CN114493375A

CN114493375A - Construction safety macroscopic evaluation system and method

Info

Publication number: CN114493375A
Application number: CN202210339975.1A
Authority: CN
Inventors: 方东平; 古博韬; 岳清瑞; 李建华; 黄玥诚; 郭红领; 王尧; 曹思涵; 刘云飞; 曹海涛
Original assignee: Tsinghua University; Beijing Urban Construction Group Co Ltd
Current assignee: Tsinghua University; Beijing Urban Construction Group Co Ltd
Priority date: 2022-04-02
Filing date: 2022-04-02
Publication date: 2022-05-13
Anticipated expiration: 2042-04-02
Also published as: CN114493375B

Abstract

The invention relates to a construction safety macroscopic evaluation system and a method, wherein the system comprises: the data processing module is used for acquiring a data image of a construction site, marking and identifying the data image so as to determine each worker and the engineering machinery of the construction site and a central point coordinate and a contour coordinate corresponding to each worker and each engineering machinery; the safety evaluation module is used for determining an unsafe behavior value, a worker risk consciousness value and a management agility value of each worker without wearing a safety helmet according to each worker and the engineering machinery on a construction site and a central point coordinate and a contour coordinate which correspond to each worker and the engineering machinery; the training module is used for training by adopting a fuzzy neural network algorithm according to the unsafe behavior value, the worker risk consciousness value and the management agility value in the preset time period and the corresponding preset construction safety macroscopic evaluation value to obtain a construction safety macroscopic evaluation model; and the evaluation module is used for carrying out construction safety macroscopic evaluation on the target construction site by using the construction safety macroscopic evaluation model.

Description

Construction safety macroscopic evaluation system and method

Technical Field

The invention relates to the technical field of data processing of construction sites, in particular to a construction safety macroscopic evaluation system and a construction safety macroscopic evaluation method.

Background

The ubiquitous unsafe behavior on a construction site is the biggest challenge in the construction stage of the current construction industry. According to the theory of accident causes, unsafe behaviors of people and unsafe states of things are direct causes of asset loss and casualties. At present, the main method for reducing the safety accidents on the construction site is to patrol safety related personnel and identify potential safety hazards to avoid risks by observing unsafe behaviors and unsafe states on the site by manpower. But there are deeper system reasons behind the frequent occurrence of unsafe behavior and unsafe conditions. For the safety problem in construction, researchers find that three key elements are worth paying attention, namely safety force, safety culture and safety behavior.

Safety decision power is defined as the interactive process that a decision maker carries out with its subordinate in order to achieve an organizational safety goal, in which process the decision maker can exert influence on followers depending on the combined action of organizational and personal factors. This definition is widely recognized by researchers. The safety decision power is considered a sub-concept of decision power. Many studies have demonstrated that active revolutionary or contractual security decision-making efforts may affect subordinate managers or base workers: the subjective aspect realizes good perception of the superior managers on the safety attitude and the safety commitment, and builds a correct safety value view and positive acceptance on a safety target; and objectively prompting the subordinate or the worker to avoid unsafe production activities by setting reward and punishment measures and regulation programs based on the safety performance.

A security culture is defined as the view and belief of all members of an organization as to the problems of risk, accidents, and health faced by the organization at the time of production. The security culture is a sub-concept of the organizational culture, belongs to the organizational culture, and has main connotations including beliefs, attitudes and value views. The safety culture can be condensed into a combination of safety beliefs and value views constructed on people, events and things in a collective, and reflects the belief and behavior habits on life and safety which are agreed by all people in the collective. The safety culture of the construction project is expressed in three aspects of the behavior of workers, the subjective perception of the workers, the objective environment of the operation of the workers and the like.

Safety behavior is defined as the management behavior and manipulation behavior of a person during interaction with the person, thing, subject to safety objective driven or safety requirements. Traditional building safety research shows that unsafe behaviors of workers are main causes of accidents. A great deal of research is carried out on classification and identification of unsafe behaviors of a construction site, and many researches summarize the typical worker safety behavior (or unsafe behavior) categories of the construction site from different angles. The most common site unsafe behavior is over-access to dangerous sources and the incorrect wearing of personal protective equipment.

By researching the connotation and the interaction relationship of the safety decision power, the safety culture and the safety behavior, a building safety decision power-culture-behavior (LCB) method which takes decision-driven cultural development and behavior control as the core is provided. The LCB method emphasizes the function of safety decision, not only directly reduces unsafe behaviors, but also fundamentally changes the reasons of the unsafe behaviors through the development of safety culture, and finally achieves the aims of continuously reducing the unsafe behaviors and preventing accidents.

Disclosure of Invention

In order to overcome the problems in the related art, the invention provides a construction safety macroscopic evaluation system and method based on an LCB theory.

According to a first aspect of embodiments of the present invention, there is provided a construction safety macro evaluation system, including:

the data processing module is used for acquiring data images of a construction site, labeling and identifying the data images so as to determine each worker and each engineering machine of the construction site and the center point coordinate and the outline coordinate corresponding to each worker and each engineering machine;

the safety evaluation module is used for determining an unsafe behavior value, a worker risk consciousness value and a management agility value of each worker without wearing a safety helmet according to each worker and each engineering machine on a construction site and a central point coordinate and a contour coordinate which correspond to each worker and each engineering machine;

the training module is used for training by adopting a fuzzy neural network algorithm according to the unsafe behavior value, the worker risk consciousness value and the management agility value in the preset time period and the corresponding preset construction safety macroscopic evaluation value to obtain a construction safety macroscopic evaluation model;

and the evaluation module is used for carrying out construction safety macroscopic evaluation on the target construction site by using the construction safety macroscopic evaluation model.

In one embodiment, preferably, the data processing module includes:

the data acquisition unit is used for acquiring a data image of a construction site through the camera device;

a calibration unit for calibrating the image pickup apparatus every day;

the marking unit is used for marking a special road of the engineering machinery and a high-risk area of construction operation according to the received marking instruction before daily construction;

the identification unit is used for training workers, mobile engineering machinery and a tower crane based on Mask-RCNN and Deepsort algorithms through construction images collected and labeled in advance, identifying contour coordinates of the workers, the mobile engineering machinery and the tower crane in a data image of a construction site by taking minutes as a unit, and tracking each identification object; respectively counting the total number of recognized workers, mobile engineering machinery and tower cranes in a database by taking minutes as a unit; when the outlines of workers are identified, whether the workers wear safety helmets or not is identified, the number of the corresponding unworn safety helmets is counted, each identified worker, each mobile engineering machine and each tower crane are numbered, and the corresponding outline coordinates and the corresponding central point coordinates are synchronously stored into the corresponding data sets of the workers, the mobile engineering machines and the tower cranes.

In one embodiment, preferably, the security evaluation module is configured to:

storing the number of unworn helmets appearing per minute in a database as an array of time seriesUB _i，i = 0,1,2...n-1, orderW _i = UB _i+1 - UB _i , W ₀ = 0, where n is the time length of the time series, ifW _iIf < 0 then orderW _i = 0, construct an arrayW _iCalculating the daily unsafe behavior value using the following first calculation formula:

wherein the content of the first and second substances,N _ithe total number of workers on the construction site at the moment i;

counting the number of workers exposed to the operation in the risk area at each moment according to the center point coordinate and the contour coordinate corresponding to each worker and each engineering machine in the database, and calculating a first worker risk consciousness score by adopting the following second calculation formula:

wherein the content of the first and second substances,NW _irepresenting the number of workers exposed to the work in the risk area at time i,N _ithe total number of workers on the construction site at the moment i,

a value representing the number of workers exposed to the work in the risk area on a daily basis, i.e. a first worker risk awareness score;

counting the number of workers exposed on a special road of the worker machine at each moment according to the center point coordinate and the contour coordinate corresponding to each worker and each engineering machine in the database, and calculating a second worker risk consciousness score by adopting the following third calculation formula:

wherein the content of the first and second substances,NRW _irepresenting the number of workers exposed to the worker machine specific road at time i,N _ithe total number of workers on the construction site at the moment i,N _rwa second worker risk awareness score representing a number of workers who are exposed on a worker machine specific road on a daily basis, wherein the worker risk awareness score includes the first worker risk awareness score and the second worker risk awareness score;

respectively counting the time length of the worker not wearing the safety helmet every dayT _whTime of worker entering road special for engineering machineryT _wrDuration of occupied road for engineering machineryT _rThe time length unit is minute, and finally the management agility value T = (is output)T _wh，T _wr，T _r）。

In one embodiment, preferably, the training module comprises:

a normalization unit for performing normalization processing on the unsafe behavior value, the worker risk consciousness value and the management agility value by adopting a MinMaxScaler normalization algorithm to obtain the normalized unsafe behavior value, worker risk consciousness value and management agility value

The conversion unit is used for converting the unsafe behavior value, the worker risk consciousness value and the management agility value into values in the range of 0-1 and acquiring a construction safety macroscopic evaluation value preset by an expert on a construction site;

the training unit is used for training the unsafe behavior value, the worker risk consciousness value and the management agility value which are subjected to normalization processing as input x of the fuzzy neural network algorithm, and the corresponding preset construction safety macroscopic evaluation value as output y of the fuzzy neural network algorithm to obtain a construction safety macroscopic evaluation model;

the construction safety macroscopic evaluation model comprises a fuzzy layer, a rule layer, a regularization layer, a secondary fuzzy calculation layer and a deblurring layer;

the fuzzy layer fuzzifies each input through three membership functions, calculates a corresponding membership value and outputs the membership value to the rule layer;

the rule layer multiplies all the membership values to output the activation degree of each rule to the regularization layer;

the regularization layer carries out regularization calculation on the activation degree of each rule so as to output the regularized activation degree to the secondary fuzzy calculation layer;

the secondary fuzzy calculation layer calculates a new membership value through the activation degree after regularization so as to output the new membership value to the deblurring layer;

and the deblurring layer remaps the new membership value to an accurate value in the talkback point set so as to output a construction safety macroscopic evaluation value.

In one embodiment, preferably, the evaluation module is configured to:

acquiring a target data image of a target construction site, and carrying out marking and identification processing to determine each worker and each engineering machine of the target construction site and a central point coordinate and a contour coordinate corresponding to each worker and each engineering machine;

determining a target unsafe behavior value, a target worker risk consciousness value and a target management agility value of each worker without wearing a safety helmet according to each worker and each engineering machine of a target construction site and a central point coordinate and a contour coordinate corresponding to each worker and each engineering machine;

and calculating to obtain a target construction safety macroscopic evaluation value corresponding to the target construction site according to the target unsafe behavior value, the target worker risk consciousness value, the target management agility value and the construction safety macroscopic evaluation model.

According to a second aspect of the embodiments of the present invention, there is provided a construction safety macro evaluation method for a construction safety macro evaluation system, the method including:

acquiring a data image of a construction site, and carrying out labeling and identification processing to determine each worker and each engineering machine of the construction site, and a center point coordinate and a contour coordinate corresponding to each worker and each engineering machine;

determining an unsafe behavior value, a worker risk consciousness value and a management agility value of each worker without wearing a safety helmet according to each worker and each engineering machine on a construction site and a center point coordinate and a contour coordinate corresponding to each worker and each engineering machine;

training by adopting a fuzzy neural network algorithm according to the unsafe behavior value, the worker risk consciousness value and the management agility value in the preset time period and the corresponding preset construction safety macroscopic evaluation value to obtain a construction safety macroscopic evaluation model;

and performing construction safety macroscopic evaluation on the target construction site by using the construction safety macroscopic evaluation model.

In one embodiment, preferably, acquiring a data image of a construction site and performing labeling and identification processing to determine center point coordinates and contour coordinates of each worker and each work machine on the construction site and corresponding to each worker and each work machine includes:

acquiring a data image of a construction site through a camera device;

calibrating the camera device every day;

before construction every day, marking a special road of the engineering machinery and a high-risk area of construction operation according to a received marking instruction;

training workers, mobile engineering machinery and a tower crane based on Mask-RCNN and Deepsort algorithms through construction images collected and labeled in advance, respectively identifying contour coordinates of the workers, the mobile engineering machinery and the tower crane in a data image of a construction site by taking minutes as a unit, and tracking each identified object; respectively counting the total number of the identified workers, the mobile engineering machinery and the tower crane in a database by taking minutes as a unit; when the outlines of workers are identified, whether the workers wear safety helmets or not is identified, the number of the corresponding unworn safety helmets is counted, each identified worker, the mobile engineering machine and the tower crane are numbered, and the corresponding outline coordinates and the corresponding center point coordinates are stored into the data sets of the workers, the mobile engineering machines and the tower cranes corresponding to the moment.

In one embodiment, it is preferable that the determination of the daily unsafe behavior value, the worker risk awareness value and the management agility value of the worker who does not wear the helmet is performed based on the center point coordinates and the contour coordinates of each worker and each work machine on the construction site, respectively, including:

storing the number of unworn helmets appearing per minute in a database as an array of time seriesUB _i，i = 0,1,2...n-1, orderW _i = UB _i+1 - UB _i , W ₀ = 0, wherein n is the time length of the time series, ifW _iIf < 0 then orderW _i = 0, construct an arrayW _iCalculating the daily unsafe behavior value using the following first calculation formula:

wherein, the first and the second end of the pipe are connected with each other,NW _irepresenting the number of workers exposed to the work in the risk area at time i,N _ithe total number of workers on the construction site at the moment i,

In one embodiment, preferably, the training is performed by using a fuzzy neural network algorithm according to the unsafe behavior value, the worker risk awareness value, the management agility value and the corresponding preset construction safety macroscopic evaluation value in the preset time period to obtain a construction safety macroscopic evaluation model, which includes:

the MinMaxScaler normalization algorithm is adopted to normalize the unsafe behavior value, the worker risk consciousness value and the management agility value to obtain the normalized unsafe behavior value, worker risk consciousness value and management agility value

Converting the unsafe behavior value, the worker risk consciousness value and the management agility value into values in the range of 0-1, and acquiring a construction safety macroscopic evaluation value preset by an expert on a construction site;

taking the unsafe behavior value, the worker risk consciousness value and the management agility value after the normalization processing as input x of the fuzzy neural network algorithm, taking the corresponding preset construction safety macroscopic evaluation value as output y of the fuzzy neural network algorithm, and training to obtain a construction safety macroscopic evaluation model;

In one embodiment, preferably, the construction safety macro-evaluation of the target construction site using the construction safety macro-evaluation model includes:

According to a third aspect of the embodiments of the present invention, there is provided a construction safety macro evaluation apparatus, including:

a processor;

a memory for storing processor-executable instructions;

wherein the processor is configured to:

The technical scheme provided by the embodiment of the invention can have the following beneficial effects:

the invention classifies the identification results of unsafe behaviors of construction sites, people and objects on three aspects of safety decision power, safety culture and safety behaviors based on an LCB theory, describes expert evaluation as an automatic algorithm by a fuzzy neural network method, thereby realizing the automatic evaluation of the macroscopic safety level of a construction project and reflecting the actual safety level of the construction site through objective data. In the dimension of safety decision-making power, the management level of a safety management layer of a construction site is represented by measuring the duration of unsafe behaviors of people and unsafe states of objects on the basis of a computer vision technology, and the final result is T. In the safety culture dimension, the risk consciousness value of a construction worker group is reflected by measuring the number of operation workers exposed to risks based on a computer vision technology, so that the safety culture level of a construction site is reflected, and the result N is finally obtained. And in the safety behavior dimension, identifying the frequency of occurrence of unsafe behaviors through a computer vision technology based on deep learning to depict the safety behavior level of a construction site, wherein the final result is F. And constructing a fuzzy neural network by taking the management time effective value T, the worker risk consciousness value N and the unsafe behavior value F as input, and finally forming a macroscopic safety level S of the construction site.

Therefore, the construction site macroscopic safety assessment system is constructed based on the decision-making power-culture-behavior LCB theory and by combining the computer vision and fuzzy neural network method. And evaluating the safety management state of the construction project from three aspects of unsafe behaviors, worker risk awareness, management agility value and the like through the global image data of the construction site. By objectively collecting data of a construction site and combining with expert knowledge, the safety management capability of the projects of interest-related persons such as construction project managers and owners is improved.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention.

FIG. 1 is a block diagram illustrating a construction safety macro-evaluation system in accordance with an exemplary embodiment.

FIG. 2 is a block diagram illustrating data processing modules in a construction safety macro evaluation system in accordance with an exemplary embodiment.

FIG. 3 is a block diagram illustrating data processing modules in a construction safety macro evaluation system in accordance with an exemplary embodiment.

FIG. 4 is a flow diagram illustrating a construction safety macro-assessment method in accordance with an exemplary embodiment.

Detailed Description

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the invention, as detailed in the appended claims.

As shown in fig. 1, according to a first aspect of the embodiments of the present invention, there is provided a construction safety macro evaluation system, including:

the data processing module 11 is used for acquiring data images of a construction site, labeling and identifying the data images so as to determine each worker and each engineering machine of the construction site and a center point coordinate and a contour coordinate corresponding to each worker and each engineering machine;

the safety evaluating module 12 is used for determining an unsafe behavior value, a worker risk consciousness value and a management agility value of the daily worker without wearing the safety helmet according to each worker and each engineering machine on the construction site and the center point coordinate and the contour coordinate corresponding to each worker and each engineering machine;

the training module 13 is used for training by adopting a fuzzy neural network algorithm according to the unsafe behavior value, the worker risk consciousness value and the management agility value in the preset time period and the corresponding preset construction safety macroscopic evaluation value to obtain a construction safety macroscopic evaluation model;

and the evaluation module 14 is used for carrying out construction safety macroscopic evaluation on the target construction site by using the construction safety macroscopic evaluation model.

As shown in fig. 2, in one embodiment, the data processing module 11 preferably includes:

the data acquisition unit 21 is used for acquiring a data image of a construction site through a camera device; specifically, the data acquisition unit comprises five major submodules, namely a mooring airship, a high-definition camera, a holder, a picture transmission module and a remote control module. In order to ensure that the picture covers the construction site, the high-definition camera is installed at a height of at least 200 meters, and the inclination angle of the camera is based on the picture covering the whole construction site. In order to ensure the definition of the picture, the invention selects a high-definition camera of 6K or more to monitor the construction site in real time. The holder module adopts a three-axis self-stabilizing holder, and mainly performs attitude calibration on a camera lens on the holder through an IMU (inertial measurement unit) and a motor magnetic encoder, so that the stability of the lens is ensured. The remote control module is used for remotely controlling the orientation of the lens. The image transmission module is used for transmitting the images acquired by the high-definition camera to a ground industrial personal computer in real time.

The construction site lacks a high-altitude location to install the camera, so the tethered airship is selected to provide the location for the camera installation. The main reason for choosing the mooring airship here is to consider that it can be powered uninterruptedly so as to work at high altitude for a long time, and the price is more economical than that of an unmanned aerial vehicle (the unmanned aerial vehicle, an industrial balloon and other equipment can also meet the requirements here in cooperation with a mooring system). In specific implementation, firstly, based on a construction drawing, a proper position is selected on site to lift the airship to the height of 100 meters on the premise of stable power supply through the mooring system. And then the orientation of the high-definition camera is adjusted by connecting the remote control module and the image transmission module on the ground through an industrial personal computer. The image transmission module transmits the pictures shot by the high-definition camera to a ground industrial personal computer in real time. In addition, in order to ensure long-time work of the data acquisition module, the mooring airship is connected with a power supply of a construction site (which can be a living area power supply) and supplies power for the whole data acquisition module.

A calibration unit 22 for calibrating the imaging apparatus every day;

after the mooring airship is lifted off for the first time, the high-definition camera shooting area is adjusted through the ground industrial personal computer, so that the picture can cover all construction sites, and the numerical value of the IMU sensor is reset to zero at the moment. The IMU sensor is used for measuring the attitude of the high-definition camera, and the IMU sensing module forms a Cartesian coordinate system through a combined unit formed by 3 accelerometers and 3 gyroscope groups C. Having an x-axis, a y-axis, and a z-axis, the sensor is capable of measuring linear motion in the directions of the respective axes, as well as rotational motion about the respective axes. After zeroing, the linear movement and the rotation angle are adjusted every half hour so that the values are again 0. Because of the accumulated error of IMU, the IMU value is reset to zero after manual calibration at 7 am every day. A2000 mm by 2000mm purple square plastic plate was placed on site as a calibration color block before use, and the complete presence of the color block was confirmed in a high-definition camera. Traversing pixels in a frame image, and finding pixel blocks meeting the following requirements:

H ∈ [125,155],S ∈ [43,255],V ∈[46,255]

and recording pixel block coordinates corresponding to the maximum value and the minimum value in the horizontal direction and pixel block coordinates corresponding to the maximum value and the minimum value in the vertical direction, calculating Euclidean distances between two pixel points in the vertical direction and two pixel points in the horizontal direction pairwise respectively, adding the obtained 4-segment distances, calculating the arithmetic mean, wherein the corresponding real distance is 2 meters, and obtaining the scale of the shooting picture.

A labeling unit 23, configured to label, before construction every day, a dedicated road of the engineering machine and a high risk area of the construction work according to the received marking instruction;

because the construction site environment is complex, high-altitude falling is the most frequently occurring accident type, and the direct occurrence reason is that workers work at the edge of the high altitude, so that the marking of the real risk edge in the complex environment is the problem to be solved by the marking unit. Before labeling, a user needs to fully patrol the site and record the risk edge to be labeled. When labeling, the risk edge is drawn by a straight line or a curve, and then the range related to the risk area is drawn. The method comprises the following specific steps of firstly clicking the direction of a risk area by using a mouse to judge whether the risk area is generated inwards or outwards, then inputting the real width of the risk area, for example, 5 meters, then converting the real width into the distance in an image according to a scale, taking a drawn risk edge as a reference, making parallel lines in the selected direction, wherein the distance is the distance obtained by conversion, and finally highlighting and displaying the area related to the risk area in the image.

The human-computer collision accident is taken as a main construction accident source, and the essential reason is that workers appear in a risk area of construction operation, so that the risk area of the engineering machinery needs to be calibrated in advance. The risk area of the engineering machinery comprises the risk area around the engineering machinery and the preset engineering machinery road. In the calibration module, specific engineering machinery moving roads, including a transport road bearing heavy engineering machinery, a construction road and a temporary road walking normal engineering machinery, need to be calibrated at the beginning of construction, and the filling color algorithm uses the fillPoly algorithm in OpenCV. The calibration mode is similar to the calibration risk margin, and the calibration area needs to be ensured to be complete and closed. Finally, the area of the construction of the day is depicted by a polygonal line frame before the construction every day.

The identification unit 24 is used for training workers, mobile engineering machinery and a tower crane based on Mask-RCNN and Deepsort algorithms through construction images collected and labeled in advance, identifying contour coordinates of the workers, the mobile engineering machinery and the tower crane in a data image of a construction site by taking minutes as a unit, and tracking each identification object; respectively counting the total number of the identified workers, the mobile engineering machinery and the tower crane in a database by taking minutes as a unit; when the outlines of workers are identified, whether the workers wear safety helmets or not is identified, the number of the corresponding unworn safety helmets is counted, each identified worker, the mobile engineering machine and the tower crane are numbered, and the corresponding outline coordinates and the corresponding center point coordinates are stored into the data sets of the workers, the mobile engineering machines and the tower cranes corresponding to the moment.

The identification unit is mainly used for identifying workers and engineering machinery on a construction site and obtaining image coordinate points corresponding to the outer contours of the workers and the engineering machinery. According to the method, construction site data are collected and labeled in advance, constructors, typical mobile engineering machinery (an excavator, a mobile crane, a forklift, a pile driver and the like) and a tower crane are trained on the basis of Mask-RCNN and Deepsort algorithms, outlines of the constructors, the mobile engineering machinery and the tower crane are segmented in an image by taking minutes as a unit, and an identification object is tracked. For each identified object, the number of categories is incremented by 1 at the location of the minute in the database. And (4) identifying the contour of the constructor, identifying the safety helmet, and if the safety helmet is not identified, adding 1 to the number of people without the safety helmet corresponding to the moment in the database. And numbering each identified worker, the mobile engineering machine and the tower crane, and storing the corresponding outline coordinate and the central point coordinate into the data sets of the worker, the mobile engineering machine and the tower crane corresponding to the moment, wherein the outline storage schematic of each element in the database is shown in table 1. Wherein w represents a worker, c represents a mobile crane, f represents a forklift, e represents an excavator, t represents a tower crane, and tr represents a truck.

TABLE 1

In one embodiment, preferably, the security evaluation module is configured to:

wherein the content of the first and second substances,NRW _irepresenting the number of workers exposed to the worker machine specific road at time i,N _ithe total number of workers on the construction site at the moment i,N _rwindicating the number of workers exposed daily on the worker's machine specific road, i.e. the second worker risk awareness score,wherein the worker risk awareness value comprises a first worker risk awareness score and a second worker risk awareness score;

in this embodiment, the risks include high fall risks and man-machine collision risks. Since workers are required for work and must sometimes work in high risk areas, such workers exposed to high risk areas do not represent a macroscopic safety level of the construction project low, and therefore such a number of workers needs to be excluded when counting the number. Workers should in principle not be present on roads specifically planned for the construction machine, and therefore the number of workers present on the transport, construction and temporary roads reflects their awareness of risks. In summary, the worker risk awareness algorithm calculates two types of worker numbers, namely, the number of workers exposed to the peak area and the number of workers exposed to the road dedicated to the construction machine, and the specific algorithm is as follows.

And 5 meters are set as the risk distance of human-computer collision, and the 5 meters are converted into the pixel point distance in the graph through a scale of the calibration module. The number of workers exposed in the high risk area is calculated by calling the contour data and the center point data of the engineering machinery per minute identified by the identification unit, calculating the difference value of the coordinates of each contour point and the coordinates of the center point in the contour data of the engineering machinery (taking a mobile crane as an example here), and adding the distance between the maximum value and the corresponding pixel point of 5 meters to obtain the radiusR _ctkm, k = 1,2,...n，mN, where ct represents a mobile crane, k represents a crane number, m represents an m-th minute, and then centered on the identified work machine center point coordinate,R _ctkmdrawing a circle for the radius, then counting the number of workers exposed in the risk area coordinates of the edge of the engineering machinery circle and the calibration module in the minute, and subtracting the number of workers exposed in the construction area in the minute to obtain an array of a time sequence (the number of workers in the construction area in the minute is shown in the specification and the description of the invention)UNW _i), i = 1,2...nThen, counting in the database the number of workers with a continuous retention time in the risk zone of less than or equal to 5 minutes, counting the number of such workers present per minute (UNW _i), i = 1,2...nRevised (plus workers who stay in the risk area for no more than 5 minutes per minuteQuantity), get a revised time series (NW _i), i = 1,2...n，NW _iRepresenting the number of workers exposed to the work in the risk area at the moment i, and finally calculating the number value of the workers exposed to the risk area in the day

WhereinN _iThe total number of the construction site at time i.

The number of workers exposed on the construction machine exclusive road is calculated by counting the number of workers appearing on the construction machine exclusive road per minute (calibration by the calibration module), as follows

；

WhereinN _iThe total number of people at the site at time i,NRW _ithe number of workers on the road special for the engineering machinery at the moment i. Finally outputting daily N = (N _hw，N _rw)。

The management agility value mainly counts the unsafe state of a person and the duration of unsafe behaviors of the person, including the time when a worker appears on a road special for the engineering machinery and a road special for the engineering machinery is blocked without wearing a safety helmet.

In the construction process, the construction materials often occupy the road of the engineering machinery, so that the engineering machinery is in an unsafe state when entering and exiting. The method comprises the steps of firstly extracting a calibrated range of the special road for the engineering machinery from a collected image of a high-definition camera per minute, then subtracting the range, corresponding to the engineering machinery and workers, identified by an identification unit, of the engineering machinery, then counting coordinates of object points in a variable foreground by adopting a background difference algorithm for the rest images, then clustering the points in the foreground, determining the number of categories by using the maximum value of a contour coefficient, counting the maximum value of the distance between the coordinates of a clustering center point and the coordinates of the points of the categories, comparing the maximum value with the corresponding pixel distance of 1 meter, and recording the categories as the occupation of the special road for the engineering machinery if the distance is greater than 1 meter.

As shown in fig. 3, in one embodiment, preferably, the training module 13 includes:

the normalization unit 31 is configured to perform normalization processing on the unsafe behavior value, the worker risk awareness value and the management agility value by using a minmaxscale normalization algorithm to obtain the normalized unsafe behavior value, worker risk awareness value and management agility value;

the conversion unit 32 is used for converting the unsafe behavior value, the worker risk consciousness value and the management agility value into values in the range of 0-1 and acquiring a construction safety macroscopic evaluation value preset by an expert on a construction site;

the training unit 33 is used for training the unsafe behavior value, the worker risk consciousness value and the management agility value after the normalization processing as input x of the fuzzy neural network algorithm and the corresponding preset construction safety macroscopic evaluation value as output y of the fuzzy neural network algorithm to obtain a construction safety macroscopic evaluation model; the training algorithm employs hybrid training method in the ANFIS toolbox using matlab, for example, 15 data points may be randomly selected from 20 samples as training data, another 5 data points as test set, the training error acceptance range is set to 0.001, and the maximum training number is set to 1000.

models in common (F，N _hw，N _rw，T _wh，T _wr，T _r) Six inputs, each input passing through three membership functions, output a single value SA _i1 , A _i2 , A _i3Fuzzification is carried out, which respectively represents the high level, the middle level and the low level of a certain safety performance of the day on a construction site, wherein the first subscript represents the sequence of three membership function, and the second i represents the sequence of six inputs. ANFIS can adaptively learn fuzzy logic similar to that of humans when reasoning through a dataset of (F, N, T, S):

the membership function fuzzifies the classical set theory and indicates how well the mathematical properties required by the input for a set are met. The common membership function includes triangle and Gaussian, the invention adopts Gaussian membership function, the calculation formula of fuzzy layer output is:

。

whereinc _iAndσ _iis the membership function shape parameter and x is the raw data received by the ambiguity layer, i.e. the input data. Using the ANFIS toolbox of matlab, the collected data sets are loaded first, and then the number of membership functions corresponding to each input is set to 3And selecting the category of the membership function as gausssf. The structural design of ANFIS needs to design the number and the types of membership functions in the ANFIS tool box of matlab.

The rule layer multiplies all the membership values to output the activation degree of each rule to the regularization layer; wherein, the calculation formula of the rule layer is as follows:

，

the output of which represents the degree of activation of each if rule.

The regularization layer carries out regularization calculation on the activation degree of each rule so as to output the regularized activation degree to the secondary fuzzy calculation layer; the calculation formula is as follows:

。

the secondary fuzzy calculation layer calculates a new membership value through the activation degree after regularization so as to output the new membership value to the deblurring layer; the calculation formula is as follows:

，

whereinx _iFor the (i) th input(s),p _iare the corresponding weights.

And the deblurring layer remaps the new membership value to an accurate value in the talkback point set so as to output a construction safety macroscopic evaluation value. The calculation formula is as follows:

。

in one embodiment, the evaluation module 14 is preferably configured to:

determining a target unsafe behavior value, a target worker risk consciousness value and a target management agility value of daily workers without wearing safety helmets according to each worker and each engineering machine of a target construction site and a central point coordinate and a contour coordinate corresponding to each worker and each engineering machine;

As shown in fig. 4, according to a second aspect of the embodiment of the present invention, there is provided a construction safety macro evaluation method for a construction safety macro evaluation system, the method including:

step S401, collecting data images of a construction site, and carrying out labeling and identification processing to determine each worker and each engineering machine of the construction site, and a center point coordinate and a contour coordinate corresponding to each worker and each engineering machine;

step S402, determining unsafe behavior values, worker risk consciousness values and management agility values of daily workers without wearing safety helmets according to each worker and each engineering machine on the construction site and center point coordinates and contour coordinates corresponding to each worker and each engineering machine;

step S403, training by adopting a fuzzy neural network algorithm according to the unsafe behavior value, the worker risk consciousness value and the management agility value in the preset time period and the corresponding preset construction safety macroscopic evaluation value to obtain a construction safety macroscopic evaluation model;

and S404, performing construction safety macroscopic evaluation on the target construction site by using the construction safety macroscopic evaluation model.

In one embodiment, preferably, the step S401 includes the following steps:

acquiring a data image of a construction site through a camera device;

calibrating the camera device every day;

In one embodiment, preferably, the step S402 includes:

In one embodiment, step S403 preferably includes:

In one embodiment, step S404 preferably includes:

a processor;

a memory for storing processor-executable instructions;

wherein the processor is configured to:

It is further understood that the term "plurality" means two or more, and other terms are analogous. "and/or" describes the association relationship of the associated objects, meaning that there may be three relationships, e.g., a and/or B, which may mean: a exists alone, A and B exist simultaneously, and B exists alone. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. The singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

It will be further understood that the terms "first," "second," and the like are used to describe various information and that such information should not be limited by these terms. These terms are only used to distinguish one type of information from another and do not denote a particular order or importance. Indeed, the terms "first," "second," and the like are fully interchangeable. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present invention.

It is further to be understood that while operations are depicted in the drawings in a particular order, this is not to be understood as requiring that such operations be performed in the particular order shown or in serial order, or that all illustrated operations be performed, to achieve desirable results. In certain environments, multitasking and parallel processing may be advantageous.

Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.

It will be understood that the invention is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the invention is limited only by the appended claims.

Claims

1. A construction safety macro evaluation system is characterized by comprising:

2. The construction safety macro evaluation system of claim 1, wherein the data processing module comprises:

a calibration unit for calibrating the image pickup apparatus every day;

the identification unit is used for training workers, mobile engineering machinery and a tower crane based on Mask-RCNN and Deepsort algorithms through construction images collected and labeled in advance, identifying contour coordinates of the workers, the mobile engineering machinery and the tower crane in a data image of a construction site by taking minutes as a unit, and tracking each identification object; respectively counting the total number of the identified workers, the mobile engineering machinery and the tower crane in a database by taking minutes as a unit; when the outlines of workers are identified, whether the workers wear safety helmets or not is identified, the number of the corresponding unworn safety helmets is counted, each identified worker, each engineering machine and each tower crane are numbered, and the corresponding outline coordinates and the corresponding central point coordinates are synchronously stored into the corresponding data sets of the workers, the mobile engineering machines and the tower cranes.

3. The construction safety macroscopic evaluation system of claim 2, wherein the safety evaluation module is configured to:

according to the center point coordinate and the contour coordinate corresponding to each worker and each engineering machine in the database, counting the number of workers exposed on a special road of the worker machine at each moment, and calculating a second worker risk consciousness score by adopting the following third calculation formula:

4. The construction safety macro evaluation system of claim 1, wherein the training module comprises:

the construction safety macroscopic evaluation model comprises a fuzzy layer, a regular layer, a regularization layer, a secondary fuzzy calculation layer and a defuzzification layer;

5. The construction safety macro evaluation system of claim 1, wherein the evaluation module is configured to:

6. A construction safety macroscopic evaluation method is characterized by comprising the following steps:

7. The macroscopic evaluation method for construction safety according to claim 6, wherein the step of collecting data images of a construction site and performing labeling and identification processing to determine coordinates of a center point and coordinates of a contour of each worker and each construction machine on the construction site, respectively, comprises the steps of:

acquiring a data image of a construction site through a camera device;

calibrating the camera device every day;

training workers, mobile engineering machinery and a tower crane based on Mask-RCNN and Deepsort algorithms through construction images collected and labeled in advance, respectively identifying contour coordinates of the workers, the mobile engineering machinery and the tower crane in a data image of a construction site by taking minutes as a unit, and tracking each identified object; respectively counting the total number of the identified workers, the mobile engineering machinery and the tower crane in a database by taking minutes as a unit; when the outlines of workers are identified, whether the workers wear safety helmets or not is identified, the number of the corresponding unworn safety helmets is counted, each identified worker, the mobile engineering machine and the tower crane are numbered, and the corresponding outline coordinates and the corresponding central point coordinates are synchronously stored into the corresponding data sets of the workers, the mobile engineering machines and the tower crane.

8. The macroscopic evaluation method for construction safety according to claim 7, wherein the determination of the unsafe behavior value, the worker risk awareness value and the management agility value of the daily worker not wearing the safety helmet according to the coordinates of the center point and the contour coordinates of each worker and each construction machine on the construction site comprises:

9. The construction safety macroscopic evaluation method of claim 6, wherein the training is performed by using a fuzzy neural network algorithm according to the unsafe behavior value, the worker risk consciousness value and the management agility value in the preset time period and the corresponding preset construction safety macroscopic evaluation value to obtain a construction safety macroscopic evaluation model, comprising:

carrying out normalization processing on the unsafe behavior value, the worker risk consciousness value and the management agility value by adopting a MinMaxScaler normalization algorithm to obtain the unsafe behavior value, the worker risk consciousness value and the management agility value after the normalization processing;

10. The construction safety macro-evaluation method of claim 6, wherein the construction safety macro-evaluation of the target construction site using the construction safety macro-evaluation model comprises: