A construction site safety information supervision system based on the Internet of Things

Abstract

The present invention provides a construction site safety information supervision system based on the Internet of Things, which relates to the field of data processing technology. The present invention dynamically optimizes the hierarchical early warning strategy through deep reinforcement learning, combines multi-source data compression with a federated learning framework, significantly improves the accuracy of risk assessment, and suppresses data noise interference; constructs a risk coupling model based on complex network theory, analyzes the inter-regional transmission path and chain effect, optimizes the global early warning plan, and enhances the system's adaptability to complex construction environments; realizes continuous optimization of the model through a dynamic weight update mechanism, maintains high adaptability under changes in the construction site environment, and adopts a dynamic resource allocation strategy to effectively solve the problem of low efficiency in high-concurrency processing and improve the overall accident early warning accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a construction site safety information monitoring system based on the Internet of Things. Background Art

[0002] Construction site safety information monitoring systems originated in the late 20th century. With the expansion of the construction industry and the increasing frequency of accidents, traditional manual monitoring models were no longer sufficient. Early systems relied on simple sensors and paper records. However, with the development of the Internet of Things (IoT), big data, and artificial intelligence (AI) in the early 21st century, these systems gradually evolved into intelligent monitoring platforms.

[0003] In the prior art, the publication number is CN114723388A, and the name is a smart construction site comprehensive safety operation information management system. This invention aims to solve the problem of relatively independent information and lack of information flow among various departments during the construction process. The method includes a user layer, an application layer, a support layer, a data layer, a transmission layer, a device layer, and a control layer. The user layer conveniently provides users with multifunctional smart construction site platform services. The subsystems in the support platform can operate both collaboratively and independently, ensuring the integrity of the system and advancing in stages. The control layer can effectively reduce the temperature in the house, collect every rainfall, make full use of non-traditional water resources, save every drop of tap water as much as possible, recycle rainwater, and use it for on-site car washing, concrete maintenance, on-site spraying for dust and haze removal, and can be used for linkage with roof water spray cooling control systems and for cooling board houses in workers' living areas, saving water and electricity.

[0004] However, the above invention has the following technical disadvantages when applied to big data analysis and risk prediction:

[0005] 1. Data noise interference is large, affecting the accuracy of prediction;

[0006] 2. Insufficient real-time computing capabilities and low efficiency in high-concurrency processing;

[0007] 3. The model has weak generalization ability and poor adaptability to dynamic environments.

[0008] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0009] The purpose of the present invention is to provide a construction site safety information monitoring system based on the Internet of Things to solve the problems raised in the above background technology.

[0010] To achieve the above object, the present invention provides the following technical solutions:

[0011] A construction site safety information supervision system based on the Internet of Things, including an area monitoring module, a data acquisition module, a risk calculation module, and an intelligent early warning module;

[0012] Regional monitoring module: used to provide different regional settings for the construction preparation stage and construction stage of the target construction site;

[0013] During the construction preparation phase, the target construction site is divided into multiple monitoring sub-areas;

[0014] During the construction phase, the construction plan and operation type of each monitoring sub-area are obtained and analyzed, and based on the analysis results, these monitoring sub-areas are divided into a core monitoring area set and an auxiliary monitoring area set;

[0015] Data acquisition module: used to collect multi-source safety monitoring data of each monitoring sub-area during the construction phase;

[0016] Based on the topological relationship of the monitoring sub-areas, the collection frequency of the core monitoring area set and the auxiliary monitoring area set is adjusted, and the communication quality is monitored at the same time;

[0017] Risk calculation module: used to build a regional risk assessment model using federated learning during the construction phase. This regional risk assessment model is used to calculate the risk entropy value Re of each monitoring sub-area within the auxiliary monitoring area set. The risk entropy value Re is used to assess whether the corresponding monitoring sub-area is abnormal;

[0018] Calculate the risk diffusion coefficient Rd of the abnormal monitoring sub-area. The risk diffusion coefficient Rd is used to evaluate the risk diffusion trend to the core monitoring area set;

[0019] Intelligent early warning module: used to calculate and evaluate the risk intervention priority Rp based on the risk diffusion coefficient Rd; then use reinforcement learning to optimize the hierarchical early warning strategy, dynamically adjust the sampling frequency and resource allocation of the core monitoring area set and the auxiliary monitoring area set, and at the same time build an inter-regional risk coupling model to optimize the global early warning strategy.

[0020] Preferably, the regional monitoring module includes a sub-region division unit;

[0021] The sub-area division unit is used to obtain the total area of ​​the construction site during the construction preparation stage, and to divide the site in combination with the construction drawings. The construction site is divided into independent monitoring sub-areas and assigned unique identifiers, including {1, 2, 3, ..., n}, where n represents the total number of monitoring sub-areas and n is the index mark of the nth monitoring sub-area. The center point coordinates and boundary information of each monitoring sub-area are also recorded.

[0022] Preferably, the regional monitoring module further includes a monitoring area classification unit;

[0023] The monitoring area classification unit is used to divide all monitoring sub-areas into a core monitoring area set and an auxiliary monitoring area set. Based on the construction plan and operation type, the personnel density Za, structural stability Zb, and equipment impact area Zc of each monitoring sub-area are obtained in real time. After dimensionless processing, the comprehensive monitoring priority value Zhj is calculated using the following formula: In the formula, the range of the personnel density Za, structural stability Zb, and equipment influence area Zc after dimensionless processing is [0,1], and the larger the comprehensive monitoring priority value Zhj, the higher the risk; m is the correction coefficient of the equipment influence area Zc;

[0024] According to the comprehensive monitoring priority value Zhj of each monitoring sub-area, all monitoring sub-areas are sorted in descending order from high to low, and the monitoring sub-areas in the top 50% are divided into the core monitoring area set, and the rest are divided into the auxiliary monitoring area set.

[0025] Preferably, the data acquisition module includes a multi-source data acquisition unit and a dynamic control unit;

[0026] The multi-source safety monitoring data collected by the multi-source data acquisition unit includes equipment status, personnel location, environmental parameters, and video stream data. A compressed sensing algorithm is used to compress and optimize the data. A hierarchical storage structure is then established based on the data types of equipment status, personnel location, environmental parameters, and video stream data. Equipment status and personnel location data are stored in a high-frequency access database, while environmental parameters and video streams are stored in a time series database. A time tag is then added to each piece of data, using a UTC timestamp accurate to milliseconds, to extract the construction activity intensity Pa within the monitored sub-area.

[0027] The dynamic control unit is used to adjust the collection frequency of the core monitoring area set and the auxiliary monitoring area set based on the topological relationship of the monitoring sub-area, and monitor the communication quality to optimize the transmission strategy. Specifically, when the communication quality is lower than the preset threshold, the local data cache and frequency reduction transmission mechanism are started; at the same time, the communication quality Pb ​​within the monitoring sub-area is extracted.

[0028] Preferably, the risk calculation module includes a risk modeling unit;

[0029] The risk modeling unit is used to build a regional risk assessment model using a federated learning framework. The model is trained through distributed node collaboration and integrates multi-source security monitoring data from all monitored sub-regions.

[0030] The local model parameters of each monitoring sub-area are then privacy-protected using a homomorphic encryption algorithm. The encrypted parameters are then aggregated and updated on the central server using a federated averaging algorithm. A dynamic weight update mechanism is also implemented to automatically adjust the model update cycle based on the frequency of environmental changes, including equipment movement and work surface expansion, at the construction site.

[0031] Preferably, the risk calculation module further includes an entropy value calculation unit;

[0032] The entropy value calculation unit is used to preferentially calculate the risk entropy value Re of each monitoring sub-area in the auxiliary monitoring area set, and use the regional risk assessment model to perform real-time analysis on the multi-source safety monitoring data to obtain the abnormal gathering of personnel Ea, density equipment abnormality Eb and vibration amplitude environmental mutation index Ec. After dimensionless processing, the risk entropy value Re is calculated using the following formula: ; In the formula, the value range of abnormal personnel gathering Ea is Ea∈[0,5], and the larger the value, the more abnormal the personnel gathering; the value range of density equipment abnormality Eb is Eb∈[0,3], and the larger the value, the more serious the abnormal equipment distribution; the value range of vibration amplitude environmental mutation index Ec is Ec∈[0,4], and the larger the value, the more severe the environmental vibration fluctuation;

[0033] Collect historical construction data, extract the dimensionless personnel abnormal aggregation index Ea, density equipment abnormality index Eb, and vibration amplitude environmental mutation index Ec, obtain the mean μRe and standard deviation σRe, and use the following formula to calculate the risk entropy threshold E: ; In the formula, k is the adjustment coefficient;

[0034] The risk entropy threshold E is compared and evaluated with each risk entropy value Re one by one. The specific evaluation contents are as follows:

[0035] When the risk entropy value Re is less than the risk entropy threshold E, it indicates that the security risk status in the current monitoring sub-area is normal and the current monitoring strategy remains unchanged;

[0036] When the risk entropy value Re ≥ the risk entropy threshold E, it means that the security risk status in the current monitoring sub-area is abnormal, and the current monitoring sub-area is marked red.

[0037] Preferably, the auxiliary monitoring area concentrated risk entropy value Re ≥ risk entropy threshold E of the abnormal monitoring sub-region is recorded as i, and i∈{1, 2, 3, ..., n};

[0038] When there are two or more adjacent monitoring sub-areas marked in red around the monitoring sub-area i, it means that regional risk diffusion is formed. At this time, the local average risk entropy value is calculated. , and judge whether it exceeds the local threshold E i : Where N(i) represents the number of monitoring sub-regions adjacent to monitoring sub-region i; j represents the index of the adjacent monitoring sub-region; Re j is the risk entropy value of the adjacent monitoring sub-region j; is the local average risk entropy value of the monitoring sub-area i; the same calculation method as the risk entropy threshold E is used to determine The local threshold is E i ;

[0039] When the local average risk entropy Less than the local threshold E i , indicating that the risk of the current monitoring sub-area i spreading to the adjacent monitoring sub-area is small;

[0040] When the local average risk entropy Greater than or equal to the local threshold E i , indicating that the risk of the current monitoring sub-region i spreading to the adjacent monitoring sub-region is high, and the risk diffusion coefficient Rd of all monitoring sub-regions marked in red is calculated.

[0041] Preferably, the risk calculation module further includes a diffusion assessment unit;

[0042] The abnormal monitoring sub-region with the concentrated risk entropy value Re≥ risk entropy threshold E in the core monitoring area is recorded as i0, and i0∈{1, 2, 3, ..., n};

[0043] The diffusion assessment unit is used to calculate the risk diffusion coefficient Rd and evaluate the risk diffusion trend to the core monitoring area set; Where max represents the maximum risk entropy increment between the monitoring sub-area i and the adjacent monitoring sub-area i0 in the core monitoring area set, which is used to measure the strongest local risk diffusion trend; the adjacent monitoring sub-area set is Contains all monitoring sub-areas directly connected to the monitoring sub-area i0, used to find the maximum risk diffusion direction; is the index of the adjacent monitoring sub-area; Re i represents the risk entropy value of monitoring sub-region i, The risk entropy value of the monitoring sub-area representing the abnormal concentration in the core monitoring area; the historical risk diffusion data of each monitoring sub-area during the construction process are collected, the mean and standard deviation of the risk diffusion coefficient Rd are calculated, the risk diffusion threshold D is obtained, and a comparative evaluation is performed with the risk diffusion coefficient Rd. The specific evaluation content is as follows: When the risk diffusion coefficient Rd ≥ the risk diffusion threshold D: it indicates that the risk diffusion trend of the current monitoring sub-area is abnormal and will soon affect the core monitoring area set; adjust the monitoring frequency and data collection frequency of the current monitoring sub-area and its adjacent monitoring sub-areas; and at the same time, additional intervention measures need to be taken, including strengthening construction safety management or optimizing personnel scheduling;

[0044] When the risk diffusion coefficient Rd is less than the risk diffusion threshold D:

[0045] This indicates that the risk diffusion trend in the current monitoring sub-area is normal and the impact on the core monitoring area set is negligible. At this time, the normal monitoring frequency is maintained and the risk change trend in the current monitoring sub-area is continued to be observed. The intelligent early warning module includes a risk intervention decision unit.

[0046] The risk intervention decision unit is used to calculate and evaluate the risk intervention priority Rp. The construction activity intensity Pa and communication quality Pb ​​in the data acquisition module are extracted and dimensionlessly processed with the risk diffusion coefficient Rd. The risk intervention priority Rp is calculated using the following formula: Among them, −1≤Rp<1;

[0047] The numerical value of the risk intervention priority Rp is evaluated as follows:

[0048] When the risk intervention priority Rp is less than or equal to 0 and is closer to -1, it means that the risk diffusion in the current monitoring sub-area is weaker, the construction activities are less, or the communication quality is better; when the risk intervention priority Rp is greater than 0 and is closer to 1, it means that the risk diffusion in the current monitoring sub-area is faster, the construction activity intensity is higher, and the communication quality is more unstable, and intervention is performed at this time.

[0049] Preferably, the intelligent early warning module further includes a dynamic optimization execution unit;

[0050] The dynamic optimization execution unit applies the DRL deep reinforcement learning algorithm to continuously optimize the graded early warning strategy; including: adjusting the sampling frequency of the core monitoring area set to 3 to 5 times that of the auxiliary monitoring area set according to the real-time risk situation; automatically calculating and allocating computing resources according to the risk level, and adjusting server resource allocation; building a risk coupling model based on complex network theory, analyzing the risk transmission path between regions, predicting chain reaction effects, and optimizing the global early warning response plan.

[0051] Compared with existing technologies, the present invention has the following beneficial effects: it dynamically adjusts monitoring frequency and computing resource allocation based on the real-time risk situation at the construction site, optimizes the hierarchical early warning strategy by introducing a deep reinforcement learning (DRL) algorithm, and effectively solves the problem of data noise interference; it collects multi-source safety monitoring data and performs compression optimization, combined with a federated learning framework, to improve the accuracy of the regional risk assessment model and effectively suppress prediction bias caused by data noise; the present invention also constructs a risk coupling model based on complex network theory, analyzes inter-regional risk transmission paths and chain reaction effects, and optimizes the global early warning response plan, thereby improving the system's adaptability to complex construction environments, significantly improving real-time computing capabilities, and overcoming the problem of low efficiency in high-concurrency processing;

[0052] The present invention ensures continuous training and optimization of the model through federated learning and dynamic weight update mechanisms, enhancing the model's generalization ability. Under dynamic environmental changes at the construction site, the system can adjust its strategy in a timely manner and maintain high adaptability, thus solving the problems of weak model generalization ability and poor adaptability to dynamic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the overall system framework of the present invention. DETAILED DESCRIPTION

[0054] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0055] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0056] Example 1: Please refer to Figure 1 , the present invention provides a technical solution: a construction site safety information supervision system based on the Internet of Things, including a regional monitoring module, a data acquisition module, a risk calculation module and an intelligent early warning module;

[0057] Regional monitoring module: used to provide different regional settings for the construction preparation stage and construction stage of the target construction site;

[0058] During the construction preparation phase, the target construction site is divided into multiple monitoring sub-areas;

[0059] During the construction phase, the construction plan and operation type of each monitoring sub-area are obtained and analyzed, and based on the analysis results, these monitoring sub-areas are divided into a core monitoring area set and an auxiliary monitoring area set;

[0060] Data acquisition module: used to collect multi-source safety monitoring data of each monitoring sub-area during the construction phase;

[0061] Based on the topological relationship of the monitoring sub-areas, the collection frequency of the core monitoring area set and the auxiliary monitoring area set is adjusted, and the communication quality is monitored at the same time;

[0062] Risk calculation module: used to build a regional risk assessment model using federated learning during the construction phase. This regional risk assessment model is used to calculate the risk entropy value Re of each monitoring sub-area within the auxiliary monitoring area set. The risk entropy value Re is used to assess whether the corresponding monitoring sub-area is abnormal;

[0063] Calculate the risk diffusion coefficient Rd of the abnormal monitoring sub-area. The risk diffusion coefficient Rd is used to evaluate the risk diffusion trend to the core monitoring area set;

[0064] Intelligent early warning module: used to calculate and evaluate the risk intervention priority Rp based on the risk diffusion coefficient Rd; then use reinforcement learning to optimize the hierarchical early warning strategy, dynamically adjust the sampling frequency and resource allocation of the core monitoring area set and the auxiliary monitoring area set, and at the same time build an inter-regional risk coupling model to optimize the global early warning strategy.

[0065] In this embodiment, based on industry standards and experimental data, actual engineering verification data and academic research results, the regional monitoring module dynamically divides the core monitoring area set and the auxiliary monitoring area set, and deploys sensor nodes in each monitoring sub-area, and at the same time establishes the topological relationship between these sensor nodes; it realizes accurate coverage of the construction area and key risk monitoring; the data acquisition module uses compressed sensing technology with a compression rate of ≥60% to collect multi-source safety monitoring data in real time, including equipment status, personnel positioning, environmental parameters and video stream data; the collection frequency is adaptively optimized based on the topological relationship; the risk calculation module constructs a risk assessment model through a federated learning framework, where the model aggregation cycle is ≤30min, and calculates the risk entropy value Re and diffusion coefficient Rd to predict the risk transmission trend; the intelligent early warning module optimizes the hierarchical early warning strategy based on the DRL deep reinforcement learning algorithm, dynamically adjusts resource allocation and constructs a risk coupling model to achieve rapid response to global early warnings and improve the overall accident early warning accuracy.

[0066] Example 2

[0067] The regional monitoring module includes a sub-region division unit;

[0068] The sub-area division unit is used to obtain the total area of ​​the construction site during the construction preparation stage, and to divide the site in combination with the construction drawings. The construction site is divided into independent monitoring sub-areas and assigned unique identifiers, including {1, 2, 3, ..., n}, where n represents the total number of monitoring sub-areas and n is the index mark of the nth monitoring sub-area. The center point coordinates and boundary information of each monitoring sub-area are also recorded.

[0069] The regional monitoring module also includes a monitoring area classification unit;

[0070] The monitoring area classification unit is used to divide all monitoring sub-areas into a core monitoring area set and an auxiliary monitoring area set. Based on the construction plan and operation type, the personnel density Za, structural stability Zb, and equipment impact area Zc of each monitoring sub-area are obtained in real time. After dimensionless processing, the comprehensive monitoring priority value Zhj is calculated using the following formula: In the formula, the value range of personnel density Za, structural stability Zb and equipment impact area Zc after dimensionless processing is [0,1], and the larger the comprehensive monitoring priority value Zhj, the higher the risk; m is the correction coefficient of the equipment impact area Zc; according to the comprehensive monitoring priority value Zhj of each monitoring sub-area, all monitoring sub-areas are sorted in descending order from high to low, and the monitoring sub-areas in the top 50% are divided into the core monitoring area set, and the rest are divided into the auxiliary monitoring area set. In this embodiment, the regional monitoring module realizes the equal-area grid division of the construction site through the sub-area division unit, establishing a structured foundation for accurate monitoring; the monitoring area classification unit is based on the three core parameters of personnel density Za, structural stability Zb and equipment impact area Zc collected in real time, and after dimensionless processing, it is classified by the comprehensive priority value. The calculation formula realizes dynamic risk assessment, intelligently dividing the top 50% high-risk sub-areas into core monitoring areas and the rest into auxiliary monitoring areas. Among them, the personnel density Za is used to reflect the risk of personnel gathering, the structural stability Zb is used to characterize the safety status of the support system, and the equipment influence area Zc is used to indicate the dangerous range of mechanical operation.

[0071] This design not only enables automatic identification of risk areas (accuracy ≥ 90%), but also significantly improves the efficiency of monitoring resource allocation through parametric grading (resource utilization increased by 35%), providing a precise spatial positioning basis for subsequent data acquisition modules. At the same time, Zhj's continuous update mechanism (cycle ≤ 15 minutes) ensures that the system can dynamically adapt to changes in the construction environment (response delay ≤ 8 seconds), and overall builds a complete mapping relationship from physical space to digital twin.

[0072] Example 3

[0073] The data acquisition module includes a multi-source data acquisition unit and a dynamic control unit;

[0074] The multi-source safety monitoring data collected by the multi-source data acquisition unit includes equipment status, personnel location, environmental parameters, and video stream data. A compressed sensing algorithm is used to compress and optimize the data. A hierarchical storage structure is then established based on the data types of equipment status, personnel location, environmental parameters, and video stream data. Equipment status and personnel location data are stored in a high-frequency access database, while environmental parameters and video streams are stored in a time series database. A time tag is then added to each piece of data, using a UTC timestamp accurate to milliseconds, to extract the construction activity intensity Pa within the monitored sub-area.

[0075] The dynamic control unit is used to adjust the collection frequency of the core monitoring area set and the auxiliary monitoring area set based on the topological relationship of the monitoring sub-area, and monitor the communication quality to optimize the transmission strategy, specifically including starting the local data cache and frequency reduction transmission mechanism when the communication quality is lower than the preset threshold; and extracting the communication quality Pb ​​in the monitoring sub-area at the same time. In this embodiment, the data acquisition module realizes the full-factor collection of equipment status, personnel positioning, environmental parameters and video stream data through the multi-source data acquisition unit, optimizes the transmission efficiency by using the compressed sensing algorithm with a compression ratio of ≥65%, and establishes a hierarchical storage architecture to achieve accurate data traceability through spatiotemporal tags; the dynamic control unit dynamically adjusts the collection frequency based on the topological relationship, and starts the local cache with a storage capacity of ≥72 hours and frequency reduction transmission with a bandwidth reduction of 40% when the communication quality Pb ​​does not meet the standard; at the same time, the time tag is used to bind the monitoring sub-area ID and three-dimensional coordinates (x, y, z), and the storage space is optimized through the data compression algorithm; the construction activity intensity Pa is obtained by analyzing the video stream data including the movement frequency of personnel and machinery to obtain activity images. The device status data, including the number of mechanical starts and stops, is quantified to obtain the device operation status value, which is obtained by weighted calculation using active pixels and device operation status values. The specific formula is: Pa = q1 × video activity index + q2 × device operation index, where q1 and q2 are the weight coefficients of active pixels and device operation status values, and q1 + q2 = 1. The communication quality Pb ​​is obtained by comprehensive calculation of three indicators: real-time monitoring of signal strength RSSI, packet loss rate, and transmission delay. The specific formula is Pb = w1 × (1-packet loss rate) + w2 × RSSI normalized value + w3 × delay normalized value, where w1, w2, and w3 are the weight coefficients of signal strength RSSI, packet loss rate, and transmission delay, and w1 + w2 + w3 = 1.

[0076] When the communication quality falls below the preset threshold, the local data cache and down-conversion transmission mechanisms are activated as follows:

[0077] 1) Local cache uses a ring buffer for storage;

[0078] 2) Reduce the frame rate of the video stream and close non-critical data channels;

[0079] 3) Resource reallocation prioritizes core monitoring areas while suspending non-essential collection in auxiliary areas;

[0080] The above mechanism uses an adaptive PID controller to evaluate the channel status every 30 seconds and dynamically adjust the reduction ratio to ensure the integrity of key data and control transmission delay within the preset range before communication is restored. The specific preset threshold and preset range are determined by the user according to the situation.

[0081] This module not only enables efficient fusion processing of multi-source heterogeneous data, but also ensures the stable operation of the system in complex construction environments through intelligent control mechanisms. It provides real-time, complete, and low-redundancy data support for upper-level risk calculations, significantly improving the overall reliability of the system. In addition, the above data is based on national standards and industry specifications, laboratory test data, and engineering evidence.

[0082] Among them, starting the local data cache and frequency reduction transmission mechanism is to prioritize the data integrity of the core monitoring area set.

[0083] Example 4

[0084] The risk calculation module includes a risk modeling unit;

[0085] The risk modeling unit is used to build a regional risk assessment model using a federated learning framework. The model is trained through distributed node collaboration and integrates multi-source security monitoring data from all monitored sub-regions.

[0086] The local model parameters of each monitoring sub-area are then privacy-protected using a homomorphic encryption algorithm. The encrypted parameters are then aggregated and updated on the central server using a federated averaging algorithm. A dynamic weight update mechanism is also implemented to automatically adjust the model update cycle based on the frequency of environmental changes, including equipment movement and work surface expansion, at the construction site.

[0087] The risk calculation module also includes an entropy value calculation unit; the entropy value calculation unit is used to preferentially calculate the risk entropy value Re of each monitoring sub-area in the auxiliary monitoring area set, and use the regional risk assessment model to perform real-time analysis on multi-source safety monitoring data to obtain abnormal personnel aggregation Ea, density equipment abnormality Eb and vibration amplitude environment mutation index Ec. After dimensionless processing, the risk entropy value Re is calculated using the following formula: In the formula, the value range of abnormal personnel gathering Ea is Ea∈[0,5], and the larger the value, the more abnormal the personnel gathering; the value range of density equipment abnormality Eb is Eb∈[0,3], and the larger the value, the more serious the abnormal equipment distribution; the value range of vibration amplitude environment mutation index Ec is Ec∈[0,4], and the larger the value, the more severe the environmental vibration fluctuation;

[0088] Collect historical construction data, extract the dimensionless personnel abnormal aggregation index Ea, density equipment abnormality index Eb, and vibration amplitude environmental mutation index Ec, obtain the mean μRe and standard deviation σRe, and use the following formula to calculate the risk entropy threshold E: Where k is the adjustment coefficient;

[0089] The risk entropy threshold E is compared and evaluated with each risk entropy value Re one by one. The specific evaluation contents are as follows:

[0090] When the risk entropy value Re is less than the risk entropy threshold E, it indicates that the security risk status in the current monitoring sub-area is normal and the current monitoring strategy remains unchanged;

[0091] When the risk entropy value Re ≥ the risk entropy threshold E, it means that the security risk status in the current monitoring sub-area is abnormal, and the current monitoring sub-area is marked red.

[0092] The abnormal monitoring sub-region with the concentrated risk entropy value Re≥ the risk entropy threshold E in the auxiliary monitoring area is recorded as i, and i∈{1, 2, 3, ..., n}; when there are two or more adjacent monitoring sub-regions marked in red around the monitoring sub-region i, it means that regional risk diffusion is formed, and the local average risk entropy value is calculated at this time , and judge whether it exceeds the local threshold E i : Where N(i) represents the number of monitoring sub-regions adjacent to monitoring sub-region i; j represents the index of the adjacent monitoring sub-region; Re j is the risk entropy value of the adjacent monitoring sub-region j; is the local average risk entropy value of the monitoring sub-area i; the same calculation method as the risk entropy threshold E is used to determine The local threshold is E i ;

[0093] When the local average risk entropy Less than the local threshold E i , indicating that the risk of the current monitoring sub-area i spreading to the adjacent monitoring sub-area is small;

[0094] When the local average risk entropy Greater than or equal to the local threshold E i , indicating that the risk of the current monitoring sub-region i spreading to the adjacent monitoring sub-region is high, and the risk diffusion coefficient Rd of all monitoring sub-regions marked in red is calculated.

[0095] The risk calculation module also includes a diffusion assessment unit; the abnormal monitoring sub-area where the risk entropy value Re≥ the risk entropy threshold E in the core monitoring area is recorded as i0, and i0∈{1, 2, 3, ..., n};

[0096] The diffusion assessment unit is used to calculate the risk diffusion coefficient Rd and evaluate the risk diffusion trend to the core monitoring area set; Where max represents the maximum risk entropy increment between the monitoring sub-area i and the adjacent monitoring sub-area i0 in the core monitoring area set, which is used to measure the strongest local risk diffusion trend; the adjacent monitoring sub-area set is Contains all monitoring sub-areas directly connected to the monitoring sub-area i0, used to find the maximum risk diffusion direction; is the index of the adjacent monitoring sub-area; Re irepresents the risk entropy value of monitoring sub-region i, The risk entropy value of the monitoring sub-area with concentrated abnormalities in the core monitoring area is shown;

[0097] Collect historical risk diffusion data of each monitoring sub-area during the construction process, calculate the mean and standard deviation of the risk diffusion coefficient Rd, obtain the risk diffusion threshold D, and compare and evaluate it with the risk diffusion coefficient Rd. The specific evaluation contents are as follows:

[0098] When the risk diffusion coefficient Rd ≥ the risk diffusion threshold D, it indicates that the risk diffusion trend in the current monitoring sub-area is abnormal and will soon affect the core monitoring area set. The monitoring frequency and data collection frequency of the current monitoring sub-area and its adjacent monitoring sub-areas should be adjusted. At the same time, additional intervention measures may be required, including strengthening construction safety management or optimizing personnel scheduling.

[0099] When the risk diffusion coefficient Rd is less than the risk diffusion threshold D:

[0100] This indicates that the risk diffusion trend in the current monitoring sub-area is normal, and the impact on the core monitoring area set is negligible. At this time, maintain the normal monitoring frequency and continue to observe the risk change trend in the current monitoring sub-area.

[0101] The intelligent early warning module includes a risk intervention decision-making unit; the risk intervention decision-making unit is used to calculate and evaluate the risk intervention priority Rp. The construction activity intensity Pa and communication quality Pb ​​from the data acquisition module are extracted and dimensionlessly processed with the risk diffusion coefficient Rd. The risk intervention priority Rp is calculated using the following formula: Among them, −1≤Rp<1;

[0102] The numerical value of the risk intervention priority Rp is evaluated as follows:

[0103] When the risk intervention priority Rp is less than or equal to 0 and is closer to -1, it means that the risk diffusion in the current monitoring sub-area is weaker, the construction activities are less, or the communication quality is better;

[0104] When the risk intervention priority Rp is greater than 0 and approaches 1, it means that the risk in the current monitoring sub-area spreads faster, the construction activity intensity is higher, and the communication quality is more unstable, and intervention is performed at this time.

[0105] The intelligent early warning module also includes a dynamic optimization execution unit;

[0106] The dynamic optimization execution unit applies the DRL deep reinforcement learning algorithm to continuously optimize the hierarchical warning strategy; including:

[0107] Based on the real-time risk situation, the sampling frequency of the core monitoring area set is adjusted to 3 to 5 times that of the auxiliary monitoring area set; computing resources are automatically calculated and allocated according to the risk level, and server resource allocation is adjusted; a risk coupling model based on complex network theory is constructed to analyze the risk transmission path between regions, predict chain reaction effects, and optimize the global early warning response plan.

[0108] In this embodiment, the risk calculation module uses the federated learning framework and homomorphic encryption technology to build a dynamic risk assessment model through the risk modeling unit to achieve secure collaborative computing of multi-source data; the entropy value calculation unit quantifies the regional risk through the formula of the risk entropy value Re based on the three core parameters of abnormal personnel aggregation Ea, equipment abnormality Eb and environmental mutation Ec, and dynamically calculates the risk threshold E based on historical data to evaluate the risk entropy value Re; calculates the local average risk entropy value In the process of regional risk diffusion, intelligent identification and quantitative assessment are realized. Its core significance lies in dynamically calculating the local average risk entropy value. and with the local threshold E i By comparison, we can accurately determine the risk transmission trend when the local average risk entropy value is Greater than or equal to the local threshold E i When a risk is detected, the diffusion coefficient Rd calculation is triggered, forming a three-level early warning mechanism of "single point red mark → regional diffusion → global response". This design significantly improves the system's foresight of chain risks, reduces the false alarm rate through topological association analysis, optimizes resource allocation, and increases the monitoring frequency only in the actual diffusion area, thereby speeding up the emergency response in the core area. Ultimately, it achieves full-dimensional security management and control from isolated risk points to regional linkage, effectively solving the problem of traditional monitoring systems' insensitivity to the transmission of hidden risks.

[0109] Furthermore, the local threshold E i It is derived from the risk entropy threshold E, and the calculation logic is exactly the same, but the difference is that the local threshold E i As a local judgment standard, it is used to judge whether the current area has a risk diffusion possibility higher than the normal range; and the local threshold E i The calculation formula is: , The historical risk entropy value Re of the adjacent sub-region set N(i) j The mean of is the historical risk entropy value Re of the adjacent sub-region set N(i) jThe standard deviation of the risk is given by [ ], and k is the adjustment coefficient, typically set to 1 ≤ k ≤ 3, based on engineering experience. The diffusion assessment unit uses the risk diffusion coefficient Rd to locate the direction of maximum risk transmission and compares it with the diffusion threshold D to determine whether it affects the core area. The intelligent early warning module calculates the intervention priority Rp based on the construction activity intensity Pa, communication quality Pb, and Rd. Based on the assessment of intervention priority Rp, it dynamically optimizes the monitoring strategy and constructs a complex network model to analyze the risk transmission path. Ultimately, it achieves a closed-loop management chain from risk identification and assessment to early warning, enabling the system to significantly improve emergency response efficiency while maintaining data privacy. In the formula for risk intervention priority Rp, Rd × Pa means that if a region has a high risk diffusion trend and high construction activity intensity, it is more likely to cause accidents, so the risk intervention priority Rp needs to be increased. The significance of −Pb is that if the communication quality Pb ​​in the region is high, it indicates that the monitoring data transmission is stable and the system can effectively monitor and respond, so the intervention priority can be appropriately lowered.

[0110] It should be noted that all calculation formulas in this application document utilize, including but not limited to, regression analysis within machine learning algorithms to deeply analyze the collected parameters and identify their natural trends and interrelationships. Professional software, such as Python's Scikit-learn library or the R language, is used to automatically generate mathematical models that match the data. Model performance is then objectively evaluated through methods such as cross-validation, combined with continuous feedback and optimization to ensure that the created formulas truly reflect the inherent laws of the data, thereby guaranteeing their validity and accuracy, and ensuring that the calculation process complies with the constraints of natural laws rather than being based on artificially set rules.

[0111] The technical solution of the present invention, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk, or optical disk, and includes instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention. The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device, or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, device, or apparatus), or for use in conjunction with such instruction execution systems, devices, or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by an instruction execution system, device, or apparatus, or in conjunction with such instruction execution systems, devices, or apparatuses.

[0112] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A construction site safety information monitoring system based on the Internet of Things, characterized by: It includes regional monitoring module, data acquisition module, risk calculation module and intelligent early warning module; Regional monitoring module: used to provide different regional settings for the construction preparation stage and construction stage of the target construction site; During the construction preparation phase, the target construction site is divided into multiple monitoring sub-areas; During the construction phase, the construction plan and operation type of each monitoring sub-area are obtained and analyzed, and based on the analysis results, these monitoring sub-areas are divided into a core monitoring area set and an auxiliary monitoring area set; Data acquisition module: used to collect multi-source safety monitoring data of each monitoring sub-area during the construction phase; Based on the topological relationship of the monitoring sub-areas, the collection frequency of the core monitoring area set and the auxiliary monitoring area set is adjusted, and the communication quality is monitored at the same time; Risk calculation module: used to build a regional risk assessment model using federated learning during the construction phase. This regional risk assessment model is used to calculate the risk entropy value Re of each monitoring sub-area within the auxiliary monitoring area set. The risk entropy value Re is used to assess whether the corresponding monitoring sub-area is abnormal; Calculate the risk diffusion coefficient Rd of the abnormal monitoring sub-area. The risk diffusion coefficient Rd is used to evaluate the risk diffusion trend to the core monitoring area set; Intelligent early warning module: includes a risk intervention decision unit, which is used to calculate and evaluate the risk intervention priority Rp based on the risk diffusion coefficient Rd. The specific calculation and evaluation process of the risk intervention priority Rp is as follows: The construction activity intensity Pa and communication quality Pb ​​in the data acquisition module are extracted and dimensionlessly processed with the risk diffusion coefficient Rd. The risk intervention priority Rp is calculated using the following formula: ; Among them, −1≤Rp<1; The numerical value of the risk intervention priority Rp is evaluated as follows: When the risk intervention priority Rp is less than or equal to 0 and approaches -1, it indicates that the risk diffusion in the current monitoring sub-area is weaker, the construction activity is less, or the communication quality is better. When the risk intervention priority Rp is greater than 0 and approaches 1, it indicates that the risk diffusion in the current monitoring sub-area is faster, the construction activity intensity is higher, and the communication quality is more unstable, and intervention is required at this time. Then, reinforcement learning is used to optimize the hierarchical early warning strategy, dynamically adjust the sampling frequency and resource allocation of the core monitoring area set and the auxiliary monitoring area set, and at the same time build an inter-regional risk coupling model to optimize the global early warning strategy.

2. The construction site safety information monitoring system based on the Internet of Things according to claim 1 is characterized by: The regional monitoring module includes a sub-region division unit; The sub-area division unit is used to obtain the total area of ​​the construction site during the construction preparation stage, and to divide the site in combination with the construction drawings. The construction site is divided into independent monitoring sub-areas and assigned unique identifiers, including {1, 2, 3, ..., n}, where n represents the total number of monitoring sub-areas and n is the index mark of the nth monitoring sub-area. The center point coordinates and boundary information of each monitoring sub-area are also recorded.

3. The construction site safety information monitoring system based on the Internet of Things according to claim 2 is characterized by: The regional monitoring module also includes a monitoring area classification unit; The monitoring area classification unit is used to divide all monitoring sub-areas into a core monitoring area set and an auxiliary monitoring area set. Based on the construction plan and operation type, the personnel density Za, structural stability Zb, and equipment impact area Zc of each monitoring sub-area are obtained in real time. After dimensionless processing, the comprehensive monitoring priority value Zhj is calculated using the following formula: Wherein, the dimensionless values ​​of personnel density Za, structural stability Zb, and equipment impact area Zc range from [0 to 1], and a larger comprehensive monitoring priority value Zhj indicates a higher risk. m is the correction coefficient for the equipment impact area Zc. Based on the comprehensive monitoring priority value Zhj of each monitoring sub-area, all monitoring sub-areas are sorted in descending order from high to low. The top 50% of monitoring sub-areas are divided into the core monitoring area set, and the rest are divided into the auxiliary monitoring area set.

4. The construction site safety information monitoring system based on the Internet of Things according to claim 3 is characterized by: The data acquisition module includes a multi-source data acquisition unit and a dynamic control unit; The multi-source safety monitoring data collected by the multi-source data acquisition unit includes equipment status, personnel positioning, environmental parameters and video stream data, and the compressed sensing algorithm is used to compress and optimize the data. Then, based on the data types including equipment status, personnel positioning, environmental parameters and video stream data, a hierarchical storage structure is established. The equipment status and personnel positioning data are stored in a high-frequency access database, and the environmental parameters and video streams are stored in a time series database; secondly, a time tag is attached to each data and the UTC timestamp is used to be accurate to milliseconds to extract the construction activity intensity Pa in the monitoring sub-area; the dynamic control unit is used to adjust the collection frequency of the core monitoring area set and the auxiliary monitoring area set based on the topological relationship of the monitoring sub-area, and monitor the communication quality to optimize the transmission strategy, specifically including starting the local data cache and frequency reduction transmission mechanism when the communication quality is lower than the preset threshold; at the same time, the communication quality Pb ​​in the monitoring sub-area is extracted.

5. The construction site safety information monitoring system based on the Internet of Things according to claim 4 is characterized in that: The risk calculation module includes a risk modeling unit. The risk modeling unit is used to build a regional risk assessment model using a federated learning framework. The model is trained through distributed node collaboration and integrates multi-source security monitoring data from all monitoring sub-regions. The local model parameters of each monitoring sub-area are then privacy-protected using a homomorphic encryption algorithm. The encrypted parameters are then aggregated and updated on the central server using a federated averaging algorithm. A dynamic weight update mechanism is also implemented to automatically adjust the model update cycle based on the frequency of environmental changes, including equipment movement and work surface expansion, at the construction site.

6. The construction site safety information monitoring system based on the Internet of Things according to claim 5 is characterized by: The risk calculation module also includes an entropy value calculation unit; The entropy value calculation unit is used to preferentially calculate the risk entropy value Re of each monitoring sub-area in the auxiliary monitoring area set, and use the regional risk assessment model to perform real-time analysis on the multi-source safety monitoring data to obtain the abnormal gathering of personnel Ea, density equipment abnormality Eb and vibration amplitude environmental mutation index Ec. After dimensionless processing, the risk entropy value Re is calculated using the following formula: In the formula, the value range of abnormal personnel gathering Ea is Ea∈[0,5], and the larger the value, the more abnormal the personnel gathering; the value range of density equipment abnormality Eb is Eb∈[0,3], and the larger the value, the more serious the abnormal equipment distribution; the value range of vibration amplitude environment mutation index Ec is Ec∈[0,4], and the larger the value, the more severe the environmental vibration fluctuation; Collect historical construction data, extract the dimensionless personnel abnormal aggregation index Ea, density equipment abnormality index Eb, and vibration amplitude environmental mutation index Ec, obtain the mean μRe and standard deviation σRe, and use the following formula to calculate the risk entropy threshold E: Where k is the adjustment coefficient; The risk entropy threshold E is compared and evaluated with each risk entropy value Re one by one. The specific evaluation content is as follows: when the risk entropy value Re is less than the risk entropy threshold E, it means that the security risk status in the current monitoring sub-area is normal and the current monitoring strategy remains unchanged; When the risk entropy value Re ≥ the risk entropy threshold E, it means that the security risk status in the current monitoring sub-area is abnormal, and the current monitoring sub-area is marked red.

7. The construction site safety information monitoring system based on the Internet of Things according to claim 6, characterized in that: The abnormal monitoring sub-region with the concentrated risk entropy value Re≥ the risk entropy threshold E in the auxiliary monitoring area is recorded as i, and i∈{1, 2, 3, ..., n}; when there are two or more adjacent monitoring sub-regions marked in red around the monitoring sub-region i, it means that regional risk diffusion is formed, and the local average risk entropy value is calculated at this time , and judge whether it exceeds the local threshold E i : Where N(i) represents the number of monitoring sub-regions adjacent to monitoring sub-region i; j represents the index of the adjacent monitoring sub-region; Re j is the risk entropy value of the adjacent monitoring sub-region j; is the local average risk entropy value of monitoring sub-region i; Use the same calculation method as the risk entropy threshold E to determine The local threshold is E i ; When the local average risk entropy Less than the local threshold E i , indicating that the risk of the current monitoring sub-area i spreading to the adjacent monitoring sub-area is small; When the local average risk entropy Greater than or equal to the local threshold E i , indicating that the risk of the current monitoring sub-region i spreading to the adjacent monitoring sub-region is high, and the risk diffusion coefficient Rd of all monitoring sub-regions marked in red is calculated.

8. The construction site safety information monitoring system based on the Internet of Things according to claim 7, characterized in that: The risk calculation module also includes a diffusion assessment unit; The abnormal monitoring sub-region with the concentrated risk entropy value Re≥ risk entropy threshold E in the core monitoring area is recorded as i0, and i0∈{1, 2, 3, ..., n}; The diffusion assessment unit is used to calculate the risk diffusion coefficient Rd and evaluate the risk diffusion trend to the core monitoring area set; Where max represents the maximum risk entropy increment between the monitoring sub-area i and the adjacent monitoring sub-area i0 in the core monitoring area set, which is used to measure the strongest local risk diffusion trend; the adjacent monitoring sub-area set is Contains all monitoring sub-areas directly connected to the monitoring sub-area i0, used to find the maximum risk diffusion direction; is the index of the adjacent monitoring sub-area; Re i represents the risk entropy value of monitoring sub-region i, The risk entropy value of the monitoring sub-region representing the concentrated anomaly in the core monitoring area; Collect historical risk diffusion data of each monitoring sub-area during the construction process, calculate the mean and standard deviation of the risk diffusion coefficient Rd, obtain the risk diffusion threshold D, and compare and evaluate it with the risk diffusion coefficient Rd. The specific evaluation content is as follows: When the risk diffusion coefficient Rd ≥ the risk diffusion threshold D: This indicates that the risk diffusion trend in the current monitoring sub-area is abnormal and will soon affect the core monitoring area set. The monitoring frequency and data collection frequency of the current monitoring sub-area and its adjacent monitoring sub-areas should be adjusted. At the same time, additional intervention measures may be required, including strengthening construction safety management or optimizing personnel scheduling. When the risk diffusion coefficient Rd is less than the risk diffusion threshold D, it indicates that the risk diffusion trend in the current monitoring sub-area is normal and the impact on the core monitoring area set is negligible. At this time, the normal monitoring frequency is maintained and the risk change trend in the current monitoring sub-area continues to be observed.

9. The construction site safety information monitoring system based on the Internet of Things according to claim 1, characterized in that: The intelligent early warning module also includes a dynamic optimization execution unit; The dynamic optimization execution unit applies the DRL deep reinforcement learning algorithm to continuously optimize the hierarchical warning strategy; including: Adjust the sampling frequency of the core monitoring area set to 3 to 5 times that of the auxiliary monitoring area set based on the real-time risk situation; Automatically calculate and allocate computing resources according to risk levels, and adjust server resource allocation; build a risk coupling model based on complex network theory, analyze risk transmission paths between regions, predict chain reaction effects, and optimize global early warning response plans.

Images