A multi-source fusion near-electric construction area dynamic safety early warning method and system

Abstract

A multi-source fusion near electric construction area dynamic safety early warning method and system, comprising: dividing the construction area into different voltage levels based on an electric field sensor, and obtaining the positions of each live body; using a laser radar to detect and calculate the positions and speeds of reflectors, and comparing the positions and speeds of each person and equipment identified by a camera, matching the reflectors with the identified persons and equipment; calculating the relative angle, relative angle change rate, distance, distance change rate and direction transformation entropy, and calculating the invasion probability of persons and equipment; and performing hierarchical early warning according to the number of persons, the number of equipment and the invasion probability of persons and equipment in the area of different voltage levels. The present application considers the influence of the uncertainty of the running track on the safety evaluation, realizes hierarchical early warning by combining the area voltage level with the invasion probability, and improves the early warning accuracy of construction safety management.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of near electric construction early warning, and more particularly relates to a multi-source fusion near electric construction area dynamic safety early warning method and system. BACKGROUND

[0002] In the field of near electric construction, especially in the work carried out in high-voltage and super-high-voltage live areas, personnel and equipment mistakenly entering dangerous areas is one of the main risks that can lead to serious safety accidents. Traditional safety warning methods, such as physical isolation barriers, manual monitoring, or single distance measuring alarms, have significant drawbacks: they cannot dynamically perceive the real-time positions and motion trends of multiple targets in complex environments, making it difficult to quantify risks and even more difficult to provide differentiated intelligent warnings based on voltage levels and behavior patterns. They are easily disturbed by the environment and cannot comprehensively judge intentions.

[0003] CN119600786A proposes a near electric construction work safety early warning method and system, which belongs to the field of near electric construction work equipment warning. It uses three-dimensional laser scanning to scan power transmission and transformation engineering construction sites to form a high-precision laser point cloud three-dimensional data model. In the three-dimensional laser point cloud model, live bodies and non-live bodies are quickly divided, and the safety distance range is attribute-hung to the live bodies. RTK positioning sensors are installed on the workers and construction machinery, and Beidou RTK is used to achieve high-precision real-time positioning and spatial distance measurement of workers and machinery. Based on the safety distance detection algorithm, the real-time spatial distance between workers, machinery, and live bodies is monitored, and different modes of reminders such as safety warning and prohibition alarm are set according to the safety operation regulations to ensure that workers, machinery, and other equipment maintain a sufficient safety distance from live bodies. However, this invention uses positioning sensors to locate all personnel and equipment, which is costly and cannot provide early warnings if new personnel enter or non-construction personnel mistakenly enter. Moreover, it only considers the current position and does not consider motion trends, intentions, etc., resulting in a single risk assessment dimension and an inability to predict trends. SUMMARY

[0004] To address the deficiencies in the prior art, the present application provides a multi-source fusion near electric construction area dynamic safety early warning method and system.

[0005] The present application employs the following technical solutions.

[0006] The first aspect of the present application proposes a multi-source fusion near electric construction area dynamic safety early warning method, comprising:

[0007] dividing the construction area into different voltage levels according to the electric field intensity of each electric field sensor measuring point and the electric field intensity threshold of each voltage level set, and obtaining the positions of each live body;

[0008] The camera is used to identify each person and equipment in the construction area and the position and speed of each person and equipment through a target recognition algorithm; laser radar detection is performed and the position and speed of the reflector are calculated, which are compared with the position and speed of each person and equipment identified by the camera, the reflector is matched with the identified each person and equipment, and the position and speed of each person and equipment measured by the laser radar are obtained;

[0009] According to the position and speed of each person and equipment measured by the laser radar at the current moment and the position of each live body, the relative angle, the relative angle change rate, the distance and the distance change rate of each person and equipment and each live body at the current moment are calculated, and the direction transformation entropy of each person and equipment is calculated according to the speed of each person and equipment measured by the laser radar in a long-term period; the intrusion probability of each person and equipment is calculated according to the relative angle, the relative angle change rate, the distance, the distance change rate and the direction transformation entropy;

[0010] According to the number of persons and the number of equipment in the area of different voltage levels and the intrusion probability of persons and equipment, a hierarchical early warning is performed.

[0011] Preferably, the position of each live body is obtained, specifically:

[0012] According to the electric field intensity of each electric field sensor measuring point, the electric field gradient vector of each electric field sensor measuring point is calculated and normalized, and each electric field sensor measuring point is clustered according to the normalized electric field gradient vector; for all normalized electric field gradient vectors of each class, the charge quantity and position of each corresponding live body are calculated by using a single-source positioning algorithm.

[0013] Preferably, the objective function of the single-source positioning algorithm is:

[0014]

[0015] wherein, is the u th class of the clustering; is the th normalized electric field gradient vector in the v th class; is the electric constant; is the charge quantity of the u th live body; is the distance from the th electric field sensor measuring point to the v th live body in the u th class; is the direction vector of the th electric field sensor measuring point to the v th live body in the u th class; is the square of the norm.

[0016] Preferably, the camera identifies each person and equipment in the construction area and the position and speed of each person and equipment through a target recognition algorithm; laser radar detection is performed and the position and speed of the reflector are calculated, specifically as follows:

[0017] The camera is a binocular camera, and the target type recognition is performed through a target recognition algorithm. The target type includes a person and equipment. The z-axis coordinate of the target is the focal length of the camera multiplied by the camera baseline and then divided by the pixel parallax of the target in the left and right images of the camera. The x-axis coordinate of the target is the difference between the pixel horizontal coordinate of the target in the left image and the pixel horizontal coordinate of the camera optical center in the image, multiplied by the z-axis coordinate divided by the focal length of the camera. The y-axis coordinate of the target is the difference between the pixel vertical coordinate of the target in the left image and the pixel vertical coordinate of the camera optical center in the image, multiplied by the z-axis coordinate divided by the focal length of the camera. The x, y and z-axis coordinates of the target are located in a coordinate system with the optical center of the binocular camera as the reference center. The speed is calculated by dividing the position change of the target between two time points by the frame length. The speed is a three-dimensional vector, and each element represents the speed of a coordinate axis.

[0018] The laser radar directly obtains the distance between the target and the laser radar by calculating the round-trip time of the reflection signal, records the azimuth angle and the pitch angle of each laser pulse, and calculates the three-dimensional coordinates of the reflector according to the distance between the target and the laser radar, the azimuth angle and the pitch angle. The three-dimensional coordinates are located in a Cartesian coordinate system with the phase center of the laser radar as the reference center. The speed is calculated by dividing the position change of the target between two time points by the time interval.

[0019] The position and speed of each person and equipment recognized by the camera and the position and speed of the reflector measured by the laser radar are converted into the same coordinate system.

[0020] Preferably, the reflector is corresponded to the recognized each person and equipment, specifically as follows:

[0021] After the position and speed of the reflector are normalized, they are spliced into a same reflector vector. The reflectors in the same cluster are the same target. The position and speed corresponding to the cluster center are obtained. The comprehensive consistency between each cluster center and all the persons and equipment recognized by the camera is calculated.

[0022] The comprehensive consistency is: the Euclidean distance between the position of the cluster center and the position of all the persons and equipment recognized by the camera at the same time is calculated as the position consistency; the Euclidean distance between the speed of the cluster center and the speed of all the persons and equipment recognized by the camera at the same time is calculated as the speed consistency; the position consistency and the speed consistency are normalized and then weighted and summed according to the set weight as the comprehensive consistency.

[0023] All the overall consistency is combined into a matrix as the cost matrix. The Hungarian algorithm is used to obtain the personnel or equipment that match each cluster center. If the cost between a cluster center and the personnel or equipment matched by the Hungarian algorithm exceeds the cost threshold, the cluster center is removed. The position and velocity of each cluster center are used as the position and velocity of the corresponding matched personnel or equipment measured by the lidar.

[0024] Preferably, the relative angle, the rate of change of the relative angle, the distance, and the rate of change of the distance are as follows:

[0025] For a person or device and a charged object, obtain the direction vector of the person's or device's velocity and the direction vector of the person or device pointing towards the charged object, with the included angle being the angle between the direction vector of the velocity and the direction vector pointing towards the charged object; the included angle The range is relative angle for ;

[0026] The rate of change of the relative angle is the relative angle at the current moment minus the relative angle at the previous moment, and then divided by the time interval.

[0027] Distance refers to the distance between a person or equipment and a live conductor;

[0028] The rate of change of distance is the distance at the current moment minus the distance at the previous moment, divided by the time interval.

[0029] Preferably, the step of calculating the directional transformation entropy of personnel and equipment based on the speeds of personnel and equipment measured by the lidar within a set long-term period specifically involves:

[0030] For a piece of equipment or a person, Divide the data into multiple angle intervals; obtain the angle interval containing the angle between the person or equipment and the charged body at each moment within a set long-term period, and count the number of changes in the angle interval containing the angle between any two adjacent moments within the set long-term period; calculate the direction transformation entropy. for:

[0031]

[0032] in, The included angle is the first n The angle range changes to the first... i Conditional probability of a range of angles. The included angle is the first i The probability of change within a range of angles. , , The angle between two adjacent moments from the first n The angle range changes to the first...i the number of times of changing from the first angle interval to the second angle interval, the number of times of changing from the second angle interval to the third angle interval, the number of times of changing from the third angle interval to the fourth angle interval, and the number of times of changing from the fourth angle interval to the fifth angle interval. i the number of times of changing from the first angle interval to the second angle interval, the number of times of changing from the second angle interval to the third angle interval, the number of times of changing from the third angle interval to the fourth angle interval, and the number of times of changing from the fourth angle interval to the fifth angle interval. m the number of times of changing from the first angle interval to the second angle interval, the number of times of changing from the second angle interval to the third angle interval, the number of times of changing from the third angle interval to the fourth angle interval, and the number of times of changing from the fourth angle interval to the fifth angle interval. n the number of times of changing from the first angle interval to the second angle interval, the number of times of changing from the second angle interval to the third angle interval, the number of times of changing from the third angle interval to the fourth angle interval, and the number of times of changing from the fourth angle interval to the fifth angle interval. m the number of times of changing from the first angle interval to the second angle interval, the number of times of changing from the second angle interval to the third angle interval, the number of times of changing from the third angle interval to the fourth angle interval, and the number of times of changing from the fourth angle interval to the fifth angle interval. the number of times of changing from the first angle interval to the second angle interval, the number of times of changing from the second angle interval to the third angle interval, the number of times of changing from the third angle interval to the fourth angle interval, and the number of times of changing from the fourth angle interval to the fifth angle interval.

[0033] Preferably, the intrusion probability of the personnel and the equipment is calculated according to the relative included angle, the relative included angle change rate, the distance, the distance change rate, and the direction transformation entropy, specifically:

[0034]

[0035] wherein, is the intrusion probability; is the intrusion factor; , , , , are all set weights; , , , , are the relative included angle, the relative included angle change rate, the distance, the distance change rate, and the direction transformation entropy, respectively.

[0036] Preferably, the hierarchical early warning is performed according to the personnel density, the equipment density, and the intrusion probability of the personnel and the equipment in the area of different voltage levels, specifically:

[0037] The voltage levels are divided into high risk, medium risk, low risk, and no risk; different personnel density thresholds and equipment density thresholds are set for the areas of different voltage levels; the personnel density and the equipment density of the areas of different voltage levels are calculated according to the positions of the personnel and the equipment measured by the laser radar;

[0038] If the personnel density in the area of high risk is greater than the corresponding personnel density threshold, or the equipment density is greater than the corresponding equipment density threshold, a high-level early warning is performed;

[0039] The personnel density of the area of other voltage levels outside the high risk is multiplied by the maximum value of the intrusion probability of all personnel to all live bodies in the corresponding area, and if the calculation result is greater than the corresponding personnel density threshold, or the equipment density of the area of other voltage levels outside the high risk is multiplied by the maximum value of the intrusion probability of all equipment to all live bodies in the corresponding area, and if the calculation result is greater than the corresponding equipment density threshold, a low-level early warning is performed.

[0040] The second aspect of the application proposes a multi-source fusion near electric construction area dynamic safety warning system based on the method of the first aspect of the application, comprising: a region division and live body positioning module, a target monitoring module, an intrusion probability calculation module and a hierarchical warning module, specifically:

[0041] The region division and live body positioning module is used to divide the construction area into different voltage levels according to the electric field intensity of each electric field sensor measuring point and the set electric field intensity threshold of each voltage level, and obtain the position of each live body;

[0042] The target monitoring module uses a camera to identify each person and equipment in the construction area through a target recognition algorithm, as well as the position and speed of each person and equipment; laser radar detection is performed and the position and speed of the reflector are calculated, which are compared with the position and speed of each person and equipment identified by the camera, the reflector is matched with the identified each person and equipment, and the position and speed of each person and equipment measured by the laser radar are obtained;

[0043] The intrusion probability calculation module calculates the relative angle, relative angle change rate, distance and distance change rate between each person and equipment and each live body at the current time according to the position and speed of each person and equipment measured by the laser radar at the current time and the position of each live body, and calculates the direction transformation entropy of each person and equipment according to the speed of each person and equipment measured by the laser radar in a long-term period; the intrusion probability of each person and equipment is calculated according to the relative angle, relative angle change rate, distance, distance change rate and direction transformation entropy;

[0044] The hierarchical warning module performs hierarchical warning according to the number of personnel, the number of equipment and the intrusion probability of personnel and equipment in the area of different voltage levels.

[0045] The beneficial effects of the application are that, compared with the prior art, the application accurately divides the voltage level area and locates the live body through the electric field sensor, ensures accurate identification of the hazard source; the type of personnel and equipment is efficiently obtained through camera visual recognition, the position and speed information of the personnel and equipment are obtained by matching the more accurate laser radar detection with the visual recognition result, the accuracy of target positioning and motion state tracking is improved; the intrusion probability is calculated based on multi-dimensional dynamic parameters (relative angle, distance and its change rate, direction transformation entropy), which enhances the comprehensiveness and adaptability of safety evaluation, and the direction transformation entropy considers the influence of the uncertainty of the running track on safety evaluation; finally, hierarchical warning is realized by combining the voltage level of the area and the intrusion probability, which improves the warning accuracy of construction safety management. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 The flowchart of the application;

[0047] Figure 2A laser radar map used in the present application. DETAILED DESCRIPTION

[0048] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions will be described clearly and completely below with reference to the accompanying drawings of the embodiments of the present application. The embodiments described in the present application are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those skilled in the art without creative efforts based on the spirit of the present application shall fall within the protection scope of the present application.

[0049] As shown in Figure 1 , a first aspect of the present application proposes a multi-source fusion near electric construction area dynamic safety early warning method, comprising:

[0050] S1, according to the electric field intensity of each electric field sensor measuring point and the set electric field intensity threshold of each voltage level, the construction area is divided into different voltage levels, and the positions of each charged body are obtained;

[0051] Specifically, the voltage level is divided into high risk, medium risk, low risk and no risk; three electric field intensity thresholds are set; when the electric field intensity is greater than or equal to the first electric field intensity threshold, it is considered to be in high risk; when the electric field intensity is less than the first electric field intensity threshold and greater than or equal to the second electric field intensity threshold, it is considered to be in medium risk; when the electric field intensity is less than the second electric field intensity threshold and greater than or equal to the third electric field intensity threshold, it is considered to be in low risk; when the electric field intensity is less than the third electric field intensity threshold, it is considered to be in no risk. The three electric field intensity thresholds are 10kV / m, 8.0kV / m and 3.0kV / m respectively.

[0052] In the embodiment, the positions of each charged body are preferably obtained as follows:

[0053] According to the electric field intensity of each electric field sensor measuring point, the electric field gradient vector of each electric field sensor measuring point is calculated and normalized, and each electric field sensor measuring point is clustered according to the normalized electric field gradient vector; for all normalized electric field gradient vectors of each class, the charge quantity and position of each corresponding charged body of the class are calculated by using a single-source positioning algorithm.

[0054] In the embodiment, the objective function of the single-source positioning algorithm is preferably as follows:

[0055]

[0056] wherein, is the i-th normalized electric field gradient vector in the j-th class; u is the i-th normalized electric field gradient vector in the j-th class; v is the i-th normalized electric field gradient vector in the j-th class; is the electrostatic constant;​​ the charge amount of the u the distance from the v u the direction vector of the v u the square of the norm.

[0057] In addition, it needs to be explained that if there are specific drawings and geographic information including live equipment information on the site, the position of the live equipment can be directly determined using these information, but the live equipment on the construction site may change due to rain and construction progress, so the position of the live equipment needs to be obtained according to the electric field intensity, and the electric field sensor used in the application is a high-precision flexible sensor.

[0058] S2, using a camera to identify each person and equipment in the construction area and the position and speed of each person and equipment through a target recognition algorithm; performing laser radar detection and calculating the position and speed of the reflector, and comparing the position and speed of each person and equipment identified by the camera, matching the reflector with the identified each person and equipment, and obtaining the position and speed of each person and equipment measured by the laser radar;

[0059] Preferably, in the embodiment, the camera is used to identify each person and equipment in the construction area and the corresponding position and speed through a target recognition algorithm; laser radar detection is performed, and the position and speed of the reflector are calculated according to the received reflection signal, specifically:

[0060] The camera is a binocular camera, and target type recognition is performed through a target recognition algorithm. The target type includes personnel and equipment. The z-axis coordinate of the target is the focal length of the camera multiplied by the camera baseline and then divided by the pixel disparity of the target in the left and right images of the camera. The x-axis coordinate of the target is the difference between the pixel horizontal coordinate of the target in the left image and the pixel horizontal coordinate of the camera optical center in the image, multiplied by the z-axis coordinate divided by the focal length of the camera. The y-axis coordinate of the target is the difference between the pixel vertical coordinate of the target in the left image and the pixel vertical coordinate of the camera optical center in the image, multiplied by the z-axis coordinate divided by the focal length of the camera. The x, y and z axis coordinates of the target are all located in the coordinate system with the optical center of the binocular camera as the reference center. The speed is calculated by dividing the position change of the target between two time points by the frame length. The speed is a three-dimensional vector, and each element represents the speed of a coordinate axis.

[0061] The laser radar is like Figure 2 ​​​​​​​As shown, the laser radar directly obtains the distance between the target and the laser radar by calculating the round-trip time of the reflected signal, records the azimuth angle and the elevation angle of each laser pulse, calculates the three-dimensional coordinates of the reflector according to the distance between the target and the laser radar, the azimuth angle and the elevation angle, the three-dimensional coordinates are all located in a Cartesian coordinate system with the phase center of the laser radar as the reference center, and the speed is calculated by dividing the position change of the target between two time points by the time interval;

[0062] The positions and speeds of each person and equipment recognized by the camera and the positions and speeds of the reflector measured by the laser radar are converted into the same coordinate system.

[0063] Specifically, the target recognition algorithm used in the embodiment is YOLOv8.

[0064] It should be noted that the positions and speeds mentioned in the embodiment below are all in the same coordinate system; the speeds calculated in the embodiment are all average speeds calculated by dividing the position change of the target between two time points by the time interval, and are not instantaneous speeds; the detection of the radar on the coordinates is more accurate than that of the binocular camera, but direct speed measurement can only measure the radial speed, so the change of the coordinates is used to calculate the speed.

[0065] Preferably, the correspondence between the reflector and the recognized each person and equipment is specifically:

[0066] The positions and speeds of the reflectors are normalized and spliced into the same reflector vector, all reflectors are clustered according to the reflector vector, the reflectors in the same cluster are the same target, the positions and speeds corresponding to the cluster centers are obtained, and the comprehensive consistency between each cluster center and each person and equipment recognized by all cameras is calculated.

[0067] It should be noted that the clustering adopts K-medoids clustering, and the cluster center is a sample, that is, one of the reflectors, so the position and speed corresponding to the cluster center can be obtained.

[0068] The comprehensive consistency is: the Euclidean distance between the position of the cluster center and the position of each person and equipment recognized by all cameras at the same time is calculated as the position consistency;

[0069] The Euclidean distance between the speed of the cluster center and the speed of each person and equipment recognized by all cameras at the same time is calculated as the speed consistency;

[0070] The position consistency and the speed consistency are normalized and weightedly summed according to the set weight as the comprehensive consistency;

[0071] All the comprehensive consistency is combined into a matrix as a cost matrix, and the personnel or equipment matched with each cluster center is obtained by using the Hungarian algorithm, and if the cost between the cluster center and the personnel or equipment matched by the Hungarian algorithm exceeds the cost threshold, the cluster center is removed; and the position and speed of each cluster center are taken as the position and speed of the corresponding matched personnel or equipment measured by the laser radar.

[0072] It should be noted that if there is an ultra-high voltage area that does not allow access in the vicinity of the electric field, the viewing angle of some of the binocular cameras can be adjusted to the voltage level area of the ultra-high voltage area, and if a person or equipment is identified in the ultra-high voltage area, an alarm is directly given; in special cases (for example, when personnel perform high-voltage facility maintenance and equipment such as unmanned aerial vehicles perform inspection), personnel or equipment need to enter the ultra-high voltage area, the personnel must be professional personnel with protective equipment, and the equipment must be special equipment certified by the absolute recognition, and a positioning sensor needs to be installed on the equipment to detect the trajectory, and if the trajectory is different from the set safety path, an alarm is given.

[0073] S3, according to the position and speed of each personnel and equipment measured by the laser radar at the current time and the position of each live body, the relative angle, the relative angle change rate, the distance and the distance change rate of the personnel and equipment and each live body at the current time are calculated, and the direction transformation entropy of the personnel and equipment is calculated according to the speed of each personnel and equipment measured by the laser radar in a long-term period; the invasion probability of the personnel and equipment is calculated according to the relative angle, the relative angle change rate, the distance, the distance change rate and the direction transformation entropy;

[0074] In the embodiment, the relative angle, the relative angle change rate, the distance and the distance change rate are preferably:

[0075] For a person or equipment and a live body, the direction vector of the speed of the person or equipment and the direction vector of the person or equipment pointing to the live body are obtained, and the included angle between the direction vector of the speed and the direction vector pointing to the live body is the included angle; the included angle is in the range of ; the relative angle is ;

[0076] The relative angle change rate is the relative angle at the current time minus the relative angle at the last time divided by the time interval;

[0077] The distance is the distance between the person or equipment and the live body;

[0078] The distance change rate is the distance at the current time minus the distance at the last time divided by the time interval.

[0079] The embodiment preferably calculates the direction transformation entropy of the personnel and the equipment according to the speed of each personnel and equipment measured by the laser radar in a set long-term period, specifically:

[0080] For one equipment or personnel, the direction transformation entropy is calculated as: dividing a plurality of angle intervals; obtaining the angle interval in which the included angle between the personnel or equipment and the live body at each time in a set long-term period is located, and counting the number of changes of the angle interval in which the included angle between two adjacent times in the set long-term period is located; and calculating the direction transformation entropy as:

[0081]

[0082] wherein, is the conditional probability that the included angle changes from the i-th angle interval to the j-th angle interval, n i is the probability that the included angle is in the i-th angle interval, i are respectively the number of times that the included angle between two adjacent times changes from the i-th angle interval to the j-th angle interval, from the i-th angle interval to the j-th angle interval, and from the i-th angle interval to the j-th angle interval; n i i m n m is the total number of angle intervals.

[0083] The embodiment preferably calculates the intrusion probability of the personnel and the equipment according to the relative included angle, the relative included angle change rate, the distance, the distance change rate and the direction transformation entropy, specifically:

[0084]

[0085] wherein, is the intrusion probability; is the intrusion factor; are all set weights; are respectively the relative included angle, the relative included angle change rate, the distance, the distance change rate and the direction transformation entropy.

[0086] ​​​​​​​​​​​​​​​​​​​It should be noted that a small relative angle indicates that the person or equipment is moving directly towards the charged object. A negative rate of change indicates that the direction of movement is becoming increasingly direct towards the charged object, and the larger the negative value of the rate of change, the faster the rate of change of direction. A small distance from the charged object indicates that the person or equipment is close to the charged object, and the probability of intrusion is high. A large negative value of the distance indicates that the person or equipment is rapidly approaching the charged object. The direction change entropy indicates the randomness of the direction change during the movement of the person or equipment. A large direction change entropy indicates that the movement is chaotic rather than following a regular trajectory. Therefore, the direction change entropy indicates that it is more likely to be suspicious behavior and may be more likely to cause accidental intrusion. Specifically... , , , , The value can be obtained by training based on historical data, where the distance weights are... Weight greater than the relative angle It has a higher weight than the others.

[0087] S4. Classified early warning is issued based on the number of people and equipment in areas with different voltage levels, as well as the probability of intrusion by people and equipment.

[0088] The tiered early warning system, based on the population density, equipment density, and intrusion probability of personnel and equipment in areas with different voltage levels, specifically includes:

[0089] Voltage levels are categorized as high-risk, medium-risk, low-risk, and no-risk; different personnel density thresholds and equipment density thresholds are set for areas with different voltage levels; personnel density and equipment density are calculated for areas with different voltage levels based on the locations of personnel and equipment detected by lidar.

[0090] If the population density or equipment density in a high-risk area exceeds the corresponding population density threshold, a high-level warning will be issued.

[0091] Calculate the personnel density of areas other than high-risk voltage levels by multiplying it by the maximum probability of all personnel in the corresponding area intruding into all live conductors. If the calculated result is greater than the corresponding personnel density threshold; or calculate the equipment density of areas other than high-risk voltage levels by multiplying it by the maximum probability of all equipment in the corresponding area intruding into all live conductors. If the calculated result is greater than the corresponding equipment density threshold, then issue a low-level warning.

[0092] It should be noted that the threshold for high risk is lower than that for medium risk, the threshold for medium risk is lower than that for low risk, and the threshold for no risk should be infinite. In actual operation, it is set to a relatively large value.

[0093] Embodiment 2 of the present application proposes a multi-source fusion near electric construction area dynamic safety early warning system based on the method described in embodiment 1 of the present application, comprising: a region division and live body positioning module, a target monitoring module, an intrusion probability calculation module and a hierarchical early warning module, specifically:

[0094] The region division and live body positioning module is used to divide the construction area into different voltage levels according to the electric field intensity of each electric field sensor measuring point and the set electric field intensity threshold of each voltage level, and obtain the position of each live body;

[0095] The target monitoring module uses a camera to identify each person and equipment in the construction area through a target recognition algorithm, as well as the position and speed of each person and equipment; performs laser radar detection and calculates the position and speed of the reflector, and compares the position and speed of each person and equipment identified by the camera, matches the reflector with the identified each person and equipment, and obtains the position and speed of each person and equipment measured by the laser radar;

[0096] The intrusion probability calculation module calculates the relative angle, relative angle change rate, distance and distance change rate of each person and equipment and each live body at the current time according to the position and speed of each person and equipment measured by the laser radar at the current time and the position of each live body, and calculates the direction transformation entropy of each person and equipment according to the speed of each person and equipment measured by the laser radar in the set long-term period; and calculates the intrusion probability of each person and equipment according to the relative angle, relative angle change rate, distance, distance change rate and direction transformation entropy;

[0097] The hierarchical early warning module performs hierarchical early warning according to the number of personnel, the number of equipment and the intrusion probability of each person and equipment in the area of different voltage levels.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and not to limit it, although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced, without departing from the spirit and scope of the present application, any modification or equivalent replacement thereof should be covered within the protection scope of the claims of the present application.

Claims

1. A method for dynamic safety early warning of a near-electric construction area by multi-source fusion, characterized in that, The method comprises the following steps: According to the electric field intensity of each electric field sensor measuring point and the set electric field intensity threshold value of each voltage level, the construction area is divided into different voltage levels, and the positions of each live body are obtained; Using a camera, each person and equipment in the construction area is identified by a target recognition algorithm, and the positions and speeds of each person and equipment are obtained; laser radar detection is performed, and the positions and speeds of the reflecting bodies are calculated; the positions and speeds of each person and equipment identified by the camera are compared with the positions and speeds of the reflecting bodies calculated by the laser radar, and the reflecting bodies are matched with the identified persons and equipment, so that the positions and speeds of each person and equipment measured by the laser radar are obtained; According to the positions and speeds of each person and equipment measured by the laser radar at the current time and the positions of each live body, the relative angle, the relative angle change rate, the distance and the distance change rate between each person and equipment and each live body at the current time are calculated, and the direction transformation entropy of each person and equipment is calculated according to the speed of each person and equipment measured by the laser radar in a long-term period; the invasion probability of each person and equipment is calculated according to the relative angle, the relative angle change rate, the distance, the distance change rate and the direction transformation entropy; According to the number of persons and equipment in the area of different voltage levels and the invasion probability of persons and equipment, a hierarchical early warning is performed.

2. The multi-source fusion near-electricity construction area dynamic safety early warning method according to claim 1, wherein: The positions of each live body are obtained by the following steps: According to the electric field intensity of each electric field sensor measuring point, the electric field gradient vector of each electric field sensor measuring point is calculated and normalized, and each electric field sensor measuring point is clustered according to the normalized electric field gradient vector; for all normalized electric field gradient vectors of each class, the single-source positioning algorithm is used to calculate the charge quantity and position of each live body corresponding to each class.

3. The multi-source fusion near-electricity construction area dynamic safety early warning method according to claim 2, wherein: The objective function of the single-source positioning algorithm is: wherein, is the first normalized electric field gradient vector of the u cluster; is the first normalized electric field gradient vector of the cluster; v is the first normalized electric field gradient vector of the cluster; is the charge amount of the first u charged body; is the distance from the first electric field sensor measurement point to the first v charged body; u is the direction vector of the first electric field sensor measurement point to the first charged body; v is the direction vector of the first u electric field sensor measurement point to the first charged body; is the square of the norm.

4. The multi-source fusion near-electricity construction area dynamic safety early warning method according to claim 1, wherein: Each person and equipment in the construction area is identified by a target recognition algorithm using a camera, and the corresponding positions and speeds are obtained; laser radar detection is performed, and the positions and speeds of the reflecting bodies are calculated by the following steps: The camera is a binocular camera, and the target type recognition is performed by a target recognition algorithm; the target type includes persons and equipment; the z-axis coordinate of the target is the focal length of the camera multiplied by the camera baseline and then divided by the pixel parallax of the target in the left and right images of the camera; the x-axis coordinate of the target is the difference between the pixel horizontal coordinate of the target in the left image and the pixel horizontal coordinate of the camera optical center, multiplied by the z-axis coordinate and then divided by the camera focal length; the y-axis coordinate of the target is the difference between the pixel vertical coordinate of the target in the left image and the pixel vertical coordinate of the camera optical center, multiplied by the z-axis coordinate and then divided by the camera focal length; the x, y and z-axis coordinates of the target are located in the coordinate system with the optical center of the binocular camera as the reference center; the speed is calculated by dividing the position change of the target between two time points by the frame length; the speed is a three-dimensional vector, and each element represents the speed of a coordinate axis. ​ The laser radar directly obtains the distance between the target and the laser radar by calculating the round-trip time of the reflected signal, records the azimuth angle and the elevation angle of each laser pulse, and calculates the three-dimensional coordinates of the reflector according to the distance between the target and the laser radar, the azimuth angle and the elevation angle, wherein the three-dimensional coordinates are located in a Cartesian coordinate system with the phase center of the laser radar as the reference center, and the speed is calculated by dividing the position change of the target between two time points by the time interval; The positions and speeds of the personnel and equipment recognized by the camera and the positions and speeds of the reflectors measured by the laser radar are converted into the same coordinate system.

5. The multi-source fusion near-electric construction area dynamic safety warning method according to claim 4, wherein: The corresponding relationship between the reflectors and the recognized personnel and equipment is as follows: The positions and speeds of the reflectors are normalized and spliced into a same reflector vector, all reflectors are clustered according to the reflector vector, the reflectors in a same cluster are a same target, the position and speed corresponding to the cluster center are obtained, and the comprehensive consistency between each cluster center and the personnel and equipment recognized by all cameras is calculated; The comprehensive consistency is calculated by taking the Euclidean distance between the position of the cluster center and the positions of the personnel and equipment recognized by all cameras at a same time as the position consistency, taking the Euclidean distance between the speed of the cluster center and the speeds of the personnel and equipment recognized by all cameras at a same time as the speed consistency, and taking the weighted sum of the position consistency and the speed consistency after normalization as the comprehensive consistency according to a set weight; All comprehensive consistencies are combined into a matrix as a cost matrix, the personnel or equipment matched with each cluster center is obtained through the Hungarian algorithm, if the cost between the cluster center and the personnel or equipment matched by the Hungarian algorithm exceeds a cost threshold, the cluster center is removed, and the position and speed of each cluster center are taken as the position and speed of the corresponding matched personnel or equipment measured by the laser radar.

6. The multi-source fusion near-electric construction area dynamic safety warning method according to claim 1, wherein: The relative included angle, the relative included angle change rate, the distance and the distance change rate are as follows: For a person or device and a live body, obtaining a direction vector of a speed of the person or device, and a direction vector of the person or device pointing to the live body, and an included angle between the direction vector of the speed and the direction vector pointing to the live body; the included angle ranges from ; relative angle to ; The relative included angle change rate is the current relative included angle minus the relative included angle at the previous time point and then divided by the time interval; The distance is the distance between the personnel or equipment and the live body; The distance change rate is the current distance minus the distance at the previous time point and then divided by the time interval.

7. The multi-source fusion near-electric construction area dynamic safety warning method according to claim 6, wherein: The direction transformation entropy of the personnel and equipment is calculated according to the speeds of the personnel and equipment measured by the laser radar in a set long-term period, and the direction transformation entropy is calculated as follows: For a device or personnel, the angle between the device or personnel and the live body is obtained dividing multiple angle intervals; obtaining the angle interval where the angle between the device or personnel and the live body is located at each time in a set long-term period, and counting the number of changes of the angle interval where the angle between two adjacent times in the set long-term period is located; calculating the direction transformation entropy is: wherein, is the conditional probability that the angle changes from the n th angle interval to the i th angle interval, is the probability that the angle changes from the i th angle interval to the th angle interval, , are the number of times that the angle changes from the n th angle interval to the i th angle interval, from the i th angle interval to the m th angle interval, from the n th angle interval to the m th angle interval, respectively, at two adjacent time instants; is the total number of angle intervals.

8. The multi-source fusion near-electric construction area dynamic safety warning method according to claim 7, wherein: The intrusion probability of the personnel and equipment is calculated according to the relative included angle, the relative included angle change rate, the distance, the distance change rate and the direction transformation entropy, and the intrusion probability is calculated as follows: wherein, is an intrusion probability; is an intrusion factor; , , , , are each a set weight; , , , , are each a relative angle, a relative angle change rate, a distance, a distance change rate, and a direction change entropy, respectively.

9. The multi-source fusion near-electric construction area dynamic safety warning method according to claim 8, wherein: The personnel density, equipment density, and invasion probability of personnel and equipment in the area according to different voltage levels are graded and prewarned, specifically: The voltage levels are divided into high risk, medium risk, low risk, and no risk; different personnel density thresholds and equipment density thresholds are set for areas with different voltage levels; the personnel density and equipment density in areas with different voltage levels are calculated according to the positions of each personnel and equipment measured by the laser radar; If the personnel density in a high-risk area is greater than the corresponding personnel density threshold, or the equipment density is greater than the corresponding equipment density threshold, a high-level prewarning is performed; The personnel density in areas with voltage levels other than high risk is multiplied by the maximum value of the invasion probability of all personnel to all live bodies in the corresponding area, and if the calculation result is greater than the corresponding personnel density threshold; or the equipment density in areas with voltage levels other than high risk is multiplied by the maximum value of the invasion probability of all equipment to all live bodies in the corresponding area, and if the calculation result is greater than the corresponding equipment density threshold, a low-level prewarning is performed.

10. A multi-source fusion-based near-electric construction area dynamic safety early warning system based on the method of any one of claims 1-9, comprising: The area division and live body positioning module, target monitoring module, invasion probability calculation module, and graded prewarning module are characterized in that: The area division and live body positioning module: used to divide the construction area into different voltage levels according to the electric field intensity of each electric field sensor measuring point and the set electric field intensity threshold of each voltage level, and obtain the positions of each live body; The target monitoring module: uses a camera to identify each personnel and equipment in the construction area, as well as the positions and speeds of each personnel and equipment, through a target recognition algorithm; performs laser radar detection and calculates the positions and speeds of reflectors, and compares them with the positions and speeds of each personnel and equipment identified by the camera, matches the reflectors with the identified each personnel and equipment, and obtains the positions and speeds of each personnel and equipment measured by the laser radar; The invasion probability calculation module: according to the positions and speeds of each personnel and equipment measured by the laser radar at the current time and the positions of each live body, calculates the relative angle, relative angle change rate, distance, and distance change rate between personnel and equipment and each live body at the current time, and according to the speed of each personnel and equipment measured by the laser radar in a long-term period, calculates the direction transformation entropy of personnel and equipment; according to the relative angle, relative angle change rate, distance, distance change rate, and direction transformation entropy, the invasion probability of personnel and equipment is calculated; The graded prewarning module: according to the number of personnel, the number of equipment, and the invasion probability of personnel and equipment in areas with different voltage levels, graded prewarning is performed.

Images