Large-scale engineering construction potential safety hazard checking and analyzing system

Abstract

The invention discloses a large-scale engineering construction potential safety hazard checking and analyzing system, and belongs to the technical field of construction safety. The system specifically comprises the following steps: S1, a data acquisition and processing module acquires various data required in the operation process of the potential safety hazard checking and analyzing system in a multi-source data acquisition mode, and pre-processes the acquired data to reduce the timestamp error of the data acquired by multiple devices; processing the collected data to eliminate sensor noise in the data, and transmitting the processed data to a risk identification and early warning module. By integrating multi-protocol conversion and intelligent data cleaning, a traditional data island is broken, real-time synchronization and deep fusion of multi-source data such as a BIM model, a sensor and video monitoring are achieved, a unified and high-credibility data basis is provided for subsequent analysis, and information integration efficiency and comprehensiveness of decision basis are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of construction safety, and in particular to a system for detecting and analyzing safety hazards in large-scale engineering construction. Background Art

[0002] Safety management is crucial during construction. Construction sites are complex environments, feature numerous personnel, and employ a wide variety of equipment. These sites present a variety of potential safety hazards, such as unsafe personnel behavior, equipment failure, and abnormal environmental factors, all of which can lead to serious accidents. Therefore, effectively identifying and analyzing these safety hazards has become a crucial issue in construction safety management.

[0003] Traditional methods for managing construction safety hazards rely primarily on manual inspections and on-site monitoring, which have numerous limitations, including high labor costs, low efficiency, and strong subjectivity. The development of technologies such as the Internet of Things, big data, and artificial intelligence has provided new approaches and methods for managing construction safety hazards. By deploying a variety of sensors at construction sites, such as temperature and humidity sensors, gas detectors, video surveillance, and RFID personnel location systems, real-time data on environmental parameters, equipment status, and personnel behavior can be collected, providing data support for the identification, assessment, and management of safety hazards. However, relying solely on sensor-collected data is insufficient for effective safety hazard management, lacking comprehensive evaluation of hazard response timeliness, resource matching, and long-term management. Summary of the Invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a large-scale engineering construction safety hazard inspection and analysis system; it can solve the problem of lack of comprehensive evaluation of hazard response timeliness, resource matching and long-term management by relying solely on data collected by sensors.

[0005] Technical solution: To solve the above technical problems, according to one aspect of the present invention, more specifically, a large-scale engineering construction safety hazard investigation and analysis system, the system specifically includes the following steps: S1. The data acquisition and processing module collects various data required during the operation of the safety hazard investigation and analysis system through multi-source data acquisition, pre-processes the collected data to reduce the timestamp error of data collected by multiple devices, processes the collected data to eliminate sensor noise in the data, and transmits the processed data to the risk identification and early warning module; S2, the risk identification and early warning module, uses deployed edge nodes to analyze pre-processed data in real time, identifying violations at the construction site, capturing key frames and key video information of violations, and issuing early warnings for identified violations at the construction site; S3. The disposal plan generation module predicts the evolution path of compound risks and generates disposal plans based on the warnings issued by the risk identification module and the knowledge graph and risk propagation simulation engine; S31. Integrate construction specifications and historical accident data to form a four-dimensional correlation network of personnel, machinery, environment, and management to build a knowledge graph; S32. Use algorithms to calculate the influence of risk nodes, predict the expansion path of compound risks, and simulate risk propagation; S33. Automatically generate a treatment plan based on the simulation results, and push the generated treatment plan; S4. The rectification record storage module uses SHA-256 hashing to link the rectification records to an unalterable record, allowing the supervisory unit to verify the authenticity of the rectification without exposing the original data. S5. The analysis and processing module sets relevant thresholds in the analysis and processing process, analyzes and processes various relevant data collected by the data acquisition and processing module during the hidden danger rectification process to obtain a hidden danger response index for the current construction site, and performs corresponding operations based on the comparison results between the hidden danger response index for the current construction site and the first and second hidden danger response index thresholds; S6. The hidden dangers, rectification status and disposal plans of the current site are displayed through the visualization module by constructing a virtual model that is proportional to the current construction site, and warning information is pushed to remind.

[0006] Furthermore, when preprocessing data, the data acquisition and preprocessing module uses a clock synchronization algorithm based on the network time protocol to reduce the timestamp error of the collected data and eliminates sensor noise through sliding window filtering.

[0007] Furthermore, the risk identification and warning module deploys an improved YOLOv7 algorithm during the risk identification process, adds a shadow compensation convolution layer, and processes the pre-processed data at the edge node, including but not limited to the Cambrian MLU270 chip invisible Kalman filter algorithm.

[0008] Furthermore, the disposal solution recommended by the disposal solution generation module includes a requirements list and an operation flow chart.

[0009] Furthermore, the rectification record stored in the rectification record storage module includes a timestamp, GPS coordinates, images, various data during the rectification process, and various rectification tracking data after the rectification is completed.

[0010] Furthermore, during the analysis and processing process, the analysis and processing module obtains a hidden danger rectification timeliness score by analyzing and processing the preset standard processing time and the actual processing time: ; Among them, SP is the rectification effectiveness score, The preset standard processing time for each hidden danger, The actual handling time of each hidden danger is the total time from the discovery of the hidden danger to the acceptance. n is the total number of current hidden dangers. The quality score of hidden danger rectification is obtained through comprehensive analysis of the number of hidden danger rectification acceptances, the total number of rectifications, the measured deviation value after rectification, and the allowable deviation threshold: ; Among them, ZP is the quality score of hidden danger rectification, YC is the number of times the hidden danger rectification has passed the acceptance, ZC is the total number of rectifications, GP is the measured deviation value after rectification, and XP is the allowable deviation threshold; The recurrence rate of the same type of hidden dangers at the same location, the total number of occurrences of the same type of hidden dangers at the same location, the average number of days between two adjacent similar hidden dangers at the same location, and the set threshold days are comprehensively analyzed to obtain the recurrence rate of the hidden dangers: ; Among them, CL is the recurrence rate of hazards, FC is the number of recurrences of the same type of hazards at the same location, GC is the total number of occurrences of the same type of hazards at the same location, GD is the average number of days between two adjacent hazards of the same type at the same location, and YD is the set threshold number of days; The process delay index is obtained by comprehensively analyzing the delay time of each process and the standard processing time of the process: ; Among them, YZ is the process delay index, The delay time for each process, is the standard processing time of the process, and m is the total number of processes; The actual resource investment in hidden danger rectification is obtained through comprehensive analysis of the actual number of people invested, standard number of people invested, actual number of equipment invested, standard number of equipment invested, actual investment funds and standard budget funds: ; Among them, ST is the actual resource investment in hidden danger rectification, SR is the number of people invested in hidden danger rectification, BR is the number of people invested in the standard hidden danger rectification, SS is the actual number of equipment invested, BS is the number of standard equipment invested, SZ is the time investment amount, and BY is the standard budget amount; The hazard response index of the current construction site is obtained through comprehensive analysis of the hazard rectification timeliness score, hazard rectification quality score, recurring hazard rate, process delay index, actual resource investment for each hazard rectification, and the hazard benchmark resource investment: ; Among them, XZ is the hidden danger response index of the current construction site, The actual resources invested in each hidden danger rectification, This is the baseline resource investment for the hidden danger rectification.

[0011] Furthermore, when the analysis and processing module performs corresponding operations according to the comparison results between the hidden danger response index of the current construction site and the first threshold value of the hidden danger response index and the second threshold value of the hidden danger response index, when When the hidden danger response index reaches the first threshold, it means that the hidden danger response efficiency meets the standard and the construction site operates normally; When the first threshold of the hidden danger response index When the hidden danger response index reaches the second threshold, it indicates that the response efficiency is insufficient and the response process and resource allocation need to be optimized; when When the hidden danger response index reaches the second threshold, it means that the response capability is seriously insufficient, there is a risk of major accidents, and work needs to be stopped immediately for rectification.

[0012] Furthermore, the analysis and processing module sets the hidden danger response index, the first threshold value of the hidden danger response index, and the second threshold value of the hidden danger response index according to the requirements of different construction sites. Beneficial effects

[0013] 1. By integrating multi-protocol conversion and intelligent data cleaning, it breaks the traditional data silos and realizes the real-time synchronization and deep integration of multi-source data such as BIM models, sensors, and video surveillance, providing a unified and highly reliable data foundation for subsequent analysis, significantly improving the efficiency of information integration and the comprehensiveness of decision-making basis.

[0014] 2. Through locally deployed intelligent algorithms and anti-interference processing, efficient risk identification and instantaneous response in complex scenarios are achieved, reducing dependence on cloud transmission, ensuring real-time monitoring capabilities in unstable network environments, while reducing data transmission load and improving the overall robustness of the system.

[0015] 3. Based on knowledge graphs and simulation technology, it accurately identifies the evolution patterns of complex risks and provides scientific decision-making support. By simulating the chain reactions of different risk scenarios, it generates optimized disposal plans, helping managers predict potential crises and develop targeted prevention and control strategies.

[0016] 4. Through the quantitative assessment model, the timeliness of hidden danger response, rectification quality and resource adaptability are dynamically measured. Combined with the graded early warning mechanism, the standardization of hidden danger handling processes and the rationalization of resource scheduling are promoted, effectively improving safety management efficiency and systemic risk prevention and control capabilities.

[0017] 5. Utilize distributed ledger and smart contract technologies to ensure the immutability and transparent traceability of rectification process data, simplify the acceptance process, enhance trust in multi-party collaboration, and provide reliable digital certificate support for security management.

[0018] 6. Through high-precision virtual mapping and immersive interaction, the construction safety situation can be visualized and emergency plans can be dynamically simulated, helping managers to intuitively locate risks, optimize resource allocation, and verify the effectiveness of disposal plans in advance, thereby improving the scientific nature and foresight of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the system principle. DETAILED DESCRIPTION

[0020] In order to make the technical solution of the present invention clearer, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. Example 1

[0021] The data acquisition and preprocessing module uses multi-source data acquisition to collect all the data required during the operation of the safety hazard investigation and analysis system, including: Environmental perception data: Real-time environmental parameters collected by temperature and humidity sensors, gas detectors such as CO2, and dust concentration sensors are used to identify environmental safety hazards such as insufficient ventilation and excessive harmful gases; Equipment operation data: RFID tags and IoT sensors collect operating status data such as load, speed, and vibration frequency of construction machinery such as tower cranes and elevators to predict equipment failure risks; Personnel behavior data: High-definition cameras deployed at the construction site are used to collect personnel work behavior images through video surveillance. Combined with location information obtained by the RFID personnel positioning system, this is used to identify illegal operations such as not wearing a hard hat or not wearing a safety belt when working at height; Structural design and construction process data in the Building Information Model (BIM) are integrated with real-time on-site data to construct a dynamic digital twin of the construction environment; Structured data such as historical hazard records, rectification plans, and accident cases are used to preprocess the collected data using a clock synchronization algorithm based on the Network Time Protocol to reduce timestamp errors in data collected by multiple devices, and the collected data is filtered through a sliding window to eliminate sensor noise in the data.

[0022] The risk identification and warning module uses the deployed edge nodes to perform real-time analysis on the pre-processed data, identify violations at the construction site, capture key frames of violations and key video information, and issue warnings for identified violations at the construction site. During the risk identification process, the improved YOLOv7 algorithm is deployed, and a shadow compensation convolution layer is added to improve image recognition accuracy under complex lighting conditions and reduce false alarms or missed alarms due to environmental interference. The edge nodes include but are not limited to the Cambrian MLU270 chip invisible Kalman filter algorithm, and the pre-processed data is processed to predict violation trends in advance, such as people approaching dangerous areas. Once a violation is identified, the system immediately captures key frames and video clips, generates warning signals locally through the edge nodes, and simultaneously pushes them to the manager's mobile terminal, solving the delay problem of traditional cloud processing and improving the timeliness of hidden danger response.

[0023] The disposal plan generation module builds a knowledge graph based on the knowledge graph, integrating construction specifications and historical accident data to form a four-dimensional association network among personnel, machinery, environment, and management. For example, it associates the behavior of "not wearing a safety belt" with "falling from height" and "overloading of a tower crane" with "risk of equipment overturning" to clarify the risk transmission path. Through risk propagation simulation, an algorithm is used to calculate the influence of risk nodes and predict the expansion path of compound risks. The PageRank algorithm is used to calculate the influence of risk nodes, combined with a Bayesian network to predict the expansion path of compound risks, such as the probability of fire risk spreading to adjacent areas. The resource requirements of different disposal plans, such as personnel input, equipment scheduling, and time cost, are quantified. The risk propagation simulation engine automatically recommends disposal plans based on simulation results. It predicts the evolution path of compound risks and generates disposal plan packages including a list of requirements and an operation flow chart. Based on the simulation results, the module automatically recommends disposal plan packages that include a list of requirements, such as the number of safety officers required, the type of emergency equipment, and an operation flow chart. For example, for the hidden danger of "improper scaffolding erection," the resource combination of "two scaffolders, safety belts, and wrenches" and a standardized rectification process are recommended to avoid resource waste or under-allocation.

[0024] Rectification record storage module: used to link the rectification record through SHA-256 hash operation to make it tamper-proof, including timestamp, GPS coordinates, images, various data during the rectification process, and rectification tracking data after the rectification is completed. The record content includes: timestamp accurate to the second, GPS coordinate positioning of the rectification location, image comparison before and after rectification, rectification process work order including responsible person, time consumption, acceptance report, etc., which supports the supervision unit to verify the authenticity of the rectification without exposing the original data. The supervision unit can verify the authenticity of the rectification through blockchain smart contracts, such as comparing the hash values ​​of images before and after rectification, without exposing the original data. It supports historical hidden danger tracing and management responsibility tracing, solves the problem of traditional paper records being easily lost and difficult to trace, and strengthens the long-term management mechanism.

[0025] Analysis and processing module: Set relevant thresholds in the analysis and processing process, analyze and process various relevant data in the hidden danger rectification process to obtain the hidden danger response index of the current construction site, and perform corresponding operations based on the comparison results between the hidden danger response index of the current construction site and the first threshold value of the hidden danger response index and the second threshold value of the hidden danger response index. The analysis and processing module sets the hidden danger response index and the first threshold value of the hidden danger response index and the second threshold value of the hidden danger response index according to the requirements of different construction sites.

[0026] During the analysis and processing process, the analysis and processing module obtains the hidden danger rectification timeliness score through analysis and processing of the preset standard processing time and the actual processing time: ; Among them, SP is the rectification effectiveness score, The preset standard processing time for each hidden danger, The actual processing time for each hidden danger is the total time from the discovery of the hidden danger to the acceptance. n is the total number of current hidden dangers. The preset standard processing time for each hidden danger is According to the type of hidden dangers such as lack of edge protection, overload of tower crane, etc., combined with industry specifications such as the "Construction Safety Inspection Standard", the preset standard processing time for the hidden danger of "not wearing a safety belt during high-altitude work" is set to 2 hours, reflecting the rigid demand of safety management for "quick response". The actual processing time for each hidden danger is The whole process time from hidden danger discovery to acceptance directly reflects the on-site response speed. , indicating a delayed response, which may lead to the escalation of hidden dangers into accidents. The hidden danger rectification timeliness score is used to measure the corresponding speed from the occurrence of hidden dangers to the completion of rectification. The shorter the actual processing time, the higher the hidden danger rectification timeliness score. The hidden danger response index is improved, and the logarithmic function is used to strengthen the short-term improvement effect. The sensitivity of the logarithmic function to numerical changes decreases as the input value increases, which amplifies the improvement effect of "from delay to standard". It encourages sites to prioritize shortening the processing time of overdue hidden dangers, which is in line with the core goal of "time is safety"; The quality score of hidden danger rectification is obtained through comprehensive analysis of the number of hidden danger rectification acceptances, the total number of rectifications, the measured deviation value after rectification, and the allowable deviation threshold: ; Among them, ZP is the quality score of hidden danger rectification, YC is the number of times the hidden danger rectification has passed the acceptance, ZC is the total number of rectifications, GP is the measured deviation value after rectification, XP is the allowable deviation threshold, and the rectification pass rate is Measuring the completion of rectification work. If the pass rate is low, it means that half of the hidden dangers have not been resolved, which directly reflects management loopholes. To control deviations, measure the accuracy of rectification quality. If the measured deviation after rectification exceeds the allowable threshold, even if the pass rate is high, the rectification quality score will still be significantly reduced. The hidden danger rectification quality score is used to comprehensively evaluate the rectification pass rate and project compliance. It requires both a high rectification pass rate and a small deviation. When the deviation after rectification exceeds the allowable value, the hidden danger rectification quality score is significantly reduced. A low rectification quality score may cause secondary hidden dangers. For example, unqualified rectification leads to insufficient structural stability. The rectification quality score is used to force the rectification to be not only completed but also marked, to avoid the recurrence of hidden dangers due to insufficient rectification quality, and to support the goal of "long-term management"; The recurrence rate of the same type of hidden dangers at the same location, the total number of occurrences of the same type of hidden dangers at the same location, the average number of days between two adjacent similar hidden dangers at the same location, and the set threshold days are comprehensively analyzed to obtain the recurrence rate of the hidden dangers: ; Among them, CL is the recurrence rate of the same type of hidden danger at the same location, FC is the number of recurrences of the same type of hidden danger at the same location, GC is the total number of times the same type of hidden danger occurs at the same location, GD is the average number of days between two adjacent hidden dangers of the same type at the same location, and YD is the set threshold number of days. Recurrence ratio: If the number of hidden dangers of the same type at the same location accounts for a high proportion of the total number of hidden dangers, it means that the management measures have not addressed the root cause. Interval penalty: If the average interval between two similar hazards at the same location is shorter than the set threshold, the square root operation amplifies the negative impact of short-term recurrences. The recurrence rate reflects the recurrence frequency and interval of similar hazards at the same location. The shorter the interval, the more significant the negative impact on the hazard response index. A high recurrence rate indicates loopholes in the management process. In the hazard response index, the recurrence rate has a weight of -0.4, which is higher than the weight of the process delay index. This indicates that recurring hazards are a more serious systemic problem than process delays and require priority treatment, in line with the principle that "long-term management requires eliminating the root cause." The process delay index is obtained by comprehensively analyzing the delay time of each process and the standard processing time of the process: ; Among them, YZ is the process delay index, The delay time for each process, is the standard processing time of the process, m is the total number of processes, and the delay time of each process covers "hazard discovery → early warning push → rectification implementation → acceptance confirmation". Delays in each link, such as delayed supervision acceptance and overtime in the allocation of rectification materials, are directly related to the response timeliness. The standard processing time of this process is set according to the construction process specifications. The process delay index reflects the efficiency deviation of a single link and quantifies the overall process loss after accumulation. It represents the comprehensive proportion of delay time in each link, including discovery, rectification, rectification, and acceptance delays. The longer the delay time, the greater the process delay index, and the more obvious the negative impact on the hidden danger response index. In the hidden danger response index, the weight of the process delay index is -0.3, which is lower than the weight of the recurring hidden danger rate. It reflects the risk classification of "occasional process delays" and "systemic recurring hidden dangers", avoids excessive negation of overall management effectiveness due to local delays, and guides the priority resolution of root causes. The actual resource investment in hidden danger rectification is obtained through comprehensive analysis of the actual number of people invested, standard number of people invested, actual number of equipment invested, standard number of equipment invested, actual investment funds and standard budget funds: ; Among them, ST is the actual resource investment in hidden danger rectification, SR is the number of people invested in hidden danger rectification, BR is the number of people invested in the standard hidden danger rectification, SS is the actual number of equipment invested, BS is the number of standard investment equipment, SZ is the time investment amount, BY is the standard budget amount. The actual resource investment is obtained by weighting the three parts of manpower, equipment, and funds. Prioritize manpower investment, match construction equipment demand to avoid rectification interruptions due to equipment shortages, control cost rationality, quantify the adequacy of actual investment resources, and reflect the resource guarantee capacity of on-site response. The hazard response index of the current construction site is obtained through comprehensive analysis of the hazard rectification timeliness score, hazard rectification quality score, recurring hazard rate, process delay index, actual resource investment for each hazard rectification, and the hazard benchmark resource investment: ; Among them, XZ is the hidden danger response index of the current construction site, The actual resources invested in each hidden danger rectification, As the benchmark resource investment for this hidden danger rectification, the rectification effectiveness score is given a weight of 0.8, reflecting the core management concept that time is safety. The shorter the processing time, the lower the probability of an accident. The hidden danger rectification quality score is given a weight of 0.6. The probability of an accident resulting from incomplete rectification is much higher than that resulting from timely rectification. Unqualified rectification may cause secondary hidden dangers. The weight slightly lower than the rectification effectiveness score reflects the guarantee of rapid processing while ensuring the quality of rectification. The recurrence hidden danger rate is given a weight of -0.4. Recurrence hidden dangers reflect fundamental defects in the management system. The cost of handling the same hidden danger twice will increase. The negative weight strengthens the constraint on management inertia. The process delay index is given a weight of -0.3. Each link is positively correlated with the probability of an accident. The weight lower than the recurrence hidden danger rate reflects the hierarchical treatment of systemic problems. The positive items of hidden danger rectification timeliness score and hidden danger rectification quality score represent the "contribution factor", and the negative items of repeated hidden danger rate and process delay index represent the "penalty factor". Through weight sorting, they reflect the management priority: response speed > rectification quality > eradication of recurrence > process optimization. The hidden danger rectification timeliness score uses a logarithmic function to amplify short-term improvements and forcibly shorten the time for handling overdue hidden dangers, which meets the needs of "time-sensitive safety management"; the actual resource input is quantified through a three-dimensional weighted model to determine the rationality of manpower, equipment, and funds. The power function avoids resource dominance and emphasizes "adaptation rather than redundancy"; the repeated hidden danger rate and process delay index reflect long-term management, and the negative weight of the recurrence rate is used to punish systemic loopholes. The quality score eliminates "superficial rectification". The two indicators support the eradication of hidden dangers and the optimization of the management system; the process delay index reflects the systematic nature of the process, quantifies the loss of the entire process, and uses a low weight to distinguish between occasional problems and root defects, guiding management resources to prioritize the resolution of key contradictions.

[0027] pass Mapping the linear combination to (0, 1) intuitively reflects the response efficiency, through When resources are sufficient, the hidden danger response index will be slightly increased; when resources are insufficient, the hidden danger response index will be moderately reduced. When the index is 0.2, the correction will be mild to avoid the result of this dominating hidden danger response index.

[0028] When the analysis and processing module performs corresponding operations based on the comparison results between the hidden danger response index of the current construction site and the first threshold value of the hidden danger response index and the second threshold value of the hidden danger response index, when When the hidden danger response index reaches the first threshold, it means that the hidden danger response efficiency meets the standard and the construction site operates normally; When the first threshold of the hidden danger response index When the hidden danger response index reaches the second threshold, it indicates that the response efficiency is insufficient and the response process and resource allocation need to be optimized; when When the hidden danger response index reaches the second threshold, it means that the response capability is seriously insufficient, there is a risk of major accidents, and work needs to be stopped immediately for rectification. Example 2

[0029] When calculating the hidden danger response index, when SP=0.5, ZP=0.9, CL=0.2, YZ=0.8, When: ; At this time, 0.66 is compared with the first threshold of the hidden danger response index and the second threshold of the hidden danger response index: when When the hidden danger response index reaches the first threshold, it means that the hidden danger response efficiency meets the standard and the construction site operates normally; When the first threshold of the hidden danger response index When the hidden danger response index reaches the second threshold, it indicates that the response efficiency is insufficient and the response process and resource allocation need to be optimized; when When the hidden danger response index reaches the second threshold, it means that the response capability is seriously insufficient, there is a risk of major accidents, and work needs to be stopped immediately for rectification.

[0030] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A large-scale engineering construction safety hazard investigation and analysis system, characterized by: The system specifically includes the following steps: S1. The data acquisition and processing module collects various data required during the operation of the safety hazard investigation and analysis system through multi-source data acquisition, pre-processes the collected data to reduce the timestamp error of data collected by multiple devices, processes the collected data to eliminate sensor noise in the data, and transmits the processed data to the risk identification and early warning module; S2, the risk identification and early warning module, uses deployed edge nodes to analyze pre-processed data in real time, identifying violations at the construction site, capturing key frames and key video information of violations, and issuing early warnings for identified violations at the construction site; S3. The disposal plan generation module predicts the evolution path of compound risks and generates disposal plans based on the warnings issued by the risk identification module and the knowledge graph and risk propagation simulation engine; S31. Integrate construction specifications and historical accident data to form a four-dimensional correlation network of personnel, machinery, environment, and management to build a knowledge graph; S32. Use algorithms to calculate the influence of risk nodes, predict the expansion path of compound risks, and simulate risk propagation; S33. Automatically generate a treatment plan based on the simulation results, and push the generated treatment plan; S4. The rectification record storage module uses SHA-256 hashing to link the rectification records to an unalterable record, allowing the supervisory unit to verify the authenticity of the rectification without exposing the original data. S5. The analysis and processing module sets relevant thresholds in the analysis and processing process, analyzes and processes various relevant data collected by the data acquisition and processing module during the hidden danger rectification process to obtain a hidden danger response index for the current construction site, and performs corresponding operations based on the comparison results between the hidden danger response index for the current construction site and the first and second hidden danger response index thresholds; S6. The hidden dangers, rectification status and disposal plans of the current site are displayed through the visualization module by constructing a virtual model that is proportional to the current construction site, and warning information is pushed to remind.

2. A large-scale engineering construction safety hazard investigation and analysis system according to claim 1, characterized in that: When preprocessing data, the data acquisition and preprocessing module uses a clock synchronization algorithm based on the network time protocol to reduce the timestamp error of the collected data and eliminates sensor noise through sliding window filtering.

3. A large-scale engineering construction safety hazard investigation and analysis system according to claim 1, characterized in that: During the risk identification process, the risk identification and warning module deploys an improved YOLOv7 algorithm, adds a shadow compensation convolution layer, and processes the preprocessed data at the edge node, including but not limited to the Cambrian MLU270 chip invisible Kalman filter algorithm.

4. A large-scale engineering construction safety hazard investigation and analysis system according to claim 1, characterized in that: The disposal solution recommended by the disposal solution generation module includes a requirements list and an operation flow chart.

5. A large-scale engineering construction safety hazard investigation and analysis system according to claim 1, characterized in that: The rectification record storage module stores rectification records including timestamps, GPS coordinates, images, various data during the rectification process, and various rectification tracking data after the rectification is completed.

6. A large-scale engineering construction safety hazard investigation and analysis system according to claim 5, characterized in that: During the analysis and processing process, the analysis and processing module obtains the hidden danger rectification timeliness score through analysis and processing of the preset standard processing time and the actual processing time: ； Among them, SP is the rectification effectiveness score, The preset standard processing time for each hidden danger, The actual handling time of each hidden danger is the total time from the discovery of the hidden danger to the acceptance. n is the total number of current hidden dangers. The quality score of hidden danger rectification is obtained through comprehensive analysis of the number of hidden danger rectification acceptances, the total number of rectifications, the measured deviation value after rectification, and the allowable deviation threshold: ； Among them, ZP is the quality score of hidden danger rectification, YC is the number of times the hidden danger rectification has passed the acceptance, ZC is the total number of rectifications, GP is the measured deviation value after rectification, and XP is the allowable deviation threshold; The recurrence rate of the same type of hidden dangers at the same location, the total number of occurrences of the same type of hidden dangers at the same location, the average number of days between two adjacent similar hidden dangers at the same location, and the set threshold days are comprehensively analyzed to obtain the recurrence rate of the hidden dangers: ； Among them, CL is the recurrence rate of hazards, FC is the number of recurrences of the same type of hazards at the same location, GC is the total number of occurrences of the same type of hazards at the same location, GD is the average number of days between two adjacent hazards of the same type at the same location, and YD is the set threshold number of days; The process delay index is obtained by comprehensively analyzing the delay time of each process and the standard processing time of the process: ； Among them, YZ is the process delay index, The delay time for each process, is the standard processing time of the process, and m is the total number of processes; The actual resource investment in hidden danger rectification is obtained through comprehensive analysis of the actual number of people invested, standard number of people invested, actual number of equipment invested, standard number of equipment invested, actual investment funds and standard budget funds: ； Among them, ST is the actual resource investment in hidden danger rectification, SR is the number of people invested in hidden danger rectification, BR is the number of people invested in the standard hidden danger rectification, SS is the actual number of equipment invested, BS is the number of standard equipment invested, SZ is the time investment amount, and BY is the standard budget amount; The hazard response index of the current construction site is obtained through comprehensive analysis of the hazard rectification timeliness score, hazard rectification quality score, recurring hazard rate, process delay index, actual resource investment for each hazard rectification, and the hazard benchmark resource investment: ； Among them, XZ is the hidden danger response index of the current construction site, The actual resources invested in each hidden danger rectification, This is the baseline resource investment for the hidden danger rectification.

7. A large-scale engineering construction safety hazard investigation and analysis system according to claim 6, characterized in that: When the analysis and processing module performs corresponding operations according to the comparison results between the hidden danger response index of the current construction site and the first threshold value of the hidden danger response index and the second threshold value of the hidden danger response index, when When the hidden danger response index reaches the first threshold, it means that the hidden danger response efficiency meets the standard and the construction site operates normally; When the first threshold of the hidden danger response index When the hidden danger response index reaches the second threshold, it indicates that the response efficiency is insufficient and the response process and resource allocation need to be optimized; when When the hidden danger response index reaches the second threshold, it means that the response capability is seriously insufficient, there is a risk of major accidents, and work needs to be stopped immediately for rectification.

8. The large-scale engineering construction safety hazard investigation and analysis system according to claim 1 is characterized by: The analysis and processing module sets the hidden danger response index, the first threshold value of the hidden danger response index, and the second threshold value of the hidden danger response index according to the requirements of different construction sites.

Images