Gymnasium evacuation scheme intelligent pushing method based on group movement simulation technology

Abstract

The invention discloses a gymnasium evacuation scheme intelligent pushing method based on a group movement simulation technology, and relates to the technical field of building safety simulation. The method comprises the following steps: firstly, determining the current position and the current speed of personnel in each gymnasium in real time by adopting a multi-target tracking algorithm and a cross-camera multi-view scene target continuous tracking technology according to field video data, and then fusing three-dimensional virtual scene data of a target gymnasium and an emergency evacuation scheme, a three-dimensional virtual emergency evacuation scene is obtained, then a crowd dynamics simulation model used for realizing a crowd movement simulation technology is initialized based on the evacuation scene and the current position and the current speed of each person, and the model is started to simulate the current crowd movement simulation duration for evacuating all persons in a hall out of a target gymnasium. And finally, the current availability of the emergency evacuation scheme is determined according to the comparison result of the simulation duration and a preset duration threshold value, and the purpose of dynamically adapting to the actual real-time situation in the gymnasium is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of building safety simulation, and particularly relates to an intelligent push method for a gymnasium evacuation plan based on group movement simulation technology. Background Art

[0002] Modern stadiums are public buildings with a large number of occupants and a high population density. Therefore, the safety evacuation design of the stadiums becomes particularly important. In recent years, a composite function stadium that can meet various needs such as art performances, exhibitions, fitness, and entertainment in addition to competitions has gradually been built. When a stadium has a composite function, the arrangement of the positions of the audience inside the stadium under different usage functions will be somewhat different, making it necessary to have different emergency evacuation plans for the stadium in case of emergencies such as fires.

[0003] In order to cope with different personnel distribution situations inside the stadium, multiple different emergency evacuation plans are generally pre-designed for the stadium. However, the existing evacuation effect testing methods for emergency evacuation plans are generally achieved through theoretical verification and / or fire drill tests. Among them, the former may have a disconnection between theory and actual situation, resulting in the problem that the evacuation effect test results do not match the actual situation (especially seriously do not match the actual real-time situation), while the latter can match some actual preset situations, but the cost is high and the time consumption is long. At the same time, both of them cannot dynamically adapt to the actual real-time situation inside the stadium, resulting in the risk of using an inapplicable stadium emergency evacuation plan for evacuation in case of emergencies such as fires. Therefore, how to consider the actual real-time situation inside the stadium and quickly and low-costly complete the evacuation test task of the stadium emergency evacuation plan in order to determine the dynamic feasibility of the plan is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent push method, device, computer device, and computer-readable storage medium for a gymnasium evacuation plan based on group movement simulation technology, so as to solve the problem that the existing evacuation effect testing methods for emergency evacuation plans cannot dynamically adapt to the actual real-time situation inside the stadium, resulting in the risk of using an inapplicable stadium emergency evacuation plan for evacuation in case of emergencies such as fires.

[0005] In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0006] In the first aspect, an intelligent push method for a gymnasium evacuation plan based on group movement simulation technology is provided, including:

[0007] Obtaining on-site video data collected in real time by multiple monitoring cameras of a target stadium;

[0008] According to the on-site video data, the multi-object tracking algorithm and the cross-camera multi-view scenario object continuous tracking technology are used to track in real time each person inside the target stadium, and the tracking result of the people inside the stadium is obtained;

[0009] According to the tracking result of the people inside the stadium, the current three-dimensional coordinate positions and current moving speeds of each person inside the stadium in the target stadium are determined in real time;

[0010] The three-dimensional virtual scene data of the target stadium and the emergency evacuation plan are obtained, wherein the emergency evacuation plan includes multi-dimensional feature data of at least one available evacuation passage preset for the target stadium, and the multi-dimensional feature data includes three-dimensional coordinate data and three-dimensional model data of the corresponding passage in the target stadium;

[0011] According to the three-dimensional coordinate data of each available evacuation passage in the at least one available evacuation passage, the three-dimensional model data of each available evacuation passage is respectively fused into the three-dimensional virtual scene data to obtain the three-dimensional virtual emergency evacuation scene of the target stadium;

[0012] Initialize the crowd dynamics simulation model for implementing the group motion simulation technology: initialize the application scenario of the crowd dynamics simulation model as the three-dimensional virtual emergency evacuation scene, and for each particle in the crowd dynamics simulation model that corresponds one-to-one with each person inside the stadium, determine the initial position of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current three-dimensional coordinate position of the corresponding person, and also determine the initial speed of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current moving speed of the corresponding person;

[0013] Start the crowd dynamics simulation model to simulate and update the real-time positions and real-time speeds of each particle leaving the three-dimensional virtual emergency evacuation scene, and stop the simulation until all the particles have left the three-dimensional virtual emergency evacuation scene, and obtain the current group motion simulation duration for evacuating all the people inside the stadium out of the target stadium;

[0014] Judge whether the current group motion simulation duration is less than or equal to a preset duration threshold;

[0015] If so, push the emergency evacuation plan as the current available evacuation plan of the target stadium to the decision-making terminal.

[0016] Based on the above invention content, a new solution is provided that can sense the actual real-time situation in the stadium and quickly and low-costly complete the evacuation test task of the stadium emergency evacuation plan. That is, first, according to the on-site video data collected in real time by multiple monitoring cameras of the target stadium, using multi-object tracking algorithms and cross-camera multi-view scene object continuous tracking technology, the current three-dimensional coordinate positions and current moving speeds of each person in the stadium are determined in real time. Then, the three-dimensional virtual scene data of the target stadium and the emergency evacuation plan are fused to obtain a three-dimensional virtual emergency evacuation scene. Then, based on the three-dimensional virtual emergency evacuation scene and the current three-dimensional coordinate positions and current moving speeds of each person in the stadium, the crowd dynamics simulation model for realizing the group movement simulation technology is initialized, and the model is started to simulate the current group movement simulation duration for evacuating all people in the stadium out of the target stadium. Finally, the current availability of the emergency evacuation plan is determined according to the comparison result between the current group movement simulation duration and the preset duration threshold. In this way, it can dynamically adapt to the actual real-time situation in the stadium, so that even in the event of an emergency such as a fire, it is possible to timely make a decision to adopt a suitable stadium emergency evacuation plan for evacuation, reduce the losses of stadium safety accidents, and facilitate practical application and promotion.

[0017] In a possible design, the multi-object tracking algorithm uses the deepsort object tracking algorithm. Among them, the deepsort object tracking algorithm includes: for unmatched tracking objects, it is judged whether they leave the target stadium according to the corresponding current predicted position. If so, the tracking object is discarded during the next position prediction; otherwise, the Kalman filtering method is still used to predict the corresponding position during the next position prediction.

[0018] In a possible design, obtaining the emergency evacuation plan of the target stadium includes:

[0019] When receiving a fire alarm message from a fire alarm, according to the known installation position of the fire alarm, at least one available evacuation passage that bypasses the known installation position is screened out from multiple preset evacuation passages for the target stadium, and then the multi-dimensional feature data of the at least one available evacuation passage is used as the emergency evacuation plan of the target stadium. Among them, the fire alarm is pre-installed in the target stadium, and the multi-dimensional feature data includes but is not limited to the three-dimensional coordinate data and three-dimensional model data of the corresponding passage in the target stadium.

[0020] In a possible design, when the multiple monitoring cameras include an entrance monitoring camera whose field of view covers the entrance area of the target stadium, the method further includes:

[0021] Extract the in - museum video images of each person in the museum from the on - site video data collected by the in - museum entrance monitoring camera;

[0022] For each person in the museum, identify the corresponding individual heterogeneity features according to the corresponding in - museum video image;

[0023] In the process of starting the crowd dynamics simulation model to simulate and update the real - time positions and real - time speeds of each particle leaving the three - dimensional virtual emergency evacuation scene, first, for each person in the museum, according to the corresponding individual heterogeneity features, assign time - varying physiological coefficients and psychological coefficients to the corresponding particles. Then, for each particle, based on the corresponding physiological coefficients and psychological coefficients, determine the corresponding real - time expected speed that varies with time. Finally, for each particle, according to the corresponding real - time expected speed and real - time speed, determine the corresponding real - time acceleration, and use this real - time acceleration to update the corresponding real - time speed and real - time position.

[0024] In a possible design, for each person in the museum, identifying the corresponding individual heterogeneity features according to the corresponding in - museum video image includes:

[0025] For each person in the museum, import the corresponding in - museum video image into a pre - trained multi - dimensional individual heterogeneity feature recognition model based on a convolutional neural network to obtain the corresponding multi - dimensional individual heterogeneity feature recognition result, where the multi - dimensional individual heterogeneity feature recognition model includes an individual gender feature recognition sub - model, an individual height feature recognition sub - model, an individual disability feature recognition sub - model, an individual age feature recognition sub - model, an individual weight feature recognition sub - model, and / or an individual personality feature recognition sub - model.

[0026] In a possible design, when there are multiple emergency evacuation plans, the method further includes:

[0027] Determine an emergency evacuation plan with the shortest current group movement simulation duration from multiple emergency evacuation plans, and push the certain emergency evacuation plan as the current available evacuation plan of the target stadium to the decision - making terminal.

[0028] In a possible design, when the number of emergency evacuation plans exceeds a preset value, determining an emergency evacuation plan with the shortest current group movement simulation duration from multiple emergency evacuation plans includes the following steps S91 - S98:

[0029] S91. Initialize the population: Set the number of gray wolves to N and the number of iterations to t maxNext, initialize the search range of the number of available evacuation channels and the search range of multi-dimensional feature data in the emergency evacuation plan, and then execute step S92, where N represents a positive integer greater than or equal to 3, tmax represents a positive integer, and (t max +1)×N is less than the number of the emergency evacuation plans;

[0030] S92. Initialize the grey wolves: Randomly select three grey wolves from the N grey wolves as the initial α wolf, β wolf, and δ wolf, and initialize and set the individual position vectors of each grey wolf in the N grey wolves within the search range of the number of available evacuation channels and the search range of the multi-dimensional feature data, and then execute step S93, where the individual position vector includes the search value of the number of available evacuation channels and the search value of the multi-dimensional feature data;

[0031] S93. For each grey wolf, first determine a most similar emergency evacuation plan from the multiple emergency evacuation plans according to the corresponding current individual position vector, then based on the most similar emergency evacuation plan, use the crowd dynamics simulation model to simulate and obtain the corresponding current group movement simulation duration, and finally use the current group movement simulation duration as the corresponding individual fitness value, and then execute step S94;

[0032] S94. Take the grey wolf with the minimum individual fitness value as the new α wolf, the grey wolf with the second minimum individual fitness value as the new β wolf, and the grey wolf with the third minimum individual fitness value as the new δ wolf, and determine whether the current iteration number reaches t max times. If so, execute step S98; otherwise, execute step S95;

[0033] S95. Calculate the convergence factor coordination vector and coordination vector Then execute step S96, where the convergence factor the coordination vector and the coordination vector The calculation formulas are as follows:

[0034]

[0035] In the formula, t represents the current iteration number, tanh() represents the hyperbolic tangent function, and respectively represent random vectors in [0,1];

[0036] S96. For each ω wolf, calculate the corresponding individual position vector Then, step S97 is executed, where the individual position vector is calculated according to the following formula:

[0037]

[0038] In the formula, represents the current individual position vector of the new α wolf, represents the current individual position vector of the new β wolf, represents the current individual position vector of the new δ wolf, represents the individual position vector in the t-th iteration, and respectively represent the collaborative vectors and respectively represent the collaborative vectors

[0039] S97. Increment the iteration count by 1, and then return to execute step S93;

[0040] S98. Determine the most similar emergency evacuation plan corresponding to the new α wolf as an emergency evacuation plan with the shortest current group movement simulation duration.

[0041] Second, an intelligent push device for stadium evacuation plans based on group movement simulation technology is provided, including a video data acquisition module, an in-venue personnel tracking module, a current information determination module, an evacuation plan acquisition module, an evacuation scenario fusion module, a simulation model initialization module, a simulation duration determination module, a judgment module, and an available plan push module;

[0042] The video data acquisition module is used to acquire on-site video data collected in real time by multiple monitoring cameras of the target stadium;

[0043] The in-venue personnel tracking module is communicatively connected to the video data acquisition module and is used to, according to the on-site video data, use a multi-target tracking algorithm and cross-camera multi-view scene target continuous tracking technology to track in real time each in-venue personnel appearing inside the target stadium and obtain the in-venue personnel tracking result;

[0044] The current information determination module is communicatively connected to the in-venue personnel tracking module and is used to, according to the in-venue personnel tracking result, determine in real time the current three-dimensional coordinate positions and current moving speeds of the respective in-venue personnel in the target stadium;

[0045] The evacuation plan acquisition module is used to acquire the three-dimensional virtual scene data of the target stadium and the emergency evacuation plan, where the emergency evacuation plan includes multi-dimensional feature data of at least one available evacuation route preset for the target stadium, and the multi-dimensional feature data includes three-dimensional coordinate data and three-dimensional model data of the corresponding route in the target stadium;

[0046] The evacuation scene fusion module, communicatively connected to the evacuation plan acquisition module, is used to respectively fuse the three-dimensional model data of each available evacuation route in the at least one available evacuation route into the three-dimensional virtual scene data according to the three-dimensional coordinate data of each available evacuation route, so as to obtain the three-dimensional virtual emergency evacuation scene of the target stadium;

[0047] The simulation model initialization module, communicatively connected to the current information determination module and the evacuation scene fusion module respectively, is used to initialize the crowd dynamics simulation model for implementing the group movement simulation technology: initialize the application scenario of the crowd dynamics simulation model as the three-dimensional virtual emergency evacuation scene, and for each particle in the crowd dynamics simulation model corresponding one-to-one to each person inside the stadium, determine the initial position of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current three-dimensional coordinate position of the corresponding person, and also determine the initial speed of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current moving speed of the corresponding person;

[0048] The simulation duration determination module, communicatively connected to the simulation model initialization module, is used to start the crowd dynamics simulation model to simulate and update the real-time positions and real-time speeds of each particle leaving the three-dimensional virtual emergency evacuation scene, and stop the simulation until all the particles have left the three-dimensional virtual emergency evacuation scene, so as to obtain the current group movement simulation duration for evacuating all the people inside the stadium from the target stadium;

[0049] The judgment module, communicatively connected to the simulation duration determination module, is used to judge whether the current group movement simulation duration is less than or equal to a preset duration threshold;

[0050] The available plan push module, communicatively connected to the judgment module, is used to, when it is determined that the current group movement simulation duration is less than or equal to the preset duration threshold, push the emergency evacuation plan as the current available evacuation plan of the target stadium to the decision-making terminal.

[0051] In a third aspect, the present invention provides a computer device, including a memory, a processor, and a transceiver that are communicatively connected in sequence. Wherein, the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer programs and execute the intelligent push method for the stadium evacuation plan as described in the first aspect or any possible design in the first aspect.

[0052] In a fourth aspect, the present invention provides a computer-readable storage medium, on which instructions are stored. When the instructions run on a computer, they execute the intelligent push method for the stadium evacuation plan as described in the first aspect or any possible design in the first aspect.

[0053] In a fifth aspect, the present invention provides a computer program product containing instructions. When the instructions run on a computer, the computer is made to execute the intelligent push method for the stadium evacuation plan as described in the first aspect or any possible design in the first aspect.

[0054] Beneficial effects of the above solutions:

[0055] (1) The present invention creatively provides a new solution that can sense the actual real-time situation in the stadium and quickly and low-costly complete the evacuation test task of the stadium emergency evacuation plan. That is, first, according to the on-site video data collected in real time by multiple monitoring cameras of the target stadium, using multi-object tracking algorithms and cross-camera multi-view scene object continuous tracking technology, the current three-dimensional coordinate positions and current moving speeds of each person in the stadium are determined in real time. Then, the three-dimensional virtual scene data of the target stadium and the emergency evacuation plan are fused to obtain a three-dimensional virtual emergency evacuation scene. Then, based on the three-dimensional virtual emergency evacuation scene and the current three-dimensional coordinate positions and current moving speeds of each person in the stadium, the crowd dynamics simulation model for realizing the group movement simulation technology is initialized, and the model is started to simulate the current group movement simulation duration for evacuating all people in the stadium out of the target stadium. Finally, according to the comparison result between the current group movement simulation duration and the preset duration threshold, the current availability of the emergency evacuation plan is determined. In this way, it can dynamically adapt to the actual real-time situation in the stadium, so that even in the event of an emergency such as a fire occurring currently, it is possible to timely make a decision to adopt a suitable stadium emergency evacuation plan for evacuation, reduce the losses of stadium safety accidents, and facilitate practical application and promotion. Description of the Drawings

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0057] Figure 1 It is a schematic flowchart of the intelligent push method for the stadium evacuation plan based on the group motion simulation technology provided by the embodiments of the present application.

[0058] Figure 2 It is a schematic structural diagram of the intelligent push device for the stadium evacuation plan based on the group motion simulation technology provided by the embodiments of the present application.

[0059] Figure 3 It is a schematic structural diagram of the computer device provided by the embodiments of the present application. Detailed implementation manners

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the present invention in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the drawing structures is only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings. It should be noted here that the description of these embodiment modes is used to help understand the present invention, but does not constitute a limitation to the present invention.

[0061] It should be understood that although terms such as first and second etc. may be used herein to describe various objects, these objects should not be limited by these terms. These terms are only used to distinguish one object from another. For example, the first object can be called the second object, and similarly, the second object can be called the first object, without departing from the scope of the exemplary embodiments of the present invention.

[0062] It should be understood that for the term "and / or" that may appear in this article, it is only a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, B exists alone, or A and B exist simultaneously, etc. three situations; another example, A, B and / or C can represent any one of A, B and C or any combination of them; for the term " / and" that may appear in this article, it is a description of another association object relationship, indicating that two relationships can exist. For example, A / and B can represent: A exists alone or A and B exist simultaneously, etc. two situations; in addition, for the character " / " that may appear in this article, generally it means that the front and rear associated objects are an "or" relationship.

[0063] Example:

[0064] As Figure 1 shown, the intelligent push method for the stadium evacuation plan provided in the first aspect of this embodiment and based on the group movement simulation technology can be, but is not limited to, executed by a computer device with certain computing resources, such as an electronic device such as a platform server in the stadium security center. As Figure 1 shown, the intelligent push method for the stadium evacuation plan can be, but is not limited to, including the following steps S1 to S9.

[0065] S1. Obtain the on-site video data collected in real time by multiple monitoring cameras of the target stadium.

[0066] In step S1, the multiple monitoring cameras are mainly used for remote monitoring and data retention inside the stadium. The lens fields of view of the multiple monitoring cameras will cover the entire area inside the stadium, and are used to collect video frame images of the area inside the target stadium in real time, obtaining on-site video data containing a number of continuous video frame images. For the purpose of facilitating the subsequent accurate determination of the current three-dimensional coordinate positions of each person inside the stadium in the target stadium, the monitoring cameras preferably adopt binocular cameras, so that each pixel point has three-dimensional coordinates in the camera coordinate system, and then based on multiple pixel points corresponding to the people inside the stadium and the known installation positions of the binocular cameras in the target stadium, the three-dimensional coordinate positions of the people inside the stadium in the target stadium are obtained through conventional conversion. In addition, the multiple monitoring cameras can transmit the collected data to local devices through conventional means.

[0067] S2. According to the on-site video data, use the multi-target tracking algorithm and the cross-camera multi-view scene target continuous tracking technology to continuously track each person inside the target stadium in real time, and obtain the tracking results of the people inside the stadium.

[0068] In the step S2, the multi-object tracking algorithm is used to detect the images of the people inside the museum located in the video frame image, and perform video tracking on the detected people inside the museum (during the tracking process, unique person numbers can be assigned to each detected person inside the museum, such as the numerical numbers 1, 2, 3, or 4, etc.). The multi-object tracking algorithm preferably adopts the deepsort object tracking algorithm. Among them, the specific process of the deepsort object tracking algorithm is as follows: First, the object detector detects the object bounding box bbox, and generates detection target information detections based on the object bounding box bbox (which is used to save all the detected targets in the current frame image). Then, the Kalman filtering method is used to predict the positions of the tracking target information tracks (which is used to save all the targets tracked in the previous frame image) in the current frame image. Then, the Mahalanobis distance based on the appearance features is used to calculate the cost matrix of the tracking target and the detection target. Then, the tracking target and the detection target are successively subjected to cascade matching and IOU (Intersection over Union) matching. Finally, all the matching pairs, unmatched tracking targets, and unmatched detection targets in the current frame image are obtained, and for each successfully matched tracking target, its corresponding detection target is used for position update, as well as processing the unmatched tracking targets and detection targets. In addition, in the deepsort object tracking algorithm, the object detector, the Kalman filtering method, the appearance features, the Mahalanobis distance, the cosine distance, the cost matrix, the cascade matching, and the IOU matching are all existing terms or technical features, and those skilled in the art can routinely obtain the specific process details of the deepsort object tracking algorithm. In addition, the cross-camera multi-view scene object continuous tracking technology refers to the cross-camera tracking of moving objects in multiple different monitoring scenes, which can be specifically implemented by using existing technologies.

[0069] This embodiment considers that the unmatched tracking targets may include occluded targets that are not detected in the current frame image. Therefore, in the deepsort target tracking algorithm, it preferably includes: for the unmatched tracking targets, determining whether they leave the target stadium according to the corresponding current predicted position (that is, judging whether they still appear in the frame image according to the position relationship between the current predicted position and the boundary of the frame image. If so, it is determined that they do not leave the target stadium, otherwise it is determined that they leave the target stadium). If so, discard the tracking target during the next position prediction, otherwise still use the Kalman filter method to predict the corresponding position during the next position prediction. In this way, when it is found that a tracking target is occluded (that is, when the tracking target is not matched with a detection target and the current predicted position still appears in the current frame image), the tracking target can be marked as occluded (masked), and then still use the Kalman filter method to predict the position in the next frame image until it is matched with a detection target or it is found that the tracking target exceeds the range of the image, thereby avoiding premature termination of video tracking due to occlusion compared with general discard processing methods and ensuring long-term continuity of tracking.

[0070] S3. According to the in-stadium personnel tracking results, determine the current three-dimensional coordinate positions and current moving speeds of the respective in-stadium personnel in the target stadium in real time.

[0071] In the step S3, the current three-dimensional coordinate positions can be obtained by conventional conversion based on the known installation positions of the monitoring cameras in the target stadium and the coordinate positions of all pixel points of the corresponding in-stadium personnel in the camera coordinate system; the current moving speeds can be calculated conventionally based on the three-dimensional coordinate positions of the corresponding in-stadium personnel at the previous moment, the current three-dimensional coordinate positions, and the time difference between the current moment and the previous moment.

[0072] S4. Obtain the three-dimensional virtual scene data of the target stadium and the emergency evacuation plan, where the emergency evacuation plan includes, but is not limited to, multi-dimensional feature data of at least one available evacuation channel preset for the target stadium, and the multi-dimensional feature data includes, but is not limited to, three-dimensional coordinate data and three-dimensional model data of the corresponding channel in the target stadium, etc.

[0073] In the step S4, the three-dimensional virtual scene data can be obtained by pre-conventional modeling based on the design drawings of the target stadium or the results of drone oblique photography of the target stadium. The emergency evacuation plan can be prefabricated manually before the activities in the stadium or formed temporarily according to the occurrence of emergencies in the stadium. Specifically, obtaining the emergency evacuation plan of the target stadium includes, but is not limited to: when receiving a fire alarm message from a fire alarm, according to the known installation location of the fire alarm, screening at least one available evacuation passage that bypasses the known installation location from multiple evacuation passages preset for the target stadium, and then using the multi-dimensional feature data of the at least one available evacuation passage as the emergency evacuation plan of the target stadium. Among them, the fire alarm is pre-installed in the target stadium, and the multi-dimensional feature data includes, but is not limited to, the three-dimensional coordinate data and three-dimensional model data of the corresponding passage in the target stadium. In addition, the three-dimensional model data can be specifically, but not limited to, model data such as passage length, passage width, passage slope, and passage height.

[0074] S5. According to the three-dimensional coordinate data of each available evacuation passage in the at least one available evacuation passage, respectively fuse the three-dimensional model data of each available evacuation passage into the three-dimensional virtual scene data to obtain the three-dimensional virtual emergency evacuation scene of the target stadium.

[0075] In the step S5, the specific fusion method is to place the three-dimensional models of the available evacuation passages into the three-dimensional virtual scene of the target stadium to obtain the three-dimensional virtual emergency evacuation scene.

[0076] S6. Initialize the crowd dynamics simulation model for implementing the group movement simulation technology: initialize the application scenario of the crowd dynamics simulation model as the three-dimensional virtual emergency evacuation scene, and for each particle in the crowd dynamics simulation model corresponding one-to-one to each person in the stadium, determine the initial position of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current three-dimensional coordinate position of the corresponding person, and also determine the initial velocity of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current moving speed of the corresponding person.

[0077] In the step S6, the group motion simulation technology is to study the motion of real groups under specific environments and specific constraint conditions, analyze the motion characteristics and laws of the groups, thereby establishing a simulation model of large-scale group motion, and using relevant simulation tools to generate the real group motion process in the computer virtual environment. Therefore, the crowd dynamics simulation model can be specifically implemented by using existing models. For example, the Social Force Model (SFM, which is a widely used crowd dynamics simulation model at present) can be used to simulate the motion patterns of crowds in specific scenarios (in this embodiment, it is the three-dimensional virtual emergency evacuation scenario).

[0078] S7. Start the crowd dynamics simulation model to simulate and update the real-time positions and real-time speeds of the respective particles leaving the three-dimensional virtual emergency evacuation scenario, and stop the simulation until all the particles have left the three-dimensional virtual emergency evacuation scenario, so as to obtain the current group motion simulation duration for evacuating all the people in the venue out of the target stadium.

[0079] In the step S7, the foregoing simulation of updating the real-time positions and real-time speeds of the respective particles leaving the three-dimensional virtual emergency evacuation scenario is the essential work of the crowd dynamics simulation model. Therefore, the current group motion simulation duration corresponding to the emergency evacuation plan can be routinely obtained based on the existing group motion simulation technology. In addition, the current group motion simulation duration does not necessarily need to be equal to the real duration from the start to the stop of the crowd dynamics simulation model, because the real-time positions and real-time speeds can be updated at a multiple speed after starting the crowd dynamics simulation model.

[0080] S8. Determine whether the current group motion simulation duration is less than or equal to a preset duration threshold.

[0081] In the step S8, the preset duration threshold can be routinely obtained in advance according to the security design requirements of the target stadium, for example, it is 10 minutes.

[0082] S9. If so, push the emergency evacuation plan as the current available evacuation plan of the target stadium to the decision-making terminal.

[0083] In step S9, when the current group movement simulation duration is less than or equal to the preset duration threshold, it indicates that the emergency evacuation plan meets the security design requirements of the target stadium. Therefore, it can be pushed to the decision-making terminal as the current available evacuation plan for the target stadium, so that in case of an emergency such as a fire, a timely decision can be made for application. Otherwise, it is not the case. In addition, when there are multiple emergency evacuation plans, in order to achieve the purpose of preferentially recommending the best emergency evacuation plan, preferably, the method further includes, but is not limited to: determining an emergency evacuation plan with the shortest current group movement simulation duration from multiple emergency evacuation plans, and pushing the emergency evacuation plan as the current available evacuation plan for the target stadium to the decision-making terminal. The specific way to determine an emergency evacuation plan with the shortest current group movement simulation duration from multiple emergency evacuation plans includes, but is not limited to: first, for each emergency evacuation plan, using the aforementioned steps S5 to S7 to simulate and obtain the corresponding current group movement simulation duration, and then selecting an emergency evacuation plan corresponding to the shortest current group movement simulation duration.

[0084] Based on the intelligent push method of the stadium evacuation plan described in the aforementioned steps S1 to S9, a new solution is provided that can sense the actual real-time situation in the stadium and quickly and low-costly complete the evacuation test task of the stadium emergency evacuation plan. That is, first, according to the on-site video data collected in real time by multiple monitoring cameras of the target stadium, using the multi-object tracking algorithm and the cross-camera multi-view scene target continuous tracking technology, the current three-dimensional coordinate positions and current moving speeds of each person in the stadium are determined in real time. Then, the three-dimensional virtual scene data of the target stadium and the emergency evacuation plan are fused to obtain a three-dimensional virtual emergency evacuation scene. Then, based on the three-dimensional virtual emergency evacuation scene and the current three-dimensional coordinate positions and current moving speeds of each person in the stadium, the crowd dynamics simulation model for implementing the group movement simulation technology is initialized, and the model is started to simulate the current group movement simulation duration for evacuating all people in the stadium out of the target stadium. Finally, according to the comparison result between the current group movement simulation duration and the preset duration threshold, the current availability of the emergency evacuation plan is determined. In this way, it can dynamically adapt to the actual real-time situation in the stadium, so that even in case of an emergency such as a fire, a timely decision can be made to adopt a suitable stadium emergency evacuation plan for evacuation, reducing the losses of stadium safety accidents and facilitating practical application and promotion.

[0085] Based on the technical solution of the foregoing first aspect, this embodiment further provides a possible design 1 for how to perceive the individual differences of different in - museum personnel and apply them to group motion simulation. That is, when the multiple monitoring cameras include an in - museum entrance monitoring camera whose field of view covers the entrance area of the target gymnasium, the method further includes but is not limited to the following steps S101 to S103.

[0086] S101. Crop the in - museum video images of each in - museum personnel from the on - site video data collected by the in - museum entrance monitoring camera.

[0087] In step S101, the specific cropping method of the in - museum video image may include but is not limited to: first, perform personnel recognition processing on each video frame image in the on - site video data collected by the in - museum entrance monitoring camera based on a target detection algorithm (which is an existing algorithm), then regard the recognized personnel as in - museum personnel, and for each recognized in - museum personnel, crop the image within the corresponding personnel bounding box as the corresponding in - museum video image.

[0088] S102. For each in - museum personnel, identify the corresponding individual heterogeneity features according to the corresponding in - museum video image.

[0089] In the step S102, specifically, for each person in the museum, corresponding individual heterogeneity features are identified based on the corresponding in-museum video images, including but not limited to: for each person in the museum, the corresponding in-museum video image is imported into a multi-dimensional individual heterogeneity feature recognition model based on a convolutional neural network and pre-trained, and a corresponding multi-dimensional individual heterogeneity feature recognition result is obtained. Among them, the multi-dimensional individual heterogeneity feature recognition model includes but not limited to an individual gender feature recognition sub-model, an individual height feature recognition sub-model, an individual disability feature recognition sub-model, an individual age feature recognition sub-model, an individual weight feature recognition sub-model, and / or an individual personality feature recognition sub-model, etc. The convolutional neural network (Convolutional Neural Networks, CNN) is a type of feedforward neural network (Feedforward Neural Networks) that contains convolutional calculations and has a deep structure. It is one of the representative algorithms of deep learning (deeplearning). The convolutional neural network has the ability of representation learning and can perform shift-invariant classification on the input information according to its hierarchical structure. Therefore, it is also called "Shift-Invariant Artificial Neural Networks (SIANN)". Therefore, based on a certain amount of sample images and individual heterogeneity feature labels marked for the sample images, the individual gender feature recognition sub-model, the individual height feature recognition sub-model, the individual disability feature recognition sub-model, the individual age feature recognition sub-model, the individual weight feature recognition sub-model, and / or the individual personality feature recognition sub-model, etc. can be trained through conventional model training methods, so that the multi-dimensional individual heterogeneity feature recognition result includes but not limited to the identified individual gender feature (such as male or female), individual height feature (such as tall, medium or short), individual disability feature (such as no disability, single-leg disability and double-leg disability), individual age feature (such as toddler, teenager, youth, middle-aged or elderly), individual weight feature (such as fat, medium or thin), and / or individual personality feature (such as introverted personality or extroverted personality), etc.

[0090] S103. In the process of starting the population dynamics simulation model to simulate and update the real-time positions and real-time velocities of the respective particles leaving the three-dimensional virtual emergency evacuation scene, first, for each person in the hall, according to the corresponding individual heterogeneity characteristics, physiological coefficients and psychological coefficients that vary with time are assigned to the corresponding particles. Then, for each particle, based on the corresponding physiological coefficients and psychological coefficients, a corresponding real-time expected velocity that varies with time is determined. Finally, for each particle, according to the corresponding real-time expected velocity and real-time velocity, a corresponding real-time acceleration is determined, and the real-time velocity and real-time position are updated using this real-time acceleration.

[0091] In step S103, the specific means of assigning the physiological coefficients and psychological coefficients, the specific means of determining the real-time expected velocity, and the specific means of determining the real-time acceleration are all prior art means and will not be elaborated here.

[0092] Based on the foregoing possible design one, it is also possible to sense the individual differences of different people in the hall according to the on-site video data collected by the entrance monitoring cameras in the hall, and apply the individual differences of the different people in the hall to the group movement simulation, so as to further conform to the actual real-time situation in the stadium and ensure the simulation accuracy of the group movement simulation duration.

[0093] Based on the technical solution of the foregoing first aspect, this embodiment also provides a possible design two on how to complete the optimal recommendation of the best emergency evacuation plan with low resource requirements, that is, when the number of the emergency evacuation plans exceeds a preset value, an emergency evacuation plan with the shortest current group movement simulation duration is determined from multiple emergency evacuation plans, including but not limited to the following steps S91 - S98.

[0094] S91. Initialize the population: Set the number of grey wolves to N, the number of iterations to t max max times, and initialize the search range of the number of available evacuation channels and the search range of multi-dimensional feature data in the emergency evacuation plan, and then execute step S92, where N represents a positive integer greater than or equal to 3, and tmax represents a positive integer and (tmax + 1) × N is less than the number of the emergency evacuation plans.

[0095] In step S91, the grey wolf is a concept in the Grey Wolf Optimizer (GWO). The grey wolf optimization algorithm is an optimization algorithm inspired by the hunting law of grey wolves in the natural environment (it mainly imitates the mechanism of wolves communicating during hunting and the social status among wolves, which are respectively reflected as hunting and hierarchical systems), and has the following algorithm principle:

[0096] Assume that there are four different status levels of wolves in a wolf pack, namely alpha wolves, beta wolves, delta wolves, and omega wolves from top to bottom. The wolves with higher status give instructions to those with lower status. First, they surround the prey. The algorithm for this part is as follows:

[0097]

[0098] In the formula, represents the distance between a gray wolf individual and the prey, t represents the current iteration number, represents the position vector of the prey, represents the position vector of a gray wolf individual, and respectively represent the cooperation vectors, which are calculated as follows:

[0099]

[0100] In the formula, represents the convergence factor, which linearly decreases from 2 to 0 during the iteration process, and respectively represent random vectors in [0, 1]. Secondly, the hunting process is carried out. During hunting, the alpha wolves, beta wolves, and delta wolves lead the omega wolves to hunt, that is, the positions of the alpha wolves, beta wolves, and delta wolves remain unchanged, and the omega wolves perform iterations. The algorithm is as follows:

[0101]

[0102] In the formula, and respectively represent the position vectors of the alpha wolves, beta wolves, and delta wolves in this iteration, and respectively represent the cooperation vectors obtained by random calculation and respectively represent the cooperation vectors obtained by random calculation and respectively represent the distances between other individuals in the group and the alpha wolves, beta wolves, and delta wolves, represents the individual position vector in the t-th iteration, It represents the individual position vector in the (t + 1)-th iteration. Thus, the grey wolf optimization algorithm can be applied to this embodiment to achieve the purpose of iterative optimization for the best emergency evacuation plan. In addition, the search range of the number of available evacuation channels can be conventionally determined according to the value range of the number of available evacuation channels in multiple emergency evacuation plans, and the search range of multi-dimensional feature data can be conventionally determined according to the value range of each dimension of feature data in multiple emergency evacuation plans. For example, for the individual age feature data, since the value range includes infants, juveniles, young people, middle-aged people, and the elderly, etc., the corresponding search range can be determined as [0, 4], where 0 represents infants, 1 represents juveniles, 2 represents young people, 3 represents middle-aged people, and 4 represents the elderly; and so on.

[0103] S92. Initialize grey wolves: Randomly select three grey wolves from N grey wolves as the initial α wolf, β wolf, and δ wolf, and initialize and set the individual position vectors of each grey wolf in the N grey wolves within the search range of the number of available evacuation channels and the search range of multi-dimensional feature data, and then execute step S93, where the individual position vector includes the search value of the number of available evacuation channels and the search value of multi-dimensional feature data.

[0104] In step S92, the individual position vector includes multiple values corresponding one-to-one to the number of available evacuation channels and multi-dimensional feature data: the number of available evacuation channels and the search values of each dimension of feature data, which can be randomly obtained from the corresponding search ranges during initialization.

[0105] S93. For each grey wolf, first determine a most similar emergency evacuation plan from multiple emergency evacuation plans according to the corresponding current individual position vector, then based on this most similar emergency evacuation plan, use the crowd dynamics simulation model to simulate and obtain the corresponding current group movement simulation duration, and finally use this current group movement simulation duration as the corresponding individual fitness value, and then execute step S94.

[0106] In step S93, since the emergency evacuation plan can also be regarded as a vector, the emergency evacuation plan with the closest distance can be conventionally determined as the most similar emergency evacuation plan based on this vector and the current individual position vector. In addition, the same steps S5 - S7 are also used to simulate and obtain the corresponding current group movement simulation duration based on this most similar emergency evacuation plan using the crowd dynamics simulation model.

[0107] S94. Take the grey wolf with the minimum individual fitness value as the new α wolf, the grey wolf with the second smallest individual fitness value as the new β wolf, and the grey wolf with the third smallest individual fitness value as the new δ wolf, and determine whether the current iteration number has reached t maxNext, if so, execute step S98; otherwise, execute step S95.

[0108] In step S94, for example, if the first four individual fitness values in the front are 10 minutes, 12 minutes, 15 minutes, and 16 minutes in sequence, the gray wolf with 10 minutes can be used as the new α wolf, the gray wolf with 12 minutes can be used as the new β wolf, and the gray wolf with 15 minutes can be used as the new δ wolf.

[0109] S95. Calculate the convergence factor coordination vector and coordination vector Then execute step S96, where the convergence factor the coordination vector and the coordination vector are calculated according to the following formulas respectively:

[0110]

[0111] In the formula, t represents the current iteration number, tanh() represents the hyperbolic tangent function, and represent random vectors in [0, 1] respectively.

[0112] In step S95, considering the defect that the convergence factor of the traditional gray wolf algorithm is purely linear, in order to non - linearize the convergence factor and facilitate the algorithm to achieve the purpose of global optimization, this embodiment is influenced by the image of the tanh activation function in the neural network (that is, selecting the image of the function in the range of [-3, 3], and performing stretching, symmetry, translation and other transformation operations successively), and substituting the iteration number into the function, and the above - mentioned convergence factor expression is obtained by improvement.

[0113] S96. For each ω wolf, calculate the corresponding individual position vector at the (t + 1) - th iteration according to the current individual position vectors of the new α wolf, β wolf and δ wolf Then execute step S97, where the individual position vector is calculated according to the following formula:

[0114]

[0115] In the formula, represents the current individual position vector of the new α wolf, represents the current individual position vector of the new β wolf, represents the current individual position vector of the new δ wolf, represents the individual position vector at the t - th iteration, and respectively represent the collaborative vectors obtained by random calculation and respectively represent the collaborative vectors obtained by random calculation

[0116] In step S96 of the present embodiment, a weighted assignment is also performed on the gray wolf position update strategy, that is, the weight coefficients w1, w2, and w3 are calculated respectively, so as to form an improvement point of the gray wolf optimization algorithm together with the new convergence factor expression, and through the performance under 10 common international standard test functions, it is found that it has good accuracy and convergence speed.

[0117] S97. Increment the iteration count by 1, and then return to execute step S93.

[0118] S98. Determine the most similar emergency evacuation plan corresponding to the new α wolf as an emergency evacuation plan with the shortest current group movement simulation duration.

[0119] Based on the foregoing possible design one, when the number of emergency evacuation plans exceeds a preset value, the maximum number of group movement simulations can be (t max +1)×N times, so that it is not necessary to simulate each emergency evacuation plan once, effectively reducing the demand for computing resources required for simulation, and achieving the purpose of preferentially recommending the best emergency evacuation plan with low resource requirements.

[0120] As Figure 2 shown, in the second aspect of the present embodiment, a virtual device for implementing the intelligent push method of the stadium evacuation plan described in the first aspect or any possible design is provided, including a video data acquisition module, an in-venue personnel tracking module, a current information determination module, an evacuation plan acquisition module, an evacuation scenario fusion module, a simulation model initialization module, a simulation duration determination module, a judgment module, and an available plan push module;

[0121] The video data acquisition module is used to acquire on-site video data collected in real time by multiple monitoring cameras of the target stadium;

[0122] The in-venue personnel tracking module is communicatively connected to the video data acquisition module, and is used to use a multi-target tracking algorithm and cross-camera multi-view scene target continuous tracking technology to track each in-venue personnel appearing inside the target stadium in real time according to the on-site video data, and obtain the in-venue personnel tracking result;

[0123] The current information determination module is communicatively connected to the in-venue personnel tracking module, and is used to determine the current three-dimensional coordinate position and current moving speed of each in-venue personnel in the target stadium in real time according to the in-venue personnel tracking result;

[0124] The evacuation plan acquisition module is configured to acquire the three-dimensional virtual scene data of the target stadium and the emergency evacuation plan, where the emergency evacuation plan includes multi-dimensional feature data of at least one available evacuation route preset for the target stadium, and the multi-dimensional feature data includes three-dimensional coordinate data and three-dimensional model data of the corresponding route in the target stadium;

[0125] The evacuation scene fusion module is communicatively connected to the evacuation plan acquisition module, and is configured to respectively fuse the three-dimensional model data of each available evacuation route in the at least one available evacuation route into the three-dimensional virtual scene data according to the three-dimensional coordinate data of each available evacuation route, so as to obtain the three-dimensional virtual emergency evacuation scene of the target stadium;

[0126] The simulation model initialization module is communicatively connected to the current information determination module and the evacuation scene fusion module respectively, and is configured to initialize the crowd dynamics simulation model for implementing the group movement simulation technology: initialize the application scene of the crowd dynamics simulation model as the three-dimensional virtual emergency evacuation scene, and for each particle in the crowd dynamics simulation model corresponding one-to-one to each person in the stadium, determine the initial position of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current three-dimensional coordinate position of the corresponding person, and further determine the initial speed of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current movement speed of the corresponding person;

[0127] The simulation duration determination module is communicatively connected to the simulation model initialization module, and is configured to start the crowd dynamics simulation model to simulate and update the real-time position and real-time speed of each particle leaving the three-dimensional virtual emergency evacuation scene, and stop the simulation until all the particles have left the three-dimensional virtual emergency evacuation scene, so as to obtain the current group movement simulation duration for evacuating all the people in the stadium out of the target stadium;

[0128] The judgment module is communicatively connected to the simulation duration determination module, and is configured to judge whether the current group movement simulation duration is less than or equal to a preset duration threshold;

[0129] The available plan push module is communicatively connected to the judgment module, and is configured to, when it is determined that the current group movement simulation duration is less than or equal to the preset duration threshold, push the emergency evacuation plan as the current available evacuation plan of the target stadium to the decision-making terminal.

[0130] For the working process, working details and technical effects of the foregoing device provided in the second aspect of this embodiment, reference may be made to the intelligent push method for the stadium evacuation plan described in the first aspect or any possible design, which will not be elaborated herein.

[0131] As Figure 3 shown, in the third aspect of this embodiment, a computer device for executing the intelligent push method of the stadium evacuation plan described in the first aspect or any possible design includes a memory, a processor, and a transceiver that are communicatively connected in sequence. Among them, the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer programs and execute the intelligent push method of the stadium evacuation plan described in the first aspect or any possible design. Specifically, for example, the memory may include, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a flash memory, a first input first output (FIFO), and / or a first input last output (FILO), etc.; the processor may include, but is not limited to, a microprocessor of the STM32F105 series. In addition, the computer device may also include, but is not limited to, a power module, a display screen, and other necessary components.

[0132] For the working process, working details, and technical effects of the foregoing computer device provided in the third aspect of this embodiment, reference may be made to the intelligent push method of the stadium evacuation plan described in the first aspect or any possible design, and details will not be elaborated here.

[0133] In the fourth aspect of this embodiment, a computer-readable storage medium storing instructions for the intelligent push method of the stadium evacuation plan described in the first aspect or any possible design is provided, that is, instructions are stored on the computer-readable storage medium, and when the instructions run on a computer, the intelligent push method of the stadium evacuation plan described in the first aspect or any possible design is executed. Among them, the computer-readable storage medium refers to a carrier for storing data, and may include, but is not limited to, computer-readable storage media such as floppy disks, optical discs, hard disks, flash memories, USB flash drives, and / or memory sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0134] For the working process, working details, and technical effects of the foregoing computer-readable storage medium provided in the fourth aspect of this embodiment, reference may be made to the intelligent push method of the stadium evacuation plan described in the first aspect or any possible design, and details will not be elaborated here.

[0135] In the fifth aspect of this embodiment, a computer program product containing instructions is provided. When the instructions run on a computer, the computer is caused to execute the intelligent push method for the stadium evacuation plan as described in the first aspect or any possible design. Among them, the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0136] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent push method for stadium evacuation plans based on group movement simulation technology, characterized in that, Including: Obtain the on-site video data collected in real time by multiple monitoring cameras of the target stadium; According to the on-site video data, use the multi-object tracking algorithm and the cross-camera multi-view scene object continuous tracking technology to track in real time each person inside the target stadium, and obtain the tracking result of the people inside the stadium; According to the tracking result of the people inside the stadium, determine in real time the current three-dimensional coordinate positions and current moving speeds of the respective people inside the stadium in the target stadium; Obtain the three-dimensional virtual scene data of the target stadium and the emergency evacuation plan, wherein the emergency evacuation plan includes multi-dimensional feature data of at least one available evacuation route preset for the target stadium, and the multi-dimensional feature data includes three-dimensional coordinate data and three-dimensional model data of the corresponding route in the target stadium; According to the three-dimensional coordinate data of each available evacuation route in the at least one available evacuation route, fuse the three-dimensional model data of each available evacuation route into the three-dimensional virtual scene data respectively, and obtain the three-dimensional virtual emergency evacuation scene of the target stadium; Initialize the crowd dynamics simulation model for implementing the group motion simulation technology: initialize the application scenario of the crowd dynamics simulation model as the three-dimensional virtual emergency evacuation scene, and for each particle in the crowd dynamics simulation model that corresponds one-to-one with each person inside the stadium, determine the initial position of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current three-dimensional coordinate position of the corresponding person, and also determine the initial speed of the corresponding particle in the three-dimensional virtual emergency evacuation scene according to the current moving speed of the corresponding person; Start the crowd dynamics simulation model to simulate and update the real-time positions and real-time speeds of the respective particles leaving the three-dimensional virtual emergency evacuation scene, and stop the simulation until all the particles have left the three-dimensional virtual emergency evacuation scene, and obtain the current group motion simulation duration for evacuating all the people inside the stadium out of the target stadium; Judge whether the current group motion simulation duration is less than or equal to a preset duration threshold; If so, push the emergency evacuation plan as the current available evacuation plan of the target stadium to the decision-making terminal.

2. The intelligent push method for the stadium evacuation plan according to claim 1, wherein The multi-object tracking algorithm adopts the deepsort target tracking algorithm, wherein the deepsort target tracking algorithm includes: for an unmatched tracking target, judge whether it leaves the inside of the target stadium according to the corresponding current predicted position. If so, discard the tracking target during the next position prediction, otherwise still use the Kalman filter method to predict the corresponding position during the next position prediction.

3. The intelligent push method for the stadium evacuation plan according to claim 1, characterized in that Obtaining the emergency evacuation plan of the target stadium includes: When receiving a fire alarm message from a fire alarm device, according to the known installation location of the fire alarm device, at least one available evacuation passage that bypasses the known installation location is screened out from multiple evacuation passages preset for the target stadium, and then the multi-dimensional feature data of the at least one available evacuation passage is used as the emergency evacuation plan for the target stadium, where the fire alarm device is pre-installed in the target stadium, and the multi-dimensional feature data includes three-dimensional coordinate data and three-dimensional model data of the corresponding passage in the target stadium.

4. The intelligent push method for the stadium evacuation plan according to claim 1, wherein When the multiple monitoring cameras include an entrance monitoring camera whose field of view covers the entrance area of the target stadium, the method further includes: Cropping out the entrance video images of each in-stadium person from the on-site video data collected by the entrance monitoring camera; For each in-stadium person, identifying the corresponding individual heterogeneity features according to the corresponding entrance video image; In the process of starting the crowd dynamics simulation model to simulate and update the real-time positions and real-time speeds of the particles leaving the three-dimensional virtual emergency evacuation scene, first, for each in-stadium person, according to the corresponding individual heterogeneity features, physiological coefficients and psychological coefficients that change with time are assigned to the corresponding particles, then for each particle, based on the corresponding physiological coefficients and psychological coefficients, a corresponding real-time expected speed that changes with time is determined, and finally, for each particle, according to the corresponding real-time expected speed and real-time speed, a corresponding real-time acceleration is determined, and the real-time speed and real-time position are updated using the real-time acceleration.

5. The intelligent push method for the stadium evacuation plan according to claim 4, wherein For each in-stadium person, identifying the corresponding individual heterogeneity features according to the corresponding entrance video image includes: For each in-stadium person, importing the corresponding entrance video image into a multi-dimensional individual heterogeneity feature recognition model based on a convolutional neural network and pre-trained, to obtain a corresponding multi-dimensional individual heterogeneity feature recognition result, where the multi-dimensional individual heterogeneity feature recognition model includes an individual gender feature recognition sub-model, an individual height feature recognition sub-model, an individual disability feature recognition sub-model, an individual age feature recognition sub-model, an individual weight feature recognition sub-model, and / or an individual personality feature recognition sub-model.

6. The intelligent push method for the stadium evacuation plan according to claim 1, wherein When the number of the emergency evacuation plans is multiple, the method further includes: Determining an emergency evacuation plan with the shortest current group movement simulation duration from the multiple emergency evacuation plans, and pushing the emergency evacuation plan as the current available evacuation plan for the target stadium to the decision-making terminal.

7. An intelligent push device for stadium evacuation plans based on group movement simulation technology, characterized in that, Including a video data acquisition module, an in-stadium personnel tracking module, a current information determination module, an evacuation plan acquisition module, an evacuation scene fusion module, a simulation model initialization module, a simulation duration determination module, a judgment module, and an available plan push module; The video data acquisition module is used to acquire on-site video data collected in real time by multiple monitoring cameras of the target stadium; The in - venue personnel tracking module, communicatively connected to the video data acquisition module, is used to track each in - venue personnel present inside the target stadium in real time according to the on - site video data, adopting a multi - target tracking algorithm and cross - camera multi - view scene target continuous tracking technology, and obtaining the in - venue personnel tracking result; The current information determination module, communicatively connected to the in - venue personnel tracking module, is used to determine the current three - dimensional coordinate position and current moving speed of each in - venue personnel in the target stadium in real time according to the in - venue personnel tracking result; The evacuation plan acquisition module is used to acquire the three - dimensional virtual scene data of the target stadium and the emergency evacuation plan. Among them, the emergency evacuation plan includes multi - dimensional feature data of at least one available evacuation passage preset for the target stadium, and the multi - dimensional feature data includes the three - dimensional coordinate data and three - dimensional model data of the corresponding passage in the target stadium; The evacuation scene fusion module, communicatively connected to the evacuation plan acquisition module, is used to fuse the three - dimensional model data of each available evacuation passage in the at least one available evacuation passage into the three - dimensional virtual scene data respectively according to the three - dimensional coordinate data of each available evacuation passage, and obtain the three - dimensional virtual emergency evacuation scene of the target stadium; The simulation model initialization module, communicatively connected to the current information determination module and the evacuation scene fusion module respectively, is used to initialize the crowd dynamics simulation model for implementing the group movement simulation technology: initialize the application scene of the crowd dynamics simulation model as the three - dimensional virtual emergency evacuation scene, and for each particle in the crowd dynamics simulation model corresponding one - to - one with each in - venue personnel, determine the initial position of the corresponding particle in the three - dimensional virtual emergency evacuation scene according to the current three - dimensional coordinate position of the corresponding personnel, and also determine the initial speed of the corresponding particle in the three - dimensional virtual emergency evacuation scene according to the current moving speed of the corresponding personnel; The simulation duration determination module, communicatively connected to the simulation model initialization module, is used to start the crowd dynamics simulation model to simulate and update the real - time position and real - time speed of each particle leaving the three - dimensional virtual emergency evacuation scene, and stop the simulation until all the particles have left the three - dimensional virtual emergency evacuation scene, and obtain the current group movement simulation duration for evacuating all in - venue personnel out of the target stadium; The judgment module, communicatively connected to the simulation duration determination module, is used to judge whether the current group movement simulation duration is less than or equal to a preset duration threshold; The available plan push module, communicatively connected to the judgment module, is used to, when it is determined that the current group movement simulation duration is less than or equal to the preset duration threshold, push the emergency evacuation plan as the current available evacuation plan of the target stadium to the decision - making terminal.

8. A computer device, characterized in that, It includes a memory, a processor, and a transceiver that are communicatively connected in sequence. Among them, the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer programs and execute the intelligent push method for the stadium evacuation plan as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that , Instructions are stored on the computer-readable storage medium. When the instructions run on a computer, they execute the intelligent push method for the stadium evacuation plan as described in any one of claims 1 to 6.

Images