A construction safety monitoring method and system based on a large CV model

Abstract

The present invention discloses a construction safety monitoring method based on a CV large model, which relates to the technical field of construction monitoring, and includes: obtaining construction video data and using the CV large model to identify the behavior characteristics of construction personnel; comparing the behavior characteristics with the standard behavior characteristic library in vectors to identify high-risk construction monitoring personnel; establishing a multi-behavior joint risk analysis model to determine the peak monitoring period; performing cluster analysis on the peak monitoring period to obtain the largest cluster, and judging its significance of difference; if the difference is significant, establishing a spatial analysis model to identify the aggregation effect, evaluating the lag degree of self-correction of risk behaviors, locating high-risk areas, and realizing collaborative safety monitoring; a construction safety monitoring system based on a CV large model, which includes: a feature recognition module, a behavior comparison module, a peak extraction module, a cluster analysis module, and a lag evaluation module, and is used to implement the above method; the present invention can improve the level of construction safety monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of construction monitoring, and particularly relates to a construction safety monitoring method and system based on a CV large model. Background Art

[0002] With the development of computer vision (CV) technology, its application in construction safety monitoring has gradually attracted attention. In the field of construction safety monitoring, there are many deficiencies in the existing technologies. Compared with the construction safety monitoring method based on the CV large model, there are gaps in risk warning, resource utilization, risk analysis, and collaborative management.

[0003] The existing technologies have limited capabilities in identifying the behavioral characteristics of construction workers. They cannot automatically extract the attention behavioral characteristics, spatial position characteristics, and safety equipment characteristics of construction workers. If the construction safety monitoring technology based on the CV large model is adopted, it can comprehensively and real-time monitor and analyze the behaviors of construction workers, improve the timeliness and accuracy of risk behavior identification, and effectively reduce the probability of accidents.

[0004] The safety management in the time and space dimensions of the existing technologies is disjointed and cannot achieve collaborative monitoring. In the time dimension, the existing technologies have not established an effective multi-behavior joint risk analysis model and cannot comprehensively process the multi-behavior data of construction workers to determine the peak monitoring period when risk behaviors are concentrated. By calculating the behavior deviation rate and constructing a multi-behavior joint risk analysis model, the peak monitoring period is determined to improve the accuracy of safety management in the time dimension;

[0005] In the space dimension, the existing technologies do not have the spatial analysis model and the ability to identify the aggregation effect of personnel positions, cannot identify the aggregation effect of the positions of construction monitoring personnel, and it is difficult to determine high-risk areas. If the established spatial analysis model is adopted to collect and analyze the personnel position information in real time, the aggregation effect and high-risk areas can be identified, which is convenient for safety management personnel to reasonably allocate resources and improve the safety management efficiency in the space dimension. Summary of the Invention

[0006] The purpose of the present invention is to provide a construction safety monitoring method and system based on a CV large model to solve at least one of the above-mentioned existing technical problems.

[0007] In the first aspect, the present invention provides a construction safety monitoring method based on a CV large model, including the following steps:

[0008] Step 1: Obtain the construction video data of deep foundation pit excavation and support operations, and identify the behavioral characteristics of construction workers through the CV large model;

[0009] Step 2: Perform vector comparison and analysis on the behavioral characteristics of construction workers with a pre-constructed library of standard behavioral characteristics, identify risk behaviors, and mark the construction workers with the highest behavioral risk level as construction monitoring personnel;

[0010] Step 3: Obtain the behavioral characteristics of self-correction of construction monitoring personnel during the monitoring period, calculate the behavior deviation rates of different construction monitoring personnel, and establish a multi-behavior joint risk analysis model to extract the peak monitoring period;

[0011] Step 4: Conduct cluster analysis on the peak monitoring periods of all construction monitoring personnel to obtain the largest cluster, and determine whether the difference in the largest cluster is significant;

[0012] Step 5: If the difference is significant, use it to establish a spatial analysis model to identify the aggregation effect of the positions of construction monitoring personnel and evaluate the degree of lag in self-correction of the risk behaviors of construction monitoring personnel.

[0013] In the second aspect, the present invention provides a construction safety monitoring system based on a CV large model, including the following modules:

[0014] Feature recognition module: used to obtain construction video data of deep foundation pit excavation and support operations, and identify the behavioral characteristics of construction workers through the CV large model;

[0015] Behavior comparison module: used to perform vector comparison and analysis on the behavioral characteristics of construction workers with a pre-constructed library of standard behavioral characteristics, identify risk behaviors, and mark the construction workers with the highest behavioral risk level as construction monitoring personnel;

[0016] Peak extraction module: used to obtain the behavioral characteristics of self-correction of construction monitoring personnel during the monitoring period, calculate the behavior deviation rates of different construction monitoring personnel, and establish a multi-behavior joint risk analysis model to extract the peak monitoring period;

[0017] Cluster analysis module: used to conduct cluster analysis on the peak monitoring periods of all construction monitoring personnel to obtain the largest cluster, and determine whether the difference in the largest cluster is significant;

[0018] Lag evaluation module: if the difference is significant, used to establish a spatial analysis model to identify the aggregation effect of the positions of construction monitoring personnel and evaluate the degree of lag in self-correction of the risk behaviors of construction monitoring personnel.

[0019] Advantages of the present invention:

[0020] 1. Utilize the image recognition ability of the CV large model to extract the attention behavior characteristics, spatial location characteristics, and safety equipment characteristics of construction workers, enabling the early detection of construction workers' behavior characteristics and preventing construction safety accidents. Compare the behavior characteristics of construction workers with the normative behavior feature library through vector comparison, calculate the Manhattan distance, and identify risk behaviors. Divide the behavior risk levels and mark the construction monitoring personnel to achieve focused attention on high-risk individuals and reduce the likelihood of accidents. At the same time, allocate safety monitoring resources targeted according to the risk levels.

[0021] 2. Establish a multi-behavior joint risk analysis model by calculating the behavior deviation rate to determine the peak monitoring period. Locate the time period when risk behaviors are most concentrated during the construction process, providing a basis for safety management personnel to intervene in a timely manner. Evaluate the concentration degree of risk behaviors by analyzing the KL divergence within the peak monitoring period, adjust the management strategy in a timely manner, and increase the control of high-risk time periods and areas.

[0022] 3. Conduct cluster analysis on the peak monitoring periods of all construction monitoring personnel to obtain the largest cluster. Integrate construction workers with similar risk characteristics, grasp the risk behavior patterns as a whole, and identify the key groups and time periods when risk behaviors occur concentratedly. Judge whether the difference of the largest cluster is significant, evaluate the reliability and effectiveness of the clustering results, and provide a scientific basis for subsequent safety decisions. If the difference is significant, a special management strategy can be formulated for this group.

[0023] 4. Used to establish a spatial analysis model to identify the aggregation effect of the positions of construction monitoring personnel, and further determine high-risk areas from the spatial dimension. Evaluate the self-correction lag degree of the risk behaviors of construction monitoring personnel by calculating the regional hysteresis index. According to the evaluation results, decide to increase the supervision frequency, adjust the construction process, or strengthen warning measures, accurately invest limited resources in high-risk areas, improve the overall safety management efficiency, and ensure construction safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0025] Figure 1 is the flowchart of the construction monitoring behavior analysis of the present invention;

[0026] Figure 2 is the flowchart of the construction safety time and space analysis provided in the second embodiment of the present invention;

[0027] Figure 3It is a module diagram of a construction safety monitoring system based on a CV large model provided in Embodiment 3 of the present invention. Detailed implementation manners

[0028] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

[0029] Embodiment 1

[0030] Deep foundation pit operations involve links such as soil excavation and support structure installation, and are prone to accidents such as collapses, object strikes, and mechanical injuries; at the same time, the distance between personnel and the edge of the foundation pit, the mechanical operation radius, etc. can be directly quantified through visual positioning; monitoring and analyzing the construction scenarios of deep foundation pit excavation and support operations is beneficial to focusing on the core pain points of construction safety;

[0031] As Figure 1 shown, a construction safety monitoring method based on a CV large model provided in the embodiment of the present invention includes the following steps:

[0032] Step 1: Obtain the construction video data of deep foundation pit excavation and support operations, and identify the behavior characteristics of construction workers through the CV large model;

[0033] In some embodiments, a plurality of camera acquisition devices are installed around the foundation pit and on the top of the tower crane to obtain the construction video data in the construction scenarios of deep foundation pit excavation and support operations;

[0034] Identify the construction workers in the construction video data through the CV large model, and extract the behavior characteristics of the construction workers;

[0035] Among them, the behavior characteristics include: attention behavior characteristics, spatial position characteristics, and safety equipment characteristics;

[0036] It should be further noted that the attention behavior characteristics reflect the alertness of construction workers to the risks of the operation environment; by analyzing the line-of-sight direction of construction workers through the CV large model, calculate the included angle between the line-of-sight direction and the normal line of the foundation pit, which is used to timely detect the distracted state of construction workers and prevent accidents such as soil collapses and mechanical collisions caused by reaction delays, so as to ensure the safety of construction operations;

[0037] Spatial position features quantify the relative position relationship between construction workers and dangerous areas such as the edge of the foundation pit and the mechanical operation area through a large CV model; calculate the distance from the edge of the foundation pit based on the monocular vision ranging formula, and reconstruct the robotic arm trajectory through multiocular vision to identify behaviors of illegally entering the construction warning area or the mechanical blind area, trigger real-time alarms or emergency stops of equipment, reduce the probability of accidents such as falls and squeezes, and optimize the safety control of the construction area;

[0038] Safety equipment features verify the wearing compliance of protective equipment such as safety helmets and seat belts through a large CV model; use the Faster R-CNN algorithm to detect the fitting degree of the safety helmet and analyze the spatial relationship between the seat belt buckle and the human skeleton; the proper wearing of seat belts and safety helmets can directly reduce the risk of head impacts, high-altitude falls, etc. At the same time, as a visual indicator of safety management compliance, it promotes the implementation of on-site safety regulations;

[0039] It should be noted that the role of extracting the behavioral characteristics of construction workers is as follows:

[0040] Function 1: Achieve risk early warning. Through the analysis of attention behavioral characteristics, the alert state of construction workers is monitored in real time to prevent reaction delays caused by distraction; the spatial position features quantify the distance from the dangerous area and trigger real-time alarms or emergency stops of equipment;

[0041] Function 2: Provide a basis for data modeling. Extracting the behavioral characteristics of construction workers provides basic data for subsequent vector comparison and analysis.

[0042] Step 2: Conduct vector comparison and analysis on the behavioral characteristics of construction workers with a pre-constructed library of standard behavioral characteristics, identify risk behaviors, and mark the construction workers with the highest behavioral risk level as construction monitoring personnel;

[0043] In some embodiments, obtain the standard behavioral characteristics during the construction of deep foundation pit excavation and support operations, and construct a library of standard behavioral characteristics;

[0044] Exemplarily, the standard behavioral characteristics include: the included angle between the line of sight direction of the construction worker and the normal line of the foundation pit is less than 30 degrees, the distance between the construction worker and the edge of the foundation pit is greater than 1.5 meters, the working radius between the construction worker and the operating equipment is higher than 3 meters, the fitting degree of the safety helmet wearing is greater than 0.85, and the seat belt buckle is within the range of ±10 cm of the waist center line;

[0045] Extract the standard vectors based on the library of standard behavioral characteristics, and obtain the characteristic behaviors of construction workers in real time through a large CV model to construct behavioral vectors;

[0046] It should be noted that the standard vectors are constructed based on the minimum standards of the standard behavioral characteristics library;

[0047] Exemplarily, the construction methods of the specification vector and the behavior vector are as follows: Minimum value of the specified line-of-sight angle: When the line-of-sight angle is within [25°, 30°], the corresponding element of the vector is set to 0; otherwise, it is set to 1.

[0048] Minimum value of the foundation pit edge distance: If the edge distance is within [1.5m, 2.0m], the corresponding element of the vector is set to 0; otherwise, it is set to 1.

[0049] Minimum value of the mechanical operation radius: If the operation radius is within [3.0m, 3.5m], the corresponding element of the vector is set to 0; otherwise, it is set to 1.

[0050] Minimum value of the safety helmet fitting degree: If the safety helmet wearing fitting degree range is within [0.85, 0.90], the corresponding element of the vector is set to 0; otherwise, it is set to 1.

[0051] Minimum value of the safety belt buckle position: If the safety belt buckle is within the range of ±10cm from the midline of the waist, the corresponding element of the vector is set to 0; otherwise, it is set to 1.

[0052] According to the minimum requirements of the standard behavior feature library, a 5-dimensional specification vector is constructed: y = [0, 0, 0, 0, 0];

[0053] Among them, the elements of the 5-dimensional specification vector correspond in sequence to the minimum value of the foundation pit edge distance, the minimum value of the mechanical operation radius, the minimum value of the safety helmet fitting degree, and the minimum value of the safety belt buckle position;

[0054] Based on the specification vector and the behavior vector, calculate the Manhattan distance between the specification vector and the behavior vector;

[0055] If the Manhattan distance of the construction worker exceeds the preset standard distance range, mark the behavior feature of the construction worker as a risk behavior;

[0056] Obtain the time when the Manhattan distance of the construction worker exceeds the preset standard distance range, and the total monitoring time;

[0057] Calculate the deviation ratio of the time when the Manhattan distance exceeds the preset standard distance range to the total monitoring time to obtain the time deviation ratio;

[0058] Based on the different time deviation ratios of the construction workers, divide the behavior features of the construction workers into multiple behavior risk levels;

[0059] Based on different risk levels, the system sends warning notifications of different risk levels to the construction workers;

[0060] If the behavior feature of the construction worker is at the highest behavior risk level, mark the construction worker as a construction monitor;

[0061] It should be noted that the construction behavior characteristics contain 5-dimensional vectors, and Manhattan distance can effectively measure the similarity of vector directions in high-dimensional space and reduce the impact of dimensional differences on behavior recognition and judgment;

[0062] The role of construction monitoring personnel is to:

[0063] Function 1: Focus on high-risk individuals in construction safety and reduce the possibility of accidents. Marking construction monitoring personnel can focus on high-risk individuals and achieve safe construction risk prevention and control. Taking deep foundation pit operations as an example, monitoring personnel's illegal entry into dangerous areas and failure to wear safety equipment correctly can easily cause collapse, fall and other accidents. Focusing on marking and monitoring can timely detect and stop dangerous behaviors and reduce the possibility of accidents.

[0064] Function 2: Optimize the allocation of safety monitoring resources. At the foundation pit construction site, safety management resources are limited. Marking construction monitoring personnel helps to reasonably allocate these safety monitoring resources. By limiting the management focus to construction monitoring personnel, safety officers can conduct more targeted supervision and guidance, thereby improving safety management efficiency;

[0065] Function three: Provide data support for subsequent risk analysis. By obtaining the behavioral characteristics of self-correction of construction monitoring personnel during the monitoring period, calculating the behavioral inaccuracy rate, and establishing a multi-behavior joint risk analysis model to extract the peak monitoring period, and then performing cluster analysis on the peak monitoring period of all construction monitoring personnel, etc., it can deeply explore the potential risks in the construction process.

[0066] The technical solution of this embodiment is: obtain the video data of deep foundation pit excavation and support operation, identify the behavioral characteristics of construction personnel through the CV large model; conduct vector comparison analysis between the behavioral characteristics of construction personnel and the pre-built standard behavioral characteristic library, identify risky behaviors and mark the construction personnel with the highest behavioral risk level as construction monitoring personnel; divide the behavioral risk level and mark the construction monitoring personnel to focus on high-risk individuals and reduce the possibility of accidents. At the same time, safety monitoring resources are allocated in a targeted manner according to the risk level.

[0067] Embodiment 2

[0068] like Figure 2 As shown, a construction safety monitoring method based on a CV large model also includes the following steps:

[0069] Step 3: Obtain the behavioral characteristics of self-correction of construction monitoring personnel during the monitoring period, calculate the behavioral inaccuracy rates of different construction monitoring personnel, and establish a multi-behavior joint risk analysis model to extract the peak monitoring period;

[0070] During the monitoring period, the CV large model is used to obtain the behavioral characteristics of the construction monitoring personnel's self-correction of risky behaviors;

[0071] It should be noted that the self-correction of risk behaviors by construction monitoring personnel means that after receiving a warning notice, the construction monitoring personnel self-correct their own risk behaviors so that their behavior characteristics do not fall into risk behaviors;

[0072] If the monitoring personnel do not self-correct the risk behaviors during the monitoring period, or do not complete the self-correction of risk behaviors during the monitoring period;

[0073] Obtain the number of times that the construction monitoring personnel fail to complete the self-correction of risk behaviors and fail to correct the risk behaviors during the monitoring period;

[0074] Based on the number of times that the construction monitoring personnel fail to complete the self-correction of risk behaviors, calculate the frequency of the construction monitoring personnel failing to complete the self-correction of risk behaviors during the monitoring period to obtain the behavior deviation rate;

[0075] Through the formula: Construct a multi-behavior joint risk analysis model;

[0076] Where Obtain the KL divergence of different behavior characteristics of construction monitoring personnel during the monitoring period, where P(x1, x2, x3) is the actual joint distribution of the behavior deviation rate, and x1, x2, x3 respectively represent the attention behavior characteristics, spatial position characteristics, and safety equipment characteristics;

[0077] Q(x1), Q(x2), Q(x3) are the canonical joint distributions of different characteristic behaviors, which are characterized by the behavior deviation rate threshold of the risk behaviors of construction monitoring personnel;

[0078] It should be noted that the KL divergence (Kullback-Leibler Divergence) is used to quantify the difference between the actual behavior distribution and the canonical behavior distribution of construction monitoring personnel, reflecting the concentration degree of risk behaviors during the monitoring period. Q(x1), Q(x2), Q(x3) are set by those skilled in the art based on experience;

[0079] Calculate the KL divergence of each monitoring period to construct a period divergence sequence;

[0080] Conduct a mutation analysis on the KL divergence of the period divergence sequence to identify the peak period of the KL divergence;

[0081] Among them, the mutation analysis obtains the monitoring period with the KL divergence higher than 3 times the standard deviation by calculating the standard deviation of the period divergence sequence as the peak period;

[0082] Obtain all the peak periods in the period divergence sequence. If the peak periods are adjacent in the time dimension, the peak periods will be merged;

[0083] Obtain the peak period, and select the peak period with the largest time dimension span after merging processing as the peak monitoring period;

[0084] It should be noted that the function of determining the peak monitoring period is as follows:

[0085] Function 1. Locate high-risk time periods: By establishing a multi-behavior joint risk analysis model to calculate the KL divergence and performing mutation analysis on the cycle divergence sequence to determine the peak monitoring period, it is possible to accurately find the time period when risk behaviors are most concentrated during the construction process. For example, in deep foundation pit construction, changes in different construction stages and working environments may cause fluctuations in risk behaviors. The peak monitoring period can accurately lock specific periods with high risk, such as when large-scale machinery is concentrated in operation or multiple types of work are carried out in cross-operation, helping safety management personnel to pay timely attention and intervene to prevent accidents;

[0086] Function 2. Provide effective data for clustering analysis: Extract features such as the cycle duration and peak intensity of the peak monitoring period, and then use clustering algorithms to group construction monitoring personnel with similar risk characteristics into one category, which helps to discover the risk behavior patterns of different groups of construction personnel, deeply understand the distribution law of risk behaviors, and provide a basis for formulating targeted safety management measures;

[0087] Function 3. Evaluate the concentration degree of risk behaviors: The higher the KL divergence value, the greater the difference between the actual behavior and the standard behavior, and the more concentrated the risk behaviors. By determining the peak monitoring period and analyzing the KL divergence therein, safety management personnel can intuitively understand the severity of the risk, timely adjust the management strategy, and increase the control intensity of high-risk time periods and areas.

[0088] Step 4. Perform clustering analysis on the peak monitoring periods of all construction monitoring personnel to obtain the largest clustering cluster, and judge whether the difference of the largest clustering cluster is significant;

[0089] Obtain the peak monitoring periods of all construction monitoring personnel, and extract the cycle characteristics of the peak monitoring periods;

[0090] Perform Min-Max normalization processing on the cycle characteristics and construct a cycle characteristic group;

[0091] Among them, the cycle characteristics include: cycle duration, that is, the time length of the peak monitoring period; peak intensity, that is, the peak of the behavior deviation rate within the peak monitoring period;

[0092] Based on the cycle characteristic group of all construction monitoring personnel, use clustering algorithms to cluster construction monitoring personnel with similar peak monitoring periods, and obtain multiple clustering clusters;

[0093] Through the formula: Calculate the silhouette coefficient s(i) for each sample in the largest cluster, where i represents the sample number in the largest cluster, i.e., the number of construction monitoring personnel;

[0094] a(i) represents the average distance from sample i to all other samples within the largest cluster, which is used to characterize the intra-cluster compactness of the sample;

[0095] b(i) represents the minimum average distance from sample i to all other clusters, which is used to characterize the inter-cluster separation of the sample. All other clusters refer to excluding the largest cluster from different clusters;

[0096] It should be noted that a(i) is calculated by taking the average of the Euclidean distances between construction monitoring personnel i and the periodic characteristics of other construction monitoring personnel within the largest cluster;

[0097] b(i) is calculated by taking the average of the Manhattan distances between construction monitoring personnel i and the periodic characteristics of other construction monitoring personnel within all other clusters;

[0098] Calculate the mean of the silhouette coefficients of all samples within the largest cluster, and calculate the degree of closeness between the mean and the preset upper limit value of the silhouette coefficient to obtain the silhouette closeness value;

[0099] Among them, the upper limit value of the silhouette coefficient is 1;

[0100] Compare the silhouette closeness value with the preset closeness range value. If the silhouette closeness value is within the preset closeness range value, it is considered that the difference in the largest cluster is significant;

[0101] If the silhouette closeness value is not within the preset closeness range value, it is considered that the difference in the largest cluster is not significant;

[0102] It should be noted that calculating the largest cluster can integrate construction monitoring personnel with similar risk characteristics; through clustering analysis of the peak monitoring periods of all construction monitoring personnel, similar periodic characteristics are grouped into one category to form multiple clusters, where the largest cluster contains the largest number of personnel groups with similar risk characteristics; it helps to grasp the risk behavior patterns of construction personnel as a whole and identify the key personnel and time periods when risk behaviors occur concentratedly;

[0103] For example, in deep foundation pit construction, it may be found that construction personnel within a certain cluster generally have similar risk behaviors such as inattention and illegal operations during a specific construction stage, such as when the earth excavation is nearing completion, providing a clear focus direction for safety management;

[0104] Judging whether the difference in the largest clustering cluster is significant can evaluate the reliability and effectiveness of the clustering results, providing a basis for subsequent safety management decisions. If the difference is significant, it indicates that the personnel risk characteristics within the largest clustering cluster have obvious consistency and uniqueness, and the clustering results are valid. Safety management can formulate specific management strategies for this group, such as strengthening the training of the personnel in this clustering cluster and increasing the supervision frequency during specific periods.

[0105] Step 5: If the difference is significant, use it to establish a spatial analysis model to identify the aggregation effect of the positions of construction monitoring personnel and evaluate the degree of self-correction lag of the risk behaviors of construction monitoring personnel;

[0106] Obtain the spatial coordinates of construction monitoring personnel within the deep foundation pit construction area, and calculate the distance between any two construction monitoring personnel within the deep foundation pit based on the spatial coordinates;

[0107] Use the Gaussian kernel function to determine the distance attenuation weight between any two construction monitoring personnel and establish a spatial weight matrix;

[0108] Based on the spatial weight matrix, construct a spatial analysis model through the global Moran's I index to identify the aggregation effect of the positions of construction monitoring personnel;

[0109] It should be noted that the value range of the global Moran's I index is [-1, 1]. If the global Moran's I index is greater than 0, it indicates positive spatial autocorrelation, that is, there is an aggregation effect in the positions of construction monitoring personnel;

[0110] The role of judging the existence of an aggregation effect in the positions of construction monitoring personnel is as follows:

[0111] Role 1: Locate high-risk areas: After determining the peak monitoring period with concentrated risks in the time dimension, the aggregation effect further determines the construction safety monitoring risks from the spatial dimension. When it is found that there is an aggregation effect in the positions of construction monitoring personnel, it indicates that in a specific spatial area of the construction site, risk behaviors are more likely to concentrate.

[0112] For example, in the construction of a deep foundation pit, it is found that the construction monitoring personnel in a certain area are concentrated. These areas may have an increase in risk behaviors, such as illegal operations and insufficient safety distances, due to factors such as narrow space and frequent cross-operation of construction. By identifying this spatial aggregation effect, safety management personnel can locate high-risk areas and take targeted measures in a timely manner, such as setting warning signs and strengthening on-site supervision, to effectively prevent accidents from occurring;

[0113] Function 2: Achieve collaborative safety management: Considering the risk concentration in the time dimension and the aggregation effect in the space dimension, the safety management at the construction site can be more collaborative and efficient. In the aggregation area, the behaviors of construction workers influence each other. Through the analysis of the aggregation effect, safety managers can organize construction workers in the aggregation area to carry out collaborative safety management activities. Conduct collective safety training to improve the overall safety awareness of construction workers, enhance the safety management level at the construction site, and ensure the smooth progress of construction;

[0114] If the global Moran's I index is less than 0, it indicates spatial negative correlation, that is, there is a discrete effect in the positions of construction monitoring personnel;

[0115] If the global Moran's I index is equal to 0, it means that the positions of construction monitoring personnel are randomly distributed;

[0116] If there is an aggregation effect in the positions of construction monitoring personnel, according to the distance decay weight between any two construction monitoring personnel, the deep foundation pit construction areas where the construction monitoring personnel are located are merged to obtain the construction monitoring area;

[0117] It should be noted that by setting a distance decay weight threshold, if the distance decay weight between any two construction monitoring personnel is higher than the distance decay weight threshold, the deep foundation pit construction areas where the construction monitoring personnel are located are merged; the distance decay weight threshold is set by professionals in this field based on experience;

[0118] After the system sends early warning notifications of different risk levels to construction workers within the construction monitoring area, obtain the difference between the self - correction time of the dangerous behaviors of construction monitoring personnel and the monitoring time to get the behavior correction delay;

[0119] Obtain the average behavior correction delay of construction monitoring personnel within the construction monitoring area, and the average behavior correction delay of all construction monitoring areas;

[0120] Calculate the ratio of the average behavior correction delay of construction monitoring personnel within the construction monitoring area to the average behavior correction delay of all construction monitoring areas to get the regional hysteresis index;

[0121] Based on the regional hysteresis index, evaluate the self - correction hysteresis degree of the risk behaviors of construction monitoring personnel;

[0122] It should be noted that by calculating the regional hysteresis index, we can intuitively understand the speed of personnel correcting risk behaviors in different construction monitoring areas. For example, in the deep foundation pit construction, if the regional hysteresis index is high, it indicates that the personnel in this area correct themselves slowly and the safety hazards last for a long time. Based on this, the safety management department can decide to increase the supervision frequency in this area, adjust the construction process or strengthen warning measures, accurately invest limited resources in high - risk areas, improve the overall safety management efficiency, and reduce the probability of accidents.

[0123] The technical solution of this embodiment is as follows: Obtain the behavioral characteristics of self-correction of construction monitoring personnel during the monitoring period, calculate the behavior deviation rates of different construction monitoring personnel, establish a multi-behavior joint risk analysis model to extract the peak monitoring period; perform cluster analysis on the peak monitoring periods of all construction monitoring personnel to obtain the largest cluster, and determine whether the difference in the largest cluster is significant; if the difference is significant, use it to establish a spatial analysis model to identify the aggregation effect of the positions of construction monitoring personnel and evaluate the degree of lag in self-correction of the risk behaviors of construction monitoring personnel; according to the evaluation results, decide to increase the supervision frequency, adjust the construction process or strengthen warning measures, accurately invest limited resources in high-risk areas, improve the overall safety management efficiency, and ensure construction safety.

[0124] Embodiment III

[0125] As Figure 3 shown, a construction safety monitoring system based on a large CV model includes the following modules:

[0126] Feature recognition module: used to obtain the construction video data of deep foundation pit excavation and support operations, and identify the behavioral characteristics of construction personnel through the large CV model;

[0127] Identify the construction personnel in the construction video data through the large CV model, and extract the behavioral characteristics of the construction personnel;

[0128] Among them, the behavioral characteristics include: attention behavioral characteristics, spatial position characteristics, and safety equipment characteristics;

[0129] Behavior comparison module: used to perform vector comparison analysis on the behavioral characteristics of construction personnel and a pre-constructed standard behavioral characteristic library, identify risk behaviors, and mark the construction personnel with the highest behavioral risk level as construction monitoring personnel;

[0130] Obtain the standard behavioral characteristics during the construction of deep foundation pit excavation and support operations, and construct a standard behavioral characteristic library;

[0131] Extract the standard vector based on the standard characteristic behavior library, and obtain the characteristic behavior of construction personnel in real time through the large CV model to construct a behavior vector;

[0132] Based on the standard vector and the behavior vector, calculate the Manhattan distance between the standard vector and the behavior vector;

[0133] If the Manhattan distance of the construction personnel exceeds the preset standard distance range, mark the behavioral characteristics of the construction personnel as risk behaviors;

[0134] Obtain the time when the Manhattan distance corresponding to the construction personnel exceeds the preset standard distance range, and the total monitoring time;

[0135] Calculate the proportion of the deviation between the time when the Manhattan distance exceeds the preset standard distance range and the total monitoring time to obtain the time deviation ratio;

[0136] Based on the different time deviation ratios of the construction workers, divide the behavioral characteristics of the construction workers into multiple behavioral risk levels;

[0137] Based on different risk levels, the system sends early warning notifications of different risk levels to the construction workers;

[0138] If the behavioral characteristics of the construction worker are at the highest behavioral risk level, mark the construction worker as a construction monitor;

[0139] Peak extraction module: used to obtain the self-corrected behavioral characteristics of construction monitors during the monitoring period, calculate the behavioral deviation rate of different construction monitors, and establish a multi-behavior joint risk analysis model to extract the peak monitoring period;

[0140] During the monitoring period, obtain the self-corrected behavioral characteristics of construction monitors for risk behaviors through the CV large model;

[0141] If the monitor does not self-correct the risk behavior during the monitoring period, or does not complete the self-correction of the risk behavior during the monitoring period;

[0142] Obtain the number of times that the construction monitor fails to complete the self-correction of risk behaviors and fails to correct risk behaviors during the monitoring period;

[0143] Based on the number of times that the construction monitor fails to complete the self-correction of risk behaviors, calculate the frequency of the construction monitor's failure to complete the self-correction of risk behaviors during the monitoring period to obtain the behavioral deviation rate;

[0144] Through the formula: Construct a multi-behavior joint risk analysis model;

[0145] Where Obtain the KL divergence of different behavioral characteristics of construction monitors during the monitoring period, where P(x1, x2, x3) is the actual joint distribution of the behavioral deviation rate, and x1, x2, x3 respectively represent the attention behavioral characteristics, spatial position characteristics, and safety equipment characteristics;

[0146] Q(x1), Q(x2), Q(x3) are the canonical joint distributions of different characteristic behaviors, which are characterized by the behavioral deviation rate threshold of the construction monitor's risk behavior;

[0147] Calculate the KL divergence of each monitoring period and construct a period divergence sequence;

[0148] Conduct mutation analysis on the KL divergence of the period divergence sequence to identify the peak period of the KL divergence;

[0149] Among them, for mutation analysis, the standard deviation of the cycle divergence sequence is calculated to obtain the monitoring cycles with KL divergence higher than 3 times the standard deviation as the peak cycles;

[0150] All peak cycles in the cycle divergence sequence are obtained. If the peak cycles are adjacent in the time dimension, the peak cycles are merged;

[0151] The peak cycle with the largest span in the time dimension after the peak cycles are merged is obtained as the peak monitoring cycle;

[0152] Clustering analysis module: used to perform clustering analysis on the peak monitoring cycles of all construction monitoring personnel to obtain the largest clustering cluster and judge whether the difference in the largest clustering cluster is significant;

[0153] All peak monitoring cycles of all construction monitoring personnel are obtained, and the cycle characteristics of the peak monitoring cycles are extracted;

[0154] The cycle characteristics are processed by Min - Max normalization, and a cycle characteristic group is constructed;

[0155] Among them, the cycle characteristics include: cycle duration, that is, the time length of the peak monitoring cycle; peak intensity, that is, the peak of the behavior misalignment rate within the peak monitoring cycle;

[0156] Based on the cycle characteristic group of all construction monitoring personnel, construction monitoring personnel with similar peak monitoring cycles are clustered through a clustering algorithm to obtain multiple clustering clusters;

[0157] Through the formula: The silhouette coefficient s(i) of each sample in the largest clustering cluster is calculated, where i represents the sample number in the largest clustering cluster, that is, the number of the construction monitoring personnel;

[0158] a(i) represents the average distance from sample i to all other samples in the largest clustering cluster, which is used to characterize the intra - cluster compactness of the sample;

[0159] b(i) represents the minimum average distance from sample i to all other clusters, which is used to characterize the inter - cluster separation of the sample. All other clusters refer to excluding the largest clustering cluster from different clustering clusters;

[0160] It should be noted that a(i) is calculated by taking the average of the Euclidean distances of the cycle characteristics of construction monitoring personnel i and other construction monitoring personnel in the largest clustering cluster;

[0161] b(i) is calculated by taking the average of the Manhattan distances of the cycle characteristics of construction monitoring personnel i and other construction monitoring personnel in all other clusters;

[0162] Calculate the mean of the silhouette coefficients of all samples within the largest cluster, calculate the degree of proximity between the mean and a preset upper limit value of the silhouette coefficient to obtain a silhouette proximity value;

[0163] Compare the silhouette proximity value with a preset proximity range value. If the silhouette proximity value is within the preset proximity range value, it is considered that the difference in the largest cluster is significant;

[0164] If the silhouette proximity value is not within the preset proximity range value, it is considered that the difference in the largest cluster is not significant;

[0165] Hysteresis evaluation module: If the difference is significant, it is used to establish a spatial analysis model to identify the aggregation effect of the positions of construction monitoring personnel and evaluate the self-correction hysteresis degree of the risk behaviors of construction monitoring personnel;

[0166] Obtain the spatial coordinates of construction monitoring personnel within the deep foundation pit construction area, and calculate the distance between any two construction monitoring personnel within the deep foundation pit based on the spatial coordinates;

[0167] Use the Gaussian kernel function to determine the distance attenuation weight between any two construction monitoring personnel and establish a spatial weight matrix;

[0168] Based on the spatial weight matrix, construct a spatial analysis model through the global Moran's I index to identify the aggregation effect of the positions of construction monitoring personnel;

[0169] If the global Moran's I index is less than 0, it indicates spatial negative correlation, that is, there is a discrete effect in the positions of construction monitoring personnel;

[0170] If the global Moran's I index is equal to 0, it indicates that the positions of construction monitoring personnel are randomly distributed;

[0171] If there is an aggregation effect in the positions of construction monitoring personnel, merge the deep foundation pit construction areas where the construction monitoring personnel are located according to the distance attenuation weight between any two construction monitoring personnel to obtain a construction monitoring area;

[0172] Obtain the difference between the self-correction time of the dangerous behaviors of construction monitoring personnel and the monitoring time after the system sends warning notifications of different risk levels to construction personnel within the construction monitoring area to obtain a behavior correction delay;

[0173] Obtain the average behavior correction delay of construction monitoring personnel within the construction monitoring area and the average behavior correction delay of all construction monitoring areas;

[0174] Calculate the ratio of the average behavior correction delay of construction monitoring personnel within the construction monitoring area to the average behavior correction delay of all construction monitoring areas to obtain a regional hysteresis index;

[0175] Based on the regional hysteresis index, evaluate the self-correction hysteresis degree of the risk behaviors of construction monitoring personnel.

[0176] The above has described in detail an embodiment of the present invention, but the above content is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of the application of the present invention shall still fall within the scope covered by the patent of the present invention.

Claims

1. A construction safety monitoring method based on a large CV model, characterized in that, It includes the following steps: Obtain the behavioral characteristics of self - correction of construction monitoring personnel during the monitoring period, establish a multi - behavior joint risk analysis model to extract the peak monitoring period; The acquisition method of the construction monitoring personnel is as follows: Obtain the construction video data of deep foundation pit excavation and support operations, and identify the behavioral characteristics of construction personnel through a large CV model; Conduct vector comparison and analysis between the behavioral characteristics of construction personnel and the pre - constructed standard behavioral characteristic library, identify risk behaviors, and mark the construction personnel with the highest behavioral risk level as construction monitoring personnel; Conduct cluster analysis on the peak monitoring periods of all construction monitoring personnel to obtain the largest cluster, and judge whether the difference of the largest cluster is significant; The acquisition method of the peak monitoring period is as follows: During the monitoring period, obtain the behavioral characteristics of self - correction of risk behaviors of construction monitoring personnel through a large CV model; Calculate the frequency of incomplete self - correction of risk behaviors of construction monitoring personnel during the monitoring period to obtain the behavior deviation rate; Based on the behavior deviation rate, establish a multi - behavior joint risk analysis model, obtain the KL divergence of different behavioral characteristics of construction monitoring personnel during the monitoring period, and construct a period divergence sequence; Conduct mutation analysis on the KL divergence of the period divergence sequence to obtain the peak period; After merging the peak periods, select the peak period with the largest time - dimension span as the peak monitoring period; If the difference is significant, use it to establish a spatial analysis model to identify the aggregation effect of the positions of construction monitoring personnel and evaluate the self - correction lag degree of the risk behaviors of construction monitoring personnel.

2. The construction safety monitoring method based on the CV large model according to claim 1, wherein The method of conducting vector comparison and analysis is as follows: Obtain the standard behavioral characteristics in deep foundation pit excavation and support operations, and construct a standard behavioral characteristic library; Extract the standard vector based on the standard characteristic behavior library, obtain the characteristic behavior of construction personnel in real - time through a large CV model, construct a behavior vector, and calculate the Manhattan distance between the standard vector and the behavior vector; If the Manhattan distance of the construction personnel exceeds the preset standard distance range, mark the behavioral characteristics of the construction personnel as risk behaviors.

3. A construction safety monitoring method based on a CV large model according to claim 1, characterized in that The acquisition method of the peak period is as follows: Calculate the standard deviation of the period divergence sequence, and obtain the monitoring period with KL divergence higher than 3 times the standard deviation as the peak period.

4. A construction safety monitoring method based on a CV large model according to claim 1, characterized in that, The method of judging whether the difference of the largest cluster is significant is as follows: Obtain the peak monitoring periods of all construction monitoring personnel, extract the period characteristics of the peak monitoring periods, and construct a period characteristic group; Based on the period characteristic group of all construction monitoring personnel, cluster the construction monitoring personnel with similar peak monitoring periods through a clustering algorithm to obtain multiple clusters; Calculate the silhouette coefficient for multiple clusters to obtain the silhouette proximity value; If the silhouette proximity value is within the preset proximity range value, it is considered that the difference of the largest cluster is significant.

5. A construction safety monitoring method based on a CV large model according to claim 4, characterized in that, The acquisition method of the silhouette proximity value is as follows: Calculate the mean of the silhouette coefficients of all samples in the largest cluster, and calculate the proximity degree between the mean and the preset upper limit value of the silhouette coefficient to obtain the silhouette proximity value.

6. A construction safety monitoring method based on a CV large model according to claim 1, characterized in that, The method of evaluating the self - correction lag degree of the risk behaviors of the monitoring personnel is as follows: Construct a spatial analysis model through the global Moran's I index to judge the aggregation effect of the positions of construction monitoring personnel; If there is an aggregation effect, the distances of construction monitoring personnel are analyzed and combined to determine the construction monitoring area; Calculate the ratio of the average behavior correction delay of construction monitoring personnel in the construction monitoring area to the average behavior correction delay of all construction monitoring areas to obtain the area hysteresis index; Based on the area hysteresis index, evaluate the self-correction hysteresis degree of the risk behaviors of construction monitoring personnel.

7. A construction safety monitoring method based on a CV large model according to claim 1, characterized in that, The method for identifying the aggregation effect of the positions of construction monitoring personnel is as follows: Obtain the spatial coordinates of construction monitoring personnel in the deep foundation pit construction area, and calculate the distance between any two construction monitoring personnel in the deep foundation pit; Use the Gaussian kernel function to determine the distance attenuation weight between any two construction monitoring personnel, establish a spatial weight matrix, and construct a spatial analysis model through the global Moran index to identify the aggregation effect of the positions of construction monitoring personnel.

8. A construction safety monitoring system based on a large CV model, which is used to implement a construction safety monitoring method based on a large CV model according to any one of claims 1-7, characterized in that, It includes the following modules: Peak extraction module: used to obtain the behavioral characteristics of self-correction of construction monitoring personnel during the monitoring period, and establish a multi-behavior joint risk analysis model to extract the peak monitoring period; Cluster analysis module: used to perform cluster analysis on the peak monitoring periods of all construction monitoring personnel to obtain the largest cluster, and judge whether the difference of the largest cluster is significant; Hysteresis evaluation module: if the difference is significant, it is used to establish a spatial analysis model to identify the aggregation effect of the positions of construction monitoring personnel, and evaluate the self-correction hysteresis degree of the risk behaviors of construction monitoring personnel.

Images