Human-machine cooperation construction unsafe behavior identification method and system

Abstract

The embodiment of the invention discloses an unsafe behavior identification method and system for man-machine cooperation construction, and the method comprises the steps: continuously obtaining the operation data of a plurality of moments in the man-machine cooperation construction, and enabling the operation data of each moment to comprise the posture of mechanical equipment, the posture of the mechanical equipment and the posture of the mechanical equipment; the number of safety helmets of various persons around the mechanical equipment is determined; according to the posture change of the mechanical equipment, the duration between the data moments and the change of the number of safety helmets of professional technicians, identifying and removing data between single operation from the operation data at the multiple moments; according to the total duration of occurrence of a plurality of number combinations of various personnel helmets in the remaining operation data, clustering is carried out on the number combinations, and clustering results correspond to common behaviors, normal behaviors and unsafe behaviors respectively; and according to the clustering result, identifying the man-machine cooperation construction unsafe behavior. According to the embodiment of the invention, advanced identification and early warning are carried out on unsafe behaviors in man-machine cooperation construction.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of unsafe behavior identification and early warning in the construction industry, and in particular to a method and system for identifying unsafe behavior in human-machine collaborative construction. Background Art

[0002] The emergence of new generation construction equipment such as new construction machinery and construction robots has made the human-machine collaboration construction production method more diverse and common. With the development of construction machinery, construction sites are becoming increasingly complex, and safety accidents caused by the interaction between people and machines have seriously restricted the safe development of the industry.

[0003] How to proactively identify and warn of unsafe behaviors in human-machine collaborative construction is an issue that needs to be addressed urgently. Summary of the invention

[0004] The embodiments of the present invention provide a method and system for identifying unsafe behaviors in human-machine collaborative construction, which can perform advance identification and early warning of unsafe behaviors in human-machine collaborative construction.

[0005] In a first aspect, an embodiment of the present invention provides a method for identifying unsafe behaviors in human-machine collaborative construction, comprising:

[0006] Continuously obtain operation data at multiple moments in human-machine collaborative construction, where the operation data at each moment includes the posture of the mechanical equipment and the number of safety helmets of various personnel around the mechanical equipment;

[0007] According to the posture change of the mechanical equipment, the duration of the data time, and the change in the number of helmets of the professional technicians, the data between single operations are identified and removed from the operation data at the multiple time points;

[0008] According to the total duration of multiple quantity combinations of safety helmets of various types of personnel in the remaining operation data, each quantity combination is clustered, where the clustering results correspond to common behaviors, normal behaviors, and unsafe behaviors respectively;

[0009] According to the clustering results, unsafe behaviors in human-machine collaborative construction are identified.

[0010] In a second aspect, an embodiment of the present invention provides a human-machine collaborative construction unsafe behavior identification system, including: a base station module, a helmet module, a standard calculation module and an early warning module:

[0011] The standard calculation module uses the base station module and the helmet module to continuously obtain the operation data of multiple moments in the human-machine collaborative construction, wherein the operation data at each moment respectively includes the posture of the mechanical equipment and the number of helmets of various types of personnel around the mechanical equipment; according to the change of the posture of the mechanical equipment, the duration of the data moment, and the change of the number of helmets of professional and technical personnel, the data between single operations are identified and removed from the operation data of the multiple moments; and according to the total duration of the appearance of multiple quantity combinations of various types of personnel helmets in the remaining operation data, each quantity combination is clustered, wherein the clustering results correspond to common behaviors, normal behaviors and unsafe behaviors respectively;

[0012] The early warning module is used to identify unsafe behaviors in human-machine collaborative construction according to the clustering results.

[0013] In summary, this embodiment provides a method and system for identifying unsafe behaviors in human-machine collaborative construction, which reflects the on-site operation situation through the posture of mechanical equipment and the number of various safety helmets, and identifies the non-operating period between single operations based on the posture changes of mechanical equipment, the length of the data time interval, and the changes in the number of safety helmets of professional technicians, and clusters and mines the data of the remaining operating periods to automatically construct a judgment standard for unsafe behaviors in human-machine collaborative construction. This method can carry out identification and early warning before unsafe behaviors occur, and improve the behavioral safety of human-machine collaborative tasks on the construction site. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0015] Figure 1 is a schematic diagram of a single-layer UWB base station unit provided by an embodiment of the present invention;

[0016] Figure 2 is a schematic diagram of a multi-layer UWB base station unit provided by an embodiment of the present invention;

[0017] Figure 3 is a schematic diagram of a helmet module provided by an embodiment of the present invention;

[0018] Figure 4 This is a workflow diagram of a data acquisition unit provided by an embodiment of the present invention;

[0019] Figure 5 It is a flow chart of a method for identifying unsafe behaviors in human-machine collaborative construction provided by an embodiment of the present invention;

[0020] Figure 6 is a flow chart of a complete working calculation method provided by an embodiment of the present invention;

[0021] Figure 7 This is a flow chart of another method for identifying unsafe behaviors in human-machine collaborative construction provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0023] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0024] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0025] In order to provide a method for advanced identification and early warning of unsafe behaviors in human-machine collaborative construction, the unsafe behaviors in human-machine collaborative construction are first explained. The "four major hazards" defined by OSHA (collision, fall, pinch and electric shock) are all potential hidden dangers in human-machine collaborative tasks. In the investigation report of prefabricated building accidents, 27% of the accidents are related to machinery. Investigations and studies on construction safety accidents at home and abroad have shown that unsafe human behavior is the most important cause of accidents. The accident cause tree shows that the incorrect coupling relationship between man and machine in time and space is an important prerequisite for the occurrence of accidents. Therefore, identifying and warning behaviors that cause incorrect time and space coupling relationships near high-risk construction machinery can effectively reduce the occurrence of human-machine collaborative accidents and improve the level of project safety management.

[0026] There are two types of behaviors that lead to wrong time and space coupling relationships in human-machine collaborative construction, namely, the behavior of non-collaborative personnel breaking into the high-risk area of ​​machinery and the behavior of key collaborative personnel leaving their posts without permission. The key to identifying such unsafe behaviors is to determine which people should appear in the high-risk area of ​​human-machine collaborative construction accidents during mechanical operation. In the prior art, the above behaviors are identified by pre-entering a white list of personnel through sensors represented by RFID. However, due to the different number of personnel required for different links of human-machine collaborative construction tasks, and a large number of mechanical equipment and equipment operators are provided by the lessor, a fixed white list is difficult to meet the dynamic personnel adjustment needs of the construction site, and the entry of the white list increases the cost of management personnel, resulting in the difficulty of implementation and promotion of the human-machine collaborative construction unsafe behavior identification and early warning method and device. Another method is to install cameras around mechanical equipment and identify such behaviors through data mining algorithms, but the cameras are severely affected by occlusion, and computer vision algorithms show low accuracy and recall rates when the environment is occluded, the light is poor, and the personnel overlap. These methods can only identify and warn after the unsafe behavior occurs, and it is difficult to give a warning before the unsafe behavior occurs.

[0027] In view of this, an embodiment of the present invention provides a method for identifying unsafe behaviors in human-machine collaborative construction. In order to illustrate the method, a system for identifying unsafe behaviors in human-machine collaborative construction that supports the implementation of the method is first described. The system includes four parts: a base station module, a safety helmet module, a standard calculation module, and an early warning module. Among them, the base station module is installed on the mobile machinery in the construction area. The base station module is used to obtain the posture information of the mobile machinery, and as a base station, obtain the distance between the safety helmet and the machinery; the safety helmet module is fixed to the safety helmet, and the safety helmet module and the base station user provide relative distance data through signal transmission; the standard calculation module is used to calculate the judgment criteria for unsafe behaviors of personnel in high-risk areas of human-machine collaborative construction through data mining; the early warning module is used to warn of the identified unsafe behaviors.

[0028] In a specific implementation, the base station module refers to a module installed on a mobile machine to obtain the relative distance between the helmet module and the base station, the helmet module number and the machine posture in real time, including a UWB base station unit and an inertial sensor unit.

[0029] Optionally, the UWB base station unit includes a base station bracket and a UWB base station installed on the machine. Since the workers in the human-machine collaborative construction area will be determined by horizontal distance in subsequent operations and the UWB signal has strong penetration ability, it will be interfered by vertical distance. For example, the machine and collaborative personnel are working on the first floor, and some workers are working on the second floor. At this time, the workers on the second floor will cause the human-machine distance data to be displayed as less than 5 meters (assuming that the operating radius of the machine on the first floor is 5 meters). However, since the operations on the first and second floors are unrelated, the workers on the second floor are not human-machine collaborative workers. Therefore, when identifying unsafe behaviors, it is necessary to filter out non-first-floor personnel. For projects with only single-layer construction operations, UWB base station units such as Figure 1 As shown, for projects with multi-layer construction operations, the UWB base station unit is as follows Figure 2 shown.

[0030] Furthermore, the UWB base station unit may include a UWB base station, a fixing frame and a wiring steel pipe, and the installation methods include base fixing and steel pipe bundling. Figure 1 and 2 There are two types of units in the system. The base station and the data computing unit can be fixed to the fixing frame by screws. The edge computing power in the base station and the data computing unit is connected via USB. The power is provided by the mechanical battery. The wires are routed through the steel pipe to connect the battery to the computing power and the UWB base station. Optionally, the steel pipe is at least 1 meter long to ensure that the bottom surface of the UWB base station closest to the ground is at least 1.4 meters vertically away from the horizontal plane where the machine is located. The difference between single-layer and multi-layer base station units lies in the number of UWB base stations. There is 1 base station for a single layer and 2 for a multi-layer, and the distance between the 2 base stations is at least 0.5 meters. The pedestal base station can be placed directly on the surface of the machine. The diameter of the base is 40 cm, and the stability of the base station can be ensured by placing a counterweight. The bundled base station can be bundled to the surface of the machine with steel wire. When bundling, try to ensure that the steel pipe is vertical to the ground.

[0031] Optionally, the inertial sensor unit includes an inertial sensor, a mobile power supply, and a 4G data transmission component. The inertial sensor unit is installed in key mechanical operating parts, such as the hook of a mobile crane, the gripper of a robotic arm, the lifting shelf of a forklift, and other parts directly related to the materials to be moved. The power supply is provided by the unit's own power supply, and the connection with the mechanical parts is magnetic or adhesive. The power supply must be charged before installation. When the UWB base station unit is powered on, a restart command is automatically issued to the inertial sensor unit through 4G, and the inertial sensor sends data to the edge computing power in the standard computing module every second. Optionally, the inertial sensor can be turned off by default after 8 hours of power-on. If the standard computing module is still working at this time, when it is found that there is no mechanical posture data input, a restart command is issued to the inertial sensor unit. The inertial sensor eliminates the accumulated error by restarting.

[0032] In a specific embodiment, the helmet module refers to a UWB receiver mounted on the helmet and a ring + UWB receiver box structure connecting the UWB receiver and the helmet, such as Figure 3 As shown. Optionally, the diameter of the collar is 22 cm. The collar and the UWB receiver box are made of plastic material and have a certain degree of elasticity. The UWB receiver is installed in the red UWB receiver box after being fully charged. The installation method is to insert it from the top. After the UWB receiver is installed on the helmet module, the receiver is placed on the back of the helmet. The helmet module is clamped on the helmet from top to bottom to ensure that the UWB receiver does not move when shaking. The UWB receiver can be directly ejected from the UWB receiver box from bottom to top. The UWB receiver must be fully charged before each use. The UWB receiver number must correspond to the color of the helmet. For example, numbers 1-20 correspond to white helmets, which are generally supervisors or Party A management personnel, numbers 21-50 correspond to blue helmets, which are generally professional and technical personnel on site, including equipment drivers, numbers 51-70 correspond to red helmets, which are generally general contractors and subcontract management personnel, and numbers 71-200 correspond to yellow helmets, which refer to ordinary workers on site.

[0033] In a specific embodiment, the standard calculation module refers to a module that controls data collection and calculation of standards for distinguishing unsafe behaviors in human-machine collaborative construction, including a micro industrial control host installed on a base station module and a data collection unit, a standard calculation unit, and a standard update unit running in the host.

[0034] Optionally, the micro industrial computer and the edge computing power of this embodiment are limited by the network. This embodiment chooses to control, calculate and update the judgment standard at the edge. The micro industrial computer requires at least 4G memory and at least 64G hard disk space. The length and width are required to be within 30 cm. The power supply is provided by mobile machinery. The industrial computer is installed on the fixed frame of the base station unit, and a water-proof and dust-proof protective cover is installed on the outside.

[0035] Optionally, the data acquisition unit refers to a unit that controls the UWB base station unit and the inertial sensor unit to collect data. The specific workflow is as follows: Figure 4 As shown, when the machine is turned on, the UWB base station unit is started, the micro industrial computer is started, the data acquisition module issues a control command, and the remote control inertial sensor is started through 4G transmission. The UWB base station and the inertial sensor will transmit the UWB receiver number data, human-machine distance data, and posture data through USB and 4G respectively. The human-machine distance data refers to the relative distance between the UWB base station and the UWB receiver, and the posture data refers to the acceleration data (storage format is acceleration x, acceleration y, acceleration z) and relative position data collected by the inertial sensor (the default coordinates at startup are (0,0,0), and the storage format is position x, position y, position z). After receiving the data, the data acquisition unit will determine whether the data from the UWB base station unit and the inertial sensor unit exist in seconds. If there is only data from the UWB base station unit, that is, it indicates that the data is incomplete, then a restart command is issued to the inertial sensor unit again. When the machine is powered off and stops working, the data acquisition unit automatically shuts down due to power failure.

[0036] Optionally, the standard calculation unit refers to a unit that calculates and updates the criteria for judging unsafe behaviors in human-machine collaboration. The unit supports a variety of machines. After the micro-industrial computer is powered on, there are three situations. In case 1, if the judgment standard has not been calculated yet, the calculation of the new judgment standard will be automatically started, and the judgment standard will be stored as judgment standard 1; in case 2, if there is already a judgment standard, it will identify a complete mechanical operation after it is turned on and update the judgment standard; in case 3, if the unit already has a judgment standard for machine 1, the unit is installed on a new type of machine. At this time, it is necessary to connect the edge computing power to the display and keyboard before installation. By entering a new digital number in the automatically opened terminal, the judgment standard will be switched. By default, a mobile computing power only supports up to 3 different types of machines. After the mobile computing power is powered on, it will automatically start and open the terminal. At this time, the judgment standard numbered 0 will be run (if there is no judgment standard, it will start recalculation as in case 1). If the user enters 1, 2 or 3, the discrimination standard of 2 or 3 will be renumbered to 0. If there is no corresponding discrimination standard for 1, 2 or 3, the same as case 1, the number will be changed to 0 and recalculated. If the user enters a number other than 1, 2 or 3, the edge computing power will automatically shut down. If the user enters a wrong number and the standard update is confused, for example, the standard for the machine should be number 2, but the judgment standard is confused because of the input of number 3, the user can enter list in the terminal. At this time, the data set corresponding to the current number 0 will be displayed. Each line is a complete mechanical operation, corresponding to the corresponding timestamp, and numbered in chronological order starting from a. The user can enter a letter number (such as b) to delete the corresponding data, and the numbered data corresponding to number 0 will also be deleted (for example, the newly added data in number 2 should be deleted in this example).

[0037] Furthermore, in order to realize the calculation and update of the judgment standard of unsafe behavior in human-machine collaboration, the standard calculation unit can execute the complete work calculation method and the judgment standard calculation method. The specific operations of the two methods will be described in detail in subsequent embodiments.

[0038] In a specific embodiment, the early warning module refers to a module that judges whether the real-time data is risky based on the latest discrimination standard and issues a warning through sound and text. The early warning module may include an early warning judgment unit and an early warning device. Among them, the early warning judgment unit is used to identify unsafe behaviors in human-machine collaborative construction and issue an early warning based on the discrimination standard calculated by the standard calculation module, and its specific operation will be described in detail in subsequent embodiments. The early warning device refers to a device that sends out early warning signals on site, including two types of voice signals and text signals. The voice signal is sent through a speaker and a buzzer. The device can be directly installed on the industrial computer of the standard calculation module, or it can be installed in a fixed position on the construction site, and the data is transmitted via 4G. The text signal is sent to the mobile terminal via 4G, and the user can view it through the APP or the APP in the instant messaging software (such as the WeChat applet).

[0039] Based on the above system, Figure 5 This is a flow chart of a method for identifying unsafe behaviors in human-machine collaborative construction provided by an embodiment of the present invention. This method is applicable to the situation of advanced identification and early warning of unsafe behaviors related to movement in human-machine collaborative construction scenarios in the construction industry. It can be executed by the standard calculation module and early warning module in the above system, or by other electronic devices. Figure 5 As shown, the method specifically includes:

[0040] S110, continuously acquiring operation data at multiple moments in the human-machine collaborative construction, wherein the operation data at each moment includes the posture of the mechanical equipment and the number of safety helmets of various types of personnel around the mechanical equipment.

[0041] This step continuously obtains human-machine distance data, machine posture data, and helmet number data at multiple moments, and calculates the number of helmets of various types of personnel within the operating radius R of the mechanical equipment as the data source of the entire method.

[0042] Optionally, in the above system, the data acquisition unit may control the UWB base station unit and the inertial sensor unit to collect the above data, and the specific operation has been described in detail in the above system embodiment. In other electronic devices, the above data may be obtained by other sensors, and this embodiment does not make specific restrictions. Table 1 exemplarily shows the acquired raw data:

[0043] Table 1 Original data record table

[0044] serial number <![CDATA[Starting time T i > <![CDATA[Acceleration A i > White blue red yellow 0 20240801091030 0.1 2 1 0 3 1 20240801091539 0.1 3 1 1 3 … … … … … … … t 20240802175601 3 3 2 2

[0045] Each line in Table 1 is the operation data at a certain time. The number is the time sequence number of each data. The starting time is T i This is the starting time of collecting the data, acceleration A iIt is the acceleration of the mechanical posture change collected by the inertial sensor. The white, blue, red, and yellow columns represent the number of helmets of each color, that is, the number of each type of personnel.

[0046] S120, based on the posture changes of the mechanical equipment, the duration of the data time, and the changes in the number of safety helmets worn by professional technicians, identify and remove the data between single operations from the operation data at the multiple time points.

[0047] Among them, a single operation refers to a basic operation unit of a mechanical equipment operation. For example, a mobile crane completes a lifting of a component, which means that it completes a complete single operation. The data between single operations refers to the non-operation data between two basic operation units. This step obtains the complete data of at least one basic operation unit by identifying and removing the non-operation data between single operations. When the above system executes the method of this embodiment, this step corresponds to the complete work calculation method in the system.

[0048] Specifically, for any high-risk machinery, after it is in place at a specific location on the site, there are two possible most common non-operation situations: there is no obvious data change in the inertial sensors of the key parts of the machinery, and the location data of professional technicians is collected from the base station module or no data of any role can be collected from the base station module. However, non-operation does not mean that a basic operation unit has been completed. It may only be a rest within the operation unit. The difference between the two is that the rest time within a single operation is generally much shorter than the rest time between single operations, because the next operation generally requires advance preparation time, such as the movement of materials. Therefore, in order to identify the basic operation unit, it is necessary to use time in the non-operation data to distinguish between the non-operation data within the group (that is, the non-operation data within a single operation) and the non-operation data between groups (the non-operation data between single operations).

[0049] In a specific implementation, the process of identifying and removing inactive data between groups may include the following steps:

[0050] Step 1: Based on the posture changes of the mechanical equipment, the length of the data time, and the changes in the number of helmets worn by professional technicians, identify at least one non-operating period from the operation data at the multiple times. Optionally, before executing this step, the user may be required to input the number of basic operation units N completed per day under normal circumstances. d (i.e. the average number of times a single task is completed in a single day), the preset inter-group rest time screening threshold T s seconds (i.e. the minimum rest time between (two) single operations), day and the minimum rest time T of the equipment during the day d seconds, and the acceleration judgment threshold θm / s used to judge whether the mechanical posture is stationary or active 2(Acceleration less than θ is static, and greater than or equal to θ is active). In the above system, this process can occur after the user inputs 1, 2, or 3. If the user does not input, the above thresholds can be defaulted to 5, 600s, 2400s, and 1m / s. 2 Based on the above thresholds, the following operations can be performed for each two adjacent moments: if two adjacent moments simultaneously meet the following three conditions, the previous moment t n and the moments before that when the mechanical devices are stationary (t n-2 ,t n-1 ) is identified as the starting time of a non-operating period, and the duration between the latter of the two adjacent moments and the starting time is identified as the duration of the non-operating period:

[0051] Condition 1: The posture of the mechanical equipment changes from static to active at the two adjacent moments;

[0052] Condition 2: The duration between the two adjacent time periods is greater than or equal to the minimum rest duration between single operations, and less than the minimum rest duration between days;

[0053] Condition three: The number of safety helmets worn by professional and technical personnel at the latter moment is greater than the number of safety helmets worn by professional and technical personnel when the mechanical equipment is first started on that day.

[0054] For example, if two adjacent moments t n and t n+1 The operation data meets the above three conditions. If at t n and t n Multiple moments before t n-4 ,t n-3 ,t n-2 ,t n-1 The period of time [t n-4 ,t n-3 ,t n-2 ,t n-1 ,t n ], the posture of the mechanical equipment is always static, then the first moment t n-4 Identify as the starting time of a non-operating period, and set t n+1 -t n-4 The duration of the idle time period is identified.

[0055] Furthermore, taking the original data shown in Table 1 as an example, the following screening formula can be constructed to screen the non-operating time period:

[0056] T i =T i-1 if maxA t-1<θand maxA t <θ(i=0,1,2,3…t) (1)

[0057] Tf k =T i -T i-1 for i∈{i|i>0 and maxA i-1 <θand maxA i ≥θand T i -T i-1 ≥T s and T i -T i-1 <T d and nb i -nb0≥0}θ(k=0,1,2,3…m) (2)

[0058] Among them, i represents the index that meets the condition, and k is the sequential index of i that meets the condition, that is, k = 1, 2, 3...m. t represents the time series number of the operation combination formed by each role within the operation radius R, as shown in the following table, where the color represents the color of the helmet, and the table corresponds to the number of helmets of the color worn within the radius R at that time. i is an array representing the starting time corresponding to each number, for example, in Table 1 below, T0=20240801091030. nb t Refers to the number of people wearing blue helmets within the operating radius R during the time period numbered t, and nb0 refers to the number of people wearing blue helmets within the operating radius R when the equipment is first turned on. Tf k Indicates the duration of each non-operating period. The calculated Tf k , date, number of helmets, etc. are stored in the MySQL database in the format shown in the following table, as shown in Table 2.

[0059] Table 2 Non-work schedule

[0060] serial number <![CDATA[Non - operating time (Tf k )]]> <![CDATA[Date (T t )]]> White blue red yellow 0 1245 20240801 2 2 0 0 1 805 20240801 2 2 1 0 … … … … … … …

[0061] The number of helmets of each color in Table 2 is T i-1 (Starting time of non-working period) The number of various types of safety helmets corresponding to the date is accurate to the day.

[0062] Step 2: Preliminarily determine whether there is a non-operating period within a single operation in the at least one non-operating period based on the number of times a single operation is completed within a single day; and based on the determination result, identify the data between single operations from the operation data at the multiple times.

[0063] Optionally, if the number of non-operating time periods is less than or equal to the set multiple of the average number of times a single operation is completed in a single day, it indicates that the newly added non-operating time period corresponding to the added day is unreasonable, and it is preliminarily determined that there is no non-operating time period within a single operation in the at least one non-operating time period, and the operation data within the at least one non-operating time period are all used as the data between single operations. The set multiple is in the interval [1,2), for example, 1.5.

[0064] If the number of non-operating time periods is greater than the set multiple of the average number of times a single operation is completed in a single day, it indicates that the newly added non-operating time periods corresponding to the added day are reasonable, and it is preliminarily determined that there is a non-operating time period within a single operation in the at least one non-operating time period. At this time, the duration of each non-operating time period is clustered, and whether the clustering is valid is determined based on the clustering silhouette coefficient; if valid, the clustering results correspond to the non-operating time period within a single operation and the non-operating time period between single operations, among which the one with a larger average duration corresponds to the non-operating time period between single operations, and the operation data in these non-operating time periods are all used as the data between single operations; if invalid, the operation data in the at least one non-operating time period are all used as the data between single operations.

[0065] Optionally, in the case where the complete work calculation method and the judgment standard calculation method (corresponding to S120-S140) in the above system are executed in a daily cycle according to the work data collected daily, the above step 2 can be performed by Figure 6 The flowchart shown is shown. Figure 6 As shown, after the machine is started, the network is used to determine the current time, obtain the date RD, and compare it with the last recorded T in the database (Table 2). t Compare and determine whether they are consistent. If they are consistent, it means that the machine has been started on that day and it is necessary to determine RD and S again. t Whether the time is consistent, where S t Refers to the date corresponding to the standard generated by the judgment standard calculation method; if they are consistent, there is no need to continue the clustering algorithm. If any of the above two judgments are inconsistent, it is necessary to further judge the number of non-working time N included in the last date with data in the database t and the preset number of basic operation units to be completed daily N d Is the ratio less than 1.5? If so, all newly added non-operating data on the last date with data (i.e. all Tf recorded on that day) will be directly k Data) Add Tr k This is the data set that will be input into the discrimination standard calculation algorithm later. If it is greater than 1.5, all Tf recorded on that day will be k The data is clustered by binary K-means, and the clustering silhouette coefficient S is calculated. c , if S cIf it is greater than or equal to 0.5, it means that the clustering is effective, and the inter-group data TK with a larger time after clustering is added to Tr k , if S c If it is less than 0.5, the clustering is invalid, and all the newly added non-operating data on the previous date with data (that is, all Tf recorded on that day) will still be included. k Data) Add Tr k When adding new data, Tf k The corresponding start time (T i ) and the number of helmets of each color are stored in the database, and finally a database containing Tr k The database is as follows:

[0066] Table 3 Non-operation time table between groups (i.e. single operation time)

[0067]

[0068] The original data covering the non-operation period between groups is the data between single operations.

[0069] Step 3: Remove the data between single operations from the original data, and the remaining operation data is the data within a single operation. When there are multiple non-operation periods between groups on the same day, the remaining operation data includes at least one complete single operation data.

[0070] S130. Clustering each quantity combination according to the total duration of appearance of multiple quantity combinations of safety helmets of various types of personnel in the remaining operation data, wherein the clustering results correspond to common behaviors, normal behaviors, and unsafe behaviors, respectively.

[0071] This step removes the data between single operations from the collected raw data, and the remaining data is the data within the operation period. Clustering the data within the operation period can reflect the basic characteristics of the operation behavior. When the above system executes the method of this embodiment, this step corresponds to the first half of the judgment standard calculation method in the system.

[0072] Optionally, if the remaining operation data includes a complete single operation data, each quantity combination is clustered according to the total duration of the appearance of multiple quantity combinations of safety helmets of various types of personnel in the remaining operation data. The clustering results correspond to common behaviors, normal behaviors, and unsafe behaviors, respectively, and the total duration of the three behaviors is gradually reduced. Among them, the behavior with the longest duration in the construction operation can correspond to the common behavior. Since this type of behavior occurs frequently, it is likely to be a safe operation behavior; the behavior with the shortest duration can correspond to the abnormal behavior. Since this type of behavior is relatively rare, it is likely to be an unsafe behavior; between common behaviors and unsafe behaviors, there is also a type of behavior that does not belong to unsafe behaviors despite a relatively short duration, such as the behavior of hanging a hook in a hoisting operation. Although it lasts for a short time, it is also a normal operation operation. This embodiment refers to this behavior as normal behavior.

[0073] Optionally, in the above-mentioned system, the complete work calculation method and the judgment standard calculation method (corresponding to S120-S140) are executed in a daily cycle based on the work data collected daily. The remaining work data of the day can be added to the cumulative database of the daily remaining work data. The database must include a complete single work data. The quantity combinations in the database can be clustered according to the total duration of the appearance of multiple quantity combinations of safety helmets of various types of personnel in the database.

[0074] In a specific implementation, the number of each group of helmet colors can be regarded as a work combination, and the frequency (in times) and total duration (in seconds) of various work combinations in the remaining operation data are counted to form another database as shown in the following table:

[0075] Table 4 Work combination frequency and duration

[0076] serial number White blue red yellow Frequency F Duration 0 2 2 1 2 31 2123 1 2 2 2 2 22 698 … … … … … … 355

[0077] Then, the outlier detection algorithm is used to perform K-means three-category clustering on the duration data except for the absolute value of the Z-score within the set threshold (such as 3), and the clustering silhouette coefficient S is calculated. j If S j If S is greater than or equal to 0.5, it means that the clustering is effective, and the clustering results correspond to common behaviors, normal behaviors, and unsafe behaviors. j If it is less than 0.5, the construction of this discrimination standard is abandoned, and the previous discrimination standard formed is selected for subsequent warnings.

[0078] S140. Identify unsafe behaviors in human-machine collaborative construction based on the clustering results.

[0079] This step compares the newly collected data with the clustering results to identify unsafe behaviors in the new data. When the above system executes the method of this embodiment, this step corresponds to the second half of the judgment standard calculation method in the above system, and the specific operation of the early warning judgment unit.

[0080] Taking the above system as an example, first, the standard calculation unit performs the last step of the judgment standard calculation method and determines the warning standard according to the clustering result. The class with the smallest average total duration in the clustering result can be used as the duration warning set T s ; Take the median of the frequency of each quantity combination in the remaining operation data, and take the quantity combination set less than the median as the frequency warning set F s More specifically, when clustering is effective, the work combination corresponding to the category with the smallest average duration is used as one of the early warning criteria, which is called the early warning duration criterion, that is, T s The frequency data corresponding to the work combination is median, and the frequency less than the median is used as another warning standard, called the warning frequency standard, that is, F s If clustering is invalid, the last formed criterion is selected as T s and F s If there is no standard before, then T s and F s Is empty.

[0081] Then, the early warning judgment unit calculates the s and F s , identify unsafe behaviors in human-machine collaborative construction. Optionally, for newly collected data, if the number of helmets of various types of personnel at the time to be identified belongs to both the duration warning set and the frequency warning set, the human-machine collaborative construction behavior at the time to be identified is identified as an unsafe behavior. Optionally, if the work combination has never existed in the database (Table 4), an early warning is also issued, and the specific formula is as follows:

[0082]

[0083] Among them, X(j) is the work combination change data set constructed according to the time sequence, j is the corresponding number, and the j value is increased by 1 regardless of whether the work combination changes after the machine is turned off and then turned on, as shown in the following table:

[0084] Table 5X(j) Schematic table

[0085] serial number White blue red yellow state 1 2 2 1 2 Power on 2 2 2 2 2 run 3 2 2 2 3 Shutdown 4 2 2 2 3 Power on … … … … … …

[0086] τ(X(j)) refers to the duration of the work combination X(j), Refers to the frequency of the work combination X(j). The above-mentioned early warning judgment method can be installed in the industrial computer of the standard computing module.

[0087] In addition, when the above system executes the method of this embodiment, the entire method can also be combined with Figure 7 The flowchart shown is understood. Figure 7 In the process, the modules of the system work together to complete the three steps of building identification standards, early warning and updating identification standards. The work content of each step is as follows:

[0088] (1) Constructing identification standards: This step is used to construct standards for the advance identification and early warning of unsafe behaviors in human-machine collaborative construction. It includes three sub-steps: equipment installation, identification of human-machine collaboration, and calculation of identification standards.

[0089] a) Equipment installation on site: After the rented mobile machinery enters the human-machine collaboration area, before starting the operation, first install the UWB base station unit in the base station module on the machinery. The power supply is provided by the battery of the machinery and starts with the start of the machinery; then install the inertial sensor unit in the base station module at the location where the machinery is operating, such as the hook of the mobile crane, the gripper of the mechanical arm, the lifting shelf of the forklift and other locations directly related to the materials to be moved. The power supply is provided by the unit's own power supply, and the connection method is magnetic suction or adhesive. Turn on the inertial sensor during installation; finally, install UWB receivers in the safety helmets of all workers in the construction project. The installation method is mounting, and turn on the UWB receiver during installation. When the equipment completes construction tasks in different projects and different areas, except for the first time involving equipment installation, the rest only needs to complete charging and calibration.

[0090] b) Identify human-machine collaboration: This embodiment is used to identify unsafe behaviors during human-machine collaboration. It does not include data when the equipment is resting on site. Therefore, it is necessary to screen the human-machine distance data during human-machine collaborative construction. This is done through the base station module. First, the base station module will only power the UWB base station after the equipment is turned on. Therefore, if the equipment is only placed statically in the project operation area, there will be no UWB data; secondly, the inertial sensor unit in the base station module is used to obtain the acceleration and relative position coordinate data of the key parts of the equipment to determine whether the key parts are stationary; finally, the human-machine collaboration identification unit is used to screen the complete one-time operation time data of the equipment.

[0091] c) Calculation of identification standards: After identifying that the newly entered machine has completed a complete human-machine collaborative work, start calculating the identification standards for unsafe behaviors in human-machine collaborative construction. This step is completed through the standard calculation module, and the calculation results are input into the early warning module.

[0092] (2) Advance warning: This step is used to identify unsafe behaviors in human-machine collaborative construction in real time, including behaviors of non-collaborative personnel entering high-risk areas of machinery and behaviors of key collaborative personnel leaving their posts without permission. It includes two sub-steps, namely, identification of human-machine collaboration and advance warning.

[0093] a) Identify human-machine collaboration: First, the base station module is used to determine whether the human-machine collaboration task has started. If the human-machine collaboration task has not been identified to have stopped after it has started, the recognition result is always that the task is in progress. Then, the base station module is used to collect the distance between the worker and the base station and the UWB receiver number on the corresponding worker's helmet, and the data is transmitted to the early warning module.

[0094] b) Advance warning: Based on the judgment standard of unsafe behavior in human-machine collaborative construction calculated in the "Building Identification Standards" step, the input human-machine distance and UWB number data are analyzed, and the early warning module is used to issue early warnings for non-collaborative personnel breaking into the high-risk area of ​​the machinery and key collaborative personnel leaving their posts without permission.

[0095] (3) Updating identification standards: This step is used to update the warning standards for unsafe behaviors in human-machine collaborative construction obtained in the “building identification standards” step, thereby improving the accuracy of advance warning. It includes two sub-steps, namely, identifying human-machine collaboration and advance warning.

[0096] a) Identification of human-machine collaboration: First, the base station module determines whether the human-machine collaboration task has completed a new work task, that is, the start of the recognition task to the end of the recognition task; then the incremental human-machine distance data and UWB receiver number data compared with the end time of the previous task are transmitted to the standard calculation module.

[0097] b) Update identification standards: After the incremental human-machine collaborative task data is input, the calculation of the human-machine collaborative construction unsafe behavior identification standards is updated. The specific algorithm is the same as the step of building the identification standard, only the calculated data is different. The data includes the original data and the incremental data. This step is completed through the standard calculation module, and the calculation results are updated and input into the early warning module.

[0098] like Figure 1-3 As shown, an embodiment of the present invention also provides a system for identifying unsafe behaviors in human-machine collaborative construction. The standard calculation module in the system uses the base station module and the safety helmet module to continuously obtain the operation data of multiple moments in human-machine collaborative construction, wherein the operation data at each moment respectively includes the posture of the mechanical equipment, and the number of safety helmets of various types of personnel around the mechanical equipment; according to the changes in the posture of the mechanical equipment, the duration between data moments, and the changes in the number of safety helmets of professional and technical personnel, the data between single operations is identified and removed from the operation data at the multiple moments; and according to the total duration of the appearance of multiple quantity combinations of safety helmets of various types of personnel in the remaining operation data, each quantity combination is clustered, wherein the clustering results correspond to common behaviors, normal behaviors and unsafe behaviors, respectively. The early warning module is used to identify unsafe behaviors in human-machine collaborative construction based on the clustering results.

[0099] It should be noted that, due to the three-dimensional spatial relationship between personnel and machinery, the present invention provides two types of mechanical UWB base station units suitable for single-layer construction operations and multi-layer construction operations and corresponding personnel identification units in high-risk accident areas. The risk area warned by this embodiment is limited to the same horizontal plane and is not applicable to cross-floor risk areas. For example, when the present invention is applied to a mobile crane, it can warn of unsafe human-machine collaboration behaviors on the ground floor, but it cannot warn of unsafe human-machine collaboration behaviors on the second floor and above caused by hoisting objects.

[0100] The system in this embodiment and the above method embodiment are based on the same inventive concept, and any limitations in the above method embodiment are applicable to this embodiment.

[0101] In summary, this embodiment provides a method and system for identifying unsafe behaviors in human-machine collaborative construction, which realizes advanced identification and early warning of unsafe behaviors in human-machine collaborative construction based on ultra-wideband, inertial sensors, data mining technology and edge computing hardware. Specifically, the key to identifying and warning the behavior of non-collaborative personnel breaking into the high-risk area of ​​machinery and the behavior of key collaborative personnel leaving their posts lies in two points, namely, automatically constructing the judgment standard of unsafe behaviors in the high-risk area of ​​human-machine collaborative construction accidents and real-time identification and early warning of unsafe behaviors. This embodiment installs UWB sensors in safety helmets, installs UWB base stations and inertial sensors on mobile machinery, and reflects the on-site operation conditions by collecting data such as the posture of mechanical equipment and the number of various safety helmets. It also identifies the non-operating period between single operations based on the posture changes of mechanical equipment, the duration of the data time interval, and the changes in the number of safety helmets of professional and technical personnel, and clusters and mines the data of the remaining operation periods to automatically construct the judgment standard of unsafe behaviors in human-machine collaborative construction. Compared with the RFID method, this embodiment greatly improves the warning accuracy on the one hand, and solves the problem of needing to preset a whitelist on the other hand; compared with the computer vision method, this embodiment solves the problem of low warning accuracy and recall rate caused by ambient lighting, environmental occlusion and personnel overlap. In addition, this embodiment can carry out identification and warning before unsafe behaviors occur, which can be better promoted and implemented in construction practice, and improve the behavioral safety of human-machine collaborative tasks on construction sites.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying unsafe behaviors in human-machine collaborative construction, characterized in that: include: Continuously obtain operation data at multiple moments in human-machine collaborative construction, where the operation data at each moment includes the posture of the mechanical equipment and the number of safety helmets of various personnel around the mechanical equipment; According to the posture change of the mechanical equipment, the duration of the data time, and the change in the number of helmets of the professional technicians, the data between single operations are identified and removed from the operation data at the multiple time points; According to the total duration of multiple quantity combinations of safety helmets of various types of personnel in the remaining operation data, each quantity combination is clustered, where the clustering results correspond to common behaviors, normal behaviors, and unsafe behaviors respectively; According to the clustering results, unsafe behaviors in human-machine collaborative construction are identified.

2. The method according to claim 1, characterized in that The method of identifying the data of a single operation from the operation data at multiple moments according to the posture change of the mechanical equipment, the duration of the data moment, and the change in the number of helmets of the professional technicians includes: According to the posture change of the mechanical equipment, the duration of the data time, and the change in the number of safety helmets of professional technicians, at least one non-operating period is identified from the operation data at the multiple time points; Preliminarily determining whether there is a non-operation period within a single operation in the at least one non-operation period according to the number of times a single operation is completed within a single day; If it exists, cluster the duration of each non-operating period, and the clustering results correspond to the non-operating periods within a single operation and between single operations respectively; the operation data within the non-operating period between single operations are all regarded as the data between single operations; If it does not exist, the operation data in the at least one non-operation period will be regarded as the data of a single operation.

3. The method according to claim 2, characterized in that The step of identifying at least one non-operating period from the operation data at the plurality of times according to the posture change of the mechanical equipment, the duration of the data time, and the change in the number of safety helmets of the professional technicians includes: For every two adjacent moments, if the two adjacent moments simultaneously meet the following conditions, the first moment of the first moment of the two adjacent moments and the first moment of the multiple moments before which the mechanical equipment is stationary are identified as the starting moment of a non-operating period, and the duration between the second moment of the two adjacent moments and the starting moment is identified as the duration of the non-operating period: Condition 1: The posture of the mechanical equipment changes from static to active at the two adjacent moments; Condition 2: The duration between the two adjacent time periods is greater than or equal to the minimum rest duration between single operations, and less than the minimum rest duration between days; Condition three: The number of safety helmets worn by professional and technical personnel at the latter moment is greater than the number of safety helmets worn by professional and technical personnel when the mechanical equipment is first started on that day.

4. The method according to claim 2, characterized in that: The preliminarily determining whether there is a non-operating period within a single operation in the at least one non-operating period according to the number of times a single operation is completed within a single day includes: If the number of non-operating time periods is greater than a set multiple of the average number of times a single operation is completed in a single day, it is preliminarily determined that there is a non-operating time period within a single operation in the at least one non-operating time period; Otherwise, preliminarily determining that the at least one non-operating time period does not include a non-operating time period within a single operation; The set multiple is in the interval [1,2).

5. The method according to claim 2, characterized in that: The duration of each non-operating period is clustered, and the clustering results correspond to the non-operating periods within a single operation and between single operations, respectively, including: The duration of each non-operating period is clustered into two categories, and the effectiveness of the clustering is determined based on the clustering silhouette coefficient; If it is effective, the non-operating time period with a larger average duration will be used as the non-operating time period between single operations.

6. The method according to claim 1, characterized in that The identifying unsafe behaviors of human-machine collaborative construction according to the clustering results includes: The cluster with the smallest average total duration in the clustering results is used as the duration warning set; Taking the median of the frequency of occurrence of each quantity combination in the remaining operation data, and taking the quantity combination set smaller than the median as the frequency warning set; If the number combination of safety helmets of various types of personnel at the time to be identified belongs to both the duration warning set and the frequency warning set, the human-machine collaborative construction behavior at the time to be identified is identified as an unsafe behavior.

7. The method according to claim 1, characterized in that The continuously acquiring operation data at multiple moments in the human-machine collaborative construction includes: Through the UWB receivers fixed on the safety helmets of various personnel and the UWB base stations installed on the mechanical equipment, the number of safety helmets of various personnel within the operating radius of the mechanical equipment at multiple times during human-machine collaborative construction is continuously obtained; By installing inertial sensors on mechanical equipment, the posture of the mechanical equipment at multiple times during human-machine collaborative construction can be continuously obtained.

8. The method according to claim 1, characterized in that The clustering of each quantity combination according to the total duration of appearance of multiple quantity combinations of safety helmets of various types of personnel in the remaining operation data includes: If the remaining operation data includes a complete single operation data, each quantity combination is clustered according to the total duration of appearance of multiple quantity combinations of safety helmets of various types of personnel in the remaining operation data.

9. The method according to claim 1, characterized in that: The identifying and removing the data between single operations from the operation data at multiple times according to the posture change of the mechanical equipment, the length of the data time, and the change in the number of safety helmets of the professional and technical personnel includes: identifying and removing the data between single operations from the operation data at multiple times of the day according to the posture change of the mechanical equipment, the length of the data time, and the change in the number of safety helmets of the professional and technical personnel; Correspondingly, the method clusters each quantity combination according to the total duration of the appearance of multiple quantity combinations of safety helmets of each type of personnel in the remaining operation data, including: adding the remaining operation data of the day to the cumulative database of daily remaining operation data, and clustering each quantity combination according to the total duration of the appearance of multiple quantity combinations of safety helmets of each type of personnel in the cumulative database.

10. A human-machine collaborative construction unsafe behavior identification system, characterized in that: include: Base station module, helmet module, standard calculation module and early warning module: The standard calculation module uses the base station module and the helmet module to continuously obtain the operation data of multiple moments in the human-machine collaborative construction, wherein the operation data at each moment respectively include the posture of the mechanical equipment and the number of helmets of various types of personnel around the mechanical equipment; according to the change of the posture of the mechanical equipment, the duration of the data moment, and the change of the number of helmets of professional and technical personnel, the data between single operations are identified and removed from the operation data of the multiple moments; and according to the total duration of the appearance of multiple quantity combinations of helmets of various types of personnel in the remaining operation data, each quantity combination is clustered, wherein the clustering results correspond to common behaviors, normal behaviors and unsafe behaviors respectively; The early warning module is used to identify unsafe behaviors in human-machine collaborative construction according to the clustering results.

Images