Construction site safety risk intelligent assessment method and system

Abstract

The invention discloses a construction site safety risk intelligent assessment method and system, and belongs to the technical field of engineering supervision. The method comprises the following steps: S100, collecting multi-source data including an equipment state, an image video, a personnel state, an equipment distance, an environment parameter and an engineering text in real time; s200, performing space-time alignment, preprocessing and cross-modal feature fusion on the multi-source data; s300, outputting a risk score and a risk level label based on the dynamic risk knowledge graph and a multi-model fusion algorithm; s400, generating a hierarchical disposal strategy according to the risk scoring hierarchy, and realizing risk closed-loop management and control through rectification verification and cooperation of multiple parties; and S500, outputting a three-dimensional visual risk distribution map and a compliance report. Through technologies of multi-source data fusion, dynamic risk mapping knowledge, intelligent closed-loop management and the like, comprehensive perception, accurate evaluation, efficient management and control and compliance landing of engineering supervision safety risks are realized, the occurrence rate of safety accidents is remarkably reduced, and reliable technical support is provided for intelligent construction site construction.

Description

Technical Field

[0001] This application relates to the technical field of project supervision, and particularly to an intelligent assessment method and system for construction site safety risks. Background Art

[0002] In the field of construction projects, project supervision is a core link to ensure construction quality, safety and compliance. The safety risks of project supervision refer to the potential threats that may affect project safety, the safety of personnel's lives and property, and the interests of the supervision unit itself during the project supervision process due to various uncertain factors.

[0003] Traditional supervision modes mainly rely on manual inspections, paper records and experience judgments, and cannot detect and warn of safety risks in a timely manner. They have poor real-time performance, low efficiency and are extremely prone to safety accidents. In some existing improved project supervision systems, technologies based on Internet of Things communication have been widely applied to project construction and management. These systems use Internet of Things devices, such as sensors and monitoring cameras, to collect on-site data and achieve real-time monitoring of project progress, equipment status and working environment. The Internet of Things technology enables supervision work to be more accurate and efficient. Through real-time data analysis, supervision personnel can immediately understand various situations on the project site, so as to better guide construction and ensure project quality, and can also identify and prevent on-site safety accidents to a certain extent.

[0004] Although the above improvement schemes have improved in terms of efficiency and prevention of safety accidents, there are still the following technical defects: fragmented data collection, lack of effective integration, failure to consider comprehensive risk influencing factors, and insufficient degree of data fusion processing, resulting in one-sided risk assessment; static risk assessment models, unable to dynamically respond to the complex and changeable environment at the construction site, with untimely rule updates and insufficient adaptability; inefficient rectification of potential safety hazards, lack of closed-loop management; not deeply bound to regulations, compliance inspections rely on manual comparison of clauses, which is time-consuming and prone to omissions, difficult to implement compliance, and prominent legal risks; insufficient integration with new technologies such as BIM, unable to intuitively display the risk evolution trend. Summary of the Invention

[0005] Therefore, this application is proposed in view of the problems and requirements existing in the above-mentioned prior art. The purpose of this application is achieved through the following technical solutions:

[0006] In a first aspect, an embodiment of this application provides an intelligent assessment method for construction site safety risks, including the following steps:

[0007] S100. Real-time collect multi-source data including equipment status, image and video, personnel status, equipment distance, environmental parameters, construction progress and project texts;

[0008] S200. Perform spatio-temporal alignment, preprocessing, and cross-modal feature fusion on multi-source data;

[0009] S300. Output risk scores and risk level labels based on a dynamic risk knowledge graph and a multi-model fusion algorithm;

[0010] S400. Generate a hierarchical disposal strategy according to the risk score hierarchy, and achieve closed-loop risk control through rectification verification and multi-party collaboration;

[0011] S500. Output a 3D visual risk distribution map and a compliance report.

[0012] In the above intelligent assessment method for construction site safety risks, step S300 includes:

[0013] S310. Input multi-source feature data into the dynamic risk knowledge graph and associate the "risk type - cause - consequence" nodes in the graph;

[0014] S320. Use the XGBoost model to assign weights to static risk factors; use the LSTM model to analyze high-frequency sensor data and predict short-term risks; use the Transformer model to fuse multi-source time-series data and predict long-term risk trends;

[0015] S330. Calculate the risk score of the operation unit according to the risk characteristic parameters, calculate the overall project risk score by weighted summation of the operation unit weights and associated risk values, and map the score result to a risk level and automatically associate it with the standard terms;

[0016] S340. Pre-render a risk heat map in the BIM model and mark the locations and cause chains of high-risk areas.

[0017] In the above intelligent assessment method for construction site safety risks, step S300 also includes the update of the dynamic risk knowledge graph, specifically: through incremental learning technology, real-time update the new construction data to the knowledge graph nodes; based on the graph attention network (GAT), dynamically optimize the association weights between nodes; when a new risk pattern is detected, automatically expand the graph nodes and associated edges.

[0018] In the above intelligent assessment method for construction site safety risks, the hierarchical disposal strategy in step S400 includes:

[0019] High risk level (≥70 points): Immediately stop work, push a special rectification plan and initiate an emergency response;

[0020] Medium risk level (30 points < risk score < 70 points): Rectify within a time limit and include it in the key inspection list;

[0021] Low risk level (≤30 points): Record the log and update the monitoring plan, and conduct regular reviews.

[0022] Second, the embodiment of the present application provides an intelligent safety risk assessment system for construction sites to implement the above-mentioned intelligent safety risk assessment method for construction sites, including:

[0023] A multi-source data acquisition module, which is used to obtain real-time Internet of Things sensor data, image and video data, engineering document data, environmental parameter data, personnel biometric data, equipment distance monitoring data, and construction progress data at the construction site;

[0024] A heterogeneous data fusion module, which is used to perform spatio-temporal alignment, feature extraction, and cross-modal correlation analysis on multi-source data to generate a structured risk feature set;

[0025] A dynamic risk assessment module, which is used to output the overall project risk score, the risk score of the operation unit, and the associated risk score between operation units based on the dynamically updated risk knowledge graph and multi-model fusion algorithm, and generate a risk level label;

[0026] An intelligent closed-loop management module, which is used to generate a hierarchical disposal strategy according to the risk score level, and achieve risk closed-loop control through rectification verification and multi-party collaboration;

[0027] A visualization and compliance module, which is used to dynamically display the risk distribution through a three-dimensional visualization interface and generate a legal compliance report bound to industry standards.

[0028] In the above-mentioned intelligent safety risk assessment system for construction sites, the multi-source data acquisition module includes:

[0029] A personnel status monitoring unit, which is used to monitor the heart rate, blood oxygen saturation, and fall behavior of personnel in real time;

[0030] A visual behavior recognition unit, which is used to perform real-time analysis on the image and video stream based on a deep learning model, identify the illegal behaviors of construction personnel, and output the behavior category and confidence level;

[0031] An equipment distance monitoring unit, which is used to calculate the spatial distance between personnel and equipment in real time; and dynamically identify personnel entering the safety warning area of the equipment based on a depth perception algorithm;

[0032] An environment and construction data acquisition unit, including the following sub-units:

[0033] A structural safety monitoring sub-unit, which is used to monitor the horizontal displacement and vertical settlement rate of the scaffolding in real time; and collect the foundation pit settlement data;

[0034] The device status monitoring sub-unit is used to monitor the tower inclination angle in real time; and monitor the three-phase current balance status and leakage current value in the temporary power supply circuit;

[0035] The environmental monitoring sub-unit is used to collect wind speed, rainfall intensity, and groundwater level change data in real time;

[0036] The construction data analysis sub-unit extracts safety hazard keywords in engineering documents based on OCR and natural language processing (NLP) technologies;

[0037] The construction progress tracking sub-unit is integrated with the BIM model to compare the design parameters and the on-site construction progress in real time; and collects the deviation data of the component installation position through Internet of Things sensors to generate a progress deviation report.

[0038] In the above intelligent assessment system for construction site safety risks, the heterogeneous data fusion module includes:

[0039] The spatio-temporal alignment unit is used to align multi-source data according to a unified spatio-temporal benchmark;

[0040] The data preprocessing unit is used to perform data cleaning, filtering, and standardization on multi-source data;

[0041] The feature extraction unit extracts risk features for different data types;

[0042] The cross-modal feature interaction unit is used to map feature vectors of different dimensions to a unified N-dimensional space through a fully connected layer, and uses a multi-channel Transformer model to perform cross-modal attention calculation and fuse multi-source features.

[0043] In the above intelligent assessment system for construction site safety risks, the dynamic risk assessment module includes:

[0044] The dynamic risk knowledge graph construction unit is used to construct a "risk type - cause - consequence" association network based on historical accident cases and safety specifications;

[0045] The multi-model fusion evaluation unit is used to assign weights to static risk factors using the XGBoost model; analyze high-frequency sensor data using the LSTM model to predict short-term risks; and fuse multi-source time-series data using the Transformer model to predict long-term risk trends;

[0046] The grading and scoring unit is used to output the risk score of the operation unit, map the risk score of the operation unit to low, medium, and high risk levels according to a preset threshold; and dynamically calculate the overall risk level of the project based on the risk score of the operation unit and the associated risk value between operation units.

[0047] In the above intelligent construction site safety risk assessment system, the intelligent closed-loop management module includes:

[0048] A hierarchical disposal strategy unit that generates different disposal strategies according to the risk level, including:

[0049] High risk level (≥70 points): Immediately stop work, push a special rectification plan and initiate an emergency response;

[0050] Medium risk level (30 points < risk score < 70 points): Rectify within a time limit and include it in the key inspection list;

[0051] Low risk level (≤30 points): Record the log and update the monitoring plan, and conduct regular reviews;

[0052] A rectification verification unit that verifies the completion degree of rectification measures through OCR and image comparison technologies. If not up to standard, trigger a secondary disposal process and upgrade the risk level;

[0053] A multi-party collaboration interface unit that synchronously pushes disposal instructions to the terminals of the construction, supervision, and construction parties, records the response time and execution results; generates a responsibility traceability log, associating the time stamp, geographical location, and operator information.

[0054] In the above intelligent construction site safety risk assessment system, the visualization and compliance module includes:

[0055] A three-dimensional risk heat map annotation unit for hierarchically annotating risk areas on the BIM model;

[0056] An AR on-site navigation unit for locating potential hazard points through augmented reality devices and displaying disposal guidelines;

[0057] A compliance binding unit for automatically matching standard terms; generating a standard report containing electronic signatures, legal basis indexes, and rectification evidence chains.

[0058] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0059] The intelligent safety risk assessment method and system for construction sites of this application integrate multi-source data such as Internet of Things sensors, visual recognition, and BIM models, and adopt technologies such as spatio-temporal alignment and cross-modal feature fusion to effectively solve the problem of data fragmentation in traditional supervision. The system can comprehensively cover risk factors such as equipment status, personnel behavior, and environmental parameters, avoid the one-sidedness of evaluation caused by a single data dimension, significantly improve the accuracy of identifying potential safety hazards, and enhance the comprehensiveness of risk assessment; based on a dynamic risk knowledge graph and a multi-model fusion algorithm, the system can respond in real time to the complex changes at the construction site, achieving a leap from static rules to dynamic perception, with strong adaptability and a significant improvement in the timeliness of risk early warning; the system automatically generates a hierarchical disposal strategy according to the risk level, and verifies the rectification effect through technologies such as OCR image comparison. The multi-party collaborative interface pushes the instructions to the terminals of the construction, supervision, and construction parties in real time, and records the response time and execution results, forming a complete responsibility traceability log, greatly shortening the rectification response time and significantly improving the efficiency of closed-loop management; through the compliance binding unit, the system automatically matches the regulatory clauses, generates a standard report with electronic signatures, legal basis indexes, and rectification evidence chains, avoiding omissions in manual comparison, ensuring that the compliance inspection coverage rate reaches 100%, and reducing the legal risk by more than 90%; dynamically renders a risk heat map on the BIM model, supports AR on-site navigation, and combines a time-axis animation to display the risk trend, helping managers intuitively grasp the overall risk distribution and evolution path, and greatly improving the decision-making efficiency; through the dynamic calculation of the associated risk scores and weight coefficients between operation units, the system can quantify the cross-unit risk conduction effect, avoid the limitations of isolated evaluation, effectively improve the coverage rate of high-risk area identification, and greatly reduce the global risk assessment error rate.

[0060] In summary, through technologies such as multi-source data fusion, dynamic risk knowledge graph, and intelligent closed-loop management, this application realizes the comprehensive perception, accurate assessment, efficient control, and compliance implementation of safety risks in engineering supervision, significantly reduces the incidence of safety accidents, and provides reliable technical support for the construction of smart construction sites. Brief Description of the Drawings

[0061] By describing the embodiments of this application in more detail in conjunction with the accompanying drawings, the above and other purposes, features, and advantages of this application will become more obvious. The accompanying drawings are used to provide a further understanding of the embodiments of this application, and constitute a part of the specification. They are used together with the embodiments of this application to explain this application, and do not constitute a limitation to this application.

[0062] Figure 1 It is a schematic structural diagram of an intelligent safety risk assessment system for construction sites provided by an embodiment of this application;

[0063] Figure 2 It is a schematic step diagram of an intelligent safety risk assessment method for construction sites provided by an embodiment of this application;

[0064] Figure 3 It is a schematic diagram of the steps for outputting risk scores based on a dynamic risk knowledge graph and a multi-model fusion algorithm provided by an embodiment of the present application. Detailed implementation manners

[0065] Next, exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only a part of the embodiments of the present application, and the present application is not limited by the exemplary embodiments described herein.

[0066] Embodiment 1

[0067] In the prior art, data collection is fragmented and lacks effective integration. Comprehensive risk influencing factors are not considered, and the degree of data fusion processing is insufficient, resulting in one-sided risk assessment; the risk assessment model is static and cannot dynamically respond to the complex and changeable environment at the construction site. Rule updates are not timely and the adaptability is insufficient; the rectification of potential safety hazards is inefficient and lacks closed-loop management; it is not deeply bound to regulations, and compliance inspections rely on manual comparison of terms, which is time-consuming and prone to omissions. It is difficult to implement compliance, and legal risks are prominent; there is insufficient integration with new technologies such as BIM, and the risk evolution trend cannot be intuitively displayed.

[0068] Therefore, in view of the above problems, an embodiment of the present application provides an intelligent construction site safety risk assessment system, as Figure 1 shown, including:

[0069] A multi-source data collection module 100, configured to obtain in real time the Internet of Things sensor data, image and video data, engineering document data, environmental parameter data, personnel biometric data, equipment distance monitoring data, and construction progress data at the construction site;

[0070] A heterogeneous data fusion module 200, configured to perform spatio-temporal alignment, feature extraction, and cross-modal correlation analysis on the multi-source data to generate a structured risk feature set;

[0071] A dynamic risk assessment module 300, configured to output the overall project risk score, the risk score of the operation unit, and the associated risk score between operation units based on a dynamically updated risk knowledge graph and a multi-model fusion algorithm, and generate risk level labels;

[0072] An intelligent closed-loop management module 400, configured to generate a hierarchical disposal strategy according to the risk score hierarchy, and achieve risk closed-loop control through rectification verification and multi-party collaboration;

[0073] A visualization and compliance module 500, configured to dynamically display the risk distribution through a three-dimensional visualization interface and generate a legal compliance report bound to industry standards.

[0074] Among them, the multi-source data acquisition module 100 collects construction site data in real time through a variety of sensors and terminal devices, including the following units:

[0075] The personnel status monitoring unit 110 collects the heart rate (sampling rate 100Hz), blood oxygen saturation and fall behavior data of construction workers in real time through a PPG sensor (such as model MAX30102) and a six-axis gyroscope (such as model MPU6050) deployed on the safety helmet. When abnormal heart rate is detected (such as exceeding 120bpm for 3 consecutive detections) or the fall acceleration threshold is exceeded (the acceleration threshold can be 2.5-3g), an alarm signal is triggered. And when abnormal heart rate or excessive fatigue index of personnel is detected, a rest instruction is automatically pushed to the personal terminal and their high-risk operation permission is suspended;

[0076] The visual behavior recognition unit 120 analyzes the image and video stream collected by the camera in real time through a deployed high-definition camera (such as Hikvision DS-2CD3T86 series) based on the YOLOv8 deep learning model (input resolution 640×640, confidence threshold 0.7), and recognizes behaviors such as not wearing a safety helmet and / or not fastening the safety rope, and illegally crossing the edge protection, and outputs the behavior category and coordinate frame information;

[0077] The equipment distance monitoring unit 130 calculates the three-dimensional space distance between personnel and tower cranes and excavators in real time through UWB positioning tags (such as Decawave DW1000, accuracy ±10cm) and receivers installed at the end of the tower crane arm and the excavator cockpit using the TOF algorithm; and based on the depth perception algorithm of a binocular camera (such as Intel RealSense D435, depth resolution 1280×720), dynamically recognizes that personnel enter the warning area of the tower crane rotation radius (such as a preset safety distance of 5 meters). When personnel enter the dangerous area of the equipment, an emergency stop signal is sent to the equipment operator, and the violation event is recorded in the responsibility traceability system;

[0078] The environment and construction data acquisition unit 140 includes the following sub-units:

[0079] The structural safety monitoring sub-unit 141 monitors the horizontal displacement and vertical settlement rate in real time through displacement sensors (range ±50mm, accuracy 0.1mm) deployed at key nodes of the scaffold; and a static level (accuracy ±0.01mm) installed in the deep foundation pit collects the foundation pit settlement data based on a preset frequency (such as once every 10 minutes);

[0080] The device status monitoring sub-unit 142 is used to monitor the tower body inclination angle in real time through a tower crane inclination sensor (accuracy ±0.1°). If it exceeds the preset threshold (longitudinal inclination threshold 2° or transverse inclination threshold 1.5°), an early warning is triggered; and a current transformer (such as Honeywell CT-1000 model) in the temporary power supply circuit is used to detect the three-phase current unbalance degree (threshold ±10%) and the leakage current value (threshold 30 mA);

[0081] The environmental monitoring sub-unit 143 is used to collect wind speed, rainfall intensity, and groundwater level change data in real time through a wind speed sensor (range 0 - 60 m / s), a rainfall gauge (resolution 0.2 mm), and a groundwater level monitor (accuracy ±0.1 m), and dynamically compare them with environmental safety thresholds (such as wind speed threshold level 6, hourly rainfall threshold 30 mm);

[0082] The construction data analysis sub-unit 144 extracts safety hazard keywords (such as "collapse", "leakage", "support cracks", "not rectified", etc.) from engineering documents based on OCR and natural language processing (NLP) technologies, and performs semantic association analysis in combination with the BERT model;

[0083] The construction progress tracking sub-unit 145 is integrated with the BIM model to compare the design parameters with the on-site construction progress in real time; and collects component installation position deviation data through Internet of Things sensors (such as RFID tags (reading and writing distance 10 m), laser scanners), and generates a progress deviation report (color mapping deviation range: green <5 mm, red ≥10 mm).

[0084] Among them, the heterogeneous data fusion module 200 processes multi-source data to generate a structured risk feature set, including the following unit modules:

[0085] The spatio-temporal alignment unit 210 is used to align multi-source data according to a unified spatio-temporal reference. It adds a unified timestamp to all data using a GPS synchronous clock (timing accuracy ±1 ms) or adopts a sliding window mechanism to unify data with different sampling frequencies to the same timestamp, and realizes spatial alignment through coordinate system conversion (WGS84 to construction local coordinate system), such as registering the BIM model and UAV point cloud data using the ICP algorithm, and binding the sensor position to the BIM component ID;

[0086] The data preprocessing unit 220 is used to perform data cleaning, filtering, and standardization on multi-source data. For example, for Internet of Things sensor time-series data, noise filtering (such as Kalman filtering) and missing value interpolation processing are adopted; for video data, it is decoded into the H.264 format by FFmpeg, text data is cleaned through regular expressions, and image data is processed through image enhancement (histogram equalization);

[0087] The feature extraction unit 230 extracts risk features for different data types. For example, for sensor data, time-domain features are extracted through mean, variance, and peak values, and frequency-domain features are extracted through FFT spectrum energy. For image or video data, based on object detection using YOLOv8, "helmet wearing confidence" and "number of missing edge protections" are output. Through optical flow analysis, abnormal human movement trajectories (such as entering restricted areas) are calculated. For text data, semantic features are extracted through the BERT model (such as mapping "excessive displacement of the support structure" to a risk vector), and rule matching is performed (such as associating with the clause numbers in the "Code for High Altitude Work in Building Construction").

[0088] The cross-modal feature interaction unit 240 is used to map feature vectors of different dimensions to a unified 128-dimensional space through a fully connected layer. A multi-channel Transformer model (8 heads, hidden layer dimension 512) is adopted to perform cross-modal attention calculations on sensor time-series data, image features, text embeddings, personnel status, equipment distance, etc. (such as associating sensor data of "wind speed > level 6", a heart rate of 130 beats per minute of a person with the image behavior of "high altitude work not paused"). Multi-source features are fused to generate a structured risk feature set, and the fusion weights are adjusted according to the real-time risk level (such as the weight of sensor data is increased to 70% in a high-risk state).

[0089] Among them, the dynamic risk assessment module 300 includes:

[0090] The dynamic risk knowledge graph construction unit 310 is used to construct a "risk type - cause - consequence" association network based on historical accident cases and safety specifications. For example, a scaffold displacement rate > 5 mm / h is associated with the "collapse risk" node, the keyword "not rectified" in text data is associated with the "management dereliction of duty" node, an hourly rainfall of 50 mm (threshold 30 mm) is associated with the "rainstorm waterlogging" node, a foundation pit settlement rate of 0.8 mm / h (threshold 0.5 mm / h) is associated with the "support failure" node, a heart rate > 120 beats per minute of a person → the probability of operation error increases and the distance between the person and the tower crane jib < 5 meters → the collision risk increases, which is associated with the "personnel status - equipment distance - injury type" node, and the clauses in the "Safety Inspection Standards for Building Construction" and corresponding disposal measures are associated.

[0091] The multi-model fusion evaluation unit 320 is used to assign weights to static risk factors (such as construction plan compliance, material strength, foundation pit depth, scaffolding materials, etc.) using the XGBoost model (learning rate 0.1, tree depth 6); analyze high-frequency sensor data using the LSTM model (128 hidden units) to predict short-term risks (such as predicting the probability of sudden changes in tower crane inclination in the next 5-10 minutes, and the settlement rate of deep foundation pits in the scenario of sudden heavy rain in the next 10 minutes); use the Transformer model (4 encoder layers) to fuse multi-source time-series data to predict long-term risk trends (such as predicting the change rate of foundation pit water accumulation caused by continuous heavy rain in the next 2-6 hours).

[0092] The grading and scoring unit 330 is used to output the risk score of the operation unit, map the risk score of the operation unit to low, medium, and high risk levels according to the preset threshold; and dynamically calculate the overall risk level of the project based on the risk score of the operation unit and the associated risk value between operation units.

[0093] Specifically, the risk score of the operation unit is calculated by the following formula:

[0094]

[0095] Among them, S is the risk score of the operation unit, wi is the weight assigned by XGBoost, fi is the feature value, α is the associated risk weight coefficient, which is set with an initial value according to engineering safety specifications and historical accident cases, and is dynamically adjusted according to the construction stage, environmental complexity or historical data. S_associated is the associated risk score between operation units (for example, if operation unit A is adjacent to a high-risk unit B (such as a tower crane operation area), the associated risk of A will increase due to the influence of B). S_associated can be calculated by the following formula:

[0096]

[0097] Among them, Rj is the risk score of the jth operation unit associated with the current operation unit, and βj is the associated weight, which is calculated by the GAT model according to real-time data (for example, the higher the dependence and the closer the physical distance, the larger βj).

[0098] For the overall risk of the project, it is calculated by the following formula:

[0099]

[0100] Among them, γk is the weight of operation unit k, Sk is the risk score of operation unit k, β is the global associated risk coefficient, and ∑S_associated,k is the total associated risk value.

[0101] The risk scoring thresholds are: high risk level (≥70 points), medium risk level (30 points < risk score < 70 points), and low risk level (≤30 points). According to the thresholds, the risk scores of the operation units and the overall project risk scores are mapped to low, medium, and high risk levels.

[0102] Among them, the intelligent closed-loop management module 400 includes:

[0103] The hierarchical disposal strategy unit 410 generates different disposal strategies according to the risk level (matching the preset standardized disposal processes from the response measure library), including:

[0104] High risk level (≥70 points): Immediately stop work, push a special rectification plan, and initiate an emergency response;

[0105] Medium risk level (30 points < risk score < 70 points): Rectify within a time limit and include it in the key inspection list;

[0106] Low risk level (≤30 points): Record the log and update the monitoring plan (such as adjusting the sensor sampling frequency from 1Hz to 0.2Hz), and conduct regular reviews;

[0107] The rectification verification unit 420 verifies the completion degree of the rectification measures through OCR and image comparison technologies (such as the number of scaffolding reinforcement points and the replacement status of leakage protectors). If the standard is not met (similarity < 90%), trigger a secondary disposal process and upgrade the risk level (such as upgrading from medium risk to high risk);

[0108] The multi-party collaboration interface unit 430 synchronously pushes the disposal instructions to the terminals of the construction, supervision, and construction parties, records the response time and execution results; generates a responsibility traceability log, associating the time stamp, geographical location, and operator information.

[0109] Among them, the visualization and compliance module 500 includes:

[0110] The three-dimensional risk heat map annotation unit 510 is used to hierarchically annotate the risk areas on the BIM model (such as red corresponding to high risk, yellow corresponding to medium risk, and green corresponding to low risk);

[0111] The AR on-site navigation unit 520 is used to locate the potential hazard points through the augmented reality device and display the disposal guidelines;

[0112] The compliance binding unit 530 is used to automatically match the standard terms; generate a standard report containing electronic signatures, legal basis indexes, and rectification evidence chains.

[0113] The intelligent safety risk assessment system for construction sites in this embodiment integrates multi-source data such as Internet of Things sensors, visual recognition, and BIM models, and adopts technologies such as spatio-temporal alignment and cross-modal feature fusion to effectively solve the problem of data fragmentation in traditional supervision. The system can comprehensively cover risk factors such as equipment status, personnel behavior, and environmental parameters, avoid the one-sidedness of evaluation caused by a single data dimension, significantly improve the accuracy of safety hazard identification, and enhance the comprehensiveness of risk assessment. Based on the dynamic risk knowledge graph and multi-model fusion algorithm, the system can respond in real time to the complex changes at the construction site, realizing the leap from static rules to dynamic perception, with strong adaptability and a significant improvement in the timeliness of risk early warning. The system automatically generates a hierarchical disposal strategy according to the risk level, and verifies the rectification effect through technologies such as OCR image comparison. The multi-party collaborative interface pushes the instructions to the terminals of the construction, supervision, and construction parties in real time, and records the response time and execution results, forming a complete responsibility traceability log, greatly shortening the rectification response time and significantly improving the efficiency of closed-loop management. Through the compliance binding unit, the system automatically matches the regulatory clauses, generates a standard report with electronic signatures, legal basis indexes, and rectification evidence chains, avoiding omissions in manual comparison, ensuring that the compliance inspection coverage rate reaches 100%, and reducing the legal risk by more than 90%. Dynamically render the risk heat map on the BIM model, support AR on-site navigation, and display the risk trend in combination with the time-axis animation to help managers intuitively grasp the overall risk distribution and evolution path, and greatly improve the decision-making efficiency. Through the dynamic calculation of the associated risk scores and weight coefficients between operation units, the system can quantify the cross-unit risk conduction effect, avoid the limitations of isolated evaluation, effectively improve the coverage rate of high-risk area identification, and greatly reduce the global risk assessment error rate.

[0114] Embodiment 2

[0115] This application embodiment also provides an intelligent method for assessing construction site safety risks, as Figure 2 shown, including S100 - S500.

[0116] S100. Real-time collect multi-source data including equipment status, image and video, personnel status, equipment distance, environmental parameters, construction progress, and project text;

[0117] S200. Perform spatio-temporal alignment, preprocessing, and cross-modal feature fusion on the multi-source data;

[0118] S300. Output risk scores and risk level labels based on the dynamic risk knowledge graph and multi-model fusion algorithm;

[0119] S400. Generate a hierarchical disposal strategy according to the risk score hierarchy, and achieve risk closed-loop control through rectification verification and multi-party collaboration;

[0120] S500. Output a three-dimensional visual risk distribution map and a compliance report.

[0121] Among them, step S300, as Figure 3 shown, further includes the following sub-steps:

[0122] S310: Input multi-source feature data into the dynamic risk knowledge graph and associate the "risk type - cause - consequence" nodes in the graph;

[0123] S320: Use the XGBoost model to assign weights to static risk factors; use the LSTM model to analyze high-frequency sensor data to predict short-term risks; use the Transformer model to fuse multi-source time-series data to predict long-term risk trends;

[0124] S330: Calculate the risk score of the operation unit according to the risk characteristic parameters, calculate the overall project risk score by weighted summation of the operation unit weight and the associated risk value, and map the scoring result to the risk level and automatically associate it with the standard terms;

[0125] S340: Pre-render the risk heat map in the BIM model and mark the locations and cause chains of high-risk areas.

[0126] Taking the scenario of sudden heavy rain in a deep foundation pit as an example, environmental data such as hourly rainfall of 50 mm (exceeding the threshold of 30 mm) → associate the "rainstorm waterlogging" node, sensor data of the foundation pit settlement rate of 0.8 mm / h (exceeding the threshold of 0.5 mm / h) → associate the "support failure" node; the LSTM predicts that the settlement rate will reach 1.2 mm / h in the next 10 minutes, and the Transformer predicts that after continuous rainfall for 6 hours, the groundwater level will rise to the warning line, adding a new associated edge of "rainstorm → support failure → collapse" with an initial weight of 0.7; the operation unit of the foundation pit is calculated to have a score of 90 (high risk), bind Article 4.3.2 of the "Technical Specification for Building Foundation Pit Support", the overall project score is 75 points (high risk) after weighted calculation, and the BIM model marks the foundation pit as a red area, showing the associated risk chain "rainstorm → support failure → collapse".

[0127] Among them, the update method of the dynamic risk knowledge graph is specifically:

[0128] Through incremental learning techniques (such as online random forest algorithms), new construction data (such as hidden danger records, rectification feedback) is updated in real-time to the nodes of the knowledge graph, with triggering conditions such as the number of new data > 100 or the frequency of risk events > 3 times / day; based on the graph attention network (GAT), the association weights between nodes are dynamically optimized (such as the causal chain weight of "rainstorm → support displacement" is increased by 20%), and latent associations are mined (such as "personnel fatigue → operation error → equipment collision"); when a new risk pattern is detected, the graph nodes and associated edges are automatically expanded. For example, in the scenario of a new rainstorm causing foundation pit waterlogging, the knowledge graph automatically adds the associated edge "rainstorm → groundwater level rise → support displacement", and the weight is adjusted from 0.6 to 0.72. When a new risk of prefabricated component installation is detected, the graph nodes are expanded and associated with "hoisting plan defect → bolt loosening → overturning risk".

[0129] The specific processing procedures of each step and sub-step in the above-mentioned intelligent assessment method for construction site safety risks are introduced in detail in the above-mentioned intelligent assessment system for construction site safety risks. Therefore, the repeated description will be omitted here.

[0130] Embodiment III

[0131] Specifically, the hierarchical disposal strategy in step S400 of Embodiment II includes:

[0132] High risk level (≥70 points): Immediately stop work, push a special rectification plan and initiate an emergency response;

[0133] Medium risk level (30 points < risk score < 70 points): Rectify within a time limit and include it in the key inspection list;

[0134] Low risk level (≤30 points): Record the log and update the monitoring plan, and conduct regular reviews.

[0135] In summary, through technologies such as multi-source data fusion, dynamic risk knowledge graph, and intelligent closed-loop management, this application realizes the comprehensive perception, accurate assessment, efficient control, and compliance implementation of engineering supervision safety risks, significantly reduces the incidence of safety accidents, and provides reliable technical support for the construction of smart construction sites.

[0136] The basic principles of this application are described above in combination with specific embodiments. It should be understood that the above-disclosed specific details are only for the purpose of illustration and easy understanding, rather than limitation, and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A construction site safety risk intelligent assessment method, characterized in that: The following steps are involved: S100, real-time collection of multi-source data including equipment status, image video, personnel status, equipment distance, environmental parameters, construction progress and engineering text; S200, performing spatiotemporal alignment, preprocessing and cross-modal feature fusion on multi-source data; S300, output risk scores and risk level labels based on dynamic risk knowledge graph and multi-model fusion algorithm; S400. Generate a hierarchical disposal strategy based on the risk scoring level, and achieve risk closed-loop control through rectification verification and collaboration with multiple parties; S500, output 3D visual risk distribution map and compliance report.

2. The intelligent assessment method for construction site safety risk according to claim 1, characterized in that: The step S300 includes: S310, inputting multi-source feature data into the dynamic risk knowledge graph, and associating the "risk type-cause-consequence" node in the graph; S320, use the XGBoost model to assign weights to static risk factors; use the LSTM model to analyze high-frequency sensor data to predict short-term risks; use the Transformer model to fuse multi-source time series data to predict long-term risk trends; S330, calculating the risk score of the operation unit according to the risk characteristic parameters, calculating the overall risk score of the project according to the weighted sum of the operation unit weight and the associated risk value, and mapping the score result to the risk level and automatically associating it with the standard terms; S340. Pre-render risk heat maps in the BIM model, marking the locations of high-risk areas and their causal chains.

3. The intelligent assessment method for construction site safety risk according to claim 2, characterized in that: The step S300 also includes updating the dynamic risk knowledge graph, specifically: Through incremental learning technology, new construction data is updated to the knowledge graph nodes in real time; Based on the Graph Attention Network (GAT), the association weights between nodes are dynamically optimized; When new risk patterns are detected, graph nodes and associated edges are automatically expanded.

4. The intelligent assessment method for construction site safety risk according to claim 1, characterized in that: The hierarchical disposal strategy in step S400 includes: High risk level (≥70 points): Immediately stop work, push special rectification plan and initiate emergency response; Medium risk level (30 points < risk score < 70 points): rectification within a limited time and inclusion in the key inspection list; Low risk level (≤30 points): Record logs and update monitoring plans, and review them regularly.

5. A construction site safety risk intelligent assessment system, used to implement a construction site safety risk intelligent assessment method according to any one of claims 1 to 4, characterized in that: include: Multi-source data acquisition module, used to obtain IoT sensor data, image and video data, engineering document data, environmental parameter data, personnel biometric data, equipment distance monitoring data and construction progress data at the construction site in real time; Heterogeneous data fusion module, used to perform spatiotemporal alignment, feature extraction and cross-modal correlation analysis on multi-source data to generate a structured risk feature set; Dynamic risk assessment module, which is used to output the overall risk score of the project, the risk score of the operation unit, and the associated risk score between operation units based on the dynamically updated risk knowledge graph and multi-model fusion algorithm, and generate risk level labels; Intelligent closed-loop management module, which is used to generate hierarchical disposal strategies according to risk scoring levels, and realize risk closed-loop control through rectification verification and collaboration with multiple parties; The visualization and compliance module is used to dynamically display risk distribution through a three-dimensional visualization interface and generate legal compliance reports that are bound by industry standards.

6. The intelligent assessment system for safety risks at construction sites according to claim 5, characterized in that: The multi-source data acquisition module comprises: Personnel status monitoring unit, used to monitor personnel heart rate, blood oxygen saturation and falling behavior in real time; Visual behavior recognition unit, which is used to analyze image video streams in real time based on deep learning models, identify construction workers' illegal behaviors, and output behavior categories and confidence levels; The equipment distance monitoring unit is used to calculate the spatial distance between personnel and equipment in real time; and dynamically identify personnel entering the equipment safety warning area based on the depth perception algorithm; The environment and construction data collection unit includes the following subunits: Structural safety monitoring subunit, used to monitor the horizontal displacement and vertical settlement rate of the scaffolding in real time; and to collect foundation pit settlement data; Equipment status monitoring subunit, used to monitor the inclination angle of the tower in real time; and monitor the three-phase current balance status and leakage current value in the temporary power circuit; Environmental monitoring subunit, used to collect real-time data on wind speed, rainfall intensity, and groundwater level changes; The construction data parsing subunit extracts safety hazard keywords from engineering documents based on OCR and natural language processing (NLP) technology; The construction progress tracking sub-unit is integrated with the BIM model to compare the design parameters with the on-site construction progress in real time; it also collects component installation position deviation data through IoT sensors to generate progress deviation reports.

7. The intelligent construction site safety risk assessment system according to claim 5, characterized in that: The heterogeneous data fusion module includes: A spatiotemporal alignment unit, used to align multi-source data according to a unified spatiotemporal reference; Data preprocessing unit, used for data cleaning, filtering and standardization of multi-source data; Feature extraction unit, extracting risk features for different data types; The cross-modal feature interaction unit is used to map feature vectors of different dimensions to a unified N-dimensional space through a fully connected layer. It uses a multi-channel Transformer model to perform cross-modal attention calculations and fuse multi-source features.

8. The intelligent assessment system for safety risks at construction sites according to claim 5, characterized in that: The dynamic risk assessment module comprises: Dynamic risk knowledge graph construction unit, used to build a "risk type-cause-consequence" association network based on historical accident cases and safety regulations; The multi-model fusion evaluation unit is used to use the XGBoost model to assign weights to static risk factors; use the LSTM model to analyze high-frequency sensor data to predict short-term risks; and use the Transformer model to fuse multi-source time series data to predict long-term risk trends; The hierarchical scoring unit is used to output the risk score of the operation unit, map the risk score of the operation unit into three risk levels of low, medium and high according to the preset threshold; and dynamically weighted calculate the overall risk level of the project based on the risk score of the operation unit and the associated risk value between the operation units.

9. The intelligent construction site safety risk assessment system according to claim 5, characterized in that: The intelligent closed-loop management module comprises: The hierarchical disposal strategy unit generates different disposal strategies according to the risk level, including: High risk level (≥70 points): Immediately stop work, push special rectification plan and initiate emergency response; Medium risk level (30 points < risk score < 70 points): rectification within a limited time and inclusion in the key inspection list; Low risk level (≤30 points): record logs and update monitoring plans, and review them regularly; The rectification verification unit verifies the completion of rectification measures through OCR and image comparison technology. If the standards are not met, the secondary disposal process is triggered and the risk level is upgraded; The multi-party collaborative interface unit pushes the disposal instructions to the construction, supervision and construction party terminals synchronously, records the response time and execution results; generates a responsibility traceability log, and associates the timestamp, geographic location and operator information.

10. The intelligent assessment system for safety risks at construction sites according to claim 5, characterized in that: The visualization and compliance module includes: 3D risk thermal map annotation unit, used to mark risk areas in a graded manner on the BIM model; AR on-site navigation unit, used to locate potential hazards through augmented reality equipment and display treatment instructions; Compliance binding unit, used to automatically match standard clauses; generate standard reports with electronic signatures, legal basis indexes and rectification evidence chains.

Images