Campus abnormal behavior analysis method based on location information

Abstract

The application provides a campus abnormal behavior analysis method based on position information, and the specific steps are as follows: according to the work and rest schedule, a time period in which the campus abnormal behavior is prone to occur is divided, and a weight coefficient is set for each time period; according to the campus layout, a space region is divided, and a space weight coefficient is set for each space region; student position information is collected and obtained, the daily behavior habits of each student are statistically analyzed according to the collected student position information in a period of time, and the relationship closeness of all students is calculated; the abnormal behavior that may occur to the members in the clustering group is analyzed by fusion in combination with auxiliary information such as the time-space function attribute, the daily behavior habits of students, the relationship closeness of students and the like, and an alarm prompt is given. The campus abnormal behavior analysis method can better adapt to various occasions, and effectively avoid the exposure of student privacy.

Description

Technical Field

[0001] The present invention relates to the technical field of behavior analysis, and in particular to a method for analyzing abnormal campus behavior based on location information. Background Art

[0002] Abnormal behavior on campus is increasingly attracting public attention, and the introduction of information technology has effectively improved schools' ability to prevent and respond to such safety issues. Currently, there are two main types of methods commonly used to analyze abnormal behavior on campus: one is an intelligent analysis method based on surveillance cameras, which can identify specific action patterns such as fighting and chasing from videos, thereby quickly responding to potential safety incidents. However, this method is not suitable for hidden places such as corridors and bathrooms; the other is an intelligent analysis method based on language, which can identify sensitive words such as calls for help based on the language content, thereby triggering an alarm signal. However, this method has limited practical application in open and noisy places such as school gates and playgrounds. Based on this, the present invention proposes a method for analyzing abnormal behavior on campus based on location information. Summary of the Invention

[0003] The purpose of the present invention is to propose a method for analyzing abnormal campus behavior based on location information, which can better adapt to various occasions and effectively avoid the problem of student privacy exposure. Through the school badge with built-in indoor and outdoor positioning chip worn by students, student location information is obtained in real time. After clustering using spatiotemporal information, combined with auxiliary information such as spatiotemporal functional attributes, students' daily behavioral habits, and the degree of intimacy between students, an integrated analysis of abnormal behaviors that may occur among members of the cluster group is performed, and an alarm prompt is given.

[0004] To achieve the above objectives, the technical solution of the present invention is: a method for analyzing abnormal campus behavior based on location information, the specific steps of which are as follows:

[0005] S1: Divide the time periods when abnormal behaviors are likely to occur on campus according to the work and rest schedule, and set a weight coefficient for each time period;

[0006] S2: Divide the spatial regions according to the campus layout and set the spatial weight coefficient for each spatial region;

[0007] S3: Collect and obtain student location information, and statistically analyze each student's daily behavior habits based on the student location information collected over a period of time, and calculate the relationship intimacy of all students;

[0008] S4: Set the sliding time window Ts and the time sampling interval Δt;

[0009] S5: Determine whether the sampling time t of the current student location information is in the time period when abnormal behavior on campus is likely to occur, that is, whether ST u ≤t≤ET u, where ST u ET u Respectively represent the start time point and end time point of a certain period of time when abnormal behavior on campus is likely to occur. If yes, execute the following steps; otherwise, execute step S6;

[0010] S5.1: Continue to obtain the student location information within the time t+Ts, and summarize the student location information of all sampling moments obtained within the sliding time window according to the individual students, that is,

[0011] E1={StuL 1,1 ,StuL 1,2 ..., StuL 1,n |t1=t,t2=t+Δt,...,t n =t+Ts},

[0012] E2={StuL 2,1 ,StuL 2,2 ..., StuL 2,n |t1=t,t2=t+Δt,...,t n =t+Ts}, ...

[0014] E m ={StuL m,1 ,StuL m,2 ..., StuL m,n |t1=t,t2=t+Δt,...,t n =t+Ts}

[0015] Where, Ts=(n-1)Δt, StuL m,n Indicates that the mth student at the nth sampling time t n location information;

[0016] S5.2: If t + Ts ≤ ET u , then execute step S5.3, otherwise set t = t + Ts and return to step S5 to continue;

[0017] S5.3: According to time points t1, t2, ..., t n , cumulative statistics E1, E2, ..., E m The average distance dTs between any two students i and j i,j , and then use the unsupervised clustering algorithm to automatically cluster, and for each cluster C in the clustering result k = {E′1, E′2…} to perform potential risk analysis. If the potential risk exceeds the threshold, an alarm is given. After all clusters are analyzed, let t = t + Δt, and go to step S5.1 to enter the next time window for analysis;

[0018] S6: Enter the next sampling time, that is, t=t+Δt, and return to step S5 to continue execution.

[0019] Furthermore, the S1 is specifically as follows: divide the students' time at school according to the work and rest schedule and select the time period when abnormal behavior on campus is likely to occur. The time period when abnormal behavior on campus is likely to occur includes: entering the school in the morning, each morning break, leaving the school in the morning, entering the school in the afternoon, each afternoon break, and leaving the school in the afternoon; the selected time period is recorded as Tim u ={TName u ,ST u , E.T. u , WT u}, where TName u Indicates the time period name, ST u Indicates the starting time point; ET u Indicates the end time point; WT u Indicates the weight coefficient set for this time period.

[0020] Furthermore, the S2 is specifically as follows: divide the space area according to the campus layout, and the space area includes classrooms, passages, toilets, playgrounds, school gates, and other hidden areas. Set the space weight coefficient according to the probability of abnormal phenomena in different space areas. The greater the probability of abnormal phenomena, the greater the space weight coefficient of the space area. The divided space area is recorded as Pla v ={PName v ,Ω v , WP v}, where PName v Indicates the name of the spatial region and is unique, Ω v Indicates the specific location of the polygonal area; WP v Indicates the weight coefficient set for this space.

[0021] Furthermore, the collection and acquisition of student location information is specifically as follows: acquiring student location information according to a set sampling frequency, performing integrated positioning of the school badge with a positioning chip worn by students on campus based on BeiDou satellites and gateway devices, so as to calculate the three-dimensional coordinate position of each student at each sampling time point; recording the student location information as StuL i,l ={ID i , t l , L l}, where ID i represents the school badge number of student i, which corresponds to the student identity one by one, t l and L l They represent the time point and location information of student i respectively.

[0022] Furthermore, the daily behavior habits of each student are statistically analyzed based on the student location information collected over a period of time. Specifically, based on the student location information collected over a period of time, the probability P of each student appearing in the different spatial areas divided during each time period prone to abnormal campus behavior is calculated. i,u,v , P i,u,v = Tim, a time period when abnormal campus behavior is likely to occur u Student i appears in a certain space region Pla v Number of coordinate points / time period Tim u The number of all coordinate points of student i; the daily behavior habits of student i are recorded as StuB i ={ID i , TName u , PName v , P i,u,v}, where ID i Indicates the school badge number of student i, TName u Indicates the name of the u-th time period, PName v Indicates the name of the v-th spatial region.

[0023] Furthermore, the relationship intimacy of all students is calculated based on the student location information collected over a period of time. Specifically, an m×m matrix S is used to represent the relationship intimacy of all students, where m represents the number of students.

[0024] The matrix S in S i,i represents the abnormal behavior tendency of student i, S i,i =1 means student i likes violent classmates, S i,i =0.5 means student i has no particular preference, S i,i =0 means that student i is more likely to be subjected to violence; this value can be set manually based on the prior information provided by the head teacher. By default, all values ​​can be set to S i,i =0.5;

[0025] The matrix S in S i,j Represents the relationship between student i and student j: Calculate the probability M that student i and student j are together within a period of time i,j , that is, M i,j = the number of coordinate points where student i and student j are together during this period / the total number of coordinate points generated by the system collecting a student during this period, let S i,j =M i,j / (1+|S i,i -S j,j |); wherein, the student i and student j are together refers to the spatial position distance d of student i and student j at the same time point t i,j,tLess than the prior distance threshold D d , that is, d i,j,t <D d .

[0026] Furthermore, the average distance dTs i,j The calculation is as follows:

[0027]

[0028] Where, d i,j,t' represents the spatial distance between student i and student j at a certain sampling time t', and |Ts| represents the number of system sampling time points in the sliding time window.

[0029] Furthermore, for each cluster C in the clustering result k Perform potential risk analysis and issue warnings if the potential risk exceeds the threshold. Specifically:

[0030] For any cluster C k , when the number of student members in the cluster |C k When |>1, the potential risk R k The calculation formula is,

[0031]

[0032] Where, WT r Indicates the weight coefficient corresponding to the time period of the current time window, WP s Indicates the weight coefficient corresponding to the spatial area where the cluster center is located; Stu q ∈C k Stu i ∈C k Indicates that student q and student i belong to the same cluster C k ; DisB represents the average close relationship between student q and other students i in the same category. The smaller the value, the worse the relationship between student q and the group. q Indicates the degree of difference between student q’s current behavior and daily behavior habits, Represents cluster C k The relationship between people is poor and the behavior differences are large; at this time, if R k When the risk is greater than a certain threshold RiskP1, the system will issue an alarm and prompt the locations and personnel that may be at risk;

[0033] When |C k When |=1, that is, there is only one student member q in the cluster, its potential risk R k The calculation formula is,

[0034]

[0035] Where S q,i Indicates the close relationship between the only member student q in the cluster and any student i in the school, DisB q Indicates the degree of difference between student q’s behavior in this time window and his daily behavior habits; at this time, if R k When it is greater than a certain threshold RiskP2, the system will issue an alarm and prompt the locations and personnel that may be at risk.

[0036] Furthermore, the DisB q The calculation is as follows:

[0037]

[0038] That is, first count the probability NewP of student q appearing in each spatial area in the current time window q,r,s , and generate statistical results NewStuB q ={ID q ,NewTName r ,NewPName s , NewP q,r,s}, where NewTName r Indicates the name of the time period of the current time window, NewPName s Indicates the name of the spatial area where the student's location point is located; then use the spatiotemporal matching relationship NewTName r =TName r ,NewPName s =PName s , find out the daily behavior habits of students qStuB q The probability P of appearing in all the same time periods and places q,r,s and with NewP q,r,s For comparison, P q,r,s Indicates the daily behavior habits of student q in time period TName r Appears in the space area PName s probability;

[0039] min(NewP q,r,s ,P q,r,s ) is P q,r,s and NewP q,r,s The smaller value in , indicates the commonality between the current behavior and daily behavior habits. If the behavior of student q in the current time window is relatively consistent with the daily behavior habits, DisB q The value of DisB is smaller, otherwise q Larger values ​​indicate greater differences in behavior.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] This method uses school badges (with built-in indoor and outdoor positioning chips) worn by students to obtain real-time student location information. After clustering using spatiotemporal information, it combines auxiliary information such as spatiotemporal functional attributes, students' daily behavior habits, and the closeness of student relationships to perform a fusion analysis of possible abnormal behavior among cluster members and issue alarm prompts. This method of analyzing abnormal campus behavior based on location information is well-suited to various scenarios and effectively prevents the exposure of student privacy. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a schematic diagram of the structure of the campus abnormal behavior analysis system based on location information of the present invention;

[0043] Figure 2 This is a flow chart of the method for analyzing abnormal campus behavior based on location information of the present invention. DETAILED DESCRIPTION

[0044] The following is combined with Figure 1-2 , the technical solution of the present invention is described in detail.

[0045] The present invention proposes a method for analyzing abnormal behavior on campus based on location information, and the specific steps are as follows:

[0046] S1: Divide the time periods when abnormal behaviors are likely to occur on campus according to the work and rest schedule, and set a weight coefficient for each time period;

[0047] S2: Divide the spatial regions according to the campus layout and set the spatial weight coefficient for each spatial region;

[0048] S3: Collect and obtain student location information, and statistically analyze each student's daily behavior habits based on the student location information collected over a period of time, and calculate the relationship intimacy of all students;

[0049] S4: Set the sliding time window Ts and the time sampling interval Δt;

[0050] S5: Determine whether the sampling time t of the current student location information is in the time period when abnormal behavior on campus is likely to occur, that is, whether ST u ≤t≤ET u , where ST u ET u Respectively represent the start time point and end time point of a certain period of time when abnormal behavior on campus is likely to occur. If yes, execute the following steps; otherwise, execute step S6;

[0051] S5.1: Summarize the location information obtained at each sampling moment within the sliding time window according to the individual students, that is,

[0052] E1={StuL 1,1 ,StuL 1,2 ..., StuL 1,n |t1=t,t2=t+Δt,...,t n =t+Ts},

[0053] E2={StuL 2,1 ,StuL 2,2 ..., StuL 2,n |t1=t,t2=t+Δt,...,t n =t+Ts}, ...

[0055] E m ={StuL m,1 ,StuL m,2 ..., StuL m,n |t1=t,t2=t+Δt,...,t n =t+Ts}

[0056] Where, Ts=(n-1)Δt, StuL m,n Indicates that the mth student at the nth sampling time t n location information;

[0057] S5.2: If t + Ts ≤ ET u , then execute step S5.3, otherwise set t = t + Ts and return to step S5 to continue;

[0058] S5.3: According to time points t1, t2, ..., t n , cumulative statistics E1, E2, ..., E m The average distance dTs between any two students i and j i,j , and then use the unsupervised clustering algorithm to automatically cluster, and for each cluster C in the clustering result k = {E′1, E′2...} to perform potential risk analysis. If the potential risk exceeds the threshold, an alarm is given. After all clusters are analyzed, let t = t + Δt, and go to step S5.1 to enter the next time window for analysis;

[0059] S6: Enter the next sampling time, that is, t=t+Δt, and return to step S5 to continue execution.

[0060] In this embodiment, S1 is specifically as follows: dividing the students' time at school according to the work and rest schedule and selecting the time period when abnormal campus behavior is likely to occur. The time period when abnormal campus behavior is likely to occur includes: entering the school in the morning, each morning break, leaving the school in the morning, entering the school in the afternoon, each afternoon break, and leaving the school in the afternoon; the selected time period is recorded as Tim u ={TName u ,ST u , E.T. u , WT u}, where TName u Indicates the time period name, ST u Indicates the starting time point; ET u Indicates the end time point; WT u Represents the weight coefficient set for the time period; for example, abnormal events are more likely to occur during breaks than during class time, so the corresponding weight should be larger; it should be noted that, considering that teachers are usually present during class time and students are subject to strong temporal and spatial constraints, the probability of abnormal behavior on campus is very low, so this method focuses on non-class time periods such as school start and end, and breaks.

[0061] In this implementation, S2 is specifically as follows: divide the space area according to the campus layout, and the space area includes classrooms, passages, toilets, playgrounds, school gates, and other hidden areas. Set the space weight coefficient according to the difficulty of abnormal phenomena in different space areas. The space area that is more prone to abnormal phenomena has a larger space weight coefficient. For example, toilets are more likely to have campus abnormal events than classrooms, so the corresponding weight should be larger. The divided space area is recorded as Pla v ={PName v ,Ω v , WP v}, where PName v Indicates the name of the spatial region and is unique, Ω v Indicates the specific location of the polygonal area; WP v Indicates the weight coefficient set for this space.

[0062] In this embodiment, the step S3 of collecting and obtaining student location information specifically includes: obtaining student location information according to the set sampling frequency, performing integrated positioning of the school badge with a positioning chip worn by the students on campus based on the Beidou satellite and gateway device, and calculating the three-dimensional coordinate position of each student in real time; recording the student location information as StuL i,l ={ID i , t l , L l}, where ID i represents the school badge number of student i, which corresponds to the student identity one by one, tl and L l They represent the time point and location information of student i respectively.

[0063] The daily behavior habits of each student are statistically analyzed based on the student location information collected over a period of time. Specifically, based on the student location information collected over a period of time (e.g., one month), the probability P of each student appearing in the different spatial areas divided during each time period prone to abnormal campus behavior is calculated. i,u,v , P i,u,v = Tim, a time period when abnormal campus behavior is likely to occur u Student i appears in a certain space region Pla v Number of coordinate points / time period Tim u The number of all coordinate points of student i; the daily behavior habits of student i are recorded as StuB i ={ID i , TName u , PName v , P i,u,v}, where ID i Indicates the school badge number of student i, TName u Indicates the name of the u-th time period, PName v Indicates the name of the vth spatial region. In addition, student behavior habits can be incrementally updated according to actual needs.

[0064] The relationship intimacy of all students is statistically analyzed based on the student location information collected over a period of time. Specifically, an m×m matrix S is used to represent the relationship intimacy of all students, where m represents the number of students.

[0065] The matrix S in S i,i represents the abnormal behavior tendency of student i, S i,i =1 means student i likes violent classmates, S i,i =0.5 means student i has no particular preference, S i,i =0 means that student i is relatively introverted and is more likely to be subjected to violence; this value can be set manually based on the prior information provided by the class teacher. By default, all values ​​can be set to S i,i =0.5;

[0066] The matrix S in S i,j Represents the relationship between student i and student j: Calculate the probability M that student i and student j are together within this period of time (e.g., 1 month) i,j , that is, M i,j = the number of coordinate points where student i and student j are together during this period / the total number of coordinate points generated by the system collecting a student during this period. Considering that the perpetrator and the victim may often be together, let Si,j =M i,j / (1+|S i,i -S j,j |); wherein, the student i and student j are together refers to the spatial position distance d of student i and student j at the same time point t i,j,t Less than the prior distance threshold D d , that is, d i,j,t <D d .

[0067] There are problems with common crowd clustering methods: when clustering crowds according to sampling time intervals, the clustering results based on a single time point have no temporal persistence, which is not conducive to determining whether there are abnormal behaviors on campus (such as students going to the toilet in groups during class); when clustering crowds according to the time periods divided as mentioned above, the system needs to wait until the end of the time period before performing cluster analysis. The abnormal behavior on campus may have occurred and ended, especially for time periods with a long time span (such as going to school and leaving school). Even if an alarm is issued, it is impossible to carry out on-site disposal, which loses the practical significance of prevention.

[0068] Therefore, the present invention adopts a crowd clustering method based on a sliding time window to divide the time period Tim u Start time ST u As the starting point, divide the time period Tim u End time ET u As the end point, clustering is performed using a time sliding window (the step size is the position information sampling interval Δt). Within the sliding window Ts, the average distance between any two students is accumulated according to the time points. Where, d i,j,t' represents the spatial distance between student i and student j at a certain sampling time t', |Ts| represents the number of system sampling time points in the sliding time window, and then automatically clusters using unsupervised clustering algorithms (such as nearest neighbor propagation, iterative self-organization, etc.).

[0069] In this embodiment, for each cluster C in the clustering result, k Perform potential risk analysis and issue warnings if the potential risk exceeds the threshold. Specifically:

[0070] For any cluster C k , when the number of student members in the cluster |C k When |>1, the potential risk R k The calculation formula is,

[0071]

[0072] Where, WT r Indicates the weight coefficient corresponding to the time period of the current time window, WPs Indicates the weight coefficient corresponding to the spatial area where the cluster center is located; Stu q ∈C k Stu i ∈C k Indicates that student q and student i belong to the same cluster C k ; DisB represents the average close relationship between student q and other students i in the same category. The smaller the value, the worse the relationship between student q and the group. q Indicates the degree of difference between student q’s current behavior and daily behavior habits, Represents cluster C k The relationship between people is poor and the behavior differences are large; at this time, if R k When the risk is greater than a certain threshold RiskP1, the system will issue an alarm and prompt the locations and personnel that may be at risk;

[0073] When |C k When |=1, that is, there is only one student member q in the cluster, its potential risk R k The calculation formula is,

[0074]

[0075] Where S q,i Indicates the close relationship between the only member student q in the cluster and any student i in the school, DisB q Indicates the degree of difference between student q’s behavior in this time window and his daily behavior habits; at this time, if R k When it is greater than a certain threshold RiskP2, the system will issue an alarm and prompt the locations and personnel that may be at risk.

[0076] Wherein, the DisB q The calculation is as follows:

[0077]

[0078] That is, first count the probability NewP of student q appearing in each spatial area in the current time window q,r,s , and generate statistical results NewStuB q ={ID q ,NewTName r ,NewPName s , NewP q,r,s}, where NewTName r Indicates the name of the time period of the current time window, NewPName s Indicates the name of the spatial area where the student's location point is located; then use the spatiotemporal matching relationship NewTName r=TName r ,NewPName s =PName s , find out the daily behavior habits of students qStuB q The probability P of appearing in all the same time periods and places q,r,s and with NewP q,r,s For comparison, P q,r,s Indicates the daily behavior habits of student q in time period TName r Appears in the space area PName s probability;

[0079] min(NewP q,r,s ,P q,r,s ) is P q,r,s and NewP q,r,s The smaller value in , indicates the commonality between the current behavior and daily behavior habits. If the behavior of student q in the current time window is relatively consistent with the daily behavior habits, DisB q The value of DisB is smaller, otherwise q Larger values ​​indicate greater differences in behavior.

[0080] The above are preferred embodiments of the present invention. Any changes made according to the technical solution of the present invention, as long as the resulting functions and effects do not exceed the scope of the technical solution of the present invention, shall fall within the scope of protection of the present invention.

Claims

1. A method for analyzing abnormal campus behavior based on location information, characterized in that: The specific steps are as follows: S1: Divide the time periods when abnormal behaviors are likely to occur on campus according to the work and rest schedule, and set a weight coefficient for each time period; S2: Divide the spatial regions according to the campus layout and set the spatial weight coefficient for each spatial region; S3: Collect and obtain student location information, and statistically analyze each student's daily behavior habits based on the student location information collected over a period of time, and calculate the relationship intimacy of all students; S4: Set the time sampling interval Δt and the sliding time window Ts; S5: Determine whether the sampling time t of the current student location information is in the time period when abnormal behavior on campus is likely to occur, that is, whether ST u ≤t≤ET u , where ST u ET u Respectively represent the start time point and end time point of a certain period of time when abnormal behavior on campus is likely to occur. If yes, execute the following steps; otherwise, execute step S6; S5.1: Continue to obtain the student location information within the time t+Ts, and summarize the student location information of all sampling moments obtained within the sliding time window according to the individual students, that is, E1={StuL 1,1 ,StuL 1,2 ...,StuL 1,n |t1=t,t2=t+Δt,...,t n =t+Ts}, E2={StuL 2,1 ,StuL 2,2 …,StuL 2,n |t1=t,t2=t+Δt,…,t n =t+Ts}, ...， It is m ={StuL m,1 ,StuL m,2 ...,StuL m,n |t1=t,t2=t+Δt,...,t n =t+Ts} Where, Ts=(n-1)Δt, StuL m,n Indicates that the mth student at the nth sampling time t n location information; S5.2: If t + Ts ≤ ET u , then execute step S5.3, otherwise set t = t + Ts and return to step S5 to continue; S5.3: According to time points t1, t2, ..., t n , cumulative statistics E1, E2, ..., E m The average distance dTs between any two students i and j i,j , and then use the unsupervised clustering algorithm to automatically cluster, and for each cluster C in the clustering result k Perform potential risk analysis and issue warnings if the potential risk exceeds the threshold. Specifically: For any cluster C k , when the number of student members in the cluster |C k When |>1, the potential risk R k The calculation formula is, Where, WT r Indicates the weight coefficient corresponding to the time period of the current time window, WP s Indicates the weight coefficient corresponding to the spatial area where the cluster center is located; Stu q ∈C k Stu i ∈C k Indicates that student q and student i belong to the same cluster C k ; DisB represents the average intimacy between student q and other students i in the same category. The smaller the value of the average intimacy, the worse the relationship between student q and the group. q Indicates the degree of difference between student q’s current behavior and daily behavior habits, Represents cluster C k The relationship between people is poor and the behavior differences are large; at this time, if R k When the risk is greater than a certain threshold RiskP1, the system will issue an alarm and prompt the locations and personnel that may be at risk; When |C k When |=1, that is, there is only one student member q in the cluster, its potential risk R k The calculation formula is, Where S q,i Indicates the close relationship between the only member student q in the cluster and any student i in the school, DisB q Indicates the degree of difference between student q’s behavior in this time window and his daily behavior habits; at this time, if R k When the risk is greater than a certain threshold RiskP2, the system will issue an alarm and prompt the locations and personnel that may be at risk; After all clusters are analyzed, let t = t + Δt and proceed to step S5.1 to enter the next time window for analysis; S6: Enter the next sampling time, that is, t=t+Δt, and return to step S5 to continue execution.

2. The method for analyzing abnormal campus behavior based on location information according to claim 1, characterized in that: Specifically, S1 is as follows: divide the students' time at school according to the schedule and select the time period when abnormal behavior on campus is likely to occur. The time period when abnormal behavior on campus is likely to occur includes: entering the school in the morning, each morning break, leaving the school in the morning, entering the school in the afternoon, each afternoon break, and leaving the school in the afternoon; the selected time period is recorded as Tim u ={TName u ,ST u , E.T. u , WT u }, where TName u Indicates the time period name, ST u Indicates the starting time point; ET u Indicates the end time point; WT u Indicates the weight coefficient set for this time period.

3. The method for analyzing abnormal campus behavior based on location information according to claim 1, characterized in that: The S2 is specifically as follows: divide the space area according to the campus layout, and the space area includes classrooms, passages, toilets, playgrounds, school gates, and other hidden areas. Set the space weight coefficient according to the probability of abnormal behavior in different space areas. The greater the probability of abnormal behavior, the greater the space weight coefficient of the space area. The divided space area is recorded as Pla v ={PName v ,Ω v , WP v }, where PName v Indicates the name of the spatial region and is unique, Ω v Indicates the specific location of the space area; WP v Indicates the weight coefficient set for the space.

4. The method for analyzing abnormal campus behavior based on location information according to claim 1, characterized in that: The method of collecting and obtaining student location information is as follows: obtaining student location information according to a set sampling frequency, performing integrated positioning of the school badge with a positioning chip worn by students on campus based on BeiDou satellites and gateway devices, and calculating the three-dimensional coordinate position of each student at each sampling time point; recording the student location information as StuL i,l ={ID i , t l , L l }, where ID i represents the school badge number of student i, which corresponds to the student identity one by one, t l and L l They represent the time point and location information of student i respectively.

5. The method for analyzing abnormal campus behavior based on location information according to claim 1, characterized in that: The daily behavior habits of each student are statistically analyzed based on the student location information collected over a period of time. Specifically, based on the student location information collected over a period of time, the probability P of each student appearing in the different spatial areas divided during each time period prone to abnormal campus behavior is calculated. i,u,v , P i,u,v = Tim, a time period when abnormal campus behavior is likely to occur u Student i appears in a certain space region Pla v Number of coordinate points / time period Tim u The number of all coordinate points of student i; the daily behavior habits of student i are recorded as StuB i ={ID i , TName u , PName v , P i,u,v }, where ID i Indicates the school badge number of student i, TName u Indicates the name of the u-th time period, PName v Indicates the name of the v-th spatial region.

6. The method for analyzing abnormal campus behavior based on location information according to claim 1, characterized in that: The relationship intimacy of all students is calculated based on the student location information collected over a period of time. Specifically, the relationship intimacy of all students is represented by an m×m matrix S, where m represents the number of students. The matrix S in S i,i represents the abnormal behavior tendency of student i, S i,i =1 means student i likes violent classmates, S i,i =0.5 means student i has no particular preference, S i,i =0 means that student i is more likely to be subjected to violence; this value is set manually based on prior information; The matrix S in S i,j Represents the relationship between student i and student j: Calculate the probability M that student i and student j are together within a period of time i,j , that is, M i,j = the number of coordinate points where student i and student j are together during this period / the total number of coordinate points generated by the system collecting a student during this period, let S i,j =M i,j / (1+|S i,i -S j,j |); wherein, the student i and student j are together refers to the spatial position distance d of student i and student j at the same time point t i,j,t Less than the prior distance threshold D d , that is, d i,j,t <D d .

7. The method for analyzing abnormal campus behavior based on location information according to claim 1, characterized in that: The average distance dTs i,j The calculation is as follows: Where, d i,j,t' represents the spatial distance between student i and student j at a certain sampling time t', and |Ts| represents the number of system sampling time points in the sliding time window.

8. The method for analyzing abnormal campus behavior based on location information according to claim 1, characterized in that: The DisB q The calculation is as follows: That is, first count the probability NewP of student q appearing in each spatial area in the current time window q,r,s , and generate statistical results NewStuB q ={ID q ,NewTName r ,NewPName s , NewP q,r,s }, where NewTName r Indicates the name of the time period of the current time window, NewPName s Indicates the name of the spatial area where the student's location point is located; then use the spatiotemporal matching relationship NewTName r =TName r ,NewPName s =PName s , find out the daily behavior habits of students qStuB q The probability P of appearing in all the same time periods and places q,r,s and with NewP q,r,s For comparison, P q,r,s Indicates the daily behavior habits of student q in time period TName r Appears in the space area PName s probability; min(NewP q,r,s ,P q,r,s ) is P q,r,s and NewP q,r,s The smaller value in , indicates the commonality between the current behavior and daily behavior habits. If the behavior of student q in the current time window is relatively consistent with the daily behavior habits, DisB q The value of DisB is smaller, otherwise q Larger values ​​indicate greater differences in behavior.

Images