Public emergency detection system based on heterogeneous data fusion

Abstract

The invention relates to the technical field of data processing, and particularly discloses a public emergency detection system based on heterogeneous data fusion, comprising an information acquisition module used for acquiring various types of related information; the region division module is used for dividing a walkable region of the detection region into a plurality of second-level regions with equal walking areas and determining dangerous regions in the second-level regions; and the area analysis module is used for analyzing the pedestrian flow of the dangerous area and the pedestrian flow of the adjacent area, obtaining an analysis result and generating a coping strategy according to the analysis result, the detection area is divided into a plurality of second-level areas with equal walking areas, then the dangerous area is searched from the second-level areas, and the coping strategy is generated according to the coping strategy. According to the method, the pedestrian flow of the dangerous area and the pedestrian flow of the adjacent area are analyzed, and the coping strategy is generated according to the analysis result, so that the crowd moving state of the detection area is mastered more accurately, and the trampling area is processed in time to reduce the risk of a trampling accident.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a public emergency detection system based on heterogeneous data fusion. Background Technology

[0002] Public emergencies refer to sudden events that cause or may cause significant casualties, property damage, ecological damage, and serious social harm, thereby endangering public safety.

[0003] Public emergencies are generally classified into four levels according to their nature, severity, controllability, and scope of impact. Among them, stampedes are classified as public emergencies and are adjusted between level two and level four depending on the circumstances. Unlike other natural disaster-related public emergencies, stampedes can be avoided through human intervention.

[0004] Generally, stampedes occur in scenic areas or streets. Existing methods for preventing stampedes involve limiting visitor flow based on the maximum capacity of the scenic area or street. The maximum capacity is typically calculated by dividing the area that can be visited by the unit area required to maintain visitor safety. This method can effectively control the flow of people and greatly reduce the probability of stampedes. However, this method is not effective in more specific areas. When activities or other attractive events occur in certain areas, the flow of people in those areas will increase rapidly. Existing detection or crowd control methods cannot detect the trend of crowd changes in a timely manner. They can only send staff to restrict and guide the crowd in the area after a large number of people have accumulated. In view of this, this invention proposes a public emergency detection system based on heterogeneous data fusion. Summary of the Invention

[0005] The purpose of this invention is to provide a public emergency detection system based on heterogeneous data fusion, and to solve the following technical problems:

[0006] How to promptly detect changes in the population in specific areas and analyze their trends, and to restrict and guide the population in those areas in a timely manner to minimize the occurrence of public emergencies such as stampedes.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] A public emergency detection system based on heterogeneous data fusion includes:

[0009] The access control module includes a statistics unit for recording the number of people entering and exiting the detection area, and multiple access control units set at the entrances and exits of the detection area to restrict pedestrian access.

[0010] The information acquisition module is used to acquire various types of relevant information, including at least the number of people, direction of movement, and facial orientation.

[0011] The area division module is used to divide the walkable area of ​​the detection area into several secondary areas of equal walkable area and to identify the dangerous areas within the secondary areas;

[0012] The area analysis module analyzes pedestrian traffic in dangerous areas and their adjacent areas, obtains analysis results, and generates response strategies based on the analysis results.

[0013] The above technical solution involves dividing the detection area into several secondary zones of equal walking area, then identifying dangerous areas within these secondary zones, analyzing the pedestrian flow in and around these dangerous areas, and generating response strategies based on the analysis results. This allows for a more accurate understanding of pedestrian movement within the detection area and timely intervention in areas prone to stampedes to reduce the risk of such accidents.

[0014] As a further technical solution of the present invention: the process of determining the secondary region within the walkable area of ​​the detection area includes:

[0015] The number of pedestrians in the secondary area is acquired periodically. The number of pedestrians can be obtained through image counting software, which is existing technology and will not be elaborated here. The number of pedestrians is compared with a standard number range, which is a preset range, generally set to 65%-70% of the upper limit of the detection area.

[0016] If the number of pedestrians is less than the minimum value of the standard number range, the secondary area is deemed to be risk-free.

[0017] If the number of pedestrians is within the standard range, it is determined that there is a potential risk in the secondary area, and the time interval for obtaining the number of pedestrians in the secondary area is reduced.

[0018] If the number of pedestrians exceeds the highest value of the standard number range, the secondary area is judged as a dangerous area.

[0019] As a further technical solution of the present invention: the process of analyzing the pedestrian flow in the dangerous area and its adjacent areas includes:

[0020] During the time period t1,

[0021] Using Formula 1: Obtain the pedestrian load coefficient

[0022] Where O(t) is the function curve of the number of people leaving the danger zone during time period t1 as a function of time, I(t) is the function curve of the number of people entering the danger zone during time period t1 as a function of time, Δ is the predicted number of children under 1.2 meters tall in the danger zone, δ1 is the population flow variation coefficient, δ2 is the load replenishment coefficient, t1 is a standard detection duration, and N(t) is the function curve of the net increase in the number of people in the detection area as a function of time.

[0023] The above technical solution involves: judging the carrying capacity of the detected dangerous area by the trend of increasing number of people, and using the flow rate change coefficient obtained from the data on the direction of people in the flow and the load replenishment coefficient obtained from the data on the direction of people to assist in optimizing the judgment formula, so that the judgment process considers more variable parameters and is more accurate.

[0024] As a further technical solution of the present invention: the process by which the regional analysis module obtains analysis results and generates response strategies based on the analysis results includes:

[0025] The real-time pedestrian load coefficient will be obtained by calculating the danger zone using formula 1. With preset load range Perform a comparison;

[0026] like If the dangerous area is deemed overloaded, it is necessary to dispatch personnel to manage the situation.

[0027] like If the danger zone is deemed to be in a critical state and there is a risk of stampede, some entrances to the detection area will be closed.

[0028] like This indicates that the dangerous area is currently under control.

[0029] As a further technical solution of the present invention: the passenger flow variation coefficient δ1 is obtained by formula two. Obtain;

[0030] Where n is the number of exits from the danger zone, a k This represents the number of people who leave the danger zone via the k-th exit during time period t1. It is from a1 to a n A function to find the maximum value of all numbers.

[0031] As a further technical solution of the present invention: the load compensation coefficient δ2 is obtained through formula three. Obtain;

[0032] Where z is the number of adjacent regions to the danger zone, and R i (t) is a function curve showing the change in the number of people traveling from the danger zone to the i-th zone over time, where r iThis represents the number of people who have faced the i-th region within the time interval 0 to t1. It should be noted that r... i The number of people facing the i-th region is calculated non-repeatingly within the time interval 0 to t1. The calculation method can use a trained neural network model, which is existing technology and will not be elaborated here.

[0033] The above technical solution allows us to determine the current trend of population movement in dangerous areas by comparing the ratio of the number of people moving in the same direction as the trend to the number of people moving in a different direction. Furthermore, it allows us to determine the trend of population movement by comparing the ratio of the number of people moving to other areas adjacent to the dangerous area to the number of people moving towards those adjacent areas. This serves as a supplementary calculation to the previous method of determining the carrying capacity of dangerous areas based on the trend of increasing population numbers, resulting in a more accurate final outcome.

[0034] As a further technical solution of the present invention: the predicted quantity Δ is obtained through formula four. Obtain;

[0035] Where σ1 and σ2 are correction parameters that take values ​​of 1 or 0. If no new items attracting children are added in the danger zone during the detection time, σ1 = 1 and σ2 = 0. If new items attracting children are added in the danger zone during the detection time, σ1 = 0 and σ2 = 1. This is the attraction coefficient, which is the proportion of children who enter the detection area and remain in the danger zone, obtained based on empirical data. C is the total number of children who have entered the detection area, and N² is the number of pedestrians detected in the danger zone. A This represents the total number of pedestrians who have entered the testing area.

[0036] The above technical solution addresses the issue that image recognition is used to identify people in dangerous areas, which is ineffective or even impossible for shorter children. By predicting the number of children, the accuracy of load status prediction is further improved by adding the number of children to the load status prediction formula.

[0037] As a further technical solution of the present invention: the process of closing the entrance to the detection area includes:

[0038] Step 1: Obtain the image information of the current secondary region and obtain the entry location information of each person in the crowd in the secondary region who is in a different direction from the movement trend direction. The movement trend direction refers to the direction in which the most people are moving in the current secondary region.

[0039] Step 2: Close the entry point for obtaining location information in Step 1.

[0040] As a further technical solution of the present invention: the process of closing the entry point for obtaining location information in step one in step two further includes:

[0041] Obtain the pedestrian flow variation coefficient δ1, and find the number of closed entrances corresponding to the pedestrian flow variation coefficient δ1 in the dangerous area from the preset table. The preset table is a comparison table between the pedestrian flow variation coefficient δ1 obtained from empirical data and the number of closed entrances.

[0042] Sort all entry points for obtaining location information in descending order of the number of people located in the danger zone;

[0043] Based on the number m of closed entry points obtained from a preset table, the first m entry points are closed. The beneficial effects of this invention are:

[0044] (1) This invention divides the detection area into several secondary areas with equal walking area, then finds dangerous areas from the secondary areas, analyzes the flow of people in the dangerous areas and their adjacent areas, and generates response strategies based on the analysis results, thereby more accurately grasping the movement status of people in the detection area, and promptly handling areas where stampedes occur to reduce the risk of stampedes. At the same time, the analysis results use multiple types of data to complete the fusion of heterogeneous data, and the heterogeneous data have a high correlation.

[0045] (2) The present invention judges the carrying capacity of the detected dangerous area by the trend of increasing number of people, and uses the flow rate change coefficient obtained by the flow of people and the load replenishment coefficient obtained by the direction of people to assist in optimizing the judgment formula, so that the judgment process considers more variable parameters and is more accurate.

[0046] (3) In dangerous areas, the present invention, on the one hand, judges the current trend of population flow in dangerous areas by the ratio of the number of people moving in the same direction as the trend of movement to the number of people moving in different directions; on the other hand, judges the trend of population flow by the ratio of the number of people going to other areas adjacent to the dangerous area to the number of people heading towards the corresponding adjacent area within a period of time. This serves as a supplementary calculation to the judgment of the carrying capacity of dangerous areas by the trend of increasing population, making the final result more accurate.

[0047] (4) This invention uses image recognition to identify the number of people in dangerous areas. However, it has a problem that the identification effect of children with short stature is poor or even impossible. By predicting the number of children, the number of children is added to the prediction formula of the load status, thereby further improving the accuracy of the load status prediction. Attached Figure Description

[0048] The invention will now be further described with reference to the accompanying drawings.

[0049] Figure 1 This is a diagram showing the information flow relationship between the modules of this invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see Figure 1 As shown, in one embodiment, a public emergency detection system based on heterogeneous data fusion is provided, including:

[0052] The access control module includes a statistics unit for recording the number of people entering and exiting the detection area, and multiple access control units set at the entrances and exits of the detection area for restricting pedestrian access. The number of people entering and exiting is preferably recorded using facial recognition via camera to avoid double counting. The access control units are access control machines controlled by remote control devices, which are existing technologies and will not be described in detail.

[0053] The information acquisition module is used to acquire various types of relevant information, including at least the number of people, direction of movement, and facial orientation. The direction of movement or facial orientation is acquired through image recognition by a trained machine learning model. The main difference between the two is that the direction of movement requires two consecutive images, while the facial orientation can be acquired with one image. The specific acquisition process is existing technology and will not be elaborated here.

[0054] The area division module is used to divide the walkable area of ​​the detection area into several secondary areas with equal walkable areas and to identify the dangerous areas in the secondary areas. The equal walkable area module can unify the detection standards for dangerous areas and improve detection efficiency.

[0055] The area analysis module analyzes pedestrian traffic in dangerous areas and their adjacent areas, obtains analysis results, and generates response strategies based on the analysis results.

[0056] The above technical solution involves dividing the detection area into several secondary zones of equal walking area, then identifying dangerous areas within these secondary zones, analyzing the pedestrian flow in and around these dangerous areas, and generating response strategies based on the analysis results. This allows for a more accurate understanding of pedestrian movement within the detection area and timely intervention in areas prone to stampedes to reduce the risk of such accidents.

[0057] The process of determining the secondary region within the walkable area of ​​the detection area includes:

[0058] The number of pedestrians in the secondary area is acquired periodically. The number of pedestrians can be obtained through image counting software, which is existing technology and will not be elaborated here. The number of pedestrians is compared with a standard number range, which is a preset range, generally set to 65%-70% of the upper limit of the detection area.

[0059] If the number of pedestrians is less than the minimum value of the standard number range, the secondary area is deemed to be risk-free.

[0060] If the number of pedestrians is within the standard range, it is determined that there is a potential risk in the secondary area, and the time interval for obtaining the number of pedestrians in the secondary area is reduced.

[0061] If the number of pedestrians exceeds the highest value of the standard number range, the secondary area is judged as a dangerous area.

[0062] The process of analyzing pedestrian flow in the danger zone and its adjacent areas includes:

[0063] During the time period t1,

[0064] Using Formula 1: Obtain the pedestrian load coefficient It should be noted that before measuring the load status through the area of ​​change in tourist numbers, it is necessary to estimate the number of short children when tourists gather in large numbers, in order to correct the estimate of the load status and increase the accuracy of the estimate.

[0065] Where O(t) is the function curve of the number of people leaving the danger zone during time period t1 as a function of time, I(t) is the function curve of the number of people entering the danger zone during time period t1 as a function of time, Δ is the predicted number of children under 1.2 meters tall in the danger zone, δ1 is the population flow variation coefficient, δ2 is the load replenishment coefficient, t1 is a standard detection duration, and N(t) is the function curve of the net increase in the number of people in the detection area as a function of time.

[0066] The above technical solution involves: judging the carrying capacity of the detected dangerous area by the trend of increasing number of people, and using the flow rate change coefficient obtained from the data on the direction of people in the flow and the load replenishment coefficient obtained from the data on the direction of people to assist in optimizing the judgment formula, so that the judgment process considers more variable parameters and is more accurate.

[0067] The process by which the regional analysis module obtains analysis results and generates response strategies based on those results includes:

[0068] The real-time pedestrian load coefficient will be obtained by calculating the danger zone using formula 1. With preset load range A comparison was conducted, among which These are all constants determined by empirical data;

[0069] like If the dangerous area is deemed overloaded, it is necessary to dispatch personnel to manage the situation.

[0070] like If the danger zone is deemed to be in a critical state and there is a risk of stampede, some entrances to the detection area will be closed.

[0071] like This indicates that the dangerous area is currently under control.

[0072] The pedestrian flow variation coefficient δ1 is obtained through formula two. Obviously, the more exits a dangerous area has, the more factors will affect the final result. Taking a street with only two exits as an example, δ1 is the ratio of the number of people moving in two directions, with the direction with fewer people as the denominator. The final value of δ1 is between [0,1], which can represent the degree of internal chaos of the flow of people in the secondary area that is judged as a dangerous area. The larger the ratio, the more people are moving in opposite directions, and the load state coefficient of the area needs to be increased accordingly. If the ratio is close to zero, it means that all people are moving in the same direction, and this factor does not have a major impact on the load state and can be ignored.

[0073] Where n is the number of exits in the danger zone, which is measured by a predetermined constant based on a width that allows one person to pass at a time. For example, a gate with a width of 3 meters that can allow four people to pass at a time has four exits. k This represents the number of people who leave the danger zone via the k-th exit during time period t1. It is from a1 to a n A function to find the maximum value of all numbers.

[0074] The load compensation coefficient δ2 is obtained through formula three. Similarly, taking a street as an example, the number of adjacent areas of the dangerous area is 2. The load replenishment coefficient is the ratio of the change in the number of people moving from the dangerous area to two adjacent areas to the number of people who have looked at the two areas. This is used to predict the movement trend of the crowd. As a replenishment coefficient of the load state, it is composed of several quantities between [0,1]. The more secondary area exits the dangerous area has, the more channels the crowd has to disperse. The larger the proportion of δ2 in the final calculation process.

[0075] Where z is the number of adjacent regions to the danger zone, and R i (t) is a function curve showing the change in the number of people traveling from the danger zone to the i-th zone over time, where r iThis represents the number of people who have faced the i-th region within the time interval 0 to t1. It should be noted that r... i The number of people facing the i-th region is calculated non-repeatingly within the time interval 0 to t1. The calculation method can use a trained neural network model, which is existing technology and will not be elaborated here.

[0076] The above technical solution allows us to determine the current trend of population movement in dangerous areas by comparing the ratio of the number of people moving in the same direction as the trend to the number of people moving in a different direction. Furthermore, it allows us to determine the trend of population movement by comparing the ratio of the number of people moving to other areas adjacent to the dangerous area to the number of people moving towards those adjacent areas. This serves as a supplementary calculation to the previous method of determining the carrying capacity of dangerous areas based on the trend of increasing population numbers, resulting in a more accurate final outcome.

[0077] The predicted quantity Δ is obtained through formula four. To improve the accuracy of the estimation, when no new attractions for children are added, the number of children is estimated based on empirical data. When new attractions for children are added, the number of children is estimated based on the proportion of people in the crowd.

[0078] Where σ1 and σ2 are correction parameters that take values ​​of 1 or 0. If no new items attracting children are added in the danger zone during the detection time, σ1 = 1 and σ2 = 0. If new items attracting children are added in the danger zone during the detection time, σ1 = 0 and σ2 = 1. This is the attraction coefficient, representing the proportion of children who enter the detection area based on empirical data and will remain in the danger zone. C is the total number of children who have already entered the detection area, and N² is the number of pedestrians detected in the danger zone. A This represents the total number of pedestrians who have entered the testing area.

[0079] The above technical solution addresses the issue that image recognition is used to identify people in dangerous areas, which is ineffective or even impossible for shorter children. By predicting the number of children, the accuracy of load status prediction is further improved by adding the number of children to the load status prediction formula.

[0080] The process of closing the entrance to the detection area includes:

[0081] Step 1: Obtain the image information of the current secondary region and obtain the entry location information of each person in the crowd in the secondary region who is in a different direction from the movement trend direction. The movement trend direction refers to the direction in which the most people are moving in the current secondary region.

[0082] Step 2: Close the entry point for obtaining location information in Step 1.

[0083] The process of closing the entry point for obtaining location information in step one in step two also includes:

[0084] Obtain the pedestrian flow variation coefficient δ1, and find the number of closed entrances corresponding to the pedestrian flow variation coefficient δ1 in the dangerous area from the preset table. The preset table is a comparison table between the pedestrian flow variation coefficient δ1 obtained from empirical data and the number of closed entrances.

[0085] Sort all entry points for obtaining location information in descending order of the number of people located in the danger zone;

[0086] Based on the number m of closed entry points obtained from the preset table, close the first m entry points.

[0087] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A public emergency detection system based on heterogeneous data fusion, characterized in that, include: The access control module includes a statistics unit for recording the number of people entering and exiting the detection area, and multiple access control units set at the entrances and exits of the detection area to restrict pedestrian access. The information acquisition module is used to acquire various types of relevant information; The area division module is used to divide the walkable area of ​​the detection area into several secondary areas of equal walkable area and to identify the dangerous areas within the secondary areas; The area analysis module analyzes pedestrian traffic in hazardous areas and their adjacent areas, obtains analysis results, and generates response strategies based on the analysis results.

2. The public emergency detection system based on heterogeneous data fusion according to claim 1, characterized in that, The process of determining the secondary region within the walkable area of ​​the detection area includes: The number of pedestrians in the secondary area is periodically obtained and compared with the standard number range. If the number of pedestrians is less than the minimum value of the standard number range, the secondary area is deemed to be risk-free. If the number of pedestrians is within the standard range, it is determined that there is a potential risk in the secondary area, and the time interval for obtaining the number of pedestrians in the secondary area is reduced. If the number of pedestrians exceeds the highest value of the standard number range, the secondary area is judged as a dangerous area.

3. A public emergency detection system based on heterogeneous data fusion according to claim 1, characterized in that, The process of analyzing pedestrian flow in the danger zone and its adjacent areas includes: During the time period t1, Using Formula 1: Obtain the pedestrian load coefficient Where O(t) is the function curve of the number of people leaving the danger zone during time period t1 as a function of time, I(t) is the function curve of the number of people entering the danger zone during time period t1 as a function of time, Δ is the predicted number of children under 1.2 meters tall in the danger zone, δ1 is the population flow variation coefficient, δ2 is the load replenishment coefficient, t1 is a standard detection duration, and N(t) is the function curve of the net increase in the number of people in the detection area as a function of time.

4. A public emergency detection system based on heterogeneous data fusion according to claim 3, characterized in that, The process by which the regional analysis module obtains analysis results and generates response strategies based on those results includes: The real-time pedestrian load coefficient will be obtained by calculating the danger zone using formula 1. With preset load range Perform a comparison; like If the dangerous area is deemed overloaded, it is necessary to dispatch personnel to manage the situation. like If the danger zone is deemed to be in a critical state and there is a risk of stampede, some entrances to the detection area will be closed. like This indicates that the dangerous area is currently under control.

5. A public emergency detection system based on heterogeneous data fusion according to claim 3, characterized in that, The pedestrian flow variation coefficient δ1 is obtained through formula two. Obtain; Where n is the number of exits from the danger zone, a k This represents the number of people who leave the danger zone via the k-th exit during time period t1. It is from a1 to a n A function to find the maximum value of all numbers.

6. A public emergency detection system based on heterogeneous data fusion according to claim 3, characterized in that, The load compensation coefficient δ2 is obtained through formula three. Obtain; Where z is the number of adjacent regions to the danger zone, and R i (t) is a function curve showing the change in the number of people traveling from the danger zone to the i-th zone over time, where r i It represents the number of people who have faced the i-th region within the time interval 0 to t1.

7. A public emergency detection system based on heterogeneous data fusion according to claim 3, characterized in that, The predicted quantity Δ is obtained through formula four. Obtain; Where σ1 and σ2 are correction parameters that take values ​​of 1 or 0; This is the attraction coefficient, where C is the total number of children already in the detection area, and N² is the number of pedestrians detected in the danger zone. A This represents the total number of pedestrians who have entered the testing area.

8. A public emergency detection system based on heterogeneous data fusion according to claim 4, characterized in that, The process of closing the entrance to the detection area includes: Step 1: Obtain the image information of the current secondary region and obtain the entry location information of each person in the crowd in the secondary region who is in a different direction from the movement trend direction. The movement trend direction refers to the direction in which the most people are moving in the current secondary region. Step 2: Close the entry point for obtaining location information in Step 1.

9. A public emergency detection system based on heterogeneous data fusion according to claim 8, characterized in that, The process of closing the entry point for obtaining location information in step one in step two also includes: Obtain the pedestrian flow variation coefficient δ1, and find the number of closed entrances corresponding to the pedestrian flow variation coefficient δ1 in the dangerous area from the preset table; Sort all entry points for obtaining location information in descending order of the number of people located in the danger zone; Based on the number m of closed entry points obtained from the preset table, close the first m entry points.

Images