CN111125886A

CN111125886A - A crowd evacuation simulation system and simulation method based on three different behaviors

Info

Publication number: CN111125886A
Application number: CN201911225104.1A
Authority: CN
Inventors: 周晓晶; 蔡艳潇; 陈国华
Original assignee: Southeast University
Current assignee: Southeast University
Priority date: 2019-12-04
Filing date: 2019-12-04
Publication date: 2020-05-08

Abstract

The invention discloses a crowd evacuation simulation system and a simulation method based on three different behaviors, wherein the system comprises: the system comprises a scene modeling module, a pedestrian information loading module, a crowd behavior modeling module and a result output and analysis module; the method comprises the following steps: (1) establishing an abstract environment and carrying out layout of related facilities; (2) generating safe evacuation crowd, initializing pedestrians, determining pedestrian targets and position information, and setting behavior correlation attributes of the pedestrians; (3) establishing a crowd evacuation behavior model, and driving the pedestrian to move through the behavior decision of the pedestrian; (4) and (3) realizing the simulation result and carrying out correlation analysis on the result. The invention can provide required data support for evacuation research, reveal the evacuation rule of people and find the evacuation bottleneck, thereby having important theoretical significance and practical significance for improving the safety design level of buildings and the evacuation capacity.

Description

Crowd evacuation simulation system and simulation method based on three different behaviors

Technical Field

The invention relates to the technical field of crowd evacuation simulation, in particular to a crowd evacuation simulation system and a crowd evacuation simulation method based on three different behaviors.

Background

In recent years, with the rapid development of society, when there are many pedestrians in the environment, severe clogging often occurs. And some pedestrians do not like congestion, and can select a detour to avoid congestion when congestion ahead is found. The pedestrian's route of choice may change because of congestion ahead. Therefore, more and more people simulation researches take the road congestion condition as an important factor to influence the path selection of the pedestrians.

Currently, most congestion prediction research mainly focuses on predicting local motion of other pedestrians, and such prediction can effectively achieve collision avoidance. Steffen has proposed a concept of forecasting, i.e., inferring future population movements based on the current situation. The researches mainly consider the influence of the current and local environment congestion on the pedestrians, the researches on the remote congestion prediction model and the influence on the path selection are few, and the pedestrian classification is only one type, so that the behavior characteristics of different people are not reflected.

Disclosure of Invention

The technical problem to be solved by the invention is to provide a crowd evacuation simulation system and a simulation method based on three different behaviors, which can provide required data support for evacuation research, reveal the evacuation rule of people and find evacuation bottlenecks, thereby having important theoretical significance and practical significance for improving the safety design level of buildings and the evacuation capacity.

In order to solve the above technical problems, the present invention provides a crowd evacuation simulation system based on three different behaviors, comprising: the system comprises a scene modeling module, a pedestrian information loading module, a crowd behavior modeling module and a result output and analysis module;

the scene modeling module specifies the size of an evacuation scene, the position of an obstacle and the position of an exit;

the pedestrian information loading module finds that the behavior of the crowd to be evacuated has three characteristics according to the situation of the crowd collected by the camera in real life during evacuation: the method comprises the following steps of selecting a nearest exit for evacuation without worrying about congestion, selecting to avoid congestion and bypass barriers all the time for evacuation, comprehensively considering congestion and the nearest exit for evacuation, and dividing pedestrian behaviors into three types: waiting for crowds with congestion, bypassing the crowds with congestion and comprehensively judging the crowds, and defining the number of the crowds and the positions of people;

the crowd behavior modeling module comprehensively judges reasonable analysis of crowds during path selection, compares time consumed by bypassing congestion positions and waiting for the two behaviors, and finally selects a path with the least consumed time;

and the result output and analysis module simulates the personnel flow in the scene and analyzes the simulation result.

Preferably, the crowd waiting for congestion does not mind congestion, the crowd waiting for congestion can select to wait when the congestion occurs, the crowd waiting for congestion arrives at the destination by following the shortest path all the time, the crowd waiting for congestion does not mind congestion, the crowd waiting for congestion can bypass the congestion as long as the congestion occurs on the road, the positions of the congestion which do not come are predicted and identified and used as obstacles, and then the positions of the congestion are avoided.

Correspondingly, the crowd evacuation simulation method based on three different behaviors comprises the following steps:

(1) establishing an abstract environment and carrying out layout of related facilities;

(2) generating safe evacuation crowd, initializing pedestrians, determining pedestrian targets and position information, and setting behavior correlation attributes of the pedestrians;

(3) establishing a crowd evacuation behavior model, and driving the pedestrian to move through the behavior decision of the pedestrian;

(4) and (3) realizing the simulation result and carrying out correlation analysis on the result.

Preferably, in the step (3), a crowd evacuation behavior model is established, in the motion of the pedestrian driven by the behavior decision of the pedestrian, the congested crowd is received without thinking of congestion, the pedestrian can select to wait when encountering congestion and reach the destination by always following the shortest path, a single-terminal dynamic shortest path algorithm-DOT algorithm is selected, and the DOT algorithm is executed once for the pedestrian with the same destination, and the method comprises the following steps:

(1) initializing a network, and presetting N1 as a set of nodes occupied by pedestrians;

(2) executing DOT algorithm to calculate the shortest path from all nodes to the destination K1;

(3) for all pedestrians p1 closest to the destination currently, finding the shortest path j1 of the destination K1 through unoccupied nodes;

(4) add all nodes on path j1 to set N1;

(5) disabling routes containing any node on path j 1;

(6) the shortest path from the affected node to destination K1 is updated.

Preferably, in the step (3), a crowd evacuation behavior model is established, the pedestrian moves by means of behavior decision of the pedestrian, congestion crowd is intentionally jammed and bypassed as long as the road is jammed, an upcoming congestion position is predicted and identified and is used as an obstacle, and the congestion position is avoided, so that a congestion identification algorithm is added to the pedestrian when the pedestrian arrives at the destination along the shortest path, the crowd evacuation time consumption is the shortest path time consumption after the congestion point is regarded as the obstacle, and the method specifically comprises the following steps of:

(1) initializing a network, presetting N2 as a set of nodes occupied by pedestrians, presetting M2 as a set of congestion nodes, and presetting R2 as a total set of nodes in the set N2 and nodes around the nodes by 3 x 3;

(2) for each node in R2, calculating the total number P2 of nodes occupied by obstacles and the total Q2 of nodes occupied by pedestrians;

(3) judgment of

Whether the value of (a) is greater than or equal to the congestion coefficient epsilon 2, if the obtained value is less than epsilon 2, the node is indicated to be not congested, and if the obtained value is greater than or equal to epsilon 2, the node is added into the set M2;

(4) the shortest path from the affected node to destination K2 is updated.

Preferably, in the step (3), a crowd evacuation behavior model is established, the pedestrian moves by means of a behavior decision of the pedestrian, the congested crowd is intentionally congested, the congested crowd is bypassed as long as the road is congested, the position of the future congestion is predicted and identified and is used as an obstacle, and then the position of the congestion is avoided, so that a congestion identification algorithm is added when the pedestrian arrives at the destination along the shortest path, the time consumed for evacuating the crowd is the time consumed by the shortest path after the congestion point is regarded as the obstacle, and the method specifically comprises the following steps of:

(1) initializing a network, presetting N3 as a set of nodes occupied by pedestrians, presetting M3 as a set of congestion nodes, and presetting R3 as a total set of nodes in the row where the nodes in the set N3 are located and the nodes in the previous row;

(2) for each node in R3, calculating the total number P3 of nodes occupied by obstacles and the total Q3 of nodes occupied by pedestrians;

(3) judgment of

Whether the value of (a) is greater than or equal to the congestion coefficient epsilon 3, if the obtained value is less than epsilon 3, the node is indicated to be not congested, and if the obtained value is greater than or equal to epsilon 3, the node is added into the set M3;

(4) the shortest path from the affected node to destination K3 is updated.

Preferably, comprehensively judging reasonable analysis of the crowd during path selection, comparing time consumed by the two behaviors of bypassing the congestion position and waiting for the congestion position, and finally selecting the path with the least consumed time, wherein the method comprises the following steps:

(1) calculating the time t1 for receiving the evacuation of the congested crowd for the current grid;

(2) calculating the evacuation time t2 for the current grid to bypass the congested crowd;

(3) comparing time t1 and t2, a path that takes less time is selected.

(4) The shortest path from the affected node to destination K4 is updated.

The invention has the beneficial effects that: (1) the method comprises the steps of firstly, considering the influence of different behaviors on crowds in the process of crowd evacuation, and classifying the behaviors of people in the evacuation process; for each individual, the individual constantly senses the surrounding conditions and makes a behavior decision through own judgment, so that each individual has own behavior, all people forming the crowd constantly conduct the individual behaviors, and the individual behaviors interact with each other and influence the scene where the individual behaviors are located. The invention is based on a grid scene model, fully considers the human behaviors, and divides the human behaviors into three types: waiting for congestion, bypassing congestion and comprehensively judging, which is consistent with the actual situation, so that the evacuation situation has more variability; (2) the video camera is used for collecting video data during crowd evacuation in real life, and people evacuation behaviors are classified by selecting 5 frames of image data before and after a congestion point in a video, so that the method is more reasonable and practical; (3) in the aspect of congestion identification, the congestion identification based on an image, the congestion identification based on a hidden Markov model, the congestion identification based on a convolutional neural network and the like are mainly adopted in the prior art, and the method is based on a gridding scene, defines a congestion coefficient on the basis of the gridding scene, and can achieve the purpose of congestion identification only by judging the size of the congestion coefficient, so that the method is more efficient and simple and makes up the defects of complexity and low efficiency in the prior art; (4) the invention is suitable for evacuation processes in various complex scenes, utilizes an anti-collision principle and adds a competition algorithm, so that more than two persons cannot exist in the same position, and the simulation is more reasonable and real.

Drawings

Fig. 1 is a schematic flow chart of the functional structure of the present invention.

Fig. 2 is a schematic view of the camera acquisition situation of the present invention.

FIG. 3 is a schematic diagram of a path prediction process according to the present invention.

Fig. 4 is a schematic diagram illustrating an intra-floor congestion identification process according to the present invention.

Fig. 5 is a schematic diagram illustrating a flow of identifying congestion in stairs according to the present invention.

Fig. 6 is a schematic diagram of an evacuation simulation within a floor according to the present invention.

Fig. 7 is a schematic diagram of the simulation of evacuation inside stairs according to the present invention.

FIG. 8 is a schematic diagram of the overall process of model simulation according to the present invention.

Detailed Description

As shown in fig. 1, a crowd evacuation simulation system based on three different behaviors includes: the system comprises a scene modeling module, a pedestrian information loading module, a crowd behavior modeling module and a result output and analysis module.

The scene modeling module gridds the scene based on the rectangular grid, mainly specifies the size of an evacuation scene, the size of stairs, the position of specified obstacles, the position of the stairs and the position of an exit.

The pedestrian information loading module is used for collecting video data of people during crowd evacuation in real life by using a camera, people evacuation behaviors are classified by selecting 5 frames of image data before and after a congestion point in a video, and the behaviors of people to be evacuated are found to have three characteristics: and selecting a nearest exit for evacuation, not worrying about congestion, selecting to avoid congestion, bypassing obstacles all the time for evacuation, comprehensively considering congestion and the nearest exit for evacuation, so that the behaviors of pedestrians are divided into three types, waiting for congestion crowds, bypassing congestion crowds and comprehensively judging crowds, and then counting the total number of observed crowds and the proportion of each behavior crowd. The module also specifies the number of people and their locations, the camera acquisition being as shown in figure 2.

The crowd behavior modeling module comprises four parts for modeling: pedestrian division, behavior decision, path planning and position updating.

In the pedestrian division modeling, the pedestrian behaviors are divided into three types based on the gridding model, and different algorithms are provided for the behaviors of three different crowds.

In the behavior decision modeling, congestion crowds are received, congestion is not cared for, waiting is selected when congestion occurs, and the shortest path is always followed to reach a destination, so that a single-terminal dynamic shortest path algorithm-DOT algorithm is selected, the time consumed for evacuating the crowds is the congestion waiting time plus the path traveling time, and the congestion waiting time is integral multiple of t because position updating and congestion recognition are carried out once every unit time t. The DOT algorithm is an algorithm for updating the consumption of all nodes to destinations in a network in a decreasing order over time, thereby efficiently calculating the lowest consumption of all nodes to destinations, and thus selecting the lowest consumption path.

However, if the shortest path algorithm is performed once for each pedestrian, time and operation efficiency are wasted, and especially when the number of pedestrians is particularly large, the operation efficiency is particularly low. When a pedestrian selects a shortest path to reach a destination, only a very small part of nodes in the network will change, i.e. the nodes change from unoccupied to occupied, so a DOT algorithm is performed once for the pedestrian with the same destination, as shown in the path prediction flowchart of fig. 3, and the steps are as follows:

step 1: initializing a network, and presetting N1 as a set of nodes occupied by pedestrians;

step 2: executing DOT algorithm to calculate the shortest path from all nodes to the destination K1;

and step 3: for all pedestrians p1 closest to the destination currently, finding the shortest path j1 of the destination K1 through unoccupied nodes;

and 4, step 4: add all nodes on path j1 to set N1;

and 5: disabling routes containing any node on path j 1;

step 6: the shortest path from the affected node to destination K1 is updated.

Congestion crowds are intentionally congested by bypassing, congestion is bypassed as long as a road is congested, the congestion positions in the future are predicted and identified and used as obstacles, and then the congestion positions are avoided, so that a congestion identification algorithm is added when pedestrians reach a destination along the shortest path, and the time consumed for evacuating the crowds is the time consumed by the shortest path after a congestion point is considered as an obstacle. As shown in the flow chart of fig. 4, the congestion identification in the floor includes the following steps:

step 1: initializing a network, presetting N2 as a set of nodes occupied by pedestrians, presetting M2 as a set of congestion nodes, and presetting R2 as a total set of nodes in the set N2 and nodes around the nodes by 3 x 3;

step 2: for each node in R2, calculating the total number P2 of nodes occupied by obstacles and the total Q2 of nodes occupied by pedestrians;

and step 3: judgment of

and 4, step 4: the shortest path from the affected node to destination K2 is updated.

The evacuation rules in the stairs are different from the floors as follows:

(1) evacuation in stairs does not allow retreat, but can wait;

(2) the evacuation speed in the stairs is 3/4 of the evacuation speed in the floors;

(3) evacuation scene size inside stairs: length of stairs

H is the height of the stairs, r is the width of the stairs, the scene area s is the length m of the stairs and the evacuation width n, and finally the evacuation scene of the stairs is formed according to the size proportion of the evacuation scene unit grids in the floors.

For the identification of congestion in stairs, as shown in the flow chart of fig. 5, the steps are as follows:

step 1: initializing a network, presetting N3 as a set of nodes occupied by pedestrians, presetting M3 as a set of congestion nodes, and presetting R3 as a total set of nodes in the row where the nodes in the set N3 are located and the nodes in the previous row;

step 2: for each node in R3, calculating the total number P3 of nodes occupied by obstacles and the total Q3 of nodes occupied by pedestrians;

and step 3: judgment of

and 4, step 4: the shortest path from the affected node to destination K3 is updated.

Comprehensively judging reasonable analysis of crowds during path selection, comparing time consumed by bypassing congestion positions and waiting for the two behaviors, and finally selecting a path with the least consumed time, wherein the steps are as follows:

step 1: calculating the time t1 for receiving the evacuation of the congested crowd for the current grid;

step 2: calculating the evacuation time t2 for the current grid to bypass the congested crowd;

and step 3: comparing time t1 and t2, a path that takes less time is selected.

And 4, step 4: the shortest path from the affected node to destination K4 is updated.

In the path planning modeling, a situation that a plurality of persons compete for the same grid position exists, so a competition algorithm is needed, and when only one pedestrian takes the grid as a target, the pedestrian moves to the grid; when a plurality of people target the grid, if the people are crowds with different behaviors, the probability of bypassing the crowd crowds for competition win is larger than the probability of comprehensively judging the crowd competition win, the probability of comprehensively judging the crowd competition win is larger than the probability of receiving the crowd competition win, if the crowd with the same behavior, the people with small distance move to the grid, and if the distance is the same, the pedestrians are randomly selected to move to the grid.

In the updated position modeling, the position of the pedestrian after moving each time is updated, and each grid is only allowed to be occupied by one pedestrian or obstacle.

Several simulation examples are provided below:

the crowd evacuation simulation method comprises the steps that 3 psychological crowds are evacuated in rooms with 10 × 10 in each floor, for example, a simulation graph for in-floor evacuation is shown in fig. 6, for example, a simulation graph for in-stair evacuation is shown in fig. 7, under the condition, crowd evacuation simulation is conducted according to behavior characteristics of different crowds, the crowd with congestion is received, shortest paths are selected for evacuation, the crowd with congestion is bypassed, the crowd with congestion is selected for bypassing congestion all the time, evacuation time of the crowd with two methods is comprehensively judged, and finally the crowd with congestion waiting for ending is selected for evacuation.

The evacuation method of three populations is shown in the general flow chart of the model simulation of fig. 8.

And finally, feeding back the simulation result of the personnel flow in the scene to the user through a result output and analysis module.

The invention provides a crowd evacuation simulation system and a crowd evacuation simulation method based on three different behaviors, which are used for dividing the crowd behaviors into three types under the condition that a part of pedestrians can continuously predict long-distance congestion and can effectively avoid possible future congestion places during path selection: waiting for congested people, bypassing congested people and comprehensively judging people, which is consistent with actual human behavior and potentially reduces prediction errors in the method. The evacuation simulation method provided by the invention can provide required data support for evacuation research, reveal the evacuation rule of people and find the evacuation bottleneck, thereby having important theoretical significance and practical significance for improving the safety design level of buildings and the evacuation capability.

Claims

1. a crowd evacuation simulation system based on three different behaviors, is characterized in that, comprises: scene modeling module, loading pedestrian information module, crowd behavior modeling module and result output and analysis module;

The scene modeling module specifies the size of the evacuation scene, the location of obstacles and the location of exits;

Loading the pedestrian information module According to the crowd evacuation situation collected by the camera in real life, it is found that the behavior of the crowd to be evacuated has three characteristics: choose the nearest exit for evacuation and do not mind the congestion, choose to avoid congestion and always avoid obstacles for evacuation and Considering the congestion and the nearest exit for evacuation, the behavior of pedestrians is divided into three types: waiting for the crowded crowd, bypassing the crowded crowd and comprehensively judging the crowd, specifying the number of crowds and the location of the people;

The crowd behavior modeling module comprehensively judges that the crowd will rationally analyze the path selection, compare the time consumed by the two behaviors of bypassing the congestion location and waiting, and finally choose the path that consumes the least time;

The result output and analysis module simulates the flow of people in the scene and analyzes the simulation results.

2. The crowd evacuation simulation system based on three different behaviors as claimed in claim 1, it is characterized in that, the crowd waiting for the congestion does not mind the congestion, and will choose to wait when encountering congestion, and always follow the shortest path to the destination, bypassing the crowded crowd. Mind the congestion, as long as the road is congested, avoid it, predict, identify the future congestion location and use it as an obstacle, and then avoid the congestion location.

3. a crowd evacuation simulation method based on three different behaviors, is characterized in that, comprises the steps:

(1) Establish an abstract environment and carry out the layout of related facilities;

(2) Generate safe evacuation crowds, initialize pedestrians, determine pedestrian targets and location information, and set behavior-related attributes of pedestrians;

(3) Establish a crowd evacuation behavior model, and drive pedestrians’ movements through their behavioral decisions;

(4) The realization of the simulation results and the correlation analysis of the results.

4. the crowd evacuation simulation method based on three kinds of different behaviors as claimed in claim 3 is characterized in that, in step (3), establish crowd evacuation behavior model, in the movement of pedestrians driven by pedestrian's behavioral decision-making, receive crowded crowd Don’t mind the congestion, choose to wait when encountering congestion, always follow the shortest path to the destination, choose the single-end dynamic shortest path algorithm-DOT algorithm, and execute the DOT algorithm once for pedestrians with the same destination, including the following steps:

(1) Initialize the network, and preset N1 as the set of nodes occupied by pedestrians;

(2) Execute the DOT algorithm to calculate the shortest path from all nodes to the destination K1;

(3) For all pedestrians p1 currently closest to the destination, find the shortest path j1 to the destination K1 through unoccupied nodes;

(4) Add all nodes on the path j1 to the set N1;

(5) Routes containing any node on path j1 are prohibited;

(6) Update the shortest path of the affected node to the destination K1.

5. the crowd evacuation simulation method based on three kinds of different behaviors as claimed in claim 3, is characterized in that, in step (3), establish crowd evacuation behavior model, drive pedestrian's movement by pedestrian's behavioral decision-making, bypass congestion Crowds mind congestion, as long as the road is congested, they will bypass it, predict and identify the future congestion location and use it as an obstacle, and then avoid the congestion location. Therefore, a congestion recognition algorithm is added to such pedestrians while following the shortest path to their destination. , the evacuation time of this type of crowd is the shortest path time after the congestion point is regarded as an obstacle. For the congestion identification in the floor, the specific steps are as follows:

(1) Initialize the network, preset N2 to be the set of nodes occupied by pedestrians, preset M2 to be the set of congested nodes, preset R2 to be the total set of nodes in set N2 and 3*3 nodes around the node;

(2) For each node in R2, calculate the total number of nodes P2 occupied by obstacles and the total number of nodes occupied by pedestrians Q2;

(3) Judgment

Whether the value of is greater than or equal to the congestion coefficient ε2, if the obtained value is less than ε2, the node is not congested, if the obtained value is greater than or equal to ε2, the node is added to the set M2;

(4) Update the shortest path of the affected node to the destination K2.

6. the crowd evacuation simulation method based on three kinds of different behaviors as claimed in claim 3, it is characterized in that, in step (3), establish crowd evacuation behavior model, in the movement of pedestrians driven by pedestrian's behavioral decision-making, bypass congestion Crowds mind congestion, as long as the road is congested, they will bypass it, predict and identify the future congestion location and use it as an obstacle, and then avoid the congestion location. Therefore, a congestion recognition algorithm is added to such pedestrians while following the shortest path to their destination. , the evacuation time of this type of crowd is the shortest path time after the congestion point is regarded as an obstacle. For the congestion identification in the stairs, the specific steps include the following:

(1) Initializing the network, preset N3 is the set of nodes occupied by pedestrians, preset M3 is the set of congested nodes, preset R3 is the total set of nodes in the row where the nodes in the set N3 are located and the nodes in the previous row;

(2) For each node in R3, calculate the total number of nodes P3 occupied by obstacles and the total number of nodes occupied by pedestrians Q3;

(3) Judgment

Whether the value of is greater than or equal to the congestion coefficient ε3, if the obtained value is less than ε3, the node is not congested, if the obtained value is greater than or equal to ε3, the node is added to the set M3;

(4) Update the shortest path of the affected node to the destination K3.

7. The crowd evacuation simulation method based on three different behaviors as claimed in claim 3, characterized in that, comprehensively judging that crowds can be rationally analyzed during route selection, and comparing the cost of bypassing the congestion location and waiting for these two behaviors. time, the path that consumes the least time will eventually be selected, including the following steps:

(1) Calculate the evacuation time t1 for the current grid to receive the crowded crowd;

(2) It takes t2 to calculate the evacuation time of bypassing the crowded crowd on the current grid;

(3) Compare times t1 and t2, and select the path that takes less time.

(4) Update the shortest path of the affected node to the destination K4.