Campus safety risk prevention and control system and method based on AI and data analysis

Abstract

The invention discloses a campus safety risk prevention and control system and method based on AI and data analysis, and relates to the technical field of campus safety risk prevention and control, and the method comprises the steps: building a three-dimensional model of a campus place, and dividing the three-dimensional model into all first risk regions, second risk regions and feature risk regions; obtaining a target moving object in the monitoring picture according to the inspection route and the monitoring equipment; extracting a local area image of the target moving object in the monitoring picture, and determining a risk value of the target moving object to the inspection robot; and determining the risk degree of the target moving object to the inspection robot, and further determining the opportunity of the inspection robot for performing early warning prompt on the target moving object. According to the invention, the potential collision risk is pre-judged in advance through the actual movement track change of the student in the campus and in combination with the analysis of the monitoring equipment, and the moving object is prompted in advance, so that the common collision problem of the inspection robot in the inspection process is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of campus safety risk prevention and control, and specifically to a campus safety risk prevention and control system and method based on AI and data analysis. Background Art

[0002] Campus safety is related to the personal safety of every teacher and student and is the top priority of school work. Nowadays, inspection robots have quietly entered the campus and, with their intelligent and efficient characteristics, undertake the important task of safety patrol. They shuttle tirelessly through every corner of the campus, contributing to the protection of campus safety.

[0003] Due to the large number of students on campus, there will be situations where students do not notice the inspection robot that is patrolling and thus collide with the inspection robot. Currently, the commonly used anti-collision analysis is usually to install sensors, such as radar sensors, on the inspection robot. When a student enters the sensing range, a warning prompt is given to the student. However, the existing warning analysis does not fully consider the actual movement trajectory changes of students on campus and cannot predict potential collision risks in advance, resulting in frequent collision accidents and making it difficult to fundamentally solve the collision problem between robots and students on campus. Summary of the Invention

[0004] The purpose of the present invention is to provide a campus safety risk prevention and control system and method based on AI and data analysis to solve the problems raised in the prior art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions: A campus safety risk prevention and control method based on AI and data analysis, comprising the following steps: Step S100: Establish a three-dimensional model of the campus site, and divide all first risk areas in the three-dimensional model; obtain the collision events where the inspection robot has collided with the moving object in history, extract the corresponding second risk areas of each collision event in the three-dimensional model, and take the area where the first risk area intersects with the second risk area as the characteristic risk area; Step S200: According to the set inspection route of the inspection robot and the monitoring devices installed on campus, obtain the first warning range and the second warning range during the movement of the inspection robot, and combine the position changes of the moving object in the monitoring video to obtain the target moving object; Step S300: Extract the local area image corresponding to the local part of the target moving object in the monitoring video, and determine the risk value of the target moving object to the inspection robot according to the change of the movement trajectory of the local part; Step S400: Determine the risk level of the target moving object to the patrol robot based on the risk value and various risk areas in the three-dimensional model, and determine the timing of the patrol robot to give a warning prompt to the target moving object according to the risk level.

[0006] Further, step S100 includes: obtaining the corner position corresponding to the area at the corner in the campus venue, and taking the area with a radius of R around the corner position as the first risk area; Step S100 also includes: obtaining a number of collision events where the patrol robot has collided with a moving object in the past. A collision event is an event where the patrol robot comes into contact with a moving object during the patrol process and triggers the collision alarm of the patrol robot; extracting the moving object corresponding to the collision event, as well as the collision time, collision position, and the sensing range corresponding to the sensor deployed on the patrol robot when the collision occurs. The sensing range is a circular range; Step S120: Establish a three-dimensional model of the campus venue. According to the collision time T 1 corresponding to a certain historical collision event and the moving object A, obtain the moment when starting from the collision time T 1 and moving forward for the first time to satisfy that the position P A of the moving object A is at the edge of the sensing range corresponding to the deployed sensor, and take this moment as the warning time T 2 ; Obtain the moving distance S 2 of the moving object A between the time T 1 and the time T A , and obtain the moving speed V A =S A / (T 1 -T 2 ). If V A <V 1 and the patrol robot gives a warning prompt at the time T 2 , and V 1 is the first speed threshold, then take all the sensing ranges corresponding to the sensor between the time T 2 and the time T 1 as the second risk area, and obtain all the corresponding second risk areas in the three-dimensional model; If the moving speed of the moving object is less than the first speed threshold at this time, it means that the moving speed of the moving object is normal at this time. However, if a collision still occurs with the patrol robot in this case, it means that this part of the area on the campus is an area prone to collisions, such as places that are not easy to observe, such as the entrance and exit of the teaching building, the dormitory corridor, and the intersection of campus roads.

[0007] Further, step S200 includes: Step S210: Take the sensing range corresponding to the sensors deployed on the inspection robot as the first warning range, and the first obstacle avoidance range is the range with the position P of the inspection robot as the center and a radius of r. 0 And take the range with a radius of r with the position P as the center. 1 Also, take the range on both sides of Q meters along the inspection route starting from the position P as the first warning range. 0 That is, the range on both sides of Q meters along the inspection route starting from the position P is also taken as the first warning range. 1 And take the range on both sides of L meters along the inspection route starting from the position P as the first warning range. 1 That is, the range on both sides of L meters along the inspection route starting from the position P is also taken as the first warning range; Take the range with the position P of the inspection robot as the center and a radius of r as the second warning range. 0 And take the range with a radius of r with the position P as the center. 2 Also, take the range on both sides of Q meters along the inspection route starting from the position P as the second warning range. 0 That is, the range on both sides of Q meters along the inspection route starting from the position P is also taken as the second warning range. 2 And take the range on both sides of L meters along the inspection route starting from the position P as the second warning range. 2 That is, the range on both sides of L meters along the inspection route starting from the position P is also taken as the second warning range, where r < r, Q < Q, L < L, and divide the first warning range and the second warning range in the monitoring screen; 1 Here, the first warning range is the minimum range for the inspection robot to issue a prompt, and the second warning range is the maximum range for the inspection robot to issue a prompt. That is, when a moving object enters the first warning range, the probability of collision with the inspection robot increases greatly at this time, and a warning prompt must be issued. The second warning range is the standard for making an early prompt. When the risk level obtained by analyzing the moving object is relatively high, an early prompt needs to be made within the second warning range to prompt the moving object as early as possible to avoid late prompting and collision; 2 Step S220: Real-time obtain the position of a certain moving object in the monitoring screen, and detect whether the certain moving object enters the second warning range. If it enters the second warning range, take the moment immediately before the moment T when it first enters the second warning range as T. 1 The position of the certain moving object at the moment T is P. 2 The position at the moment T is P. 1 Take the direction from P to P as the moving direction; 2 According to the moving distance of a certain moving object within the time period D before the moment T, obtain the moving speed V of the certain moving object; if V > V and the intersection position of the moving direction and the inspection route is within the second warning range, and V is the second speed threshold, then take the certain moving object as the target moving object. Here, the first warning range is the minimum range for the inspection robot to issue a prompt, and the second warning range is the maximum range for the inspection robot to issue a prompt. That is, when a moving object enters the first warning range, the probability of collision with the inspection robot increases greatly at this time, and a warning prompt must be issued. The second warning range is the standard for making an early prompt. When the risk level obtained by analyzing the moving object is relatively high, an early prompt needs to be made within the second warning range to prompt the moving object as early as possible to avoid late prompting and collision; Step S220: Real-time obtain the position of a certain moving object in the monitoring screen, and detect whether the certain moving object enters the second warning range. If it enters the second warning range, take the moment immediately before the moment T when it first enters the second warning range as T. 3 The position of the certain moving object at the moment T is P. 4 The position at the moment T is P. 3 Take the direction from P to P as the moving direction; 3 According to the moving distance of a certain moving object within the time period D before the moment T, obtain the moving speed V of the certain moving object; if V > V and the intersection position of the moving direction and the inspection route is within the second warning range, and V is the second speed threshold, then take the certain moving object as the target moving object. 4 The position of the certain moving object at the moment T is P. 4 The position at the moment T is P. 4 Take the direction from P to P as the moving direction; 3 According to the moving distance of a certain moving object within the time period D before the moment T, obtain the moving speed V of the certain moving object; if V > V and the intersection position of the moving direction and the inspection route is within the second warning range, and V is the second speed threshold, then take the certain moving object as the target moving object. According to the moving distance of a certain moving object within the time period D before the moment T, obtain the moving speed V of the certain moving object; if V > V and the intersection position of the moving direction and the inspection route is within the second warning range, and V is the second speed threshold, then take the certain moving object as the target moving object. 3 According to the moving distance of a certain moving object within the time period D before the moment T, obtain the moving speed V of the certain moving object; if V > V and the intersection position of the moving direction and the inspection route is within the second warning range, and V is the second speed threshold, then take the certain moving object as the target moving object. 2 If V > V and the intersection position of the moving direction and the inspection route is within the second warning range, and V is the second speed threshold, then take the certain moving object as the target moving object. 2 If V > V and the intersection position of the moving direction and the inspection route is within the second warning range, and V is the second speed threshold, then take the certain moving object as the target moving object.

[0008] Further, step S300 includes: Step S310: Extract the monitoring screen R corresponding to a certain moment T within the time period D of the target moving object T , and extract the monitoring screen R T . In the monitoring screen R, extract the local area image corresponding to the local part of the target moving object; perform horizontal and vertical edge detection on the local area image through an edge detection algorithm to obtain the edge contour of the corresponding local part in the local area image; According to the monitoring screen R T , establish a two-dimensional coordinate system, randomly obtain the coordinates of several pixel points from the edge contour, and respectively obtain the average abscissa i and the average ordinate j according to the abscissa and ordinate of each pixel point, and obtain the characteristic coordinate C T (i, j) corresponding to a certain moment T; the monitoring device is a fixed-view and fixed-position bullet camera, and then mark the characteristic coordinates at each moment within the time period D according to the corresponding local area images in the monitoring screens at each moment within the time period D in the two-dimensional coordinate system; Step S320: Respectively use the characteristic coordinates corresponding to three consecutive moments a, b, and c within the time period D as C a , C b , and C c ; use the vector from C a to C b as V ab , and the vector from C b to C c as V bc , and then obtain the eigenvalue , where e is the natural exponent, x is the included angle between the vector V ab and the vector V bc , max() is to find the maximum value, min() is to find the minimum value, K 1 and K 2 are the first characteristic coefficient and the second characteristic coefficient respectively; if the eigenvalue M is less than the preset characteristic threshold, take the moment a as the characteristic moment, and according to the total number N of moments within the time period D and the number N 1 of the characteristic moments among them, obtain the risk value Y = N 1 / (N - 2).

[0009] Further, step S400 includes: Step S410: Consider the areas on campus that are not any risk areas as the first type of area with a weight of W 1 , the first risk area and the second risk area as the second type of area with a weight of W 2 , and the characteristic risk area as the third type of area with a weight of W 3 ; consider the position P3 Connect to the position P 0 to obtain a line segment L. According to the length ratio of various regions in the line segment L to the total length of the line segment L, the risk level of the target moving object to the inspection robot is obtained as , where Y is the risk value, and W q is the weight corresponding to the q-th type of region, and R q is the length ratio occupied by the q-th type of region; Step S420: If the risk level Y 0 is greater than the risk level threshold, a prompt is given at time T 3 ; otherwise, after the target moving object enters the second warning range, the risk value and the length ratio occupied by various regions are obtained again in real time, and the risk level of the target moving object to the inspection robot is obtained again, and the time when the risk level is greater than the risk level threshold is prompted; if no prompt is given before entering the first obstacle avoidance range, a prompt is given when just entering the first obstacle avoidance range.

[0010] Here, in obtaining the risk value and the length ratio occupied by various regions again in real time, it is obtained according to the real-time moment after entering the second warning range. Among them, the monitoring images at each moment within the acquisition period D are also obtained from the monitoring images within the period D intercepted forward starting from the real-time moment.

[0011] A campus security risk prevention and control system based on AI and data analysis includes a risk area division module, a target moving object determination module, a risk value calculation module, and a warning prompt timing determination module; Risk area division module: used to establish a three-dimensional model of the campus site and divide all first risk areas in the three-dimensional model; obtain collision events where the inspection robot has collided with moving objects in the past, extract the corresponding second risk areas of each collision event in the three-dimensional model, and use the intersecting area of the first risk area and the second risk area as the characteristic risk area; Target moving object determination module: used to obtain the first warning range and the second warning range of the inspection robot during movement according to the set inspection route of the inspection robot and the monitoring devices installed on the campus, and combine the position changes of the moving objects in the monitoring images to obtain the target moving objects; Risk value calculation module: used to extract the local area image corresponding to the local part of the target moving object in the monitoring image, and determine the risk value of the target moving object to the inspection robot according to the change of the moving trajectory of the local part; Early warning prompt timing determination module: used to determine the risk level of the target moving object to the inspection robot according to the risk value and various risk areas in the three-dimensional model, and determine the timing of the inspection robot to give an early warning prompt to the target moving object according to the risk level.

[0012] Furthermore, the risk area division module includes a collision event extraction unit and a risk area division unit; Collision event extraction unit: used to obtain the collision events of the inspection robot colliding with moving objects in history; extract the moving objects corresponding to the collision events, as well as the collision time, collision location when the collision occurs, and the sensing range corresponding to the sensors deployed on the inspection robot. Risk area division unit: used to establish a three-dimensional model of the campus site, divide all first risk areas in the three-dimensional model, extract the second risk areas corresponding to each collision event in the three-dimensional model, and use the area where the first risk area intersects the second risk area as the characteristic risk area.

[0013] Furthermore, the early warning prompt timing determination module includes a risk level calculation unit and an early warning prompt timing determination unit; Risk level calculation unit: used to determine the risk level of the target moving object to the inspection robot according to the risk value and various risk areas in the three-dimensional model; Early warning prompt timing determination unit: used to determine the timing of the inspection robot to give an early warning prompt to the target moving object according to the risk level.

[0014] Compared with the prior art, the beneficial effects of the present invention are: The present invention provides a campus security risk prevention and control system and method based on AI and data analysis, including: establishing a three-dimensional model of the campus site, and dividing all first risk areas, second risk areas and characteristic risk areas in the three-dimensional model; obtaining the target moving object in the monitoring screen according to the inspection route and monitoring equipment; extracting the local area image of the target moving object in the monitoring screen, and determining the risk value of the target moving object to the inspection robot; determining the risk level of the target moving object to the inspection robot, and further determining the timing of the inspection robot to give an early warning prompt to the target moving object. The present invention anticipates potential collision risks in advance through the actual movement trajectory changes of students on campus and combined with the analysis of monitoring equipment, and gives prompts to moving objects in advance, optimizing the common collision problems in the inspection process of inspection robots. Description of the Drawings

[0015] Figure 1 It is a flow schematic diagram of a campus security risk prevention and control method based on AI and data analysis of the present invention; Figure 2This is the structural diagram of a campus security risk prevention and control system based on AI and data analysis according to the present invention. Specific Embodiment

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment: As Figure 1 shown, the present invention provides a technical solution for a campus security risk prevention and control method based on AI and data analysis, including the following steps: Step S100: Establish a three-dimensional model of the campus site, and divide all first risk areas in the three-dimensional model; obtain the collision events of the patrol robot colliding with moving objects in history, extract the corresponding second risk areas of each collision event in the three-dimensional model, and use the area where the first risk area intersects with the second risk area as the characteristic risk area; Obtain the corner position corresponding to the corner area in the campus site, and use the area with a radius of R around the corner position as the first risk area.

[0018] Step S110: Obtain several collision events of the patrol robot colliding with moving objects in history. The collision event is an event where the patrol robot makes contact with a moving object during the patrol process and triggers the collision alarm of the patrol robot; extract the corresponding moving objects of the collision event, as well as the collision time, collision position at the time of collision, and the sensing range corresponding to the sensors deployed on the patrol robot. The sensing range is a circular range; Step S120: Establish a three-dimensional model of the campus site. According to the collision time T 1 corresponding to a certain collision event in history and the moving object A, obtain the starting point from the collision time T 1 , and the moment when the position P A of the moving object A first satisfies the edge of the sensing range of the deployed sensor going forward, and use the moment as the warning time T 2 ; Obtain the moving distance S 2 of the moving object A between the time T 1 and the time T A , and obtain the moving speed V A of the moving object A = S A / (T 1 - T 2 ), if V A < V 1And the inspection robot at time T 2 makes a warning prompt, V 1 is the first speed threshold, then the time period from time T 2 to time T 1 All the sensing ranges corresponding to the sensors during this period are used as the second risk areas, and all the corresponding second risk areas in the 3D model are obtained.

[0019] Step S200: According to the set inspection route of the inspection robot and the monitoring devices installed on campus, obtain the first warning range and the second warning range during the movement of the inspection robot, and combine the position changes of the moving objects in the monitoring video to obtain the target moving objects.

[0020] Step S210: Use the sensing range corresponding to the sensors deployed on the inspection robot as the first warning range. The first obstacle avoidance range is a range with the position P 0 of the inspection robot as the center and a radius of r 1 ; and the range with a distance of Q 0 meters forward along the inspection route from the position P 1 on both sides with a width of L 1 meters is also used as the first warning range; Use the range with the position P 0 of the inspection robot as the center and a radius of r 2 as the second warning range, and the range with a distance of Q 0 meters forward along the inspection route from the position P 2 on both sides with a width of L 2 meters is also used as the second warning range, where r 1 < r 2 , Q 1 < Q 2 , L 1 < L 2 , and divide the first warning range and the second warning range in the monitoring video; Here, the first warning range is the minimum range for the inspection robot to issue a prompt, and the second warning range is the maximum range for the inspection robot to issue a prompt. That is to say, when a moving object enters the first warning range, the probability of collision with the inspection robot increases greatly, and a warning prompt must be issued. The second warning range is the standard for making an early prompt. When the risk level of the moving object is analyzed to be relatively high, an early prompt needs to be made within the second warning range to prompt the moving object as early as possible and avoid late prompting and collisions.

[0021] Step S220: Continuously obtain the position of a certain moving object in the monitoring video, and detect whether the moving object enters the second warning range. If it enters the second warning range, record the time T when it just enters the second warning range3 The previous moment is taken as T 4 At moment T, a certain moving object 3 is at position P 3 At moment T 4 is at position P 4 Let P 4 point to P 3 The direction is taken as the moving direction; According to the moving distance of a certain moving object within the previous time period D at moment T, the moving speed V of the moving object is obtained; if V > V 3 and the intersection position of the moving direction and the inspection route is within the second warning range, and V 2 is the second speed threshold, then the moving object is taken as the target moving object. 2

[0022] Step S300: Extract the local area image corresponding to the local part of the target moving object in the monitoring screen, and determine the risk value of the target moving object to the inspection robot according to the change of the moving trajectory of the local part.

[0023] Step S310: Extract the monitoring screen R corresponding to a certain moment T within the time period D of the target moving object T , and extract the local area image corresponding to the local part of the target moving object in the monitoring screen R T ; perform horizontal and vertical edge detection on the local area image through the edge detection algorithm to obtain the edge contour of the corresponding local part in the local area image; In this embodiment, the edge detection algorithm is the sobel algorithm, and the specific process is the prior art and will not be elaborated here.

[0024] According to the monitoring screen R T , establish a two-dimensional coordinate system, randomly obtain the coordinates of several pixel points from the edge contour, and respectively obtain the average abscissa i and the average ordinate j according to the abscissa and ordinate of each pixel point, and obtain the characteristic coordinate C T (i, j) corresponding to a certain moment T; the monitoring device is a fixed-view and fixed-position bullet camera, and then mark the characteristic coordinates at each moment within the time period D according to the local area images corresponding to the monitoring screens at each moment within the time period D in the two-dimensional coordinate system; Step S320: Take the characteristic coordinates corresponding to three consecutive moments a, b, and c within the time period D as C a , C b and C c respectively; take the vector from C a to C b as V ab , and the vector from C b to C c ​The vector is used as V bc , and then the eigenvalue is obtained , where e is the natural exponent and x is the vector V ab and the vector V bc The included angle between them, max() is used to find the maximum value, min() is used to find the minimum value, K 1 and K 2 are the first characteristic coefficient and the second characteristic coefficient respectively; if the eigenvalue M is less than the preset characteristic threshold, the moment a is used as the characteristic moment, and according to the total number N of moments within the time period D and the number N of characteristic moments among them 1 , the risk value Y = N 1 / (N - 2).

[0025] It should be noted that in this embodiment, the moving object is a person and the local part is the ear. The main reasons for using the ear area as the basis for judgment are as follows: Since the number of ear contours is large, the ear contours can be photographed from various angles, which is easy to identify, and the monitoring device can photograph the ears from multiple directions; since the ear is one of the five sense organs of the human body, when the position of the ear changes, it means that when the human body observes the surrounding environment left and right, the observable visual field range is large. The larger the eigenvalue, the larger the observable range of the outside world, and the lower the collision threat degree to the patrol robot. Similarly, the more the number of smaller eigenvalues, the greater the collision threat degree to the patrol robot, so the greater the risk value.

[0026] Step S400: Determine the risk degree of the target moving object to the patrol robot according to the risk value and various risk areas in the three-dimensional model, and determine the timing of the patrol robot to give a warning prompt to the target moving object according to the risk degree.

[0027] Step S410: Designate the areas on the campus that are not any risk areas as the first type of area with a weight of W 1 , the first risk area and the second risk area as the second type of area with a weight of W 2 , and the characteristic risk area as the third type of area with a weight of W 3 ; Connect the position P 3 and the position P 0 to obtain a line segment L. According to the length ratio of each type of area in the line segment L to the total length of the line segment L, the risk degree of the target moving object to the patrol robot is obtained as , where Y is the risk value, W q is the weight corresponding to the qth type of area, and R q is the length ratio occupied by the qth type of area; Step S420: If the risk degree Y 0 is greater than the risk degree threshold, at the moment T 3Give a prompt; otherwise, after the target moving object enters the second warning range, re-obtain the risk value and the length ratio of each type of area in real time, and re-determine the risk level of the target moving object to the inspection robot, and give a prompt for the moments when the risk level is greater than the risk level threshold; if no prompt is given before entering the first obstacle avoidance range, give a prompt when just entering the first obstacle avoidance range.

[0028] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. A campus safety risk prevention and control method based on AI and data analysis, characterized in that: The following steps are involved: Step S100: establishing a three-dimensional model of the campus and dividing all first risk areas in the three-dimensional model; Acquire historical collision events of the inspection robot and the mobile object, extract the second risk area corresponding to each collision event in the three-dimensional model, and use the area where the first risk area intersects with the second risk area as a characteristic risk area; Step S200: According to the inspection route set by the inspection robot and the monitoring equipment installed in the campus, the first warning range and the second warning range of the inspection robot during the movement are obtained, and the target moving object is obtained in combination with the position change of the moving object in the monitoring picture; Step S300: extracting a local area image corresponding to a local part of the target moving object in the monitoring screen, and determining a risk value of the target moving object to the inspection robot according to a change in the moving trajectory of the local part; Step S400: Determine the risk level of the target moving object to the inspection robot based on the risk value and various risk areas in the three-dimensional model, and determine the timing for the inspection robot to issue an early warning prompt to the target moving object based on the risk level.

2. According to a campus safety risk prevention and control method based on AI and data analysis according to claim 1, it is characterized in that: Step S100 includes: obtaining a corner position corresponding to a corner area in a campus location, and taking an area with a radius R around the corner position as a first risk area.

3. According to a campus safety risk prevention and control method based on AI and data analysis according to claim 1, it is characterized in that: Step S100 includes: Step S110: obtaining several collision events in which the inspection robot has collided with the mobile object in history, wherein the collision event is an event in which the inspection robot contacts the mobile object during the inspection process and triggers the collision alarm of the inspection robot; extracting the mobile object corresponding to the collision event, as well as the collision time, collision position and sensing range corresponding to the sensor deployed on the inspection robot when the collision occurs, wherein the sensing range is a circular range; Step S120: Establish a three-dimensional model of the campus, and obtain the position P of the moving object A that first meets the collision time T1 based on the collision time T1 corresponding to a certain historical collision event and the moving object A. A , at the moment when the deployed sensor reaches the edge of the sensing range, and taking the moment as the warning moment T2; Obtain the moving distance S of the moving object A between time T2 and time T1 A , and obtain the moving speed V of the moving object A A =S A / (T1 - T2). If V A < V1 and the inspection robot gives a warning prompt at time T2, where V1 is the first speed threshold, then all the sensing ranges corresponding to the sensors between time T2 and time T1 are used as the second risk areas, and all the corresponding second risk areas in the three-dimensional model are obtained.

4. According to a campus safety risk prevention and control method based on AI and data analysis according to claim 3, it is characterized in that: Step S200 includes: Step S210: The sensing range corresponding to the sensor deployed on the inspection robot is used as the first warning range. The first obstacle avoidance range is a range with a radius of r1 and a position P0 of the inspection robot as the center. The range with a distance of Q1 meters forward and L1 meters on both sides along the inspection route starting from the position P0 is also used as the first warning range. The range with the inspection robot's position P0 as the center and a radius of r2 is taken as the second warning range, and the range with the position P0 as the starting point, Q2 meters forward along the inspection route and L2 meters on both sides is also taken as the second warning range, r1<r2, Q1<Q2, L1<L2, and the first warning range and the second warning range are divided in the monitoring screen; Step S220: acquiring the position of a certain moving object in the monitoring screen in real time, and detecting whether the certain moving object enters the second warning range. If the certain moving object enters the second warning range, the time before the time T3 when the certain moving object just enters the second warning range is taken as T4, the position of the certain moving object at the time T3 is P3, the position at the time T4 is P4, and the direction from P4 to P3 is taken as the moving direction; According to the moving distance of the moving object in the time period D before time T3, the moving speed V of the moving object is obtained; if V>V2 and the position where the moving direction intersects the inspection route is within the second warning range, V2 is the second speed threshold, then the moving object is taken as the target moving object.

5. According to claim 4, a campus safety risk prevention and control method based on AI and data analysis is characterized in that: Step S300 includes: Step S310: extract the monitoring image R corresponding to a certain time T of the target moving object within the time period D T , extract the monitoring picture R T In the local area image corresponding to the local part of the target moving object; performing horizontal and vertical edge detection on the local area image by an edge detection algorithm to obtain the edge contour of the corresponding local part in the local area image; According to the monitoring screen R T , establish a two-dimensional coordinate system, and randomly obtain the coordinates of several pixel points from the edge contour. According to the horizontal and vertical coordinates of each pixel point, the average horizontal coordinate i and the average vertical coordinate j are obtained respectively, and the characteristic coordinate C corresponding to a certain time T is obtained T (i, j); the monitoring device is a gun-type camera with a fixed viewing angle and position, and then the characteristic coordinates of each moment in the time period D are marked in the two-dimensional coordinate system according to the corresponding local area image in the monitoring screen at each moment in the time period D; Step S320: The feature coordinates corresponding to the three consecutive moments a, b and c in the time period D are respectively used as C a , C b and C c ; C a Point to C b The vector V ab , C b Point to C c The vector V bc , and then get the eigenvalue , where e is the natural exponent and x is the vector V ab With vector V bc The angle between them, max() is for finding the maximum value, min() is for finding the minimum value, K1 and K2 are the first characteristic coefficient and the second characteristic coefficient respectively; if the characteristic value M is less than the preset characteristic threshold, the moment a is taken as the characteristic moment, and according to the total number of moments N in the time period D, and the number of characteristic moments N1 therein, the risk value Y=N1 / (N-2) is obtained.

6. According to claim 5, a campus safety risk prevention and control method based on AI and data analysis is characterized in that: Step S400 includes: Step S410: The campus area without any risk is regarded as the first category area with a weight of W1, the first risk area and the second risk area are regarded as the second category area with a weight of W2, and the characteristic risk area is regarded as the third category area with a weight of W3; a line segment L is obtained by connecting the position P3 and the position P0, and the risk level of the target moving object to the inspection robot is obtained according to the length ratio of each category of areas in the line segment L to the total length of the line segment L. , where Y is the risk value, W q is the weight corresponding to the qth type of area, R q is the length ratio of the qth type of area; Step S420: If the risk level Y0 is greater than the risk level threshold, a prompt is given at time T3; otherwise, after the target moving object enters the second warning range, the risk value and the length ratio of each type of area are re-obtained in real time, and the risk level of the target moving object to the inspection robot is re-obtained, and a prompt is given when the risk level is greater than the risk level threshold; if no prompt is given before entering the first obstacle avoidance range, a prompt will be given when the first obstacle avoidance range is just entered.

7. A campus safety risk prevention and control system, used to execute a campus safety risk prevention and control method based on AI and data analysis as described in any one of claims 1 to 6, characterized in that: The system includes a risk area division module, a target moving object determination module, a risk value calculation module and an early warning prompt timing determination module; Risk area division module: used to establish a three-dimensional model of the campus and divide all first risk areas in the three-dimensional model; Acquire historical collision events of the inspection robot and the mobile object, extract the second risk area corresponding to each collision event in the three-dimensional model, and use the area where the first risk area intersects with the second risk area as a characteristic risk area; Target moving object determination module: used to obtain the first warning range and the second warning range of the inspection robot during its movement according to the inspection route set by the inspection robot and the monitoring equipment installed on the campus, and to obtain the target moving object in combination with the position change of the moving object in the monitoring screen; Risk value calculation module: used to extract the local area image corresponding to the local part of the target moving object in the monitoring screen, and determine the risk value of the target moving object to the inspection robot according to the change of the moving trajectory of the local part; Early warning timing determination module: used to determine the risk level of the target moving object to the inspection robot based on the risk value and various risk areas in the three-dimensional model, and determine the timing for the inspection robot to issue an early warning prompt to the target moving object based on the risk level.

8. A campus safety risk prevention and control system according to claim 7, characterized in that: The risk area division module includes a collision event extraction unit and a risk area division unit; Collision event extraction unit: used to obtain collision events in which the inspection robot has collided with a mobile object in history; extract the mobile object corresponding to the collision event, as well as the collision time, collision position and the sensing range corresponding to the sensor deployed on the inspection robot; Risk area division unit: used to establish a three-dimensional model of the campus and divide all first risk areas in the three-dimensional model, extract the second risk area corresponding to each collision event in the three-dimensional model, and use the area where the first risk area intersects with the second risk area as a characteristic risk area.

9. A campus safety risk prevention and control system according to claim 8, characterized in that: The early warning timing determination module includes a risk degree calculation unit and an early warning timing determination unit; Risk degree calculation unit: used to determine the risk degree of the target moving object to the inspection robot according to the risk value and various risk areas in the three-dimensional model; Early warning timing determination unit: used to determine the timing for the inspection robot to issue an early warning prompt to the target moving object according to the risk level.

Images