Crowd evacuation prediction method and system based on elliptical particle model

Abstract

The invention is suitable for the field of public safety, and provides a crowd evacuation prediction method and system based on an elliptical particle model. The method comprises the steps of generating pedestrian elliptical particles for all pedestrian individuals in an area where a to-be-evacuated crowd is located based on an elliptical particle model; calculating a driving force and a driving torque of each pedestrian elliptical particle in the motion field, an interaction force between each pedestrian elliptical particle and each adjacent pedestrian elliptical particle in a preset interaction range, and an interaction force between each pedestrian elliptical particle and an adjacent building wall; for each pedestrian elliptical particle, performing short-axis direction decomposition on the physical contact force between the pedestrian elliptical particle and the adjacent pedestrian elliptical particle and the physical contact force between the pedestrian elliptical particle and the adjacent building wall to obtain component forces in the short-axis direction; calculating a compression strain value of each pedestrian elliptical particle; and carrying out risk assessment according to the component force and the compression strain value in the short-axis direction. According to the invention, the crowd evacuation prediction result is more real and more reliable.

Description

Technical Field

[0001] This application belongs to the field of public safety, and in particular relates to a crowd evacuation prediction method and system based on an elliptic particle model. Background Technology

[0002] With the acceleration of urbanization and the increasing frequency of public activities, the population density in large gathering places (such as subway stations, stadiums, shopping malls, airports, tunnels, etc.) has increased significantly. How to efficiently and safely organize the evacuation of people and avoid stampedes and casualties during daily operations and in the event of emergencies has become a major global challenge.

[0003] Crowd evacuation prediction is a crucial element in achieving safe management and emergency response. By simulating and predicting crowd movement patterns, density distribution, and evacuation routes, the effectiveness of evacuation plans can be evaluated, emergency response plans optimized, and a scientific basis provided for on-site decision-making. Currently, research on crowd evacuation prediction mainly relies on computer simulation techniques, including macroscopic models (such as fluid dynamics models) and microscopic models (such as cellular automata models, social force models, and particle models). Among these, particle models have a significant advantage in representing crowd dynamics at the microscopic level because they can simulate individual behavior and interactions.

[0004] This presents unprecedented opportunities for real-time monitoring of crowd status, analysis of environmental information, and provision of intelligent guidance. By deploying various sensors and smart hardware, massive amounts of dynamic crowd data and environmental parameters can be acquired, providing data support for inputting crowd evacuation prediction models and generating guidance strategies. Therefore, combining advanced simulation and prediction technologies with intelligent monitoring and guidance methods is the future trend of crowd evacuation management.

[0005] However, despite the progress made in existing crowd evacuation prediction and management technologies, significant shortcomings remain in practical applications. Current mainstream micro-level crowd evacuation prediction models, such as cellular automata and social force models, generally simplify individuals in a crowd as non-directional, rigid circular or square particles. This oversimplification prevents the models from accurately reflecting the complex behavior of real humans in dense crowds: they ignore the significant differences between shoulder width and front-to-back width, as well as the deformation capacity under compression; they fail to simulate the differences in movement capabilities in different directions (forward, backward, left, right), and also ignore the rotational inertia of the human body, making the simulated crowd turning and obstacle avoidance behaviors inconsistent with reality. More importantly, these models struggle to accurately simulate the differences in friction and compression capabilities between individuals in different directions, and cannot predict the risk of suffocation due to chest compression under extreme crowding, severely limiting their ability to assess safety in high-density scenarios. Furthermore, although modern high-precision crowd positioning technologies can acquire richer directional information such as shoulder width, existing models, due to their inherent limitations, cannot effectively utilize this monitoring data and fail to fully exploit its value. Summary of the Invention

[0006] The purpose of this application is to provide a crowd evacuation prediction method and system based on the elliptic particle model, which aims to solve the problem that existing cellular automata models and social force models cannot realistically simulate human geometric features and complex microscopic interactions.

[0007] In a first aspect, this application provides a crowd evacuation prediction method based on an elliptic particle model, comprising the following steps:

[0008] S101. Based on the elliptical particle model, generate pedestrian elliptical particles from all pedestrians in the area where the crowd to be evacuated is located. The elliptical particle model is a pre-constructed elliptical particle model that abstracts each pedestrian as a two-dimensional ellipse to generate pedestrian elliptical particles.

[0009] S102. Detect multiple pedestrian elliptical particles and building walls adjacent to the multiple pedestrian elliptical particles within a preset interaction range;

[0010] S103. Acquire and calculate the state parameters of each pedestrian elliptical particle within a preset interaction range, wherein the state parameters include basic physical properties and physiological and psychological state properties.

[0011] S104. Calculate the driving force and driving torque of each pedestrian elliptical particle in the motion field based on the state parameters of each pedestrian elliptical particle within the preset interaction range, calculate the interaction force between each pedestrian elliptical particle and each neighboring pedestrian elliptical particle within the preset interaction range, and calculate the interaction force between each pedestrian elliptical particle and the adjacent building wall. The interaction force includes physical contact force and psychological repulsion force.

[0012] S105. For each pedestrian elliptical particle, decompose the physical contact force with neighboring pedestrian elliptical particles and the physical contact force with adjacent building walls into minor axis components to obtain the minor axis component; calculate the compressive strain value of each pedestrian elliptical particle that reflects the individual crowding risk index.

[0013] S106. Based on the component force and compressive strain value in the short axis direction, a risk assessment is performed. When the component force in the short axis direction exceeds the preset physiological safety threshold or the compressive strain value approaches the risk limit, it is determined that the pedestrian is in an extremely dangerous state and a risk warning of crowding disaster is triggered.

[0014] S107. Update the motion state of the pedestrian elliptical particles. After all pedestrian elliptical particles have been updated, return to S102 until the evacuation is completed.

[0015] Secondly, this application provides a crowd evacuation prediction system based on an elliptic particle model, the system including a server, the server performing the steps of the crowd evacuation prediction method based on the elliptic particle model as described above.

[0016] In this application, by generating pedestrian elliptical particles from all individual pedestrians in the area where the crowd to be evacuated is located based on an elliptical particle model, this elliptical particle model can more accurately reflect the asymmetrical geometric shape of the real human body (the difference between shoulder width and front-to-back width) and its deformation capacity under dense and crowded conditions. Furthermore, by calculating the driving force and driving torque of each pedestrian elliptical particle in the motion field based on the state parameters of each pedestrian elliptical particle within a preset interaction range, calculating the interaction force between each pedestrian elliptical particle and each neighboring pedestrian elliptical particle within the preset interaction range, and calculating the interaction force between each pedestrian elliptical particle and adjacent building walls, the simulation of the microscopic interaction behavior of the crowd is made more consistent with actual physical laws. This overcomes the limitations of traditional circular or square particle models in simulating high-density, complex interaction scenarios, making the crowd evacuation prediction results more realistic and reliable.

[0017] Furthermore, for each pedestrian elliptical particle, the physical contact force with neighboring pedestrian elliptical particles and the physical contact force with adjacent building walls are decomposed along the minor axis, yielding the component force along the minor axis. The compressive strain value, representing an individual crowding risk index, is calculated for each pedestrian elliptical particle. A risk assessment is performed based on the component force and compressive strain value along the minor axis. When the component force along the minor axis exceeds a preset physiological safety threshold or the compressive strain value approaches the risk limit, the pedestrian is determined to be in an extremely dangerous state, triggering a risk warning for crowding-related disasters. Therefore, it can identify potential fatal hazards in advance, providing crucial risk assessment information for emergency management, achieving high-precision congestion and life safety risk warnings, thereby effectively reducing the probability of stampedes and suffocation accidents and maximizing the protection of human life. This application is not only applicable to planar evacuation scenarios; its accurate simulation of real human geometry and interaction makes it more applicable to various complex scenarios (including areas prone to high-density congestion such as narrow passages, turns, and sloping passages). Attached Figure Description

[0018] Figure 1 This is a flowchart of a crowd evacuation prediction method based on an elliptical particle model provided in one embodiment of this application.

[0019] Figure 2 This is a schematic diagram of the shape of an elliptical particle model provided in an embodiment of this application.

[0020] Figure 3 This is a schematic diagram of the physical contact force of an elliptical particle model provided in an embodiment of this application.

[0021] Figure 4 This is a schematic diagram of the crowding tolerance limit of an elliptical particle model provided in an embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and beneficial effects of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0023] To illustrate the technical solution described in this application, specific embodiments are provided below.

[0024] Please see Figure 1 This is a flowchart of a crowd evacuation prediction method based on an elliptical particle model provided in an embodiment of this application. The crowd evacuation prediction method based on an elliptical particle model provided in an embodiment of this application includes the following steps:

[0025] S101. Based on the elliptical particle model, generate pedestrian elliptical particles from all pedestrians in the area where the crowd to be evacuated is located. The elliptical particle model is a pre-constructed elliptical particle model that abstracts each pedestrian as a two-dimensional ellipse to generate pedestrian elliptical particles.

[0026] All pedestrians in the area where the evacuation group is located are obtained using existing technologies. For example, video stream data can be obtained by deploying top-view or high-angle oblique-view cameras at a high position in the area, or point cloud data can be obtained by using depth cameras. Then, object detection algorithms (such as the YOLO algorithm) are used to identify each pedestrian in the video stream data or point cloud data, and the bounding box or segmentation contour of the pedestrian is obtained.

[0027] To maintain continuous identification of the same person across consecutive video frames, existing multi-target tracking algorithms can be employed. These algorithms assign a unique pedestrian ID to each newly appearing individual and continuously update their position in subsequent video frames.

[0028] S102. Detect multiple pedestrian elliptical particles and building walls adjacent to the multiple pedestrian elliptical particles within a preset interaction range.

[0029] In one embodiment of this application, S102 may specifically include the following steps:

[0030] S1021. At the beginning of each time step, select multiple pedestrian elliptical particles and building walls adjacent to the multiple pedestrian elliptical particles within the interaction range of the preset interaction radius.

[0031] The time step is a basic unit in deep learning for processing sequence data, used to represent the temporal relationship between data points in the sequence. The interaction radius refers to the radius of the circle formed by the interacting pedestrian elliptical particles. It can be dynamically adjusted according to the needs of the crowd simulation scenario, and the value of the interaction radius can usually be set to 2m, 3m, 5m or larger.

[0032] S1022. Obtain a set list consisting of the current pedestrian elliptical particle i, the neighboring pedestrian elliptical particles j, and the adjacent building wall W within the interaction range, and use the set list as the interaction list that has potential interaction with the current pedestrian elliptical particle i.

[0033] S103. Acquire and calculate the state parameters of each pedestrian elliptical particle within the preset interaction range, wherein the state parameters include basic physical properties and physiological and psychological state properties.

[0034] In one embodiment of this application, the basic physical properties include:

[0035] Position vector r i(t) = (x i , y i ): Represents the Cartesian coordinates of the center of the ellipse of pedestrian elliptical particle i at time t; the preliminary two-dimensional ellipse center coordinates of an individual pedestrian can be obtained by calculating the pedestrian bounding box or geometric contour and using computer vision libraries (such as OpenCV);

[0036] velocity vector v i (t) = (v ix , v iy ): Represents the instantaneous velocity of pedestrian elliptical particle i at time t; the instantaneous velocity can be calculated from the centroid changes of consecutive image frames.

[0037]

[0038] in, It is the time difference between consecutive frames of an image;

[0039] Orientation Angle : Represents the angle between the major axis of the ellipse of pedestrian elliptical particle i and the global coordinate system (such as the x-axis), and defines the orientation of pedestrian elliptical particle i at time t. The major axis of the ellipse represents the shoulder direction of pedestrian individual i.

[0040] angular velocity : Represents the instantaneous angular velocity of the pedestrian elliptical particle i at time t, that is, the rotational velocity of the elliptical particle;

[0041] Geometric dimensions (a) i , b i ): represents the semi-major axis and semi-minor axis of the ellipse of pedestrian elliptical particle i, respectively, 2a i The dynamic shoulder width representing individual pedestrian i, 2b i The thickness of the chest cavity of individual pedestrian i is typically represented by a. i > b i Please see Figure 2 ;

[0042] mass m i : Represents the mass of pedestrian elliptic particle i;

[0043] Moment of inertia I i : represents the moment of inertia of the elliptical particle i rotating about its center of mass. For an ellipse, it can be approximated as . .

[0044] In one embodiment of the present invention, for the segmented contours of individual pedestrians acquired from a top-view or high-angle oblique-view camera, a least-squares ellipse fitting algorithm is run. In the OpenCV library, the `fitEllipse()` function can be directly called. Taking the point set of the individual pedestrian segmented contours as input, this function returns an object containing ellipse information. This algorithm can fit an optimal ellipse to the segmented contour of each individual pedestrian, directly outputting the center point position and major axis dimension 2a of the ellipse. i 2b (short axis dimension) i Orientation angle .

[0045] In cases where the segmentation quality of individual pedestrians is poor or the fitting fails, the average velocity of the individual pedestrians in the most recent frames can be used. Vector direction as the individual pedestrian's orientation angle The approximation or retention of the same state as the previous frame may occur in scenes where the view is obstructed.

[0046] Physiological and psychological state attributes include:

[0047] Endurance value E i (t): A scalar in the range [0, 1], representing the physical strength level of pedestrian individual i at time t; E i The value of (t) will decrease due to continuous high-speed movement and high-intensity compression, directly affecting the expected speed of pedestrian individual i. When physical strength is exhausted, It will decrease accordingly. This represents the maximum desired speed.

[0048] ;

[0049] : Indicates the desired direction of motion of pedestrian elliptical particle i;

[0050] τ i : Indicates the reaction time of pedestrian elliptical particle i;

[0051] : Represents the expected speed of pedestrian individual i;

[0052] Psychological state parameter S i (t): Used to quantify the emotional state of pedestrian individual i at time t, which can be described as S i (t) is a set of scalars consisting of [panic coefficient and calmness coefficient], with the sum of the scalar parameters being 1.0. The larger the calmness coefficient, the higher the pedestrian's probability of avoiding obstacles during path planning. i Larger sizes lead to weaker conformity, affecting the expected velocity in the scenario. and desired direction of motion The calculation of S i (t) is dynamic. For example, prolonged exposure to a high-density environment or the perception of danger signals will cause S to change. i (t) Tends towards panic. S i (t) affects the reaction time τ i Expected speed The probability of avoidance, p i Several behavioral parameters, such as reaction time τ during panic. i Shorter, expected speed Faster, but pedestrians move more erratically and are more sensitive to psychological aversion to others, increasing the probability of avoidance p. i Smaller. The panic coefficient is mainly calculated based on an individual pedestrian's panic susceptibility and their endurance value E. i (t) and related to the spread of panic among other pedestrians within the preset interaction range.

[0053] S104. Calculate the driving force and driving torque of each pedestrian elliptical particle in the motion field based on the state parameters of each pedestrian elliptical particle within the preset interaction range, calculate the interaction force between each pedestrian elliptical particle and each neighboring pedestrian elliptical particle within the preset interaction range, and calculate the interaction force between each pedestrian elliptical particle and the adjacent building wall. The interaction force includes physical contact force and psychological repulsion force.

[0054] Calculate the driving force of pedestrian elliptical particle i in the motion field. It can be represented by the following formula:

[0055]

[0056] Where, m i τ is the mass of the pedestrian elliptic particle i; i It is the reaction time of pedestrian elliptical particle i. It is the expected speed of pedestrian individual i. v is the expected direction of motion of the pedestrian elliptical particle i. i (t) is the instantaneous velocity of the pedestrian elliptic particle i at time t.

[0057] Calculate the driving torque of pedestrian elliptical particle i in the motion field. It can be represented by the following formula:

[0058]

[0059] in, and I is a constant representing the speed at which an individual pedestrian aligns with the expected direction. Its value is related to pedestrian heterogeneity, which refers to the behavioral differences exhibited by different pedestrians in the same traffic environment due to factors such as age, gender, education level, and physical condition. i It is the moment of inertia of the pedestrian elliptical particle i rotating around its center of mass; It is the angular velocity of the pedestrian elliptical particle i; It is the orientation angle of the pedestrian elliptical particle i; The direction angle of the current velocity.

[0060] In one embodiment of this application, calculating the interaction force between each pedestrian elliptical particle and each neighboring pedestrian elliptical particle within a preset interaction range specifically includes the following steps:

[0061] Traverse the interaction list that has potential interaction with the current pedestrian elliptical particle i. Perform geometric overlap detection on each neighboring pedestrian elliptical particle j. If there is no overlap, it is determined to be a long-distance psychological interaction. Calculate the psychological repulsion force generated by pedestrian elliptical particle i on neighboring pedestrian elliptical particle j. If there is overlap, it is determined to be a close-range physical contact. Calculate the physical contact force and interaction torque between the current pedestrian elliptical particle i and neighboring pedestrian elliptical particle j. The physical contact force includes the normal force along the common normal and the tangential force perpendicular to the common normal.

[0062] In one embodiment of this application, psychological repulsion refers to the force between the relative distance between the centers of mass of pedestrian elliptical particle i and its neighboring pedestrian elliptical particle j before their surfaces have come into contact. When the distance is less than the predetermined social safety distance, psychological resistance will be generated. ,in , Denotes the mass center of the elliptical particle i of the pedestrian. Let represent the centroid of the neighboring elliptical particle j. While the calculation of psychological repulsion is based on the distance between the centroids, its magnitude is modulated by the relative orientation of the elliptical particles. For example, the repulsion force when two elliptical particles face each other is significantly greater than when they are side-by-side or back-to-back, reflecting directional sensitivity in real social interactions. When calculating psychological repulsion, an anisotropic modulation factor is multiplied from the original force magnitude, which is based solely on the distance between the centroids. The modulation factor is related to the angle difference between the pedestrian's direction of movement;

[0063] in, It is the psychological repulsive force exerted by pedestrian elliptical particle i on neighboring pedestrian elliptical particle j. It is the modulation factor. It is an exponential decay function that depends only on distance. It is the relative distance between the centers of mass of pedestrian elliptical particle i and its neighboring pedestrian elliptical particle j. It is the unit vector pointing from pedestrian elliptical particle j to pedestrian elliptical particle i.

[0064] In one embodiment of this application, please refer to Figure 3 Physical contact force refers to the force generated when the surfaces of pedestrian elliptical particle i and its neighboring pedestrian elliptical particle j come into contact or overlap. Physical contact force is decomposed into normal force along the common normal. and tangential force perpendicular to the common normal The calculation of physical contact forces is based entirely on surface geometry, not the elliptical centroid.

[0065] Among them, normal force Used to simulate compression and elastic collisions, its magnitude is an increasing function of the overlap depth, calculated using a spring-damped model to penalize overlap, as shown in the following equation:

[0066] in, It is the stiffness coefficient; d overlap It is the overlap depth, which represents the minimum distance a pedestrian elliptical particle needs to move along the common normal direction to eliminate overlap; It is the dissipation coefficient. It is the relative velocity component of the pedestrian elliptical particle i in the direction of the common normal of the two elliptical contact surfaces. This force is the direct cause of the human body feeling pushed and squeezed under high density. It is the common normal vector of the contact surface between pedestrian elliptical particle i and neighboring pedestrian elliptical particle j. The common normal vector is perpendicular to the tangent at the contact point and represents the most important direction of force.

[0067] The spring-damped model is a mathematical model used to describe the behavior of a spring and damper system. It combines the elastic characteristics of a spring with the energy dissipation characteristics of a damper and has wide applications in engineering.

[0068] Tangential force Used to simulate friction, preventing or slowing down the relative sliding between elliptical particles of pedestrians. The direction of the tangential force is related to the relative tangential velocity. Conversely, its magnitude is proportional to the normal force, but does not exceed the maximum static friction force, as shown in the following formula:

[0069]

[0070] in, It is the coefficient of friction. It is the tangential damping coefficient. It is the relative tangential velocity of the pedestrian elliptical particle i in the tangential direction of the two elliptical contact surfaces; the tangential force is crucial for simulating bottleneck jamming and arching effects in crowds.

[0071] In one embodiment of this application, the interaction torque generated when pedestrian elliptical particle i and neighboring pedestrian elliptical particle j come into contact. As shown in the following formula:

[0072]

[0073] in, It is the position vector from the center point of the pedestrian elliptical particle i to the center point of the contact position. It is the normal force. It is a tangential force.

[0074] In one embodiment of this application, calculating the interaction force between each pedestrian elliptical particle and the adjacent building wall specifically includes the following steps:

[0075] Detect whether the current pedestrian elliptical particle i has made contact with the adjacent building wall W in the interaction list where there is potential interaction. If there is no contact, it is determined to be a long-distance psychological interaction, and the psychological repulsion force generated by the current pedestrian elliptical particle on the adjacent building wall W is calculated. If contact has occurred, it is considered close-range physical contact. The physical contact force between the current pedestrian elliptical particle i and the adjacent building wall W is calculated, including the wall's normal thrust. Tangential friction with the wall ;

[0076]

[0077] in, It is the psychological repulsive force exerted by the current pedestrian elliptical particle on the adjacent building wall W. It is the modulation factor. It is an exponential decay function that depends only on distance. It is the distance between the center of mass of the pedestrian elliptical particle i and the building wall W. It is the unit vector pointing from the building wall W to the pedestrian elliptical particle i.

[0078] normal thrust of the wall The model is used to simulate the squeezing and elastic collision between a pedestrian elliptical particle i and a building wall W. Its size is an increasing function of the overlap depth and is calculated using a spring-damped model to penalize the overlap, as shown in the following formula.

[0079]

[0080] in, It is the stiffness coefficient. It is the overlap depth between the pedestrian elliptical particle and the building wall. It is the dissipation coefficient. It is the relative velocity component of the pedestrian elliptical particle i in the direction of the normal to the contact surface with the building wall.

[0081] Tangential friction of the wall Used to simulate the frictional force generated with a building wall W, its direction and relative tangential velocity Conversely, its magnitude is proportional to the normal force, but does not exceed the maximum static friction force, as shown in the following formula.

[0082] ,in, It is the coefficient of friction. It is the tangential damping coefficient. It is the relative tangential velocity of the pedestrian elliptical particle i in the tangential direction of the contact surface with the building wall.

[0083] S105. For each pedestrian elliptical particle, decompose the physical contact force with neighboring pedestrian elliptical particles and the physical contact force with adjacent building walls into minor axis components to obtain the minor axis component; calculate the compressive strain value of each pedestrian elliptical particle that reflects the individual crowding risk index.

[0084] Component of force along the minor axis It can intuitively reflect the intensity of the contact force in the most dominant direction experienced by pedestrian individual i at the current moment. The component of the force along the minor axis. In other words, the force component in the direction of the chest cavity is crucial for assessing crowding risk.

[0085] The specific calculation of the compressive strain value, which represents the individual crowding risk index for each pedestrian elliptical particle, is as follows:

[0086] Define a compressive strain value for each pedestrian elliptical particle i. As an instantaneous, quantifiable indicator of individual crowding risk, The value directly reflects the degree of compression currently experienced by pedestrian individual i:

[0087]

[0088] Among them, b i,0 The semi-minor axis length b of the pedestrian elliptical particle i i The initial value of (t), b i (t) is the semi-minor axis length calculated in real time based on the current effect of the surrounding pedestrian elliptical particles.

[0089] The semi-minor axis length b is calculated in real time based on the current influence of the surrounding pedestrian elliptical particles. i (t) Specifically includes the following steps:

[0090] Maintain the center position r of the pedestrian elliptical particle i i (t) and orientation angle Unchanged; semi-major axis a is reduced proportionally. i and semi-minor axis b i The critical tangent dimension is determined as the semi-minor axis length b, until the boundary of pedestrian elliptical particle i becomes tangent to or non-overlapping with the boundaries of all other pedestrian elliptical particles or walls that overlap with pedestrian elliptical particle i. i (t).

[0091] In one embodiment of this application, the overlap state can be initially determined using the center distance and major axis of the ellipse. If it is impossible to initially determine whether there is overlap, the parametric equation of the ellipse can be used to solve for the overlap state between the pedestrian elliptical particles. By simultaneously solving the general expressions of the ellipses corresponding to the two pedestrian elliptical particles, the two ellipses are exactly tangent when the quartic equation after elimination has a unique real root. The solution of the quartic equation can be implemented using existing code libraries (such as the NumPy library in Python).

[0092] because and b i (t) is scaled down proportionally, therefore, based on the instantaneous geometric dimensions b of the pedestrian elliptical particle i. i The compressive strain value can be calculated using (t). .

[0093] In crowded conditions, pedestrians' bodies undergo elastic deformation due to compression, causing their geometric dimensions to locally shrink to a level just large enough to fit within the current physical space, i.e., tangent to neighboring pedestrian elliptical particles or adjacent building walls. Therefore, in one embodiment of the present invention, the geometric deformation of the pedestrian elliptical particles, i.e., their geometric dimensions (a) are updated in real time. i , b i For the compression deformation between pedestrian elliptical particle i and neighboring pedestrian elliptical particle j, the calculation is based on the maximum possible deformation when both are compressed against each other.

[0094] Updating the geometric deformation of pedestrian elliptical particles specifically includes the following steps:

[0095] For each overlapping pedestrian elliptical particle i, keep the center position and orientation of pedestrian elliptical particle i unchanged, and proportionally reduce the semi-major axis dimension a. i and semi-minor axis dimension b i At critical size and b i Under condition (t), the boundary of pedestrian elliptical particle i is exactly tangent to the boundaries of other overlapping pedestrian elliptical particles. If no overlap occurs, then... and b i (t) Keep the initial value a i,0 and b i,0 .

[0096] S106. Based on the component force and compressive strain value in the short axis direction, a risk assessment is performed. When the component force in the short axis direction exceeds the preset physiological safety threshold or the compressive strain value approaches the risk limit, it is determined that the pedestrian is in an extremely dangerous state, and a risk warning for crowding-induced disaster is triggered.

[0097] The compressive strain value approaching the risk limit specifically refers to the compressive strain value Approximately 1 - Maximum Crowding Tolerance Coefficient C limit,I At that time, the maximum crowding tolerance coefficient C limit,i This is a dimensionless parameter, typically ranging from 0.85 to 1.0, representing the maximum compressibility of an individual's thoracic cavity. It is pre-set for each pedestrian elliptical particle i. Please refer to... Figure 4 .

[0098] The semi-minor axis length b of the pedestrian elliptical particle i i (t) cannot be compressed below its initial value b. i,0 With the maximum crowding tolerance coefficient C limit,i The product of, i.e. The initial value of b for the semi-minor axis length of each pedestrian elliptical particle i. i,0 With the maximum crowding tolerance coefficient C limit,i The product of α and β reflects a person's limits. It is set according to human physiology and is a key parameter for judging the limits of survival. The semi-minor axis refers to half the thickness of the chest cavity.

[0099] In one embodiment of this application, the method may further include the following steps:

[0100] Calculate the duration during which the compressive strain value approaches the risk limit. When the duration exceeds the preset value, assess the risk of asphyxiation due to prolonged chest compression.

[0101] S107. Update the motion state of the pedestrian elliptical particles. After all pedestrian elliptical particles have been updated, return to S102 until the evacuation is completed.

[0102] In one embodiment of this application, S107 specifically includes the following steps:

[0103] S1071. Calculate the total resultant force on each of the pedestrian elliptical particles i. Total resultant torque .

[0104] The formula for calculating the total net force on pedestrian elliptical particle i is as follows:

[0105]

[0106] Where, m i v is the mass of the pedestrian elliptic particle i; iIt is the velocity of the pedestrian elliptical particle i. It is the simulation time; It is the net force acting on the pedestrian elliptical particle i; It is the driving force of the pedestrian elliptical particle i in the motion field; It is the interaction force between pedestrian elliptic particle i and neighboring pedestrian elliptic particle j, including psychological repulsion and physical contact force; It is the influence of pedestrian elliptical particle i on building walls W, that is, the interaction force between pedestrian elliptical particle i and adjacent building walls W, including psychological repulsion force and physical contact force.

[0107] The formula for calculating the total resultant torque on pedestrian elliptical particle i is as follows:

[0108] Where, m i It is the mass of the pedestrian elliptic particle i; ω is the angular velocity of the elliptical particle i moving in the pedestrian motion; t is the simulation time. It is the resultant torque acting on the pedestrian elliptical particle i; It is the driving torque of the pedestrian elliptical particle i in the motion field; It is the interactive torque generated when pedestrian elliptical particle i and neighboring pedestrian elliptical particle j come into contact. It is used to simulate the torque generated by the contact between pedestrians, causing the pedestrian's body to deflect due to the force. The contact torque between the pedestrian elliptical particle i and the building wall W is the force exerted on it.

[0109] S1072. Use a high-order numerical integration method to update the motion state of all pedestrian elliptical particles i one by one, for example, the motion velocity v i angular velocity The parameters are calculated using the formula shown below.

[0110]

[0111] Update the position coordinates r of the pedestrian elliptical particle i i = (x i , y i ) and orientation angle For the next time step Prepare,

[0112]

[0113]

[0114] S1073. After all pedestrian elliptical particles i have completed the calculations of the above steps, the entire simulation time step is advanced. Return to S102 until the evacuation is complete.

[0115] In one embodiment of this application, the crowd evacuation prediction method based on the elliptical particle model may further include the following steps after S106:

[0116] Based on the results of the risk assessment, high-risk areas of chest compression are identified, and the optimal evacuation route is calculated for people in these high-risk areas, triggering a risk warning for these areas.

[0117] The method for determining high-risk areas of chest compression is as follows: mark pedestrians at high risk of compression strain based on the results of risk assessment, map the locations of all marked pedestrians at high risk of compression strain onto the scene grid map, and determine the grid cell as a high-risk area of ​​chest compression when there are pedestrians at high risk of compression strain in the grid cell near the room or exit.

[0118] An embodiment of this application also provides a crowd evacuation prediction system based on an elliptic particle model. The system includes a server that performs the steps of the crowd evacuation prediction method based on an elliptic particle model as provided in an embodiment of this application.

[0119] In this application, by generating pedestrian elliptical particles from all individual pedestrians in the area where the crowd to be evacuated is located based on an elliptical particle model, this elliptical particle model can more accurately reflect the asymmetrical geometric shape of the real human body (the difference between shoulder width and front-to-back width) and its deformation capacity under dense and crowded conditions. Furthermore, by calculating the driving force and driving torque of each pedestrian elliptical particle in the motion field based on the state parameters of each pedestrian elliptical particle within a preset interaction range, calculating the interaction force between each pedestrian elliptical particle and each neighboring pedestrian elliptical particle within the preset interaction range, and calculating the interaction force between each pedestrian elliptical particle and adjacent building walls, the simulation of the microscopic interaction behavior of the crowd is made more consistent with actual physical laws. This overcomes the limitations of traditional circular or square particle models in simulating high-density, complex interaction scenarios, making the crowd evacuation prediction results more realistic and reliable.

[0120] Furthermore, for each pedestrian elliptical particle, the physical contact force with neighboring pedestrian elliptical particles and the physical contact force with adjacent building walls are decomposed along the minor axis, yielding the component force along the minor axis. The compressive strain value, representing an individual crowding risk index, is calculated for each pedestrian elliptical particle. A risk assessment is performed based on the component force and compressive strain value along the minor axis. When the component force along the minor axis exceeds a preset physiological safety threshold or the compressive strain value approaches the risk limit, the pedestrian is determined to be in an extremely dangerous state, triggering a risk warning for crowding-related disasters. Therefore, it can identify potential fatal hazards in advance, providing crucial risk assessment information for emergency management, achieving high-precision congestion and life safety risk warnings, thereby effectively reducing the probability of stampedes and suffocation accidents and maximizing the protection of human life. This application is not only applicable to planar evacuation scenarios; its accurate simulation of real human geometry and interaction makes it more applicable to various complex scenarios (including areas prone to high-density congestion such as narrow passages, turns, and sloping passages).

[0121] It should be understood that the steps in the various embodiments of this application are not necessarily executed sequentially according to the order indicated by the step numbers. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in each embodiment may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0122] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0123] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0124] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A crowd evacuation prediction method based on an ellipse particle model, characterized in that, The method comprises the following steps: S101, generating a pedestrian ellipse particle of each pedestrian individual in a region where the crowd to be evacuated is located based on an ellipse particle model, the ellipse particle model being a pre-constructed ellipse particle model for abstracting each pedestrian individual as a two-dimensional ellipse to generate a pedestrian ellipse particle; S102, detecting a plurality of pedestrian ellipse particles in a preset interaction range and building walls adjacent to the plurality of pedestrian ellipse particles; S103, acquiring and calculating a state parameter of each pedestrian ellipse particle in the preset interaction range, the state parameter comprising basic physical attributes and physiological and psychological state attributes; S104, calculating a driving force and a driving torque of each pedestrian ellipse particle in a motion field according to the state parameter of each pedestrian ellipse particle in the preset interaction range, calculating an interaction force between each pedestrian ellipse particle and each adjacent pedestrian ellipse particle in the preset interaction range, and calculating an interaction force between each pedestrian ellipse particle and an adjacent building wall, the interaction force comprising a physical contact force and a psychological repulsion force; S105, for each pedestrian ellipse particle, respectively decomposing the physical contact force between adjacent pedestrian ellipse particles and the physical contact force between the adjacent building wall in a minor axis direction to obtain a component force in the minor axis direction, and calculating a compression strain value of each pedestrian ellipse particle as an individual crowding risk indicator; S106, performing risk assessment according to the component force in the minor axis direction and the compression strain value, determining that the pedestrian is in an extremely dangerous state when the component force in the minor axis direction exceeds a preset physiological safety threshold or the compression strain value approaches a risk limit, and triggering a risk warning of crowd disaster; S107, updating the motion state of the pedestrian ellipse particle, and returning to S102 when all pedestrian ellipse particles are updated, until evacuation is completed.

2. The method of claim 1, wherein, S102 specifically comprises the following steps: S1021, at the beginning of each time step, screening a plurality of pedestrian ellipse particles in an interaction range of a preset interaction radius and building walls adjacent to the plurality of pedestrian ellipse particles; S1022, acquiring a set list composed of a current pedestrian ellipse particle i, adjacent pedestrian ellipse particles j, and adjacent building walls W in the interaction range, and taking the set list as an interaction list of potential interactions with the current pedestrian ellipse particle i.

3. The method of claim 1, wherein the basic physical attributes comprise: The physiological and psychological state attributes comprise: Position vector r i (t) = (x i , y i ) : Cartesian coordinates of the center of the ellipse of the pedestrian ellipsoidal particle i at time t; Velocity vector v i (t) = (v ix , v iy ) : represents the instantaneous velocity of the pedestrian elliptical particle i at time t; the instantaneous velocity is calculated by the change of the mass center of the image continuous frames: wherein, is the time difference of the image continuous frames; orientation angle : represents the angle between the major axis of the ellipse of the pedestrian ellipse particle i and the global coordinate system, which defines the orientation of the pedestrian ellipse particle i at time t, and the major axis of the ellipse represents the shoulder direction of the pedestrian individual i; angular velocity : represents the instantaneous angular velocity of the pedestrian elliptical particle i at time t, i.e., the rotation speed of the elliptical particle; Geometric dimensions (a i , b i ): represent the semi-major axis and semi-minor axis of the ellipse of the pedestrian ellipse particle i, respectively, 2a i represents the dynamic shoulder width of the pedestrian individual i, 2b i represents the chest thickness of the pedestrian individual i; mass m i : denotes the mass of the pedestrian ellipsoidal particle i; moment of inertia I i : denotes the moment of inertia of the pedestrian ellipsoidal particle i about its center of mass; 4. The method of claim 1, wherein the calculation of the interaction force between each pedestrian ellipse particle and each adjacent pedestrian ellipse particle in the preset interaction range specifically comprises the following steps: Endurance value E i (t): a scalar ranging [0, 1] representing the level of energy of the pedestrian individual i at time t; E i The value of E , represents the maximum value of the desired speed, ; : denotes the desired motion direction of the pedestrian ellipsoidal particle i; τ i : represents the reaction time of the pedestrian elliptical particle i; : denotes the desired speed of the pedestrian individual i; Psychological state parameter S i (t): for quantifying the emotional state of a pedestrian individual i at time t, which can be described as S i (t) = [panic coefficient, calm coefficient] a set of scalars, the sum of the scalar parameters is 1.

0. ​ calculating a driving force of a pedestrian ellipsoidal particle i in a motion field is expressed by the following equation: where m i is the mass of the pedestrian ellipse particle i; τ i is the reaction time of the pedestrian ellipse particle i, is the desired speed of the pedestrian individual i, is the desired direction of motion of the pedestrian ellipse particle i, v i (t) is the instantaneous speed of the pedestrian ellipse particle i at time t; Computing the driving torque of a pedestrian ellipsoidal particle i in a motion field As expressed by the following formula: wherein, and are constants characterizing the speed of transition of the individual pedestrian to align with the intended direction; I i is the moment of inertia of the pedestrian ellipse particle i about its center of mass; is the angular velocity of the pedestrian ellipse particle i; is the orientation angle of the pedestrian ellipse particle i; is the direction angle of the current velocity; ​ The interaction list with potential interaction with the current pedestrian ellipse particle i is traversed, and geometric overlap detection is performed on each adjacent pedestrian ellipse particle j. If there is no overlap, it is determined that there is a long-distance psychological interaction, and the psychological repulsion force generated by the current pedestrian ellipse particle i on the adjacent pedestrian ellipse particle j is calculated. If there is overlap, it is determined that there is close-range physical contact, and the physical contact force and interaction torque between the current pedestrian ellipse particle i and the adjacent pedestrian ellipse particle j are calculated. The physical contact force includes a normal force along the common normal line and a tangential force perpendicular to the common normal line; The psychological repulsion force refers to a psychological repulsion force generated when the surface of the pedestrian ellipse particle i and the surface of the adjacent pedestrian ellipse particle j have not contacted, but the relative distance of the centers of mass is less than a predetermined social safety distance wherein , represents the center of mass of the pedestrian ellipse particle i, represents the center of mass of the adjacent pedestrian ellipse particle j; and the psychological repulsion force formula is calculated as:​ wherein, is the psychological repulsion force exerted by a pedestrian ellipse particle i on a neighboring pedestrian ellipse particle j, is a modulation factor, is a distance-dependent exponential decay function, is the center-of-mass relative distance between the pedestrian ellipse particle i and the neighboring pedestrian ellipse particle j, is a unit vector pointing from the pedestrian ellipse particle j to the pedestrian ellipse particle i; The physical contact force refers to a physical contact force generated when the surface of the pedestrian ellipse particle i and the surface of the adjacent pedestrian ellipse particle j are in contact or overlap, and the physical contact force is decomposed into a normal force along the common normal line and a tangential force perpendicular to the common normal line ; where the normal force For simulating the squeeze and elastic collision, the size is an increasing function of the overlap depth, calculated using a spring-damper model to penalize overlap, as shown in the following equation: wherein, is the stiffness coefficient; d overlap is the overlap depth, representing the minimum distance a pedestrian ellipsoid needs to move in the direction of the common normal to eliminate overlap; is the dissipation coefficient, is the relative velocity component of the pedestrian ellipsoid i in the direction of the common normal of the two ellipsoids' contact surface; is the common normal vector of the contact surface of the pedestrian ellipsoid i and the neighboring pedestrian ellipsoid j, the common normal vector is perpendicular to the tangent of the contact point, representing the main direction of force action; tangential force For simulating the friction force, which prevents or slows down the relative sliding between the pedestrian elliptical particles, the direction of the tangential force is opposite to the relative tangential velocity On the contrary, the size is proportional to the normal force, but does not exceed the maximum static friction force, as shown in the following formula: Wherein, is the friction coefficient, is the tangential damping coefficient, is the relative tangential velocity of the pedestrian elliptical particle i in the tangent direction of the two elliptical contact surfaces; torque resulting from the contact between a pedestrian ellipsoidal particle i and a neighboring pedestrian ellipsoidal particle j as shown by the following equation: wherein, is a position vector of the center point of the pedestrian ellipsoidal particle i to the center point of the contact position, is a normal force, is a tangential force; The calculation of the interaction force between each pedestrian ellipse particle and the adjacent building wall specifically includes the following steps: Detecting whether the current pedestrian ellipse particle i has contact with the adjacent building wall W in the interaction list that exists potential interaction, if not in contact, it is determined as long-distance psychological interaction, and the psychological repulsion force generated by the current pedestrian ellipse particle on the adjacent building wall W is calculated , if in contact, it is determined as close-range physical contact, and the physical contact force between the current pedestrian ellipse particle i and the adjacent building wall W is calculated, including wall normal thrust force and wall tangential friction force ; wherein, is the psychological repulsion force generated by the current pedestrian ellipse particle i to the adjacent building wall W, is a modulation factor, is a distance-dependent exponential decay function, is the distance between the centroid of the pedestrian ellipse particle i and the building wall W, is the unit vector pointing from the building wall W to the pedestrian ellipse particle i; Wall normal thrust For simulating the extrusion and elastic collision between the pedestrian elliptical particle i and the building wall W, the size is the increasing function of the overlap depth, and the spring damping model is calculated to punish the overlap, as shown in the following formula: Wherein, is the stiffness coefficient, is the overlap depth of the pedestrian elliptical particle and the building wall, is the dissipation coefficient, is the relative velocity component of the pedestrian elliptical particle i in the normal direction of the contact surface of the building wall. Wall tangential friction force For simulating the friction force generated with the building wall W, the direction is opposite to the relative tangential velocity On the contrary, the size is proportional to the normal force, but does not exceed the maximum static friction force, as shown in the following formula: Wherein, is the friction coefficient, is the tangential damping coefficient, is the relative tangential velocity of the pedestrian elliptical particle i in the tangential direction of the contact surface with the building wall.

5. The method of claim 1, wherein, The compression strain value of each pedestrian ellipse particle, which embodies the individual crowd risk indicator, is specifically: A compression strain value is defined for each pedestrian ellipse particle i as an instantaneous, quantified individual risk indicator of crowd compression, whose value directly reflects the degree of compression currently experienced by the pedestrian individual i: Among them, b i,0 The semi-minor axis length b of the pedestrian elliptical particle i i The initial value of (t), b i (t) is the semi-minor axis length calculated in real time based on the current effect of the surrounding pedestrian elliptical particles; The semi-minor axis length b is calculated in real time according to the current effect of the surrounding pedestrian elliptical particles i (t) specifically includes the following steps: r = r i + r p i (t) and the orientation angle unchanged; isometrically reduced semi-major axis a i and semi-minor axis b i until the boundary of the pedestrian elliptical particle i becomes tangent or non-overlapping with the boundary of all other pedestrian elliptical particles or wall bodies that overlap with the pedestrian elliptical particle i, thereby determining the critical tangent size as the semi-minor axis length b i (t).

6. The method of claim 1, wherein, The geometric deformation of the pedestrian ellipse particle is updated, which specifically includes the following steps: For each overlapping pedestrian ellipse particle i, keep the center position and orientation of the pedestrian ellipse particle i unchanged, and scale down the semi-major axis size a i and semi-minor axis size b i by a factor of and b i (t) at which the boundary of the pedestrian ellipse particle i is tangent to the boundary of the other overlapping pedestrian ellipse particle, if no overlap occurs, then and b i (t) remain the initial values a i,0 and b i,0 .

7. The method of claim 1, wherein, The method further includes the following steps: The duration of the compression strain value approaching the risk limit is calculated, and when the duration exceeds a preset value, the risk of suffocation caused by long-time compression of the chest cavity is evaluated.

8. The method of claim 1, wherein, S107 specifically includes the following steps: S1071、Calculate the total force on each pedestrian ellipse particle i and the total moment , The total resultant force on the pedestrian ellipse particle i is calculated according to the following formula: Wherein, m i is the mass of the pedestrian ellipse particle i; v i is the moving speed of the pedestrian ellipse particle i; is the simulation time; is the resultant force on the pedestrian ellipse particle i; is the driving force of the pedestrian ellipse particle i in the moving field; is the interaction force between the pedestrian ellipse particle i and the adjacent pedestrian ellipse particle j, including the psychological repulsion force and the physical contact force; is the influence force on the pedestrian ellipse particle i between the building walls W, that is, the interaction force between the pedestrian ellipse particle i and the adjacent building wall W, including the psychological repulsion force and the physical contact force; The formula for calculating the total resultant moment on pedestrian ellipse particle i is as follows: Wherein, m i is the mass of pedestrian ellipse particle i; is the angular velocity of motion of pedestrian ellipse particle i; t is the simulation time; is the resultant moment on pedestrian ellipse particle i; is the driving moment of pedestrian ellipse particle i in the motion field; is the interaction moment between pedestrian ellipse particle i and adjacent pedestrian ellipse particle j when in contact, used to simulate the moment of force on the pedestrian due to contact, so that the body of the pedestrian deflects due to the force; is the contact moment on pedestrian ellipse particle i between the building walls W; S1072, update the motion state of all pedestrian ellipse particles i one by one by using a high-order numerical integration method; update the position coordinates r i = (x i , y i ) and the heading angle for the next time step ; S1073, after all pedestrian ellipse particles i complete the calculation of the above steps, the entire simulation time step is advanced by Δt, and S102 is returned until the evacuation is completed.

9. The method of claim 1, wherein, After S106, the following steps are further included: According to the results of risk assessment, the chest compression high-risk area is identified, and the current optimal evacuation path for the crowd in the chest compression high-risk area is calculated, and the risk warning of the chest compression high-risk area is triggered.

10. A crowd evacuation prediction system based on an ellipse particle model, characterized by, The system includes a server, and the server performs the steps of the crowd evacuation prediction method based on the ellipse particle model according to any one of claims 1 to 9.

Images