Intelligent monitoring system and method for construction site personnel behavior safety based on deep learning

Abstract

The present invention relates to the technical field of construction site safety monitoring, and specifically to an intelligent monitoring system and method for the behavior safety of construction site personnel based on deep learning. The system divides areas into standard areas according to construction processes and operation types, collects various action data of construction site personnel based on sensors, analyzes the actions of construction site personnel in corresponding behavior monitoring areas based on deep learning technology, and accurately analyzes whether construction site personnel have intuitive dangerous behaviors in the corresponding areas. Unlike the traditional monitoring method that relies on camera components, this model is no longer limited by the layout of camera components in the construction area, and effectively avoids the problem of blind spots in monitoring due to limited monitoring range. After the construction site personnel currently do not have intuitive dangerous behaviors, further behavioral pre-warning analysis is performed on the construction site personnel, and potential pre-operation hidden dangers are deeply explored to determine whether the current personnel have pre-risks of dangerous behaviors.

Description

Technical Field

[0001] The present invention relates to the technical field of construction site safety monitoring, and more specifically, to a deep learning-based intelligent monitoring system and method for construction site personnel behavior safety. Background Art

[0002] Currently, the most common method for monitoring the safety of construction site personnel behavior relies on camera monitoring. Multiple cameras are strategically placed throughout the construction area to capture video images of the construction site. Human behavior captured in the video is then analyzed using manual review or image recognition technology to determine whether any dangerous behavior is occurring.

[0003] However, this traditional camera-based monitoring method has many limitations:

[0004] 1. Blind spots in surveillance

[0005] Because construction sites are typically large and complex, achieving comprehensive coverage is difficult due to limitations in camera placement, number, and viewing angle. Human activity in hidden corners, areas obscured by objects, or in camera blind spots often goes undetected, creating potential safety risks.

[0006] 2. Poor accuracy of behavioral analysis

[0007] Relying solely on video image analysis to determine whether human behavior is dangerous is often affected by factors such as lighting variations, image resolution, clothing, and the complexity of movements. Accurately identifying subtle, less noticeable dangerous behaviors (such as minor irregularities in posture or potentially dangerous movements) is difficult. Furthermore, most existing image recognition technologies lack the targeted analytical capabilities required for diverse construction site scenarios and complex construction processes, leading to frequent misjudgments and missed detections, and failing to truly meet the accuracy requirements for behavior monitoring required by construction site safety management.

[0008] 3. Monitoring lacks initiative and foresight

[0009] Traditional approaches often rely on post-event monitoring or late-stage intervention, responding only after dangerous behavior has already occurred or is more pronounced. These approaches lack proactive risk prediction and proactive prevention mechanisms. For example, they fail to proactively analyze impending dangerous behavior, making it impossible to effectively warn and prevent danger before it occurs. This makes it difficult to prevent safety incidents at their root, leaving safety management in a reactive state.

[0010] Based on the above, the present invention proposes a construction site personnel behavior safety intelligent monitoring system and method based on deep learning. Summary of the Invention

[0011] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a construction site personnel behavior safety intelligent monitoring system and method based on deep learning.

[0012] To achieve the above object, the present invention provides the following technical solutions:

[0013] The intelligent monitoring system for construction site personnel behavior safety based on deep learning includes

[0014] The construction site personnel behavior monitoring unit divides the construction area into multiple behavior monitoring areas and regularly generates a personnel movement status log for each construction site personnel;

[0015] The first-level behavior reminder unit generates a log of the movement status of construction site personnel each time, and further determines whether the construction site personnel have any obvious dangerous behavior. If it is determined that the construction site personnel have obvious dangerous behavior, the construction site personnel will be given a first-level reminder;

[0016] The secondary behavior reminder unit, when it is determined that the construction site personnel do not have any intuitive dangerous behavior, further determines whether the construction site personnel need to be reminded of the behavior in advance. When it is determined that the construction site personnel need to be reminded of the behavior in advance, the secondary reminder will be given to the construction site personnel.

[0017] Furthermore, the personnel movement status log includes the construction site personnel ID, movement status feature set, and the behavior monitoring area.

[0018] Furthermore, the process of determining the motion state feature set of the personnel motion state log is as follows: select a construction site worker, regularly collect various action data generated by the construction site worker within a cycle, perform feature extraction on various motion data, extract various motion features, and integrate the extracted motion features into a motion state feature set in the form of a set.

[0019] Furthermore, the process of determining whether a construction site worker has engaged in intuitively dangerous behavior is specifically as follows: obtaining the motion state feature set of the construction site worker's corresponding motion state log, determining the dangerous behavior determination model corresponding to the behavior monitoring area in the worker's motion state log, importing the motion state feature set into the intuitively dangerous behavior determination model, and deriving the intuitively dangerous behavior value of the construction site worker. When the intuitively dangerous behavior value of the construction site worker is higher than the threshold, it is determined that the construction site worker has engaged in intuitively dangerous behavior.

[0020] Furthermore, after issuing a first-level reminder to the construction site personnel, the construction site personnel are simultaneously marked as construction site personnel with dangerous behavior.

[0021] Furthermore, the specific process of determining whether construction site personnel need behavioral pre-reminders is as follows: select a construction site personnel, obtain the behavior monitoring area in the corresponding personnel movement status log of the construction site personnel, determine all construction site personnel with dangerous behaviors that appeared in the behavior monitoring area before, obtain the possible danger pre-reminder value of each dangerous construction site personnel, calculate the sum and average of the possible danger pre-reminder values ​​of all dangerous construction site personnel, calculate the possible danger pre-reminder mean, set the possible danger pre-reminder threshold mean, and when the possible danger pre-reminder mean ≥ the possible danger pre-reminder threshold mean, determine that the construction site personnel needs behavioral pre-reminders.

[0022] Furthermore, the process of obtaining the possible danger pre-value of dangerous construction site personnel is as follows: obtain A personnel motion status logs generated continuously by the construction site personnel before, sort all the personnel motion status logs of the construction site personnel in the order of generation, and mark them with digital serial numbers in sequence, select a dangerous construction site personnel, obtain A personnel motion status logs generated continuously by the dangerous construction site personnel before being marked, sort all the personnel motion status logs of the dangerous construction site personnel in the order of generation, and mark them with digital serial numbers in sequence, determine the state area consistency value of each digital serial number, sum and average all state area consistency values, calculate the average state area consistency value, sort all state area consistency values ​​in order from small to large according to the digital serial number, calculate the absolute difference between the two adjacent state area consistency values ​​after sorting, calculate the state area fluctuation difference value, sum and average all state area fluctuation difference values, calculate the average state area fluctuation difference value, perform ratio calculation between the average state area consistency value and the average state area fluctuation difference value, and calculate the possible danger pre-value of the dangerous construction site personnel.

[0023] Furthermore, the process of obtaining the status area consistency value of a digital serial number is as follows: obtain two personnel motion status logs with the same digital serial number, obtain the area consistency value, obtain the motion state reference degree of the motion state feature set of the two personnel motion status logs, and calculate the status area consistency value of the digital serial number based on the area consistency value and the motion state reference degree.

[0024] Furthermore, the deep learning-based intelligent monitoring method for construction site personnel behavior safety has the following steps:

[0025] Step 1: Division of behavior monitoring areas and regular generation of personnel movement status logs;

[0026] Step 2: First-level reminder determination by construction site personnel;

[0027] Step 3: Secondary reminder determination by construction site personnel.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] The system of the present invention divides the area into standard areas according to the construction process and the type of operation, collects the various action data of the construction site personnel based on the sensors, analyzes the actions of the construction site personnel in the corresponding behavior monitoring area based on the deep learning technology, and accurately analyzes whether the construction site personnel have intuitive dangerous behaviors in the corresponding area. Different from the traditional monitoring method that relies on camera components, this mode is no longer limited by the layout position of the camera components in the construction area, and effectively avoids the problem of blind spots in monitoring due to the limitation of the monitoring range, ensuring the comprehensive coverage of the behavior monitoring of the construction site personnel, that is, avoiding the situation of blind spots in monitoring. After the construction site personnel currently do not have intuitive dangerous behaviors, further behavioral pre-warning analysis is carried out on the construction site personnel. Through the in-depth operation analysis of all dangerous construction site personnel before the construction site personnel in the area, the potential pre-operation hidden dangers are deeply excavated, so as to determine whether the current personnel have the pre-risk of dangerous behavior.

[0030] The method of the present invention constructs a smart safety management closed loop of "comprehensive perception-immediate response-proactive prevention-continuous optimization" through the rational combination of active safety monitoring and pre-emptive safety monitoring, which significantly improves the accuracy, real-time and proactive nature of site personnel behavior monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is the principle diagram of the system of the present invention;

[0032] Figure 2 A flowchart for determining behavior pre-reminders;

[0033] Figure 3 4 is a flowchart of the method of the present invention. DETAILED DESCRIPTION

[0034] Example 1, as Figures 1 to 2 , an intelligent monitoring system for construction site personnel behavior safety based on deep learning, including a construction site personnel behavior monitoring unit, a first-level behavior reminder unit, and a second-level behavior reminder unit.

[0035] The construction site personnel behavior monitoring unit divides the construction area into multiple behavior monitoring areas (standard area divisions are performed according to construction technology and operation types, such as dividing the steel structure welding area, concrete pouring area, and high-altitude operation area into three behavior monitoring areas respectively), and regularly generates personnel movement status logs for each construction site personnel (regular periodic intervals are comprehensively set according to the characteristics of different construction technologies and operation types). The personnel movement status log includes the construction site personnel ID, movement status feature set, and the behavior monitoring area where the (construction site personnel) is located.

[0036] The specific process of determining the motion state feature set of the personnel motion state log is as follows: select a construction site worker, and regularly collect various motion data generated by the construction site worker within a cycle (the motion data includes but is not limited to the following items: head turning data, limb swing data, body motion trajectory data, all motion data are collected through sensors installed on the construction site workers' clothes and safety helmets), perform feature extraction on each motion data, extract various motion features, and integrate the extracted motion features into a motion state feature set in the form of a set.

[0037] The first-level behavioral reminder unit generates a log of the personnel movement status of each construction site worker, and further determines whether the construction site workers have any visually dangerous behaviors. Once it is determined that the construction site workers have any visually dangerous behaviors, the unit will issue a first-level reminder to the construction site workers (the first-level reminder is usually an audible and visual alarm in the behavior monitoring area, which is pushed to the safety officer), and simultaneously mark the construction site workers as dangerous construction site workers.

[0038] The specific process of determining whether a construction site worker has engaged in intuitively dangerous behavior is as follows: obtaining the motion status feature set of the construction site worker's corresponding motion status log, determining the dangerous behavior determination model corresponding to the behavior monitoring area in the worker's motion status log, importing the motion status feature set into the intuitive dangerous behavior determination model, and deriving the intuitive dangerous behavior value of the construction site worker. When the intuitive dangerous behavior value of the construction site worker is higher than a threshold (the threshold is set based on the training results of the dangerous behavior determination model and historical data), it is determined that the construction site worker has engaged in intuitively dangerous behavior.

[0039] For each behavior monitoring area, an intuitive dangerous behavior judgment model based on deep learning needs to be independently constructed (for example, the same action behavior of construction workers may have opposite safety attributes in the steel structure welding area and the concrete pouring area. For example, construction workers keep bending over for a long time in the steel structure welding area (performing welding work). This action behavior is safe and reasonable in the steel structure welding area because the welding platform is equipped with guardrails, the ground is dry and without obstacles, and long-term stillness will not lead to the risk of imbalance of the center of gravity. However, construction workers keep bending over for a long time in the concrete pouring area. This action behavior may be dangerous in the concrete pouring area because there is often concrete slurry residue on the ground in the concrete pouring area. Long-term bending over can easily lead to poor blood circulation in the lower limbs, resulting in a shift in the center of gravity when standing up suddenly, causing slips and falls). Each intuitive dangerous behavior judgment model focuses on the intuitive dangerous behavior of the corresponding behavior monitoring area. In this embodiment, for steel structure welding, Taking the steel structure welding area as an example, the construction process of the intuitive dangerous behavior judgment model is disclosed: a deep learning model is built, and multiple motion state feature sets in the steel structure welding area are collected. The motion state feature sets are used as basic data to train the built deep learning model. In this process, an intuitive dangerous behavior value is assigned to each motion state feature set. The value range of the intuitive dangerous behavior value is set between 1 and 50. The size of the intuitive dangerous behavior value has a clear meaning. The larger the value, the more dangerous the behavior of the construction site personnel in the steel structure welding area corresponding to the motion state feature set. Then, the multiple motion state feature sets are divided into training set, validation set and test set according to a specific ratio. The specific division ratio is determined to be 60%:20%:20%. Stratified sampling is used to ensure that the DBV distribution of each subset is consistent: the cumulative distribution function (CDF) of the DBV of all samples is calculated, and in each quantile interval (such as 1-10, 11-20, ...,41-50) to avoid over-concentration of high-risk samples (DBV>40) in a particular subset. High-risk samples (DBV>30) typically account for less than 10%. A sample-weighted loss function is used to improve the model's sensitivity to minority classes. The deep learning model is repeatedly trained using the training set, and its performance during the training phase is verified using the validation set. Based on the verification results, the model parameters are adjusted in a timely manner to optimize the model structure. Finally, an intuitive dangerous behavior judgment model for steel structure welding areas is established.

[0040] The secondary behavior reminder unit, when it is determined that the construction site personnel do not have any intuitively dangerous behavior, further determines whether the construction site personnel need to be reminded of the behavior in advance. When it is determined that the construction site personnel need to be reminded of the behavior in advance, the secondary reminder will be given to the construction site personnel (the secondary reminder can be given to the construction site personnel through the buzzer on the construction site personnel's clothes or safety helmet). When it is determined that the construction site personnel do not need to be reminded of the behavior in advance, no further processing will be performed.

[0041] The specific process of determining whether a construction site worker needs a behavioral pre-reminder is as follows: select a construction site worker, obtain the behavior monitoring area in the corresponding personnel movement status log of the construction site worker, determine all construction site workers with dangerous behaviors that appeared in the behavior monitoring area before, obtain the possible danger pre-reminder value of each dangerous construction site worker, calculate the mean of the possible danger pre-reminder by summing up the possible danger pre-reminder values ​​of all dangerous construction site workers, calculate the possible danger pre-reminder mean, set the possible danger pre-reminder threshold mean, and when the possible danger pre-reminder mean ≥ the possible danger pre-reminder threshold mean, determine that the construction site worker needs a behavioral pre-reminder (otherwise, determine that the construction site worker does not need a behavioral pre-reminder).

[0042] The specific process of obtaining the possible danger pre-value of dangerous construction site personnel is as follows: obtain A personnel movement status logs generated continuously by the construction site personnel before, sort all the personnel movement status logs of the construction site personnel in the order of generation, and mark them with digital serial numbers in sequence (such as marked as a, a=1, 2, ..., A), select a dangerous construction site personnel, obtain A personnel movement status logs generated continuously by the dangerous construction site personnel before being marked (as a dangerous construction site personnel), sort all the personnel movement status logs of the dangerous construction site personnel in the order of generation, and mark them with digital serial numbers in sequence (such as marked as a, a=1, 2, ..., A), determine the consistent value of the state area of ​​each digital serial number, and mark all The status area consistency values ​​are summed and averaged to calculate the average status area consistency value. All status area consistency values ​​are sorted in ascending order according to the numerical sequence. The absolute difference between the two adjacent status area consistency values ​​after sorting is calculated to calculate the status area fluctuation difference value. The status area fluctuation difference value is summed and averaged to calculate the average status area fluctuation difference value (if the final calculated average status area fluctuation difference value is 0, the average status area fluctuation difference value is corrected to 0.1 to avoid the subsequent ratio calculation being unable to be performed). The ratio of the average status area consistency value to the average status area fluctuation difference value is calculated to calculate the possible danger pre-condition value of the personnel at the dangerous construction site.

[0043] The specific process of obtaining the state area consistency value of a digital serial number is as follows: obtain two personnel motion state logs with the same digital serial number (such as two personnel motion state logs with the digital serial number 1). When the behavior monitoring area where the two personnel motion state logs are located belongs to the same behavior monitoring area, the regional consistency value P (a, a) is recorded as 2. When the behavior monitoring area where the two personnel motion state logs are located does not belong to the same behavior monitoring area, the regional consistency value P (a, a) is recorded as 0. The motion state reference degree DL (a, a) of the motion state feature set of the two personnel motion state logs is obtained through Calculate the state region consistency value of the digital sequence, where pg1 is the region coefficient and pg2 is the motion state coefficient. The value of pg1 is 1.12 and the value of pg2 is 1.15.

[0044] The motion state reference degree of the motion state feature set of the two personnel motion state logs is obtained as follows: the motion state feature set of one of the personnel motion state logs is vectorized and converted into a vector C=(c1,c2,...,c n ), n is the total number of motion features, the motion state feature set of another person's motion state log is vectorized and converted into a vector D=(d1,d2,...,d n ), Sim(C,D)= Calculate the motion state reference degree.

[0045] The system divides construction workers into standardized zones based on construction processes and types of operations. Sensors collect data on various actions taken by construction workers, and using deep learning technology to analyze their movements within the corresponding behavior monitoring areas, accurately assessing whether they are engaging in visible dangerous behaviors. Unlike traditional camera-based monitoring, this model is no longer limited by the placement of cameras within the construction area, effectively avoiding blind spots caused by limited monitoring range and ensuring comprehensive coverage of construction worker behavior monitoring. If a construction worker is currently free of visible dangerous behaviors, further behavioral preemptive alert analysis is performed. By analyzing the actions of all previous dangerous workers in the area, potential preemptive operational hazards are identified to determine whether the current worker is at risk of engaging in dangerous behaviors. By combining proactive and preemptive safety monitoring, the system establishes a closed-loop intelligent safety management system characterized by comprehensive perception, immediate response, proactive prevention, and continuous optimization, significantly improving the accuracy, real-time nature, and proactive nature of construction worker behavior monitoring.

[0046] Example 2, as Figure 3 ,The intelligent monitoring method for construction site personnel behavior safety based on deep learning ,steps are as follows:

[0047] Step 1: Division of behavior monitoring areas and regular generation of personnel movement status logs;

[0048] Step 2: First-level reminder determination by construction site personnel;

[0049] Step 3: Secondary reminder determination by construction site personnel.

[0050] The above method constructs a smart safety management closed loop of "comprehensive perception-immediate response-proactive prevention-continuous optimization" through the rational combination of active safety monitoring and pre-emptive safety monitoring, significantly improving the accuracy, real-time and proactive nature of site personnel behavior monitoring.

[0051] The above formulas are all dimensionless and numerically calculated, and the preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0052] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0053] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0054] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0055] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0056] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0057] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0058] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A construction site personnel behavior safety intelligent monitoring system based on deep learning, characterized by: include The construction site personnel behavior monitoring unit divides the construction area into multiple behavior monitoring areas and regularly generates a personnel movement status log for each construction site personnel; The first-level behavior reminder unit generates a log of the movement status of construction site personnel each time, and further determines whether the construction site personnel have any obvious dangerous behavior. If it is determined that the construction site personnel have obvious dangerous behavior, the construction site personnel will be given a first-level reminder; The secondary behavior reminder unit, when it is determined that the construction site personnel do not have any obvious dangerous behavior, further determines whether the construction site personnel need to be reminded of the behavior in advance. If it is determined that the construction site personnel need to be reminded of the behavior in advance, the secondary reminder will be issued to the construction site personnel; The specific process of determining whether a construction site worker needs to be reminded of his / her behavior in advance is as follows: select a construction site worker, obtain the behavior monitoring area in the movement status log of the corresponding construction site worker, determine all construction site workers with dangerous behaviors that have appeared in the behavior monitoring area before, obtain the possible danger pre-reminder value of each dangerous construction site worker, calculate the mean of the possible danger pre-reminder by summing up the possible danger pre-reminder values ​​of all dangerous construction site workers, calculate the possible danger pre-reminder mean, set the possible danger pre-reminder threshold mean, and when the possible danger pre-reminder mean ≥ the possible danger pre-reminder threshold mean, determine that the construction site worker needs to be reminded of his / her behavior in advance; The specific process of obtaining the possible danger pre-value of dangerous construction site personnel is as follows: obtain A personnel motion status logs generated continuously by the construction site personnel before, sort all the personnel motion status logs of the construction site personnel in the order of generation, and mark them with digital serial numbers in sequence, select a dangerous construction site personnel, obtain A personnel motion status logs generated continuously by the dangerous construction site personnel before being marked, sort all the personnel motion status logs of the dangerous construction site personnel in the order of generation, and mark them with digital serial numbers in sequence, determine the state area consistency value of each digital serial number, sum and average all state area consistency values, calculate the average state area consistency value, sort all state area consistency values ​​in order from small to large according to the digital serial number, calculate the absolute difference between the two adjacent state area consistency values ​​after sorting, calculate the state area fluctuation difference value, sum and average all state area fluctuation difference values, calculate the average state area fluctuation difference value, calculate the ratio of the average state area consistency value to the average state area fluctuation difference value, and calculate the possible danger pre-value of the dangerous construction site personnel.

2. The construction site personnel behavior safety intelligent monitoring system based on deep learning according to claim 1 is characterized in that: The personnel movement status log includes the construction site personnel ID, movement status feature set, and the behavior monitoring area.

3. The construction site personnel behavior safety intelligent monitoring system based on deep learning according to claim 2 is characterized in that: The specific process of determining the motion state feature set of the personnel motion state log is as follows: select a construction site worker, regularly collect various action data generated by the construction site worker within a cycle, perform feature extraction on each motion data, extract various motion features, and integrate the extracted motion features into a motion state feature set in the form of a set.

4. The construction site personnel behavior safety intelligent monitoring system based on deep learning according to claim 1 is characterized in that: The specific process of determining whether a construction site worker has engaged in intuitively dangerous behavior is as follows: obtaining the motion status feature set of the construction site worker's corresponding motion status log, determining the dangerous behavior determination model corresponding to the behavior monitoring area in the worker's motion status log, importing the motion status feature set into the intuitive dangerous behavior determination model, and deriving the intuitive dangerous behavior value of the construction site worker. When the intuitive dangerous behavior value of the construction site worker is higher than the threshold, it is determined that the construction site worker has engaged in intuitively dangerous behavior.

5. The construction site personnel behavior safety intelligent monitoring system based on deep learning according to claim 1 is characterized in that: After issuing a first-level warning to the construction site personnel, the construction site personnel are simultaneously marked as construction site personnel with dangerous behavior.

6. The construction site personnel behavior safety intelligent monitoring system based on deep learning according to claim 1 is characterized in that: The process of obtaining the state area consistency value of a digital serial number is as follows: obtain two personnel motion status logs with the same digital serial number, obtain the area consistency value, obtain the motion state reference degree of the motion state feature set of the two personnel motion status logs, and calculate the state area consistency value of the digital serial number based on the area consistency value and the motion state reference degree.

7. A method for intelligently monitoring the safety of construction site personnel behavior based on deep learning, applied to the intelligent monitoring system for safety of construction site personnel behavior based on deep learning according to any one of claims 1 to 6, characterized in that: Here are the steps: Step 1: Division of behavior monitoring areas and regular generation of personnel movement status logs; Step 2: First-level reminder determination by construction site personnel; Step 3: Secondary reminder determination by construction site personnel.

Images