Evacuation path optimization method based on exit tree equalization algorithm

The evacuation path is optimized through the exit tree balance algorithm, which solves the problems of high computational complexity and poor real-time performance in the existing technology, and achieves rapid balance of the exit tree clearing time, improving evacuation efficiency and security.

CN119026772APending Publication Date: 2024-11-26UNIV OF SHANGHAI FOR SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411058639.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The existing evacuation path planning method has high computational complexity and poor real-time performance in complex buildings, and is insufficiently adaptable to changes in building structures, so it is unable to effectively balance the loads of different evacuation exits, resulting in limited evacuation efficiency and safety.

Method used

The evacuation path optimization method based on the exit tree balance algorithm is adopted. By building an evacuation area network model, constructing an exit tree, performing micro-evacuation simulation and clearing time adjustment, the concept of bridge edge and potential energy difference is introduced, and the exit tree is pruned to balance the clearing time and quickly optimize it. Evacuation path.

Benefits of technology

It significantly improves the running speed of the evacuation path optimization algorithm, saves modeling time, improves the applicability and real-time nature of the evacuation simulation model, reduces evacuation time, and improves evacuation efficiency and safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026772A_ABST
    Figure CN119026772A_ABST
Patent Text Reader

Abstract

The invention provides an evacuation path optimization method based on an exit tree equalization algorithm, and the method comprises the following steps: S1, representing all passable regions in an evacuation region by corresponding edges ei, j, representing the connection of different edges by a node vi, and taking a connected graph G formed by all nodes and edges as an evacuation region network model; s2, for each node vi, selecting an exit node tk closest to the node vi, and adding the node vi and all edges and nodes on a path from the node to the exit node tk into a corresponding exit tree Tk; s3, the highest crowd density rho < i, j > of each edge ei, j in each exit tree Tk and the clearing time ck of each exit tree are calculated; and S4, according to the emptying time, adjusting the affiliation of the nodes in the exit trees, enabling the exit trees with small loads to obtain the nodes, enabling the exit trees with large loads to lose the nodes, and obtaining an optimized evacuation path.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of emergency evacuation, and in particular to an evacuation path optimization method based on an exit tree balancing algorithm. Background Art

[0002] In recent years, with the acceleration of urbanization and the popularity of high-rise buildings, emergencies such as fires and earthquakes have posed severe challenges to the evacuation of people in buildings. Especially in high-density buildings, where people are densely populated and evacuation routes are complex, once an emergency occurs, rapid and effective evacuation becomes the key to ensuring the safety of people. Therefore, how to design a reasonable evacuation route, improve evacuation efficiency, and reduce casualties has become an important research direction in building design and safety management.

[0003] At present, most traditional evacuation path planning methods are based on the shortest path algorithm, which mainly considers the length of the evacuation path and strives to find the shortest path from the starting point to the safe exit. For example, Chinese patent application No. CN201410667474.1 describes an indoor evacuation simulation optimization method based on a potential energy driven cellular ant colony algorithm, which uses the potential energy evaluation standard of the artificial potential energy field to judge the personnel path and select a better evacuation path length. However, in the actual evacuation process, the length of the path is not the only consideration. The internal structure of the building is complex, and various obstacles, bottlenecks and dangerous areas may affect the evacuation speed and safety. At the same time, the unbalanced load of different evacuation exits will also cause some exits to be too crowded, thereby prolonging the evacuation time and even causing secondary disasters.

[0004] In order to deal with the above problems, a variety of evacuation path planning methods based on different optimization objectives have been proposed in recent years. For example, a dynamic evacuation model considering congestion, a path selection method based on risk assessment, and an evacuation simulation combined with personnel behavior characteristics, etc. Chinese Patent Application No. CN202310473316.1 describes a personnel evacuation path optimization management method and system based on dynamic flow distribution, which designs a preliminary evacuation route through a dynamic flow distribution method, and optimizes the evacuation path that best suits the personnel corresponding to the display device by a display device, so that the evacuated personnel can more clearly understand the evacuation path that best suits them; Chinese Patent Application No. CN202410265670.X describes an indoor evacuation simulation method and storage medium based on an extended cellular automaton, which takes into account the behavioral characteristics of pedestrians and some factors of the actual scene on site, while adding the impact of the external environment on pedestrian evacuation, and introducing aggressiveness as a power function that combines the external environment and pedestrian perception, so that the pedestrian evacuation path of the present invention is closer to reality. These methods have improved evacuation efficiency and safety to a certain extent, but there are still some shortcomings, such as high computational complexity, poor real-time performance, and insufficient adaptability to changes in building structures. Summary of the invention

[0005] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides an evacuation path optimization method based on an exit tree balancing algorithm to improve the running speed of the evacuation path optimization algorithm, save modeling time, and improve the applicability of the method to different evacuation simulation models. The method comprises the following steps:

[0006] Step S1: Construct the evacuation area network model: All the traversable areas in the evacuation area are mapped without omission or duplication using the corresponding edges e i,j Indicates that the connection between different edges is represented by node v i It indicates that the connected graph G consisting of all nodes and edges is used as the evacuation area network model;

[0007] Step S2: Construct the export tree: for each node v i , select the nearest exit node t k , the node v i And from this node to the exit node t k All edges and nodes on the path converge to form the exit tree T k ;

[0008] Step S3: Conduct micro-evacuation simulation under the initial scheme: Set the exit tree T k The target points of all edges corresponding to the area are set as exit t k , the initial evacuation plan is obtained, and the microscopic evacuation simulation model is used to calculate each exit tree T under this plan. k Each edge e in i,j The highest population density ρ i,,j , and the clearing time c of each exit tree k ;

[0009] Step S4: According to the clearing time, the node ownership in the exit tree is adjusted so that the exit tree with light load gains nodes and the exit tree with heavy load loses nodes, thereby obtaining an optimized evacuation path.

[0010] Furthermore, the step S4 comprises:

[0011] Step S41: Find a bridge edge, where the so-called bridge edge is a path existing in the original connected graph G but a vacant edge formed due to the partitioning of the exit tree;

[0012] Step S42: Calculate the potential energy difference p corresponding to the exit trees to which the two nodes on the bridge belong h,l :

[0013] p h,l =c h -c l ,

[0014] Among them, c hDenote the clearing time of the egress tree corresponding to the larger clearing time among the two egress trees to which the two nodes constituting the bridge edge respectively belong, c l Denote the clearing time of the egress tree corresponding to the smaller clearing time among the two egress trees to which the two nodes constituting the bridge edge respectively belong;

[0015] Step S43: Sort all the bridge edges in descending order of potential difference to form a set of bridge edges {E}, and set all the bridge edges in the set to the unvisited state;

[0016] Step S44: Prune the egress tree: Select the first unvisited bridge edge in the set {E}, delete the downstream edge of the high node, and connect the high node to the low node, where the high node is the node with the longer clearing time of the corresponding egress tree among the two nodes of the bridge edge, and the low node is the node with the shorter clearing time of the corresponding egress tree among the two nodes of the bridge edge;

[0017] Step S45: Recalculate the clearing time of each pruned egress tree T k Compare the maximum clearing time m1 after pruning with the maximum clearing time m0 before pruning. If m1 < m0, return to Step S41;

[0018] If m1 ≥ m0, mark the bridge edge used in this pruning as visited in the set {E}, restore the egress tree to the state before pruning. If there are still unvisited bridge edges in the set {E}, return to Step S44; otherwise, end the optimization process.

[0019] Furthermore, in Step S45, the recalculation of the clearing time of each pruned egress tree T k is based on the highest population density ρ i,,j on each edge, and the clearing time of each pruned egress tree T k is quickly calculated using the evacuation time estimation method, specifically as follows:

[0020] For the egress tree T h to which the high node belongs, the reduction in its clearing time is Δc h For the egress tree T l to which the low node belongs, the increase in its clearing time is Δc l :

[0021] Δc l = max{Δc(v i )|v i ∈P(v l ,t l )},

[0022] Δc h = max{Δc(v i )|vi ∈Q(v h ,t h )},

[0023] Δc(v i )=max{N / (f i-1,i *d i-1,i )},

[0024] P(v l ,t l ) represents the pruned export tree T l In, from the low node v l To the exit node t l The set of all nodes on the path, Q(v h ,t h ) represents the export tree T before pruning h In the middle, from the high node v h To the exit node t h The set of all nodes on the path; Δc(v i ) represents node v i The change value of the clearing time, N represents the change in the number of people on the corresponding exit tree after pruning, f i-1,i For node v i Upstream edge e i-1,i The average flow rate on i-1,i for e i-1,i The average width, average flow rate f i,j Calculated according to the following formula:

[0025]

[0026] ρ i,,j For edge e i,j The highest crowd density is obtained by step S3.

[0027] In the present invention, (1) nodes and edges are used to represent the topological structure of the area to be evacuated, and the definition of nodes and edges is clear and the setting is flexible, so modeling is convenient and fast; (2) the concepts of bridge edges and potential energy differences are introduced, and the exit trees are pruned by comparing the clearing times between different exit trees, so as to quickly balance the exit tree clearing times, accelerate the convergence speed, and achieve the optimal evacuation path; (3) in step S45, for the adjustment of the exit tree clearing time, the path where the low node and the high node are located is used to quickly estimate the evacuation time after the path adjustment, so that the amount of calculation is greatly reduced, the optimization speed is greatly accelerated, and the real-time requirements are met. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 is a flow chart of the path optimization method of the present invention;

[0029] Figure 2It is a network diagram of building structure one;

[0030] Figure 3 It is a schematic diagram of an initial evacuation plan of a building structure;

[0031] Figure 4 It is a schematic diagram of the evacuation plan after the building structure is trimmed;

[0032] Figure 5 It is a schematic diagram of the initial evacuation plan of the building structure II;

[0033] Figure 6 It is a schematic diagram of the evacuation plan after the building structure II is trimmed;

[0034] Figure 7 It is the schematic diagram of the evacuation plan after the final optimization of building structure 2;

[0035] Figure 8 It is a schematic diagram comparing the total evacuation time of the Pathfinder algorithm and the exit balancing algorithm of the present invention. DETAILED DESCRIPTION

[0036] See also Figure 1-4 In order to introduce the main process of the evacuation path optimization method based on the exit tree balancing algorithm of the present invention in detail, an evacuation scene (building structure 1) is set, which has 6 rooms, a corridor and two exits. In combination with this scene, the main steps of the evacuation path optimization method are described in detail below.

[0037] Step S1: Construct an evacuation area network model; all traversable areas are represented by corresponding edges e i,j Indicates that the connection between different edges is represented by node v i Indicates that these nodes v i and edge i,j It forms a connected graph G. Figure 2 As shown, a total of 26 nodes v1 to v are set in the evacuation area. 26 For larger rooms, multiple nodes can be set up in the room (for example, room 6 has v 12 -v 15 Four nodes), which facilitates the accuracy of statistical data.

[0038] Step S2: According to the regional network model, for each exit node t k , construct the export tree T k , export tree T k The construction method is: for each node v i , select the nearest exit node t k , the node v i And from this node to the exit node t kAll edges and nodes on the path are added to the corresponding exit tree T k (Only one repeated edge is retained, and only one repeated node is retained.) For example, Figure 2 In the graph, there are two exits t1 and t2, and two exit trees T1 and T2 can be constructed. v1-v4 are closer to exit node t1 than exit node t2, so these nodes belong to exit tree T1. For the paths in the connected graph G that do not belong to any exit tree, they are vacant edges and are no longer part of the evacuation path. For example Figure 3 In the original node v 21 and v 22 There is a path e between 21,22 , due to v 21 In the export tree T1, v 22 In the export tree T2, therefore e 21,22 The edge is empty.

[0039] Step S3: Export the tree T k The target points of all edges corresponding to the area are set as exit t k , construct the initial evacuation plan, calculate each exit tree T k Each edge e in i,j The highest population density ρ i,j , and the clearing time c of each exit tree k , where the clearing time c k The last person to leave the exit k This step can be calculated by any evacuation simulation model that can count the maximum density of a certain area, and belongs to the prior art.

[0040] Step S4: According to the clearing time, optimize the nodes in the exit tree to obtain the evacuation path. The specific method is as follows:

[0041] Step S41: Find bridge edges. The so-called bridge edges are paths that exist in the original connected graph G, but are vacant edges formed due to the partitioning of the exit tree. For example Figure 2 e in 21 ,twenty two.

[0042] Step S42: Calculate the potential energy difference ph,l corresponding to the exit trees to which the two nodes on the bridge belong:

[0043] p h,l =c h -c l ,

[0044] where c h Indicates the clearing time of the exit tree with the larger clearing time in the two exit trees to which the two nodes of the bridge edge belong respectively, c lIndicates the clearing time corresponding to the export tree with a smaller clearing time among the two export trees to which the two nodes constituting the bridge edge respectively belong. For example Figure 3 In, the clearing time c1 of the export tree T1 corresponding to the export t1 is greater than the clearing time c2 of the export tree T2 corresponding to the export t2, then c h is c1, c l is c2. In Figure 3 are respectively represented by T h and T l to represent.

[0045] Step S43: Sort all bridge edges in descending order of potential difference to form a bridge edge set {E}, and set all bridge edges in the set to the unvisited state.

[0046] Step S44: Prune the export tree: Select the first unvisited bridge edge in the set {E}, delete the downstream edge of the high node, and connect the high node to the low node, where the high node is the node with the longer clearing time of the corresponding export tree among the two nodes of the bridge edge, and the low node is the node with the shorter clearing time of the corresponding export tree among the two nodes of the bridge edge.

[0047] See Figure 3-4 , e 21,22 is the bridge edge, v 21 is the high node, v 22 is the low node, so delete the downstream edge e 21 corresponding to v 21,20 (In the export tree T1, the downstream node of v 21 is v 20 ), and at the same time connect v 21 , v 22 .

[0048] After pruning, the export tree with a short clearing time obtains the high node and its associated area, and the export tree with a long clearing time loses the high node and its associated area, so as to achieve the purpose of balancing the use of the export, that is, pruning the export tree T h where the high node is located, and connecting the high node v 21 to the low node v l of the export tree T 22 to complete one adjustment, Figure 3 The adjusted evacuation plan is as Figure 4 shown.

[0049] Step S45: Recalculate the clearing time of each pruned export tree T k , compare the maximum clearing time m1 after pruning with the maximum clearing time m0 before pruning. If m1 < m0, it indicates that the pruning in step S44 is effective, and return to step S41;

[0050] If m1≥m0, it indicates that this pruning is invalid, and the bridge edges used for this pruning are marked as visited in the set {E}, and the exit tree is returned to the state before pruning. If there are still unvisited bridge edges in the set {E}, return to step S44; otherwise, it indicates that the previous exit tree is already the optimal solution, and the optimization process ends.

[0051] In the above step S45, after the export tree is pruned, the export tree T needs to be recalculated. k The clearing time can be calculated by the method in step S3. However, in fact, the time of most paths has been calculated. If the calculation is performed according to the method in step S3, a large number of repeated calculations will occur, and the calculation time will be long. It is difficult to meet the real-time requirements of evacuation optimization. Therefore, the following method is used for calculation, focusing on the change of clearing time after exit tree pruning:

[0052] For the exit tree to which the high node belongs, the clearing time is reduced due to pruning, and the change is Δc h :

[0053] c h =c h -Δc h ,

[0054] For the exit tree to which the low node belongs, the clearing time increases due to the addition of nodes, and the change is Δc l :

[0055] c l =c l +Δc l ,

[0056] Δc l =max{Δc(v i )|v i ∈P(v l ,t l )}

[0057] Δc h =max{Δc(v i )|v i ∈Q(v h ,t h )}

[0058] Among them, P(v l ,t l ) represents the pruned export tree T l In, from the low node v l To the exit node t l The set of all nodes on the path of Figure 4 Low and medium node v l v 21 , then P(vl ,t l ) is {v 21 ,v 22 ,v 23 ,v 24 ,v 25 ,v 26 ,t2};Q(v h ,t h ) represents the export tree T before pruning h In the middle, from the high node v h To the exit node t h The set of all nodes on the path of Figure 4 Medium and high node v h v 20 , then Q(v h ,t h ) is {v 20 ,v 19 ,v 18 ,v 17 ,v 16 ,t1};v i For any point on the path, Δc(v i ) represents node v i Clearing time changes,

[0059] Δc(v i )=max{N / (f i-1,i *d i-1,i )},

[0060] Where N represents the change in the number of people on the exit tree after pruning, f i-1,i For node v i Upstream edge e i-1,i Average flow rate on the surface (unit: person / m); d i-1,i for e i-1,i The average width of the node, the present invention believes that before and after the adjustment, assuming that the flow rate on the edge remains unchanged, the change in node clearing time is estimated by dividing the increased number of people by the average flow on the edge:

[0061]

[0062] Among them, ρ i,,j is the edge e obtained in step S3 i,j The highest density on the 2 ), from the above formula, when the maximum density is lower than 1.5 people / m 2 When the maximum density is higher than 1.5 people / m 2 , the flow rate of the edge decreases as the density increases.

[0063] The evacuation method of the present invention is introduced below with respect to a teaching building (building structure 2). Figure 5 It is to construct an evacuation area network model, abstracting the teaching building into a network composed of points and edges. All accessible rooms and corridors are represented by edges. i,j , and all possible bottlenecks, such as doors, narrow passages, terrain transitions, etc., are represented by separate edges; node v i It represents the connection of edges and does not occupy any actual physical space. The exit nodes are represented by t1, t2, and t3 respectively. According to the connectivity of different edges in the evacuation area, the edges and nodes are connected to form a connected graph G, in preparation for obtaining the initial evacuation plan.

[0064] In the connected graph G, take each exit node t1, t2, t3 as the root node, find the shortest path tree T1, T2, T3 connected to the root node (that is, any node v on the tree i to i The distances are all from v on the graph G. i to i The shortest distance); the edges in the graph G that are not included in any shortest path tree are vacant edges, which are not used during evacuation. The edges belonging to any shortest path tree are used edges. The initial evacuation plan H0 is formed by combining the usage of the edges and the evacuation direction on the edges; the set of points and edges connected to the exit node is recorded as the exit tree T k , the exit tree in the initial solution is the shortest path tree, such as Figure 5 The teaching building shown has three shortest path trees T1, T2, and T3.

[0065] The crowd evacuation simulation platform Pathfinder was used to model the scene. The simulation was carried out according to the initial evacuation plan. In order to correspond to the optimization algorithm, the corridor of the Pathfinder model was divided into small rooms. Doors were added to the rooms to maintain circulation, and thin walls were added to block the circulation of people to achieve the shortest path tree. The statistics of each edge e in the simulation process were i,j The highest population density ρ i,j , and each export tree T k Clearing time c k (i.e. the last evacuee leaves the exit t k time).

[0066] Under the initial evacuation plan H0, when an edge belongs to the graph G but does not belong to any exit tree T k , and the two vertices of the edge belong to different shortest path trees, the edge is recorded as a bridge edge, such as Figure 5 Side e 8,9 and e 18,19 That is the bridge side;

[0067] Through evacuation simulation, the exit clearing times corresponding to the shortest path trees T1, T2, and T3 in this embodiment are c1 = 79.2s, c2 = 194.6s, and c3 = 89.9s, respectively. 8,9 and e 18,19 The potential energy difference is p 8,9 =10.7, p 18,19 =104.7;

[0068] Sort the potential energy differences, e 18,19 The first to be considered, so remove the high node v 19 The downstream edge e 19,20 , and the high node v 19 Connect to low node v 18 After completing one pruning, we get a new evacuation plan H1, such as Figure 6 shown.

[0069] From scheme H0 to scheme H1, that is, from Figure 5 to Figure 6 , the clearing time of the exit tree T3 where the low node is located increases by Δc3=21.9s, and the clearing time of the exit tree T2 where the high node is located decreases by Δc2=55.56s;

[0070] At this time, in the evacuation plan H1, the clearing time c3 of the exit tree T3 is 111.8 s, the clearing time c2 of the exit tree T2 is 139.04 s, and the clearing time c1 of the exit tree T1 is 79.2 s.

[0071] The maximum evacuation time of the exit tree is c2 = 139.04s, which is lower than c2 = 194.6s of scheme H0, indicating that this pruning is effective. The next round of iteration is continued. The maximum clearing time of each exit tree in the second, third and fourth iterations of this embodiment is shown in Table 1. It can be seen that the maximum clearing time of the exit in the fourth iteration is T max If it is greater than the result of the third iteration, it means that the adjustment of the bridge edge with the maximum potential energy by this pruning is invalid, the algorithm terminates the iteration, and the fourth iteration plan is discarded. The third iteration plan is retained as the final evacuation plan. Figure 7 As shown:

[0072] Table 1 Export tree clearing time (unit: seconds)

[0073] <![CDATA[T1]]> <![CDATA[T2]]> <![CDATA[T3]]> <![CDATA[T max ]]> First Iteration 79.2 139.04 111.8 139.04 Second Iteration 92.28 139.04 103.12 139.04 Third iteration 92.28 83.49 125.02 125.02 Fourth Iteration 92.28 139.04 103.12 139.04

[0074] Figure 8 The relationship between the maximum clearing time of the exit and the number of iterations of the present invention is shown in Figure 1. The maximum clearing time (i.e., total evacuation time) of each round of iteration of the exit balancing algorithm is compared with that of the Pathfinder simulation. Figure 8It can be seen that through the optimization of the exit balance algorithm, the calculation results of each iteration of the Pathfinder simulation model (the last iteration scheme is discarded) show a downward trend, and the optimized Pathfinder evacuation time is reduced by 40.5s, indicating that the present invention can greatly reduce the evacuation time and effectively improve the evacuation efficiency.

[0075] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An evacuation path optimization method based on exit tree balancing algorithm, It is characterized in that The following steps are involved: Step S1: Construct the evacuation area network model: All the traversable areas in the evacuation area are mapped without omission or duplication using the corresponding edges e i,j Indicates that the connection between different edges is represented by node v i It indicates that the connected graph G consisting of all nodes and edges is used as the evacuation area network model; Step S2: Construct the export tree: for each node v i , select the nearest exit node t k , the node v i And from this node to the exit node t k All edges and nodes on the path converge to form the exit tree T k ; Step S3: Conduct micro-evacuation simulation under the initial scheme: Set the exit tree T k The target points of all edges corresponding to the area are set as exit t k , the initial evacuation plan is obtained, and the microscopic evacuation simulation model is used to calculate each exit tree T under this plan. k Each edge e in i,j The highest population density ρ i,,j , and the clearing time c of each exit tree k ; Step S4: According to the clearing time, the node ownership in the exit tree is adjusted so that the exit tree with light load gains nodes and the exit tree with heavy load loses nodes, thereby obtaining an optimized evacuation path.

2. The method according to claim 1, It is characterized in that The step S4 comprises: Step S41: Find a bridge edge, where the so-called bridge edge is a path existing in the original connected graph G but a vacant edge formed due to the partitioning of the exit tree; Step S42: Calculate the potential energy difference p corresponding to the exit trees to which the two nodes on the bridge belong h,l : p h,l =c h -c l , Among them, c h Indicates the clearing time of the exit tree with the larger clearing time in the two exit trees to which the two nodes of the bridge edge belong respectively, c l Indicates the clearing time corresponding to the exit tree with the smaller clearing time among the two exit trees to which the two nodes of the bridge edge belong respectively; Step S43: sorting all bridge edges from large to small according to the potential energy difference to form a bridge edge set {E}, and setting all bridge edges in the set to an unvisited state; Step S44: pruning the exit tree: select the first unvisited bridge edge in the set {E}, delete the downstream edge with the high node, and connect the high node to the low node, where the high node is the node with a longer exit tree clearing time among the two nodes of the bridge edge, and the low node is the node with a shorter exit tree clearing time among the two nodes of the bridge edge; Step S45: Calculate the emptying time of each pruned exit tree T again k , compare the maximum emptying time m1 after pruning with the maximum emptying time m0 before pruning. If m1 < m0, return to Step S41; If m1≥m0, the bridge edges used for this pruning are marked as visited in the set {E}, and the exit tree is returned to the state before pruning. If there are still unvisited bridge edges in the set {E}, return to step S44; otherwise, the optimization process ends.

3. The method according to claim 2, It is characterized in that In step S45, the process recalculates each pruned export tree T k The clearing time is based on the highest crowd density ρ on each edge i,,j , using the evacuation time estimation method to quickly calculate each pruned exit tree T k The clearing time is as follows: For the exit tree T to which the high node belongs h , the reduction in clearing time is Δc h , for the exit tree T to which the low node belongs l , the increase in the clearing time is Δc l : Δc l =max{Δc(v i )|v i ∈P(v l ,t l )}, Δc h =max{Δc(v i )|v i ∈Q(v h ,t h )}, Δc(v i )=max{N / (f i-1,i *d i-1,i )}, P(v l ,t l ) represents the pruned export tree T l In, from the low node v l To the exit node t l The set of all nodes on the path, Q(v h ,t h ) represents the export tree T before pruning h In the middle, from the high node v h To the exit node t h The set of all nodes on the path; Δc(v i ) represents node v i The change value of the clearing time, N represents the change in the number of people on the corresponding exit tree after pruning, f i-1,i For node v i Upstream edge e i-1,i The average flow rate on i-1,i for e i-1,i The average width, average flow rate f i,j Calculated according to the following formula: ρ i,,j For edge e i,j The highest crowd density is obtained by step S3.