Multi-index risk quantification method based on high-density passenger flow in rail transit stations

Abstract

The present invention provides a multi-index risk quantification method based on high-density passenger flow within rail transit stations, comprising: Step S1: obtaining basic pedestrian data; Step S2: calculating pedestrian-related indicators based on the basic pedestrian data; Step S3: describing the scene's congestion level; and measuring the level of chaos; Step S4: normalizing the density (measure of congestion level) and the entropy (measure of chaos level) to obtain congestion risk values ​​and chaos risk values, and assigning weight coefficients to the congestion risk and chaos risk based on the characteristics of the scene area, thereby categorizing crowd risk into four risk levels; Step S5: obtaining time periods corresponding to the risk levels and strengthening monitoring of time periods with high risk levels. This method can achieve risk analysis and evaluation from different perspectives using multiple indicators. In the study of stampede risk, a density entropy indicator was proposed to measure the volatility of the spatial location of crowd density under high-density conditions, enriching the role of entropy in crowd risk research.

Description

Technical Field

[0001] The present invention relates to the field of rail transit technology, and in particular to a multi-index risk quantification method based on high-density passenger flow in rail transit stations. Background Art

[0002] The large passenger flows within rail transit stations during morning and evening peak hours pose significant safety risks to station operations. Localized disturbances or chaos within dense crowds can easily trigger stampede-like accidents, resulting in casualties. Current technologies and products within the industry show that passenger flow monitoring and early warning systems at transportation hubs typically assess public area risk by quantifying the number of passengers within a given area, inferring density and congestion, and using a density-based pedestrian service level grading method. These systems rarely employ indicators that reflect the level of disorder within a crowd to identify and warn of crowd risks. Therefore, there is an urgent need to develop an evaluation indicator system that accurately describes the risks of dense crowds, enabling real-time monitoring and early warning of passenger flow risks within stations and preventing serious accidents such as stampedes.

[0003] At present, the main problems in the risk assessment of clusters in rail transit stations are as follows:

[0004] (1) Looking at the current relevant technologies and products in the industry, the passenger flow monitoring and early warning systems at transportation hubs are mostly based on quantifying the number of passengers in a certain area, inferring the density and congestion level, and using a pedestrian service level classification method based on crowd density to assess the risk level of public areas. They rarely use indicators that reflect the level of internal chaos in the crowd to identify and warn of crowd risks in actual scenarios.

[0005] (2) Under extremely high density conditions, sudden changes in local crowd speed are the main cause of safety risks, which can increase the internal forces of the crowd and cause squeezing or even trampling injuries. In addition to traditional indicators such as density, speed, and evacuation time, crowd instability is an important indicator of risk, but it is currently rarely used in crowd risk assessment in rail transit stations. Moreover, most current studies do not consider the characteristics of pedestrian facilities in rail transit stations, such as the impact of obstacle layout, exit location, etc. on potential risks. Summary of the Invention

[0006] The present invention aims to provide a multi-index risk quantification method based on high-density passenger flow in rail transit stations to solve the above problems.

[0007] The technical solution of the present invention is: a multi-index risk quantification method based on high-density passenger flow in rail transit stations, including:

[0008] Step S1: Obtain basic data of pedestrians, that is, record the basic data of the coordinates (x, y) of each pedestrian at each moment through video tracking technology;

[0009] Step S2: Calculate pedestrian-related indicators based on pedestrian basic data, that is, calculate the speed and steering angle of each pedestrian in the x and y directions at each moment, and use Voronoi to calculate the local density of pedestrians;

[0010] Step S3: The scene average density is obtained by averaging the local pedestrian density calculated by Voronoi, and the scene average density is used to define the scene congestion level; the density entropy, speed entropy, and angle entropy are calculated according to the entropy calculation formula, and the mixed entropy is calculated based on the density entropy, speed entropy, and angle entropy to measure the chaos level;

[0011] Step S4: Normalize the density (measure of crowding level) and the entropy (measure of chaos level) to obtain crowding risk value and chaos risk value. Then, assign weight coefficients (0-1) to crowding risk and chaos risk based on the characteristics of the scene area, and classify crowd risk into four risk levels within 0-1.

[0012] Step S5: Obtain the time period corresponding to the risk level, and strengthen monitoring of the time period with a high risk level.

[0013] Preferably, step S3 specifically includes the following steps:

[0014] Step S31: Use the Voronoi diagram to define the local density of pedestrians in the scene, obtain the local density value of each pedestrian at different times, average the density and obtain the congestion level R congestion , specifically including:

[0015] (1) In the Voronoi diagram, each pedestrian particle is assigned a segmented unit area, and the point in the segmented unit represents the pedestrian i at t k At the moment of the moment, the unit area contains the range closest to the pedestrian particle. The greater the density of pedestrian particles, the smaller the allocated unit area. All the divided unit areas constitute the pedestrian walking area A of the entire scene:

[0016]

[0017] A——The entire pedestrian walking area composed of all divided cells;

[0018] (2) Calculate the number of pedestrians i in the measurement area at t k The local density ρ at the moment i (t k ):

[0019]

[0020] where |Ω i (t k )| indicates that pedestrian i is at tk The local area occupied at the moment

[0021] (3) By calculating the local density ρ in the scene i (t k ) averaged to obtain t k Average scene density at the moment:

[0022]

[0023] Step S32, calculating the density entropy of pedestrians, specifically includes:

[0024] Divide the density of all pedestrians in the scene into n1 intervals of equal width and calculate the density entropy of pedestrians:

[0025]

[0026] Among them, h ρ (i) represents the number of pedestrian particles whose density belongs to the i-th interval in the scene; p ρ (i) = h ρ (i) / N represents the probability of density distribution, where N is the total number of pedestrians in the area;

[0027] Step S33, calculating the pedestrian's velocity entropy, specifically includes:

[0028] Divide the speed in the scene into n2 intervals of equal width and calculate the speed entropy of the pedestrian:

[0029]

[0030] Among them, h v (j) represents the number of pedestrians whose speed is in the jth interval, p v (j) = h v (j) / N represents the distribution probability of speed, where N is the total number of pedestrians in the area;

[0031] Step S34, calculating the pedestrian's angle entropy, specifically includes:

[0032] Divide the pedestrian turning angle in the scene into n3 intervals of equal width and calculate the pedestrian angle entropy:

[0033]

[0034] Among them, h θ (k) represents the number of pedestrian turning angles in the kth interval, p θ (k) = h θ (k) / N represents the distribution probability of the steering angle, and N is the total number of pedestrians in the area.

[0035] Preferably, step S4 specifically includes:

[0036] In step S41, the formula for the normalized processing index is:

[0037]

[0038] E n ,E m ,E d -density entropy, velocity entropy, angle entropy, range is [0,1];

[0039] max{E i}、min{E i} — Maximum and minimum values ​​of risk indicators;

[0040] E p -Crowding risk value, the F level corresponds to a value of 1

[0041] E p The density ρ is used as the measure, the maximum value is the density corresponding to the F level, the minimum value is 0, and E p * For crowding risk

[0042] E i * ——E i The standardized value of

[0043] Step S42, defining the stampede risk level:

[0044] E stamp =αE disorder +βE congestion , α+β=1 (8)

[0045]

[0046]

[0047] Where α represents the scene clutter weight coefficient, β represents the scene crowding weight coefficient, α+β=1.0, where the value range of α is 0-1, and the value range of β is 0-1;

[0048] Step S43: The crowd risk is divided into four risk ranges:

[0049] A: safe, risk range: [0, 0.3];

[0050] B: relatively safe, risk range: (0.3, 0.5];

[0051] C: critical risk, risk range: (0.5,0.7];

[0052] D: Dangerous, risk range: (0.7, 1.0).

[0053] Preferably, step S5 specifically includes:

[0054] (1) The video scene area is divided and the crowding level and chaos level of each divided unit are calculated at every 0.2s time interval, thereby calculating the crowd risk value at each unit location, which can be used for real-time monitoring of local location risks and key prevention;

[0055] (2) According to the normalization of risk indicators and the weighted processing of risk levels in the present invention, the risk of crowd trampling in actual scenes is graded and quantified, which can not only evaluate the service level of facilities at the monitored locations, but also timely identify the high-incidence periods of crowd trampling risks.

[0056] The beneficial effects of the present invention are:

[0057] This paper proposes a new risk assessment index to analyze the risk of crowding and trampling in high-density crowd experimental scenarios; this paper proposes a method suitable for quantifying the risk of high-density crowds in rail transit stations, and promotes the application of the latest pedestrian risk assessment indicators in simulation models and evacuation scenarios.

[0058] This paper comprehensively considers risk causes and accident consequences, and uses multiple indicators to conduct risk analysis and evaluation from different perspectives. In the study of stampede risk, a density entropy indicator is proposed to measure the volatility of the spatial position of crowd density under high-density conditions, enriching the role of entropy value in crowd risk research. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A technical roadmap for a multi-index risk quantification method based on high-density passenger flow in rail transit stations, provided by an embodiment of the present invention;

[0060] Figure 2 A schematic flow chart of a multi-index risk quantification method based on high-density passenger flow in a rail transit station provided by an embodiment of the present invention;

[0061] Figure 3 A schematic diagram of Voronoi grid division in a multi-index risk quantification method based on high-density passenger flow in a rail transit station provided in an embodiment of the present invention;

[0062] Figure 4 A schematic diagram of the overall risk change of the people getting on and off the train in the multi-index risk quantification method based on high-density passenger flow in a rail transit station provided by an embodiment of the present invention;

[0063] Figure 5A schematic diagram of crowd risk level classification in a multi-index risk quantification method based on high-density passenger flow in a rail transit station provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. The embodiments of the present invention are not limited thereto.

[0065] Example 1

[0066] like Figure 1 As shown, the technical solution of this invention consists of three main steps: First, we briefly propose different risk indicators and risk factors within rail transit stations. Second, we propose a method for calculating the risk of crowd stampede at different densities, combining experimental results and the changes in these indicators. Finally, we verify the effectiveness of the crowd risk quantification method by combining it with real-world scenarios.

[0067] like Figure 2 As shown in Figure 2, the multi-index risk quantification method based on high-density passenger flow in rail transit stations includes:

[0068] Step S1: Obtain basic data of pedestrians, that is, record the basic data of the coordinates (x, y) of each pedestrian at each moment through video tracking technology;

[0069] Step S2: Calculate pedestrian-related indicators based on pedestrian basic data, that is, calculate the speed and steering angle of each pedestrian in the x and y directions at each moment, and use Voronoi to calculate the local density of pedestrians;

[0070] Step S3: The scene average density is obtained by averaging the local pedestrian density calculated by Voronoi, and the scene average density is used to define the scene congestion level; the density entropy, speed entropy, and angle entropy are calculated according to the entropy calculation formula, and the mixed entropy is calculated based on the density entropy, speed entropy, and angle entropy to measure the chaos level;

[0071] Step S4: Normalize the density (measure of crowding level) and the entropy (measure of chaos level) to obtain crowding risk value and chaos risk value. Then, assign weight coefficients (0-1) to crowding risk and chaos risk based on the characteristics of the scene area, and classify crowd risk into four risk levels within 0-1.

[0072] Step S5: Obtain the time period corresponding to the risk level, and strengthen monitoring of the time period with a high risk level.

[0073] Step S3 specifically includes the following steps:

[0074] like Figure 3As shown, in step S31, the Voronoi diagram is used to define the local density of pedestrians in the scene, and the local density value of each pedestrian at different times is obtained. The density is averaged to obtain the congestion level R congestion , specifically including:

[0075] (1) In the Voronoi diagram, each pedestrian particle is assigned a segmented unit area, and the point in the segmented unit represents the pedestrian i at t k At the moment of the moment, the unit area contains the range closest to the pedestrian particle. The greater the density of pedestrian particles, the smaller the allocated unit area. All the divided unit areas constitute the pedestrian walking area A of the entire scene:

[0076]

[0077] A——The entire pedestrian walking area composed of all divided cells;

[0078] (2) Calculate the number of pedestrians i in the measurement area at t k The local density ρ at the moment i (t k ):

[0079]

[0080] where |Ω i (t k )| indicates that pedestrian i is at t k The local area occupied at the moment

[0081] (3) By calculating the local density ρ in the scene i (t k ) averaged to obtain t k Average scene density at the moment:

[0082]

[0083] Step S32, calculating the density entropy of pedestrians, specifically includes:

[0084] Divide the density of all pedestrians in the scene into n1 intervals of equal width and calculate the density entropy of pedestrians:

[0085]

[0086] Among them, h ρ (i) represents the number of pedestrian particles whose density belongs to the i-th interval in the scene; p ρ (i) = h ρ (i) / N represents the probability of density distribution, where N is the total number of pedestrians in the area;

[0087] Step S33, calculating the pedestrian's velocity entropy, specifically includes:

[0088] Divide the speed in the scene into n2 intervals of equal width and calculate the speed entropy of the pedestrian:

[0089]

[0090] Among them, h v (j) represents the number of pedestrians whose speed is in the jth interval, p v (j) = h v (j) / N represents the distribution probability of speed, where N is the total number of pedestrians in the area;

[0091] Step S34, calculating the pedestrian's angle entropy, specifically includes:

[0092] Divide the pedestrian turning angle in the scene into n3 intervals of equal width and calculate the pedestrian angle entropy:

[0093]

[0094] Among them, h θ (k) represents the number of pedestrian turning angles in the kth interval, p θ (k) = h θ (k) / N represents the distribution probability of the steering angle, and N is the total number of pedestrians in the area.

[0095] Step S4 specifically includes:

[0096] In step S41, the formula for the normalized processing index is:

[0097]

[0098] E n ,E m ,E d -density entropy, velocity entropy, angle entropy, range is [0,1];

[0099] max{E i}、min{E i} — Maximum and minimum values ​​of risk indicators;

[0100] E p -Crowding risk value, the F level corresponds to 1; E p The density ρ is used as the measure, the maximum value is the density corresponding to the F level, and the minimum value is 0. For crowding risk

[0101] E i * ——E i The standardized value of

[0102] Step S42, defining the stampede risk level:

[0103] E stamp =αE disorder +βE congestion , α+β=1 (8)

[0104]

[0105]

[0106] Where α represents the scene clutter weight coefficient, β represents the scene crowding weight coefficient, α+β=1.0, where the value range of α is 0-1, and the value range of β is 0-1;

[0107] Weight coefficients were calibrated using an expert scoring method. Before completing the questionnaire, each participant was briefed on the concepts of crowding and chaos risk. Based on their subway riding experience, participants determined the weight of each risk in each scenario relative to stampede risk. Eighteen valid questionnaires were collected, and the average weight coefficients for each location were determined, as shown in Table 1.

[0108] Table 1 Risk weight coefficients for people in rail transit stations

[0109]

[0110] Step S43: The crowd risk is divided into four risk ranges:

[0111] A: safe, risk range: [0, 0.3];

[0112] B: relatively safe, risk range: (0.3, 0.5];

[0113] C: critical risk, risk range: (0.5,0.7];

[0114] D: Dangerous, risk range: (0.7, 1.0).

[0115] The risk levels are shown in Table 2.

[0116] Table 2 Risk level description

[0117] Assessment level A B C D Level Description Safety Relatively safe critical danger Danger

[0118] Table 3 Classification of risk levels of crowding and trampling

[0119] Assessment level A B C D Risk Range [0,0.3] (0.3,0.5] (0.5,0.7] (0.7，1.0)

[0120] Step S5 specifically includes:

[0121] (1) The video scene area is divided and the crowding level and chaos level of each divided unit are calculated at every 0.2s time interval, thereby calculating the crowd risk value at each unit location, which can be used for real-time monitoring of local location risks and key prevention;

[0122] (2) According to the normalization of risk indicators and the weighted processing of risk levels in the present invention, the risk of crowd trampling in actual scenes is graded and quantified, which can not only evaluate the service level of facilities at the monitored locations, but also timely identify the high-incidence periods of crowd trampling risks.

[0123] This paper uses actual footage captured during boarding and alighting periods at Beijing Xi'erqi subway station to verify the effectiveness of the crowd risk quantification method. The actual scene dimensions were measured to be approximately 7.7m x 2.4m. The boarding and alighting process is divided into three time periods: 0-29 seconds, during which passengers wait in line; 29-50 seconds, during which passengers board and alight; and 50-75 seconds, during which passengers wait again in line.

[0124] like Figure 4 As shown, the crowd congestion level, represented by the average density of the scene, shows a clear trend. Its value during boarding and alighting periods is higher than that during waiting periods, reaching approximately 1.4 people / square meter during the waiting period and a maximum of 2.249 people / square meter during boarding and alighting periods. The disorder level, on the other hand, varies more significantly than the congestion level, ranging from approximately 0.85 during the waiting period to a peak of 3.128 during boarding and alighting periods. The stampede risk, determined using a weighted coefficient of disorder and congestion levels during boarding and alighting scenarios, shows a clear trend, reaching a maximum of 2.672. Therefore, the newly defined stampede risk quantification method better reflects the actual risk of crowds in boarding and alighting scenarios than traditional density-based methods.

[0125] like Figure 5 As shown, according to the standard normalization processing of risks in the present invention, the risks during the boarding and alighting periods at Beijing Xi'erqi subway station are divided into different levels, and a dangerous period with a risk level of D for crowded stampede can be obtained, which can be used as the focus of monitoring and management by the operator.

[0126] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of an embodiment, and the processes in the accompanying drawings are not necessarily required to implement the present invention.

Claims

1. A multi-index risk quantification method based on high-density passenger flow in rail transit stations, characterized by: include: Step S1: Obtain basic data of pedestrians, that is, record the basic data of the coordinates (x, y) of each pedestrian at each moment through video tracking technology; Step S2: Calculate pedestrian-related indicators based on pedestrian basic data, that is, calculate the speed and steering angle of each pedestrian in the x and y directions at each moment, and use Voronoi to calculate the local density of pedestrians; Step S3: The scene average density is obtained by averaging the local pedestrian density calculated by Voronoi, and the scene average density is used to define the scene congestion level; the density entropy, speed entropy, and angle entropy are calculated according to the entropy calculation formula, and the mixed entropy is calculated based on the density entropy, speed entropy, and angle entropy to measure the chaos level; Step S4: Normalize the density (measure of crowding level) and the entropy (measure of chaos level) to obtain crowding risk value and chaos risk value. Then, assign weight coefficients to crowding risk and chaos risk based on the characteristics of the scene area, and convert crowd risk into four risk levels between 0 and 1. Step S5: Obtain the time period corresponding to the risk level, and strengthen monitoring of the time period with high risk level; in, Step S3 specifically includes the following steps: Step S31: Use the Voronoi diagram to define the local density of pedestrians in the scene, obtain the local density value of each pedestrian at different times, average the density and obtain the congestion level R congestion , specifically including: (1) In the Voronoi diagram, each pedestrian particle is assigned a segmented unit area, and the point in the segmented unit represents the pedestrian i at t k At the moment of the moment, the unit area contains the range closest to the pedestrian particle. The greater the density of pedestrian particles, the smaller the allocated unit area. All the divided unit areas constitute the pedestrian walking area A of the entire scene: A——The entire pedestrian walking area composed of all divided cells; (2) Calculate the number of pedestrians i in the measurement area at t k The local density ρ at the moment i (t k ): where |Ω i (t k )| indicates that pedestrian i is at t k The local area occupied at the moment (3) By calculating the local density ρ in the scene i (t k ) averaged to obtain t k Average scene density at the moment: Step S32, calculating the density entropy of pedestrians, specifically includes: Divide the density of all pedestrians in the scene into n1 intervals of equal width and calculate the density entropy of pedestrians: Among them, h ρ (i) represents the number of pedestrian particles whose density belongs to the i-th interval in the scene; p ρ (i) = h ρ (i) / N represents the probability of density distribution, where N is the total number of pedestrians in the area; Step S33, calculating the pedestrian's velocity entropy, specifically includes: Divide the speed in the scene into n2 intervals of equal width and calculate the speed entropy of the pedestrian: Among them, h v (j) represents the number of pedestrians whose speed is in the jth interval, p v (j) = h v (j) / N represents the distribution probability of speed, where N is the total number of pedestrians in the area; Step S34, calculating the pedestrian's angle entropy, specifically includes: Divide the pedestrian turning angle in the scene into n3 intervals of equal width and calculate the pedestrian angle entropy: Among them, h θ (k) represents the number of pedestrian turning angles in the kth interval, p θ (k) = h θ (k) / N represents the distribution probability of the steering angle, and N is the total number of pedestrians in the area.

2. The method for using the multi-index risk quantification method based on high-density passenger flow in rail transit stations according to claim 1 is characterized in that: Step S4 specifically includes: In step S41, the formula for the normalized processing index is: E n ,E m ,E d -density entropy, velocity entropy, angle entropy, range is [0,1]; max{E i }、min{E i } — Maximum and minimum values ​​of risk indicators; E p -Crowding risk value, the F level corresponds to a value of 1 E p The density ρ is used as the measure, the maximum value is the density corresponding to the F level, and the minimum value is 0. For crowding risk E i * ——E i The standardized value of Step S42, defining the stampede risk level: E stamp =αE disorder +βE congestion ,α+β=1 (8) Where α represents the scene clutter weight coefficient, β represents the scene crowding weight coefficient, α+β=1.0, where the value range of α is 0-1, and the value range of β is 0-1; Step S43: The crowd risk is divided into four risk ranges: A: safe, risk range: [0, 0.3]; B: relatively safe, risk range: (0.3, 0.5]; C: critical risk, risk range: (0.5,0.7]; D: Dangerous, risk range: (0.7, 1.0).

3. The method for using the multi-index risk quantification method based on high-density passenger flow in rail transit stations according to claim 1 is characterized in that: Step S5 specifically includes: (1) The video scene area is divided and the crowding level and chaos level of each divided unit are calculated at every 0.2s time interval, thereby calculating the crowd risk value at each unit location, which can be used for real-time monitoring of local location risks and key prevention; (2) By normalizing the risk indicators and weighting the risk levels, the risk of stampede in actual scenarios can be graded and quantified, which can not only evaluate the service level of the facilities at the monitored locations, but also timely identify the high incidence period of stampede risk.

Images