the next two hours, is used as input to dynamically calculate the real-time fatigue risk index and probability of quality deviation for each 0.5m × 0.5m grid cell associated with the construction process. The fatigue risk index is based on the "Classification of Physical Labor Intensity" (GB). According to the relevant standards of GB 3869-1997, the risk level output by the model is quantified into an index from 0 to 100; the quality deviation probability is directly output as a percentage value from 0% to 100%, and a preset fatigue risk index threshold is set according to the differences in job types; the quality deviation probability threshold is set by reverse calculation based on the allowable deviation of the main control items of the corresponding sub-item project in the "Unified Standard for Acceptance of Construction Quality of Building Engineering" GB50300-2013; when the system detects that the fatigue risk index or quality deviation probability of a certain area or a certain process continuously exceeds the preset threshold for two consecutive evaluation cycles (i.e., within 20 minutes), the graded early warning mechanism is automatically triggered, and the early warning is divided into three levels: when the indicator exceeds the threshold but is below 1.2 times the threshold, a prompt-level early warning is triggered; when the indicator is between 1.2 and 1.5 times the threshold, a warning-level early warning is triggered; when the indicator exceeds 1.When the threshold is exceeded by 5 times, an emergency alert is triggered. The alert information is immediately pushed to designated regional managers, safety officers, and project managers via WeChat, SMS, and the project management platform's internal messaging channel. The push notification includes the alert level, location (associated BIM component ID), type of exceeding the standard, specific value, and duration. Simultaneously, based on analysis of the effectiveness of historical intervention measures, recommended intervention measures are automatically generated and attached. All alerts and handling suggestions are structured and recorded for model iteration optimization and accountability.

[0022] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0023] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A system for quantitative measurement of building space illumination and statistical analysis of construction time data, comprising a comprehensive management center, characterized in that, The integrated management center has the following communication connection modules: The environmental dynamic perception module is used to deploy an intelligent sensing system at the construction site, use the particle swarm algorithm to collect and analyze illumination data from multiple points in real time, generate a real-time illuminance distribution map, and dynamically identify abnormal illumination intervals. The construction time impact prediction module is used to integrate digital twin technology to build a virtual model of the construction process, and combine it with reinforcement learning algorithms to predict the fluctuation impact of real-time lighting data on construction time. The intelligent task sorting module is used to receive prediction data on the impact of work hour fluctuations, and actively intervene in the construction plan in combination with the construction task network diagram, dynamically adjusting the task order on non-critical paths when there is abnormal lighting. The resource elastic configuration module is used to dynamically adjust the configuration scheme of personnel and equipment based on the lighting environment prediction and task rescheduling results, under the premise of meeting the total project duration and process logic constraints, generate an adaptive scheduling scheme that is resistant to lighting interference, and push it to the integrated management center. The quality risk early warning module is used to establish a three-dimensional risk assessment model of light-fatigue-quality by fusing convolutional neural networks and time-series prediction models, and to provide intelligent early warning of fatigue accumulation and quality risks caused by light environment during long-term construction.

2. The system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 1, characterized in that: The environmental dynamic perception module includes a light network perception unit and an abnormal light recognition unit. The light network sensing unit is used to deploy an intelligent sensing system composed of intelligent light sensors in key work areas and personnel activity areas at the construction site to collect light data from multiple points in real time. The abnormal lighting identification unit is used to clean, correct and fuse the collected multi-point lighting data, use the particle swarm algorithm to dynamically identify and remove outliers, generate real-time global and local illuminance distribution maps, and delineate and mark the normal operating illuminance range and abnormal lighting range.

3. The system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 2, characterized in that: The execution steps of the illumination network sensing unit include: Based on the construction site layout and operational requirements, determine the lighting monitoring points for key work areas and personnel activity areas, and deploy intelligent lighting sensor nodes with wireless communication capabilities to build an intelligent sensing system covering the entire construction space. The system collects real-time illumination data, including light intensity, spectral characteristics, and timestamps, from each monitoring point using an intelligent sensing system. Combined with the Internet of Things protocol, the data is synchronously uploaded to the edge computing gateway to form a multi-source, multi-dimensional raw illumination dataset. The original illumination dataset is processed for temporal alignment and spatial registration to generate real-time ambient illumination field data based on time series and spatial coordinates.

4. The system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 2, characterized in that: The execution steps of the abnormal lighting recognition unit include: Noise filtering and data correction are performed on the received multi-point illumination data. The sliding window averaging method and sensor calibration parameters are used for data cleaning to eliminate equipment errors and instantaneous interference. The particle swarm optimization algorithm is used to perform cluster analysis and outlier detection on the cleaned multi-point illumination data, dynamically identify outlier regions in the illumination distribution, and delineate normal illumination range and abnormal illumination range according to the preset illumination operation threshold. Based on the spatiotemporal distribution characteristics of abnormal illumination intervals, global and local real-time illuminance distribution maps are generated, and abnormal areas and their deviations are marked in a visual form on the maps.

5. The system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 2, characterized in that: The time impact simulation module includes a virtual-real mapping construction unit and a time impact simulation unit; The virtual-real mapping construction unit, based on BIM, schedule and personnel information, integrates digital twin technology to construct a virtual model of the construction process that is synchronized with the physical space, and maps multi-point lighting data to the corresponding spatiotemporal location in the virtual model of the construction process in real time. The time impact estimation unit is used to run a simulation proxy model based on reinforcement learning algorithm in a digital twin environment. Based on the mapped illumination data, it dynamically simulates and predicts the changes in the work efficiency of each type of work under different illumination conditions, and calculates the actual fluctuation impact of illumination on standard working hours.

6. The system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 5, characterized in that: The execution steps of the virtual-real mapping construction unit include: Based on the construction BIM model, schedule data and personnel organization structure, a virtual model of the construction process, including geometric structure, construction procedures and resource attributes, is constructed by integrating digital twin technology. Real-time multi-point illumination data is mapped to the corresponding spatial location and construction time node in the construction process virtual model through coordinate transformation and time synchronization mechanism, forming a virtual illumination field consistent with the real environment. The spatiotemporal superposition effect of the virtual illumination field layer and construction activities is dynamically displayed in the construction process virtual model.

7. The system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 6, characterized in that: The execution steps of the time impact simulation unit include: In the virtual model of the construction process, a simulation agent model based on reinforcement learning algorithm is constructed, with the operation activities of each type of work as the intelligent agent and real-time lighting data as the environmental state input. The lighting-efficiency mapping relationship is established through reinforcement learning training. Based on the real-time mapped virtual layer data of the illumination field, the simulation agent model is driven to dynamically simulate the change process of the work efficiency of various jobs under different illumination conditions, and output the actual work efficiency coefficient under the influence of illumination. Based on the standard time quota for construction tasks, the impact of time fluctuations caused by sunlight is calculated according to the work efficiency coefficient, and a time impact report is generated by classifying work type and process.

8. The system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 5, characterized in that: The execution steps of the intelligent task sorting module include: By receiving data on the impact of work hour fluctuations and combining it with the process dependencies and critical path analysis in the construction task network diagram, a set of non-critical path tasks that are significantly affected by lighting conditions is identified. Based on the temporal prediction and spatial distribution of abnormal lighting intervals, non-critical path tasks are dynamically prioritized and their execution order or time period is adjusted while satisfying the process logic. Generate an adjusted construction task sequence and verify it through simulation in a digital twin environment to confirm the feasibility of the adjusted task sequence under the constraints of schedule and resources, and output the optimized construction schedule plan.

9. The system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 8, characterized in that: The execution steps of the resource elastic configuration module include: Based on the predicted data of the lighting environment and the optimized construction schedule, we analyze the changes in personnel and equipment demand in different time periods and regions, and identify resource allocation conflicts and redundancies. Under the condition of meeting the overall project period and construction process logic constraints, a resource scheduling optimization model is established to dynamically adjust personnel shift arrangements, equipment allocation plans and material supply rhythm, and generate an adaptive resource allocation scheme that is resistant to light interference. The optimized resource allocation plan is pushed to the integrated management center and synchronized with the resource entities in the digital twin environment.

10. A system for quantitative measurement of building space illumination and statistical analysis of construction time data according to claim 9, characterized in that: The execution steps of the quality risk early warning module include: Based on long-term collected light data, personnel fatigue monitoring data and construction quality inspection records, a three-element risk assessment model of light-fatigue-quality is constructed by integrating convolutional neural network and time series prediction model, and the correlation between light fluctuation and fatigue accumulation and quality deviation is established. Input current and predicted lighting data into the risk assessment model in real time, dynamically calculate the fatigue risk index and quality deviation probability caused by lighting environment in each area and process, and compare them with their respective preset thresholds; When the fatigue risk index or the probability of quality deviation exceeds its respective preset threshold, a graded early warning mechanism is automatically triggered to push early warning information to managers and recommend targeted intervention measures based on the risk type, including adjusting lighting, arranging intermittent rest, or strengthening process inspection.

Images