An emergency rescue path planning method and system based on adaptive risk aversion

Abstract

The application provides an emergency rescue path planning method and system with adaptive risk avoidance, comprising: establishing an emergency rescue road network structure; before an actual natural disaster occurs, performing a risk entropy driven adaptive Monte Carlo simulation analysis, performing path information statistical analysis on the Monte Carlo simulation result, and establishing a pre-planned path library; when the actual natural disaster occurs, establishing a prior risk distribution of each planned path in the pre-planned path library; using a sequential Bayesian inference to dynamically correct the prior risk distribution of each planned path in the emergency rescue process; based on the path posterior risk mean of each planned path after the sequential Bayesian inference, performing dynamic planning of the emergency rescue path, selecting an optimal rescue path, and performing an emergency rescue task until the emergency rescue activity ends. Through the path dynamic adjustment strategy, the dispersed transportation strategy, the path pre-planning strategy, and the network communication free strategy, the application ensures that the rescue resources efficiently and safely reach the target location under the natural disaster scene of flood, typhoon, and the like.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of path planning, and particularly relates to an emergency rescue path planning method and system with adaptive risk avoidance. BACKGROUND

[0002] Natural disasters such as floods and typhoons often have a serious impact on human society, not only threatening people's lives and safety, but also leading to chain reactions such as damaged infrastructure, traffic disruption, and resource scarcity. After a disaster occurs, the timeliness and effectiveness of emergency rescue directly relate to the efficiency of life rescue and the speed of post-disaster recovery in the disaster area. However, due to the complex and changeable environment during the disaster, the reasonable allocation and scheduling of rescue resources face many challenges. Under this background, emergency rescue path planning becomes a key technology that determines the efficiency of rescue, and its goal is to reasonably schedule rescue resources and plan a fast and safe route for material transportation and rescue personnel in a disaster environment, so as to ensure the smooth development of rescue operations. Therefore, it is of great practical value to design an emergency rescue path planning method that can adaptively adjust and dynamically adjust the path in a complex disaster environment.

[0003] Currently, the path planning methods for emergency rescue mainly use the shortest path algorithm (such as Dijkstra, A*, etc.), optimization methods based on graph theory, and heuristic search algorithms, but these methods have certain limitations in disaster environments. First, existing methods are often based on fixed road network structures (networks composed of nodes and road segments), and do not fully consider the dynamic changes in road traffic conditions under the influence of disasters, making it difficult to adapt to the reality of road damage or sudden changes in risk. Second, traditional path optimization often aims to minimize time or distance, and cannot effectively disperse transportation risks, which increases the uncertainty and failure rate of overall transportation and poses a risk of concentration of rescue resources on high-risk road segments. In addition, most methods are designed for static scenarios before and after disasters, and are difficult to quickly apply during the disaster emergency response process. Therefore, existing path planning methods cannot meet the needs of emergency rescue in disaster occurrence and development environments, and there is an urgent need for a rescue path dynamic planning method that can take into account dynamic risk perception and path safety adjustment. SUMMARY

[0004] In view of the defects of the prior art, the application provides an emergency rescue path planning method and system with adaptive risk avoidance, which can effectively solve the above problems.

[0005] The technical scheme adopted by the application is as follows:

[0006] The application provides an emergency rescue path planning method with adaptive risk avoidance, comprising the following steps:

[0007] Step S1: establishing an emergency rescue road network structure of a region to be planned according to actual geographic information; the emergency rescue road network structure comprises a node set N and a road section set E; each node n in the node set N represents a geographic position, and each road section e in the road section set E represents a traffic connection between two nodes, n∈N, e∈E;

[0008] Step S2: evaluating a long-term risk probability R n of each node n and a long-term risk probability R e of each road section e, and extending the long-term risk probability R n of each node n and the long-term risk probability R e of each road section e into an asymmetric Logit probability curve distribution as a prior risk distribution, and obtaining a risk entropy H n of each node n and a risk entropy H e of each road section e based on the prior risk distribution;

[0009] Step S3: before an actual natural disaster occurs, performing Monte Carlo simulation analysis based on the emergency rescue road network structure to obtain a Monte Carlo simulation result; wherein, in the Monte Carlo simulation analysis, the risk entropy H n of each node n and the risk entropy H e of each road section e are used to perform risk sampling in a risk entropy driven adaptive sampling manner;

[0010] Step S4: performing path information statistical analysis based on the Monte Carlo simulation result to establish a pre-planned path library; the pre-planned path library stores a planned path finally implemented, a path risk probability average value and a path risk probability variance;

[0011] Step S5: when the actual natural disaster occurs, establishing a prior risk distribution of each planned path in the pre-planned path library;

[0012] Step S6: using sequential Bayesian inference to dynamically correct the prior risk distribution of each planned path in an emergency rescue process to obtain a path posterior risk average value of each planned path;

[0013] Step S7: performing emergency rescue path dynamic planning based on the path posterior risk average value of each planned path after the sequential Bayesian inference, selecting an optimal rescue path and executing an emergency rescue task until the emergency rescue activity ends.

[0014] Preferably, each road section e has corresponding transportation speed and unit load capacity attributes according to a road section type; the emergency rescue road network structure composed of all nodes and road sections is an undirected graph G=(N, E), and each road section e is bidirectional in the undirected graph.

[0015] Preferably, step S2 is specifically:

[0016] Step S2.1: Evaluate the long-term risk probability R of each node n using formula (1) n :

[0017]

[0018] wherein λ H is a disaster sensitivity coefficient; ξ n is a disaster excitation intensity index of node n, specifically a shape parameter of a generalized extreme value distribution fitted from a sequence of annual maximum daily rainfall; V n is a structure vulnerability index of node n, specifically a product of a building service life of node n and a fortification grade coefficient; E n is an exposure index of node n, specifically a population and economic combination parameter of node n; ξ max is a maximum value of disaster excitation intensity indexes ξ1, ξ2,..., ξ n ,..., ξ N of all nodes; V max is a maximum value of structure vulnerability indexes V1, V2,..., V n ,..., V N of all nodes; E max is a maximum value of exposure indexes E1, E2,..., E n ,..., E N of all nodes; α, β, γ are exponential weight parameters;

[0019] wherein the calculation process of disaster excitation intensity index ξ n is as follows: let a sequence of multi-year maximum daily rainfall of node n be Y1, Y2,..., Y Y≥20, X y is a maximum daily rainfall in the yth year, Y is a number of years; perform generalized extreme value distribution fitting according to the sequence of maximum daily rainfall, and determine disaster excitation intensity index ξ n according to a fitting function, the fitting function F(x n ) being:

[0020]

[0021] μ n is a mean value determined by the fitting function, σ n is a standard deviation determined by the fitting function; x n is an independent variable of the fitting function;

[0022] the calculation process of structure vulnerability index V n is as follows: let a current year be T now , a building construction year of node n be T built,n , and an average value be taken when there are multiple buildings, a building service life be T now -Tbuilt,n The building's fortification level coefficient is C. grade,n Then V n The product of the building's service life and its fortification level coefficient is calculated using the following formula:

[0023] V n =(T now -T built,n )·C grade,n (3)

[0024] Exposure index E n The calculation process is as follows: Let the population density of node n be P. n The economic density is Q n , λ P With λ Q Let E be the population weight and the economic weight, respectively. n The calculation formula is:

[0025] E n =λ P ·P n +λ Q ·Q n (4)

[0026] Step S2.2: Using formula (5), evaluate the long-term risk probability R of each road segment e. e :

[0027]

[0028] S e =(L now -L built,e )·L e (5)

[0029] Where: R e1 and R e2 S represents the long-term risk probability of the nodes on both sides of road segment e; e The structural vulnerability index for road segment e is based on the current year L. now Year of construction of the road section L built,e With the fortification level coefficient L of the road section e It is confirmed that high-security highways, waterways, and all air transport routes L e Take 0.3, for the road and waterway L passing through the slope. e Take 0.6, for old roads and waterways susceptible to flooding (L) e Take 0.9, S max S is the structural vulnerability index for all road sections. e The maximum value of ω; ω∈[0,1] is the road segment risk coefficient;

[0030] Step S2.3: Determine the prior risk distribution for each node n and each road segment e respectively:

[0031] The long-term risk probability R of node n n The long-term risk probability R of road segment e e This is transformed into a probability density distribution defined in the interval [0,1], forming a continuous expression of risk probability; an asymmetric Logit probability curve distribution is constructed as the prior risk distribution of nodes and road segments, and the probability density function p of node n is... n The probability density function p of (x) and road segment e e (x) is:

[0032]

[0033] Where: k n and k e These are the risk coefficients for node n and road segment e, respectively, used to control the shape of the distribution curve; the larger the value, the more concentrated the distribution; k n =k max (1-R n );k e =k max (1-R e );k max This represents the maximum risk coefficient.

[0034] δ n and δ e These are the offset coefficients for node n and road segment e, respectively, used to control the skewness characteristics of the risk distribution; The variance is the estimated variance;

[0035] C n and C e are the normalization coefficients for node n and road segment e, respectively, used to ensure that the integral distributed in the interval [0,1] is 1; x∈[0,1] is the independent variable of the probability density function;

[0036] Step S2.4: Use formula (7) to determine the risk entropy H of each node n. n and the risk entropy H of each road segment e e :

[0037]

[0038] This step completes the process.

[0039] Preferably, step S3 specifically includes:

[0040] Step S3.1: Set the simulation parameters, including the number of simulations M and the simulation time T. sim ;

[0041] Step S3.2: setting disaster scenario:

[0042] At a certain time, the risk probability of the first node and all its connected road segments is raised along the self-set node sequence, the disaster process is simulated, and after reaching the set time, the risk probability of the first node is raised to decline, the disaster transfer process is simulated; the risk probability of the second node and all its connected road segments is raised, and until all nodes complete the risk probability rising and falling in sequence, which represents that the simulation of a disaster evolution process is completed;

[0043] Step S3.3: disaster and emergency rescue simulation:

[0044] Based on the set simulation parameters and the set disaster scenario, and the risk sampling results, the emergency rescue path dynamic planning is performed and the emergency rescue simulation is executed until the set simulation time T is reached sim ;

[0045] Step S3.3 is specifically:

[0046] Step S3.3.1: in each simulation of disaster evolution process, according to the risk entropy H n of each node n and the risk entropy H e of each road segment e, an adaptive sampling method driven by risk entropy is used to sample the risk of all nodes and road segments once, to determine the risk probability of all nodes and road segments in the emergency rescue road network structure; wherein the risk probability of node n is represented as R n,sim , and the risk probability of road segment e is represented as R e,sim ;

[0047] The specific way is:

[0048] The adaptive sampling probability weights ω n,mc (x) and ω e,mc (x) of node n and road segment e are determined respectively by formula (8):

[0049]

[0050] Wherein: β n and β e , are the bias amplitude coefficients of node n and road segment e respectively, the larger the value is, the higher the sampling probability of high risk value is, β = 0 ~ 3; γ n and γ e , are the bias intensity coefficients of node n and road segment e respectively, which are used to control the sensitivity of sampling to high risk area, the larger the value is, the higher the sampling probability of high risk value is, γ ≥ 1;

[0051] The sampling density functions f and

[0052]

[0053] Therefore, based on the sampling density function of each node n and road segment e and Conduct a risk sampling of all nodes and road segments;

[0054] Step S3.3.2: Based on the risk probability of all nodes and road segments in the emergency rescue road network structure obtained by risk sampling, and combined with the unit load capacity and transportation speed of each road segment, the key nodes and high-risk nodes that require emergency rescue are selected as nodes to be rescued;

[0055] The selection method for key nodes requiring emergency rescue is as follows:

[0056] For each node n in the emergency rescue road network structure, the importance I(n) of the node is calculated using the following formula:

[0057]

[0058] Among them: E n Let C be the set of road segments directly connected to node n. e Speed ​​represents the unit load capacity of road segment e. e R is the transport speed of road segment e. n,sim The risk probability of node n;

[0059] Sort all nodes in descending order of importance, select the top 3 nodes as critical nodes, and if the risk probability of a critical node is higher than the first threshold T1, then the critical node is a critical node that requires emergency rescue.

[0060] The screening method for high-risk nodes requiring emergency rescue is as follows:

[0061] Among all nodes, select the node risk probability R. n,sim Nodes exceeding the second threshold T2 are considered high-risk nodes requiring emergency rescue; where the second threshold T2 is greater than the first threshold T1.

[0062] Step S3.3.3: Adopt a distributed transportation strategy. For each node to be rescued, select three nodes with the lowest risk probability from the road network structure to schedule resources. These nodes are called output nodes.

[0063] Step S3.3.4: Perform path planning between each output node and the node to be rescued. The path planning comprehensively considers transportation time and path risk factors. All possible paths between each output node and the node to be rescued are comprehensively scored, and the path with the highest comprehensive score is selected as the rescue path between the output node and the node to be rescued.

[0064] Specifically, a set of feasible paths PS' between the output node and the node to be rescued is selected. For each path P in the set of feasible paths PS'... l ∈PS', consisting of a series of consecutive road segments e1, e2, ..., e k ,...,e K The composition is determined by the following formula, which is used to calculate its comprehensive score S(P). l ):

[0065]

[0066] Where: R(P) l ) represents path P l The risk probability is the sum of the risk probabilities of all segments in the path. For road segment e k Risk probability; T(P) l ) represents path P l Total transport time is the sum of the ratios of the length to the speed of all segments in the route. For road segment e k Length, For road segment e k The corresponding transport speed; B(P) l ) represents path P l The bifurcation degree is defined as the degree of bifurcation of path P. l The sum of the number of road segments connected to the intermediate nodes; ω1, ω2, and ω3 are weighting coefficients, with ω1 and ω2 dynamically adjusted according to the risk probability changes of each node awaiting rescue, which are the differences in risk probability for each node awaiting rescue every 5 minutes; ω1 and ω2 are determined by the following formula:

[0067]

[0068] Where: ω max With ω min The two coefficients are set to satisfy ω max >ω min ;ΔR target,s Let ΔR be the difference in risk probability for the s-th node awaiting rescue every 5 minutes. When the risk probability increases, ΔR... target,s >0, when the risk probability decreases, ΔR target,s <0; When the risk probability increases, consider rapid transportation; when the risk probability decreases, consider transportation safety.

[0069] In the feasible path set PS', based on each path P l The overall score S(P) of ∈PS' l The route with the highest overall score is selected as the rescue route;

[0070] Step S3.3.5: Transport resources and carry out rescue operations according to the node and segment sequence of the rescue route. During the rescue process, whenever a node is reached, determine whether the current segment risk probability of the next segment exceeds the acceptable safety threshold T. road If so, it is considered a road segment interruption, triggering dynamic path adjustment. The current node is regarded as the output node, and the process returns to step S3.3.4 to replan the subsequent path until the node to be rescued is reached or no passable path can be found. If not, resource transportation continues according to the node and road segment order of the rescue path.

[0071] Step S3.3.6: Check the risk probability of all nodes and road segments in the emergency rescue road network structure every 5 minutes. Filter out nodes to be rescued according to step S3.3.2, check whether there are already rescue tasks being performed, and evaluate the risk probability of nodes to be rescued after the completion of existing rescue tasks. If it is less than the first threshold T1, no new rescue tasks will be added. Otherwise, proceed to steps S3.3.3 to S3.3.5. The risk probability check will continue until the simulation time limit is reached.

[0072] Preferably, step S4 specifically includes:

[0073] After completing all Monte Carlo simulations, path statistics are performed based on all simulation results, including the final planned path after reaching the node to be rescued, the average path risk probability, and the variance of the path risk probability; a pre-planned path library is established based on the final planned path.

[0074] Let the set of planned paths in the pre-planned path library be PS. For each path P i Based on the statistical results of Monte Carlo simulation analysis, the average path risk probability of ∈PS is calculated using the following formula. and path risk probability variance σ 2 (P i ):

[0075]

[0076] Where: M i Path P for Monte Carlo analysis i The number of executions, R (j) (P i ) represents path P in the j-th simulation. i The probability of risk when M i When the value is too small, let σ be.2 (P i =0.08.

[0077] Preferably, in step S5, establishing the prior risk distribution of each planned path in the pre-planned path library specifically involves:

[0078] For each path P i Based on the average path risk probability and path risk probability variance σ 2 (P i The path P is constructed using an asymmetric Logit probability curve distribution. i The prior risk distribution, path P i probability density function p i (x) is:

[0079]

[0080] Where: k i For path P i Risk coefficient; δ i For path P i The offset coefficient.

[0081] Preferably, in step S6, sequential Bayesian inference is used to dynamically correct the prior risk distribution of each planned path during the emergency rescue process, thereby obtaining the mean posterior risk of each planned path, specifically as follows:

[0082] Step S6.1: Obtain the risk probability change value for each planned path:

[0083] During emergency rescue operations, the risk probability change values ​​of all nodes and road segments in the emergency rescue road network structure are checked every minute. This updates the risk probability change value of each planned path, which is then used as the observation variable O for Bayesian updates. t (P i ); when O t (P i ) > 0 indicates path P i The probability of risk is increasing, when O t (P i ) < 0 indicates path P i The probability of risk is decreasing, when O t (P i ) = 0 indicates path P i The probability of risk remains unchanged, O t (P i It is calculated using the following formula:

[0084] O t (P i ) = R t (Pi )-R t-1 (P i (16)

[0085] Among them: O t (P i ) represents path P at minute t. i The change in risk probability, R t (P i ) represents path P at minute t. i The probability of risk, R t-1 (P i ) represents path P at minute t-1. i The risk probability is the sum of the risk probabilities of all segments in the path;

[0086] Step S6.2: Update the posterior risk distribution for each planned path:

[0087] Based on the statistical distribution model of path risk, the posterior parameters of path risk are adjusted to update the prior risk distribution of the path to the posterior risk distribution; specifically, formula (17) is used to adjust the path P i The risk coefficient k, which is used as a priori parameter i Updated to posterior parameter k i ′, and, path P i The offset coefficient δ, which is used as a prior parameter i Updated to posterior parameter δ i ′:

[0088] k i ′=k i +O t (P i ),δ i ′=δ i +O t (P i (17)

[0089] Step S6.3: Calculate path P using formula (18). i mean posterior risk of the path

[0090]

[0091] Path P i mean posterior risk of the path As path P in emergency rescue process i The updated mean posterior risk of the path.

[0092] Preferably, in step S7, dynamic planning of emergency rescue paths is performed based on the mean posterior risk of each planned path after sequential Bayesian inference. The optimal rescue path is selected and the emergency rescue task is executed until the emergency rescue activity ends. Specifically:

[0093] Step S7.1: Based on measured data, determine the risk probability of all nodes and road segments in the emergency rescue road network structure;

[0094] Step S7.2: Based on the traffic data and node risk probability of all road segments, select the nodes to be rescued;

[0095] Step S7.3: For each selected node to be rescued, check if a rescue mission is already in progress. If not, proceed to step S7.4; if so, proceed to step S7.9.

[0096] Step S7.4: For each node to be rescued, select the three nodes with the lowest risk probability from the emergency rescue road network structure as output nodes;

[0097] Step S7.5: Between each output node and the node to be rescued, path planning is carried out based on the mean posterior risk of each planned path determined in step S6, and an optimal rescue path is selected as the rescue path.

[0098] Step S7.6: Carry out the rescue according to the order of nodes and road segments of the rescue route;

[0099] Step S7.7: During the rescue operation, whenever a node is reached, determine whether the current risk probability of the next road segment exceeds the acceptable safety threshold T. road If so, it is considered a road segment interruption, triggering dynamic path adjustment, treating the current node as an output node, and returning to step S7.5 to re-plan the subsequent path; if not, proceed to step S7.8.

[0100] Step S7.8: Determine whether the node to be rescued has been reached. If not, return to step S7.6; if yes, proceed to step S7.9.

[0101] Step S7.9: End the rescue mission.

[0102] Preferred options also include:

[0103] Based on the current traffic data, check the risk probability of all nodes and road segments in the emergency rescue road network structure every 5 minutes, filter out nodes to be rescued, check whether there are already rescue tasks being performed, and assess the risk probability of nodes to be rescued after the completion of existing rescue tasks. If it is less than the first threshold T1, no new rescue tasks will be added; otherwise, return to step S7.4, add output nodes, add new rescue tasks, and repeat this process continuously.

[0104] Step S7.5: Between each output node and the node to be rescued, path planning is performed based on the mean posterior risk of each planned path determined in Step S6, and an optimal rescue path is selected as the rescue path, specifically as follows:

[0105] First, filter the set of feasible paths PS between the output node and the node to be rescued. ” For each path P r ∈PS”, based on its mean posterior risk. Choose the path P with the lowest mean posterior risk. * As the optimal rescue route, route P * It is calculated using the following formula:

[0106]

[0107] This allows us to determine the optimal rescue route.

[0108] This invention also provides a system for an adaptive risk-avoidance emergency rescue route planning method, comprising: a Beidou short message communication module, a GIS visualization and risk assessment module, a route planning calculation module, and a rescue dispatch terminal;

[0109] The BeiDou short message communication module is used for data transmission in disaster environments to ensure information exchange during rescue missions. Installed on rescue vehicles, drones, and fixed monitoring stations, the module is responsible for sending and receiving road status information, rescue mission allocation instructions, and real-time route adjustment plans. Uplink data includes the location information of rescue teams, the execution status of rescue missions, road conditions, and risk update information; downlink data includes optimal route instructions, mission adjustment commands, and detour suggestions.

[0110] The GIS visualization and risk assessment module is used to model the location and road topology of the disaster area, convert it into an emergency rescue road network structure composed of nodes and road segments, and dynamically assess the risk probability of road segments and nodes. The GIS visualization and risk assessment module is installed in the rescue command center or on rescue vehicles. When there is a network environment, it connects to a remote server to obtain more accurate data. When there is no network environment, it runs independently by relying on Beidou short message data and historical disaster models. The road segment risk probability is dynamically updated using sequential Bayesian inference code set in the module.

[0111] The route planning and calculation module is used to dynamically plan rescue routes. It is installed on rescue vehicles and stores the route statistical analysis results obtained through Monte Carlo analysis. It can independently plan rescue routes based on the latest road network risk information sent by the GIS visualization and risk assessment module, and quickly adjust the rescue routes in the event of a sudden disaster to ensure that the rescue mission is not interrupted.

[0112] The rescue dispatch terminal is installed on transportation vehicles, rescue personnel equipment, or emergency material distribution equipment to execute rescue missions and report mission progress. The rescue dispatch terminal has path display and navigation functions, can receive path planning instructions from the path planning calculation module, and guide rescue personnel to proceed according to the path calculated by the system. At the same time, the rescue dispatch terminal has status feedback capabilities, and can send mission completion status, road damage, and risk change information to the Beidou short message communication module to ensure that the GIS visualization and risk assessment module updates the road section risk assessment based on the latest data.

[0113] The adaptive risk-avoidance emergency rescue path planning method and system provided by this invention has the following advantages:

[0114] This invention ensures that rescue resources reach their destinations efficiently and safely in natural disaster scenarios such as floods and typhoons by employing strategies such as dynamic path adjustment, decentralized transportation, path pre-planning, and network-free communication. Attached Figure Description

[0115] Figure 1 This is a framework diagram of an adaptive risk avoidance emergency rescue path planning method provided by the present invention;

[0116] Figure 2 This is a flowchart of an adaptive risk avoidance emergency rescue path planning method provided by the present invention;

[0117] Figure 3 This is a schematic diagram of the road network structure composed of nodes and road segments used in this invention;

[0118] Figure 4 This is a schematic diagram of the sampling density function based on risk entropy used in this invention;

[0119] Figure 5 This is a path planning diagram for a node to be rescued during the application of this invention;

[0120] Figure 6 This is the final implementation path diagram for a node to be rescued during the application of this invention;

[0121] Figure 7 This is a risk probability change diagram of a node awaiting rescue during the application of this invention;

[0122] Figure 8 This is a risk distribution evolution diagram of one path in the application of this invention. Detailed Implementation

[0123] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0124] Existing route planning methods are mostly based on fixed road network assumptions and fail to address the issue of transportation risk dispersion. They cannot dynamically adapt to sudden changes in the disaster environment, and are difficult to adjust routes in a timely manner after critical roads are disrupted, causing overall rescue progress to stagnate and failing to meet emergency response requirements. To address these issues, this invention proposes an adaptive risk-avoidance emergency rescue route planning method. Before a disaster occurs, based on the road network structure, risk entropy-driven adaptive Monte Carlo analysis is used to simulate various possible risk probability distribution scenarios of the road network structure. A decentralized transportation strategy is adopted to ensure the success rate of rescue, and a dynamic route adjustment strategy is introduced to avoid high-risk routes, ensuring that routes can be adaptively adjusted and improving the stability of rescue operations. Based on the Monte Carlo analysis results, statistical information of each route is obtained to form a pre-planned route library. During the disaster development process, sequential Bayesian inference is used to dynamically update the route risk distribution, and low-risk routes in the pre-planned route library are extracted as planned routes, avoiding the time consumption of direct solution. This invention also relates to an adaptive rescue route planning system integrating BeiDou short message communication, GIS visualization and risk assessment modules, route planning calculation modules, and rescue dispatch terminals. Through BeiDou short message communication, road network information is shared among rescue vehicles, drones, and monitoring stations. Combined with dynamic risk distribution analysis via GIS, the route planning calculation module can autonomously plan routes and transmit tasks to the rescue terminal, ensuring emergency response efficiency and disaster resilience even without cellular networks. Compared to traditional methods, this invention, through dynamic route adjustment strategies, decentralized transportation strategies, route pre-planning strategies, and network-free communication strategies, ensures that rescue resources reach their destinations efficiently and safely in natural disaster scenarios such as floods and typhoons.

[0125] This invention relates to an adaptive risk-avoidance emergency rescue path planning method, the framework diagram of which is shown below. Figure 1 As shown, the flowchart is as follows Figure 2 As shown, the specific implementation method is as follows: Figures 3 to 8Before a disaster occurs, an emergency rescue road network structure is established based on actual geographic information. Risk entropy-driven Monte Carlo analysis is used to simulate various possible risk probability distribution scenarios for the road network structure. A decentralized transportation strategy is adopted to ensure the success rate of rescue efforts, and a dynamic route adjustment strategy is introduced to avoid high-risk routes, ensuring that routes can adaptively adjust and improving the stability of rescue operations. Based on the Monte Carlo analysis results, statistical information for each route is obtained, forming a pre-planned route library. During the disaster's development, sequential Bayesian inference is used to dynamically update the route risk distribution, extracting low-risk routes from the pre-planned route library as planned routes, avoiding the time consumption of direct solution.

[0126] like Figure 1 and Figure 2 As shown, this embodiment of the invention provides an adaptive risk-avoidance emergency rescue path planning method, including the following steps:

[0127] Step S1: Establish the emergency rescue road network structure for the area to be planned based on actual geographic information; generalize the locations within the area to be planned into nodes, and generalize different types of transportation routes (including highways, waterways, aviation, etc.) into road segments. Therefore, the emergency rescue road network structure includes a node set N and a road segment set E; each node n in the node set N represents a geographic location, and each road segment e in the road segment set E represents a traffic connection between two nodes, n∈N, e∈E;

[0128] Specifically, each road segment e has corresponding transport speed and unit load capacity attributes based on its type. For example, road segments can be further divided into various types, including highway transport, ordinary road transport, waterway transport, and air transport. Each type corresponds to a different transport speed and unit load capacity, i.e., the load capacity of a specific vehicle, such as the weight of goods transported by a small truck. The emergency rescue road network structure composed of all nodes and road segments is an undirected graph G = (N, E), in which each road segment e in the undirected graph can be bidirectional.

[0129] In this embodiment, the established emergency rescue road network structure is shown in [reference needed]. Figure 3 It consists of 10 nodes and 17 road segments. The nodes are represented by A to J. The road segments include four types: expressway transportation, ordinary road transportation, waterway transportation and air transportation. The correspondence between road segment type, transportation speed and unit cargo capacity is shown in Table 1. The length of each road segment is shown in Table 2.

[0130] Table 1. Correspondence between road segment type, transport speed, and unit cargo capacity

[0131] Road segment type Transport speed (km / h) Unit load (tons) Highway transport 100 20 Ordinary road transport 60 10 Waterway transport 30 100 Air transport 300 2

[0132] Table 2 Lengths of each road segment in the road network structure

[0133] Road segment Road segment length (km) Road segment Road segment length (km) Road segment Road segment length (km) A-B 63.03 A-C 50.23 A-G 42.48 B-D 54.13 B-F 35.99 C-E 59.98 C-G 44.67 C-H 68.94 D-F 60.09 D-G 49.15 D-I 71.34 E-F 66.73 F-G 55.05 G-H 57.29 H-I 52.18 H-J 59.15 I-J 51.52

[0134] Step S2: Evaluate the long-term risk probability R of each node n n The long-term risk probability R of each road segment e e This is then extended to an asymmetric Logit probability curve distribution as the prior risk distribution, and the risk entropy H of each node n is obtained based on this prior risk distribution. n and the risk entropy H of each road segment e e Based on the risk entropy calculated by distribution, this prepares for subsequent risk entropy-driven adaptive Monte Carlo analysis.

[0135] This invention calculates the long-term risk probability of each node and road segment based on disaster induction intensity, structural vulnerability, and exposure indicators, thereby achieving a potential risk assessment of nodes and road segments in the road network structure. Specifically, it includes the following steps:

[0136] Step S2.1: Using formula (1), evaluate the long-term risk probability R of each node n. n :

[0137]

[0138] Where: λ H ξ is the disaster sensitivity coefficient, with a value range of 0.001 to 0.1; n V is the disaster triggering intensity index for node n, specifically the shape parameter of the generalized extreme value distribution fitted to the annual maximum daily rainfall sequence (at least 20 years); n E represents the structural vulnerability index of node n, specifically the product of the building's service life and its fortification level coefficient. n ξ represents the exposure index of node n, specifically the population-economic combination parameter of node n; max For all nodes, the disaster triggering intensity indices ξ1,ξ2,...,ξ n ,...,ξ N The maximum value of V is used for normalization; max Structural vulnerability indices V1, V2, ..., V for all nodes n ,...,V N The maximum value of E is used for normalization; max Exposure metrics E1, E2, ..., E for all nodes n ,...,E N The maximum value is used for normalization; α, β, and γ are exponential weight parameters; in this embodiment, α = 1.0, β = 0.5, and γ = 1.2.

[0139] Among them: disaster triggering intensity index ξ nThe calculation process is as follows: Let the multi-year maximum daily rainfall sequence of node n be... Y≥20, X y Let Y be the maximum daily rainfall in year y, and Y be the number of years. A generalized extreme value distribution is fitted based on the maximum daily rainfall sequence, and the disaster triggering intensity index ξ is determined based on the fitted function. n Fitting function F(x) n )for:

[0140]

[0141] μ n σ is the mean determined by the fitted function. n x is the standard deviation determined by the fitted function; n The independent variables of the fitting function;

[0142] Structural vulnerability index V n The calculation process is as follows: Let the current year be T. now The construction year of node n is T. built,n When multiple buildings exist, the average value is taken, and the service life of the buildings is T. now -T built,n The building's fortification level coefficient is C. grade,n If the value is 0.3 for level 1, 0.6 for level 2, and 0.9 for level 3, then V n The product of the building's service life and its fortification level coefficient is calculated using the following formula:

[0143] V n =(T now -T built,n )·C grade,n (3)

[0144] Exposure index E n The calculation process is as follows: Let the population density of node n be P. n Units: people / km 2 The economic density is Q n The unit is 10,000 yuan / km 2 , λ P With λ Q Let E be the population weight and the economic weight, respectively. n The calculation formula is:

[0145] E n =λ P ·P n +λ Q ·Q n (4)

[0146] Step S2.2: Long-term risk probability R of road segment e eConsidering the long-term risk probability of the nodes on both sides and the structural vulnerability index of the road segment, specifically using formula (5), the long-term risk probability R of each road segment e is evaluated. e :

[0147]

[0148] S e =(L now -L built,e )·L e (5)

[0149] Where: R e1 and R e2 S represents the long-term risk probability of the nodes on both sides of road segment e; e The structural vulnerability index for road segment e is based on the current year L. now Year of construction of the road section L built,e With the fortification level coefficient L of the road section e It is confirmed that high-security highways, waterways, and all air transport routes L e A value of 0.3 can be taken, for the road and waterway L passing through the slope. e A value of 0.6 can be taken, for old roads and waterways susceptible to flooding. e 0.9 can be taken, S max S is the structural vulnerability index for all road sections. e The maximum value of ω is used for normalization; ω∈[0,1] is the road segment risk coefficient.

[0150] In this invention, the long-term risk probability of each node and road segment is a real number in the range of [0,1]. 0 represents the probability of a node being completely destroyed or a road segment being interrupted due to a disaster of 0%, 1 represents the probability of a node being completely destroyed or a road segment being interrupted due to a disaster of 100%, and other values ​​are deduced accordingly.

[0151] Step S2.3: Determine the prior risk distribution for each node n and each road segment e respectively:

[0152] This invention measures the long-term risk probability R of node n. n The long-term risk probability R of road segment e e This is transformed into a probability density distribution defined in the interval [0,1], forming a continuous expression of risk probability to enhance the ability to model uncertainty; this invention constructs an asymmetric Logit probability curve distribution as the prior risk distribution of nodes and road segments, where the probability density function p of node n is... n The probability density function p of (x) and road segment e e (x) is:

[0153]

[0154] Where: k n and k e These are the risk coefficients for node n and road segment e, respectively, used to control the shape of the distribution curve; the larger the value, the more concentrated the distribution; k n =k max (1-R n );k e =k max (1-R e );k max This represents the maximum risk coefficient. Specifically, by establishing a mapping relationship between the risk coefficient and the long-term risk probability, the smaller the long-term risk probability, the larger the risk coefficient, the more concentrated the risk distribution, and the higher the certainty in subsequent sampling; conversely, the larger the long-term risk probability, the smaller the risk coefficient, the more dispersed the risk distribution, and the lower the certainty in subsequent sampling. max To limit extreme distributions, a value of 10 to 30 can be used.

[0155] δ n and δ e These are the offset coefficients for node n and road segment e, respectively, used to control the skewness characteristics of the risk distribution; The variance is estimated; specifically, by establishing a mapping relationship between the offset coefficient and the long-term risk probability, when the long-term risk probability R... e and R n When the probability is equal to 0.5, the risk exhibits a symmetrical distribution. When the long-term risk probability R... e and R n When the value is greater than 0.5, the distribution peak is biased towards the high-risk area, and the long-term risk probability R... e and R n When the value is less than 0.5, the peak value of the distribution is biased towards the low-risk area. In this embodiment, to estimate the variance,

[0156] C n and C e are the normalization coefficients for node n and road segment e, respectively, used to ensure that the integral distributed in the interval [0,1] is 1; x∈[0,1] is the independent variable of the probability density function;

[0157] Step S2.4: Use formula (7) to determine the risk entropy H of each node n. n and the risk entropy H of each road segment e e :

[0158]

[0159] This step completes the process.

[0160] Specifically, unlike the traditional uniform sampling Monte Carlo simulation method, the risk entropy-driven adaptive Monte Carlo simulation proposed in this invention can characterize the uncertainty of the prior risk distribution of nodes and road segments. It constructs a risk entropy H as a measure of risk variability. The larger the entropy value, the more dispersed the risk distribution. In subsequent simulations, it can be used to guide sampling to the right side of the risk distribution, increase the sampling frequency of high-risk values, and enhance the ability to identify potential extreme events.

[0161] Furthermore, the risk entropy of this invention can reflect the asymmetry and tail characteristics of risk distribution. In actual disasters, high risk often originates from extreme events on the right side of the risk distribution. Such risks are difficult to reflect in indicators such as mean and variance, but they have a decisive impact on path planning. Risk entropy is extremely sensitive to tail distribution; the larger the entropy value, the more concentrated the tail risk, which is beneficial for prioritizing high-risk areas during simulation.

[0162] Based on the entropy values ​​of all nodes and road segments, a strategy for risk sampling in Monte Carlo simulation is proposed in subsequent steps by constructing adaptive sampling probability weights.

[0163] Step S3: Before an actual natural disaster occurs, perform Monte Carlo simulation analysis based on the emergency rescue road network structure to obtain Monte Carlo simulation results; wherein, during the Monte Carlo simulation analysis, the risk entropy H of each node n is used as the basis for the simulation. n and the risk entropy H of each road segment e e Risk sampling is performed using a risk entropy-driven adaptive sampling method.

[0164] Step S3 is as follows:

[0165] Step S3.1: Set the simulation parameters, including the number of simulations M and the simulation time T. sim The number of simulations is M, meaning M random trials are performed, and the simulation time is T. sim It covers both the initial preparation and disaster evolution stages;

[0166] In this embodiment, a total of 100 simulations were performed, and the simulation time was set to 150 minutes.

[0167] Step S3.2: Set up the disaster scenario:

[0168] Considering that in reality, the spread trends of disasters such as typhoon paths and flood spread routes can often be predicted in advance through meteorological and hydrological means, the disaster scenario can be set as a single path to reduce the complexity of the simulation. Starting at a specific time, the risk probability increases from the first node and all its connecting road segments along a self-defined node sequence, simulating the disaster process at that location. After reaching the set time, the risk probability of the first node changes from increasing to decreasing, simulating the disaster transfer process. Subsequently, the risk probability increases from the second node and all its connecting road segments until all nodes have completed the increase and decrease of risk probability in sequence, representing the completion of a flood or typhoon disaster evolution process in the simulation.

[0169] In this embodiment, starting from the 30th minute, the risk probability of node J increases by 0.01 per minute, and the risk probability of all road segments connected to node J increases by 0.02 per minute. Starting from the 60th minute, the risk probability of node J changes from increasing to decreasing by 0.01 every 2 minutes and continues until the 90th minute. Starting from the 90th minute, the risk probability of node I increases by 0.01 per minute, and the risk probability of all road segments connected to node I increases by 0.02 per minute. Starting from the 120th minute, the risk probability of node I changes from increasing to decreasing by 0.01 every 2 minutes and continues until the 150th minute, at which point the simulation ends.

[0170] Step S3.3: Disaster and Emergency Rescue Simulation:

[0171] Based on the set simulation parameters and disaster scenarios, as well as the risk sampling results, dynamic planning of emergency rescue paths is performed and emergency rescue simulation is executed until the set simulation time T is reached. sim ;

[0172] Step S3.3 specifically includes:

[0173] Step S3.3.1: In each simulated disaster evolution process, based on the risk entropy H of each node n... n and the risk entropy H of each road segment e e An adaptive sampling method driven by risk entropy is used to perform risk sampling on all nodes and road segments to determine the risk probability of all nodes and road segments in the emergency rescue road network structure; where the risk probability of node n is represented by R. n,sim The risk probability of road segment e is represented by R. e,sim ;

[0174] Specifically, based on the risk entropy H of each node n n and the risk entropy H of each road segment e e Construct adaptive sampling probability weights ω n,mc (x) and ω e,mc(x), which constitutes the core parameter of the sampling guidance mechanism, is used to replace the uniform sampling strategy in traditional Monte Carlo simulation. Formula (8) is used to determine the adaptive sampling probability weights ω for node n and road segment e, respectively. n,mc (x) and ω e,mc (x):

[0175]

[0176] Where: β n and β e β and γ are the bias amplitude coefficients for node n and road segment e, respectively. The larger the value, the higher the sampling probability of high-risk values, β = 0~3; n and γ e γ and β are the bias intensity coefficients for node n and road segment e, respectively, used to control the sensitivity of sampling to high-risk areas. The larger the value, the higher the sampling probability of high-risk values, γ≥1; in this embodiment, β n and β e Set to 2, γ n and γ e Set to 2.

[0177] The sampling density function for sampling the prior risk distribution of node n and road segment e is as follows: For nodes and road segments with high risk entropy, sampling will tend to be applied to high-risk areas, thereby increasing the simulation probability of extreme events. For nodes and road segments with low risk entropy, sampling will remain stable. Specifically, formula (9) is used to determine the sampling density function for node n and road segment e respectively. and

[0178]

[0179] Therefore, based on the sampling density function of each node n and road segment e and Conduct a risk sampling of all nodes and road segments;

[0180] In this embodiment, in β n and β e Set to 2, γ n and γ e When the value is set to 2, the difference in sampling density functions corresponding to high-risk entropy and low-risk entropy is as follows: Figure 4 As shown.

[0181] In this embodiment, the risk probabilities of nodes and road segments obtained from a single risk sampling are shown in Table 3.

[0182] Table 3 Risk probabilities of nodes and road segments obtained from a single risk sampling.

[0183]

[0184] Step S3.3.2: Based on the risk probabilities of all nodes and road segments in the emergency rescue road network structure obtained from risk sampling, and combined with the traffic data of each road segment, including traffic type, unit load capacity, and transport speed, key nodes and high-risk nodes requiring emergency rescue are selected as nodes awaiting rescue. Specifically, key nodes are those that have a significant impact on the traffic network; their destruction will lead to the interruption of all road segments connected to them. These are determined by calculating the importance of the nodes. High-risk nodes are those severely affected by the disaster and require rescue. The key nodes and high-risk nodes requiring emergency rescue are referred to as nodes awaiting rescue.

[0185] The selection method for key nodes requiring emergency rescue is as follows:

[0186] For each node n in the emergency rescue road network structure, the importance I(n) of the node is calculated using the following formula:

[0187]

[0188] Where: E n Let C be the set of road segments directly connected to node n. e Speed ​​represents the unit load capacity of road segment e. e R is the transport speed of road segment e. n,sim The risk probability of node n;

[0189] Sort all nodes in descending order of importance, select the top 3 nodes as critical nodes, and if the risk probability of a critical node is higher than the first threshold T1, then the critical node is a critical node that requires emergency rescue.

[0190] The screening method for high-risk nodes requiring emergency rescue is as follows:

[0191] Among all nodes, select the node risk probability R. n,sim Nodes exceeding the second threshold T2 are considered high-risk nodes requiring emergency rescue; where the second threshold T2 is greater than the first threshold T1.

[0192] In this embodiment, T1 = 0.3, and the risk probabilities of the three key nodes F, D and H selected are 0.86, 0.81 and 0.71, respectively. Since they are all higher than the first threshold T1, F, D and H are all key nodes that require emergency rescue.

[0193] In this embodiment, T2 = 0.6, so the high-risk nodes requiring emergency rescue are B, D, F, H, and J.

[0194] Step S3.3.3: To reduce the failure rate of rescue routes, this invention adopts a distributed transportation strategy. For each node to be rescued, three nodes with the lowest risk probability are selected from the road network structure and resources are scheduled, which are called output nodes.

[0195] In this embodiment, the nodes to be rescued are B, D, F, H, and J, and the output nodes are C, I, and A.

[0196] Step S3.3.4: Perform path planning between each output node and the node to be rescued. The path planning comprehensively considers transportation time and path risk factors. All possible paths between each output node and the node to be rescued are comprehensively scored, and the path with the highest comprehensive score is selected as the rescue path between the output node and the node to be rescued.

[0197] Specifically, a set of feasible paths PS' between the output node and the node to be rescued is selected. For each path P in the set of feasible paths PS'... l ∈PS', consisting of a series of consecutive road segments e1, e2, ..., e k ,...,e K The composition is determined by the following formula, which is used to calculate its comprehensive score S(P). l ):

[0198]

[0199]

[0200] Where: R(P) l ) represents path P l The risk probability is the sum of the risk probabilities of all segments in the path. For road segment e k Risk probability; T(P) l ) represents path P l Total transport time is the sum of the ratios of the length to the speed of all segments in the route. For road segment e k Length, For road segment e k The corresponding transport speed; B(P) l ) represents path P l The bifurcation degree is defined as the degree of bifurcation of path P. l The sum of the number of road segments connected to the intermediate nodes; ω1, ω2, and ω3 are weighting coefficients. In this invention, ω1 and ω2 are dynamically adjusted according to the risk probability changes of each node awaiting rescue. The risk probability changes as the difference in risk probability of each node awaiting rescue every 5 minutes; ω1 and ω2 are determined by the following formula:

[0201]

[0202] Where: ωmax With ω min The two coefficients are set to satisfy ω max >ω min ;ΔR target,s Let ΔR be the difference in risk probability for the s-th node awaiting rescue every 5 minutes. When the risk probability increases, ΔR... target,s >0, when the risk probability decreases, ΔR target,s <0; When the risk probability increases, consider rapid transportation; when the risk probability decreases, consider transportation safety.

[0203] In this embodiment, ω min =0.2, ω max =0.8, ω3=0.1, the three paths determined after path planning for the node J to be rescued are shown below. Figure 5 .

[0204] In the feasible path set PS', based on each path P l The overall score S(P) of ∈PS' l The route with the highest overall score is selected as the rescue route;

[0205] Step S3.3.5: Transport resources and carry out rescue operations according to the node and segment sequence of the rescue route. During the rescue process, whenever a node is reached, determine whether the current segment risk probability of the next segment exceeds the acceptable safety threshold T. road If so, it is considered a road segment interruption, triggering dynamic path adjustment. The current node is regarded as the output node, and the process returns to step S3.3.4 to replan the subsequent path until the node to be rescued is reached or no passable path can be found. If not, resource transportation continues according to the node and road segment order of the rescue path.

[0206] In this embodiment, T road =0.8, see the three final implementation paths for completing emergency rescue at rescue node J. Figure 6 The HJ section of the first route was interrupted as the risk probability of the road section increased, and the route was partially adjusted to HIJ, thus completing the emergency rescue mission.

[0207] Step S3.3.6: Check the risk probability of all nodes and road segments in the emergency rescue road network structure every 5 minutes. Filter out nodes to be rescued according to step S3.3.2, check whether there are already rescue tasks being performed, and evaluate the risk probability of nodes to be rescued after the completion of existing rescue tasks. If it is less than the first threshold T1, no new rescue tasks will be added. Otherwise, proceed to steps S3.3.3 to S3.3.5. The risk probability check will continue until the simulation time limit is reached.

[0208] In this embodiment, T1 = 0.3. After each existing resource transportation task is completed, the risk probability of the node to be rescued decreases by 0.2. At the 5th minute, all rescue activities are incomplete. Therefore, the selected nodes to be rescued are the same as the initially determined nodes: B, D, F, H, and J. Since each node to be rescued already has 3 resource transportation tasks in progress, and the risk probability of all nodes to be rescued will be less than the first threshold T1 after the tasks are completed, no new tasks are added. Step S3.3.1 is then executed every 5 minutes until the simulation time limit (150 minutes) is reached. The risk probability change graph for node J to be rescued is shown below. Figure 7 All three planned paths were successfully implemented. Two batches of resources were received consecutively when the risk probability of node J began to rise, and a third batch of resources was received after the disaster subsided. However, when dynamic path planning was not used for emergency disaster relief, the risk probability of node J would reach 1, meaning the node would be completely destroyed. The comparison between the two shows the beneficial effects of this invention in emergency rescue.

[0209] Step S4: Complete all simulations, perform statistical analysis of path information based on the Monte Carlo simulation results, and establish a pre-planned path library; the pre-planned path library stores the final implemented planned path, the average path risk probability, and the variance of the path risk probability;

[0210] Step S4 is as follows:

[0211] After completing all Monte Carlo simulations, path statistics are performed based on all simulation results, including the final planned path after reaching the node to be rescued, the average path risk probability, and the variance of the path risk probability; a pre-planned path library is established based on the final planned path.

[0212] Let the set of planned paths in the pre-planned path library be PS. For each path P i Based on the statistical results of Monte Carlo simulation analysis, the average path risk probability of ∈PS is calculated using the following formula. and path risk probability variance σ 2 (P i ):

[0213]

[0214] Where: M i Path P for Monte Carlo analysis i The number of executions, R (j) (P i ) represents path P in the j-th simulation. i The probability of risk when M i When the value is too small, such as M i <5, let σ 2 (Pi =0.08.

[0215] In this embodiment, some path statistics are shown in Table 4.

[0216] Table 4. Statistical results of partial paths based on Monte Carlo analysis.

[0217] Number Final implementation path Risk average Risk variance 1 A-B 0.27 0.16 2 A-G-D 1.14 0.16 3 A-G-F 0.84 0.21 4 C-A-G-F 1.42 0.21 5 E-F-G-D 1.36 0.00 6 E-C-G-A-B 1.81 / …… ……

[0218] Step S5: When an actual natural disaster occurs, establish the prior risk distribution of each planned path in the pre-planned path library; specifically:

[0219] For each path P i Based on the average path risk probability and path risk probability variance σ 2 (P i The path P is constructed using an asymmetric Logit probability curve distribution. i The prior risk distribution, path P i probability density function p i (x) is:

[0220]

[0221] Where: k i For path P i Risk coefficient; δ i For path P i The offset coefficient.

[0222] Step S6: Use sequential Bayesian inference to dynamically correct the prior risk distribution of each planned path during the emergency rescue process, and obtain the mean posterior risk of each planned path.

[0223] Step S6 is as follows:

[0224] Step S6.1: Obtain the risk probability change value for each planned path:

[0225] During emergency rescue operations, the risk probability change values ​​of all nodes and road segments in the emergency rescue road network structure are checked every minute. This updates the risk probability change value of each planned path, which is then used as the observation variable O for Bayesian updates. t (P i ); when O t (P i ) > 0 indicates path P i The probability of risk is increasing, when O t (P i ) < 0 indicates path P i The probability of risk is decreasing, when O t (P i) = 0 indicates path P i The probability of risk remains unchanged, O t (P i It is calculated using the following formula:

[0226] O t (P i ) = R t (P i )-R t-1 (P i (16)

[0227] Among them: O t (P i ) represents path P at minute t. i The change in risk probability, R t (P i ) represents path P at minute t. i The probability of risk, R t-1 (P i ) represents path P at minute t-1. i The risk probability is the sum of the risk probabilities of all segments in the path;

[0228] Step S6.2: Update the posterior risk distribution for each planned path:

[0229] Based on the statistical distribution model of path risk, the posterior parameters of path risk are adjusted to update the prior risk distribution of the path to the posterior risk distribution; specifically, formula (17) is used to adjust the path P i The risk coefficient k, which is used as a priori parameter i Updated to posterior parameter k i ′, and, path P i The offset coefficient δ, which is used as a prior parameter i Updated to posterior parameter δ i ′:

[0230] k i ′=k i +O t (P i ),δ i ′=δ i +O t (P i (17)

[0231] Step S6.3: Calculate path P using formula (18). i mean posterior risk of the path

[0232]

[0233] Path P imean posterior risk of the path As path P in emergency rescue process i The updated mean posterior risk of the path.

[0234] In this embodiment, the risk distribution evolution diagram of path IJ (a path consisting of only one road segment) is shown below. Figure 8 At the start of the emergency rescue, the mean posterior risk of the path was 0.29. As observation data continued to be fed back, for example, at the 10th minute, the probability of path risk for IJ was 0.20, and the mean posterior risk of the path decreased to 0.28. At the 50th minute, the probability of path risk for IJ was 0.70, and the mean posterior risk of path IJ also increased to 0.35.

[0235] Step S7: Based on the mean posterior risk of each planned path after sequential Bayesian inference, perform dynamic planning of emergency rescue paths, select the optimal rescue path and execute the emergency rescue task until the emergency rescue activity ends.

[0236] Step S7 is as follows:

[0237] Step S7.1: Based on measured data, determine the risk probability of all nodes and road segments in the emergency rescue road network structure;

[0238] Step S7.2: Based on the traffic data and node risk probability of all road segments, select the nodes to be rescued;

[0239] Step S7.3: For each selected node to be rescued, check if a rescue mission is already in progress. If not, proceed to step S7.4; if so, proceed to step S7.9.

[0240] Step S7.4: For each node to be rescued, select the three nodes with the lowest risk probability from the emergency rescue road network structure as output nodes;

[0241] Step S7.5: Between each output node and the node to be rescued, path planning is carried out based on the mean posterior risk of each planned path determined in step S6, and an optimal rescue path is selected as the rescue path.

[0242] Specifically, first filter the set of feasible paths PS between the output node and the node to be rescued. For each path P r ∈PS”, based on its mean posterior risk. Choose the path P with the lowest mean posterior risk. * As the optimal rescue route, route P * It is calculated using the following formula:

[0243]

[0244] This allows us to determine the optimal rescue route.

[0245] In this embodiment, the feasible path set refers to the paths listed in Table 4 and other paths obtained through Monte Carlo analysis.

[0246] Step S7.6: Carry out the rescue according to the order of nodes and road segments of the rescue route;

[0247] Step S7.7: During the rescue operation, whenever a node is reached, determine whether the current risk probability of the next road segment exceeds the acceptable safety threshold T. road If so, it is considered a road segment interruption, triggering dynamic path adjustment. The current node is regarded as the output node, and the process returns to step S7.5 to re-plan the subsequent path. That is, each time a node is reached, the process checks whether the next road segment is interrupted. If it is interrupted, a new path from the current node to the node to be rescued is planned based on the path risk distribution after sequential Bayesian inference. If no passable path can be found, the task is terminated. If not, step S7.8 is executed.

[0248] In addition, during the rescue process, based on the current traffic data, the risk probability of all nodes and road segments in the emergency rescue road network structure is checked every 5 minutes to screen out nodes to be rescued, check whether there are already rescue tasks being performed, and assess the risk probability of nodes to be rescued after the completion of existing rescue tasks. If it is less than the first threshold T1, no new rescue task will be added; otherwise, return to step S7.4 to add output nodes and add new rescue tasks, and so on in a continuous loop.

[0249] Step S7.8: Determine whether the node to be rescued has been reached. If not, return to step S7.6; if yes, proceed to step S7.9.

[0250] Step S7.9: End the rescue mission.

[0251] This invention also provides a system for an adaptive risk-avoidance emergency rescue route planning method, comprising: a Beidou short message communication module, a GIS visualization and risk assessment module, a route planning calculation module, and a rescue dispatch terminal;

[0252] The BeiDou short message communication module is used for data transmission in disaster environments, ensuring information exchange for rescue missions even if traditional cellular networks are damaged. Installed on rescue vehicles, drones, and fixed monitoring stations, the module is responsible for sending and receiving road status information, rescue mission allocation instructions, and real-time route adjustment plans. Uplink data includes the location information of rescue teams, the execution status of rescue missions, road conditions, and risk updates; downlink data includes optimal route instructions, mission adjustment commands, and detour suggestions.

[0253] The GIS visualization and risk assessment module is used to model the location and road topology of the disaster area, converting it into an emergency rescue road network structure composed of nodes and road segments, and dynamically assessing the risk probability of road segments and nodes. This module is installed in the rescue command center or on rescue vehicles. When a network environment is available, it connects to a remote server to obtain more accurate data; when no network environment is available, it operates independently using BeiDou short message data and historical disaster models. The road segment risk probability is dynamically updated using sequential Bayesian inference code set within the module. For example, if multiple rescue vehicles report increased water accumulation on a certain road, or a drone reports intensified landslides, the GIS system will automatically increase the risk probability of that road. Through the sharing of BeiDou short message data among rescue vehicles, drones, and fixed monitoring stations, the GIS module can continuously update road risk information even without a network connection, ensuring the accuracy of rescue route selection.

[0254] The route planning and calculation module is used to dynamically plan rescue routes. This module can be installed on rescue vehicles and stores the route statistical analysis results obtained through Monte Carlo analysis. It can independently plan rescue routes based on the latest road network risk information sent by the GIS visualization and risk assessment module, and quickly adjust the rescue routes in the event of a sudden disaster to ensure that the rescue mission is not interrupted.

[0255] The rescue dispatch terminal is installed on transportation vehicles, rescue personnel equipment, or emergency material distribution equipment to execute rescue missions and report mission progress. The rescue dispatch terminal has path display and navigation functions, can receive path planning instructions from the path planning calculation module, and guide rescue personnel to proceed according to the path calculated by the system. At the same time, the rescue dispatch terminal has status feedback capabilities, and can send mission completion status, road damage, and risk change information to the Beidou short message communication module to ensure that the GIS visualization and risk assessment module updates the road section risk assessment based on the latest data.

[0256] In this embodiment, the GIS visualization and risk assessment module first performs risk modeling on the entire road network. Subsequently, the route planning and calculation module, based on the road segment risk data provided by GIS and combined with feedback information received via BeiDou short messages from rescue vehicles, drones, and fixed monitoring stations, completes route planning and sends it to the rescue dispatch terminal. During the execution of rescue missions, the rescue dispatch terminal transmits information such as transportation status and road conditions back via BeiDou short messages. The GIS visualization and risk assessment module then performs sequential Bayesian inference based on this information, updating the risk probabilities of all nodes and road segments in the entire road network structure and outputting data to the route planning and calculation module. When the risk probability of a road segment exceeds a set threshold, the route planning and calculation module quickly recalculates the route and pushes the latest planned route to the rescue dispatch terminal, ensuring that the rescue mission is not interrupted and that rescue vehicles or personnel can safely reach the target location.

[0257] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An adaptive risk-avoidance emergency rescue path planning method, characterized in that, Includes the following steps: Step S1: Establish the emergency rescue road network structure for the area to be planned based on actual geographic information; the emergency rescue road network structure includes a set of nodes. and road segment collection The set of nodes Each node in The set of road segments represents a geographical location. Each section of the road Represents the traffic connection between two nodes. , ; Step S2: Evaluate each node Long-term risk probability With each section Long-term risk probability This is then extended to an asymmetric Logit probability curve distribution as the prior risk distribution, and each node is obtained based on this prior risk distribution. Risk entropy and each section of road Risk entropy ; Step S3: Before an actual natural disaster occurs, perform Monte Carlo simulation analysis based on the emergency rescue road network structure to obtain Monte Carlo simulation results; wherein, during the Monte Carlo simulation analysis, based on each node... Risk entropy and each section of road Risk entropy Risk sampling is performed using a risk entropy-driven adaptive sampling method. Step S4: Perform statistical analysis of path information based on the Monte Carlo simulation results to establish a pre-planned path library; the pre-planned path library stores the final implemented planned path, the average path risk probability, and the variance of the path risk probability; Step S5: When an actual natural disaster occurs, establish the prior risk distribution of each planned path in the pre-planned path library; Step S6: Use sequential Bayesian inference to dynamically correct the prior risk distribution of each planned path during the emergency rescue process, and obtain the mean posterior risk of each planned path. Step S7: Based on the mean posterior risk of each planned path after sequential Bayesian inference, perform dynamic planning of emergency rescue path, select the optimal rescue path and execute the emergency rescue task until the emergency rescue activity ends. Specifically, step S2 is as follows: Step S2.1: Use formula (1) to evaluate each node. Long-term risk probability : （1） in: Disaster sensitivity coefficient; For nodes The disaster triggering intensity index is specifically the shape parameter of the generalized extreme value distribution obtained by fitting the annual maximum daily rainfall sequence; For nodes The structural vulnerability index, specifically the node The product of the building's service life and its fortification level coefficient; For nodes Exposure metrics, specifically nodes Population-economic combination parameters; Disaster triggering intensity index for all nodes The maximum value; Structural vulnerability index for all nodes The maximum value; Exposure metrics for all nodes The maximum value; , , For exponential weighting parameters; Among them: disaster triggering intensity index The calculation process is as follows: Let the node The sequence of the largest daily rainfall in many years is , , For the first The highest daily rainfall in a year, The number of years; a generalized extreme value distribution is fitted based on the maximum daily rainfall sequence, and the disaster triggering intensity index is determined based on the fitted function. Fitting function for: （2） The mean value is determined by the fitted function. The standard deviation is determined by the fitted function; The independent variables of the fitting function; Structural vulnerability index The calculation process is as follows: Let the current year be... ,node The year of construction is When multiple buildings exist, the average value is taken, and the service life of the buildings is... The building's fortification level coefficient is ,but The product of the building's service life and its fortification level coefficient is calculated using the following formula: （3） Exposure index The calculation process is as follows: Let the node The population density is Economic density is , and If the population weight and economic weight are respectively, then The calculation formula is: （4） Step S2.2: Use formula (5) to evaluate each road segment. Long-term risk probability : ； （5） in: and The road sections The long-term risk probability of the two nodes on both sides; For road section Structural vulnerability indicators, based on the current year Year of construction of road section With the fortification level coefficient of the road section Confirmed, high-security highways, waterways, and all air transport routes. Take 0.3, for roads and waterways passing through slopes. Take 0.6, old roads and waterways susceptible to flooding Take 0.9, Structural vulnerability index for all road sections The maximum value; Risk coefficient of road section; Step S2.3: Determine each node separately and each section of road Prior risk distribution: Node Long-term risk probability and road section Long-term risk probability Transform into definition in The probability density distribution of the interval forms a continuous expression of the risk probability; an asymmetric Logit probability curve distribution is constructed as the prior risk distribution of nodes and road segments. probability density function and road sections probability density function for: ； （6） in: and They are nodes and road sections The risk coefficient is used to control the shape of the distribution curve; the larger the value, the more concentrated the distribution. ; ; This represents the maximum risk coefficient. and They are nodes and road sections The offset coefficient is used to control the skewness of the risk distribution; ; ; The variance is the estimated variance; and They are nodes and road sections The normalization coefficient is used to ensure that the distribution is within the range of 1 / 2. The integral over the interval is 1; , where is the independent variable of the probability density function; Step S2.4: Use formula (7) to determine each node. Risk entropy and each section of road Risk entropy : ； （7） This step completes the process.

2. The adaptive risk-avoidance emergency rescue path planning method according to claim 1, characterized in that, Each section The road segment type has corresponding transport speed and unit cargo capacity attributes; The emergency rescue road network structure, composed of all nodes and road segments, is an undirected graph. In each segment of the undirected graph Two-way traffic.

3. The adaptive risk-avoidance emergency rescue path planning method according to claim 1, characterized in that, Step S3 is as follows: Step S3.1: Set simulation parameters, including the number of simulations. and simulation time ; Step S3.2: Set up the disaster scenario: Starting at a specific time, the risk probability of the first node and all its connected road segments increases according to a self-defined node sequence, simulating the disaster process. After reaching the set time, the risk probability of the first node changes from increasing to decreasing, simulating the disaster transfer process. Subsequently, the risk probability of the second node and all its connected road segments increases, until all nodes have completed the increase and decrease of risk probability in sequence, representing the completion of one disaster evolution process. Step S3.3: Disaster and Emergency Rescue Simulation: Based on the set simulation parameters and disaster scenarios, as well as the risk sampling results, dynamic planning of emergency rescue routes is performed and emergency rescue simulation is executed until the set simulation time is reached. ; Step S3.3 specifically includes: Step S3.3.1: In each simulated disaster evolution process, based on each node... Risk entropy and each section of road Risk entropy An adaptive sampling method driven by risk entropy is used to perform risk sampling on all nodes and road segments to determine the risk probability of all nodes and road segments in the emergency rescue road network structure; wherein, nodes The probability of risk is expressed as Road section The probability of risk is expressed as ; The specific method is as follows: Using formula (8), the nodes are determined respectively. and road sections Adaptive sampling probability weights and : ； （8） in: and , respectively, nodes and road sections The larger the bias amplitude coefficient, the higher the sampling probability of high-risk values. ; and , respectively, nodes and road sections The bias intensity coefficient is used to control the sensitivity of sampling to high-risk areas; the larger the value, the higher the sampling probability of high-risk values. ; Using formula (9), the nodes are determined respectively. and road sections sampling density function and : ； （9） Therefore, based on each node and road sections sampling density function and Conduct a risk sampling of all nodes and road sections; Step S3.3.2: Based on the risk probability of all nodes and road segments in the emergency rescue road network structure obtained by risk sampling, and combined with the unit load capacity and transportation speed of each road segment, key nodes and high-risk nodes that require emergency rescue are selected as nodes to be rescued; The selection method for key nodes requiring emergency rescue is as follows: For each node in the aforementioned emergency rescue road network structure The importance of nodes is calculated using the following formula. : （10） in: For nodes A set of directly connected road segments. For road section The unit cargo capacity For road section The speed of transportation For nodes The probability of risk; Sort all nodes in descending order of importance, and select the top 3 nodes as critical nodes. If the risk probability of a critical node is higher than a first threshold... If so, then this critical node is a critical node that requires emergency rescue; The screening method for high-risk nodes requiring emergency rescue is as follows: Filter out the node risk probability from all nodes. Above the second threshold The nodes are high-risk nodes requiring emergency rescue; among them, the second threshold Greater than the first threshold ; Step S3.3.3: Adopt a distributed transportation strategy. For each node to be rescued, select three nodes with the lowest risk probability from the road network structure to schedule resources. These nodes are called output nodes. Step S3.3.4: Perform path planning between each output node and the node to be rescued. The path planning comprehensively considers transportation time and path risk factors. All possible paths between each output node and the node to be rescued are comprehensively scored, and the path with the highest comprehensive score is selected as the rescue path between the output node and the node to be rescued. Specifically, a set of feasible paths between the output node and the node to be rescued is selected. For the set of feasible paths Each path in It consists of a series of continuous road sections. The composition is determined by the following formula, which is used to calculate the overall score. : （11） ； ； in: For path The risk probability is the sum of the risk probabilities of all segments in the path. For road section The probability of risk; For path Total transport time is the sum of the ratios of the length to the speed of all segments in the route. For road section Length, For road section The corresponding transport speed; For path The bifurcation degree is defined as the degree of bifurcation of the path. The sum of the number of road segments connected to the intermediate nodes; These are the weighting coefficients. and The risk probability is dynamically adjusted based on the change in risk probability of each node to be rescued, which is the difference in risk probability of each node to be rescued every 5 minutes. and Determined by the following formula: （12） in: and The two coefficients set satisfy the following conditions: ; For the first The difference in risk probability between each node awaiting rescue every 5 minutes; when the risk probability increases... When the probability of risk decreases When the probability of risk increases, consider rapid transportation; when the probability of risk decreases, consider transportation safety. In the set of feasible paths In the middle, according to each path Overall rating The route with the highest overall score is selected as the rescue route; Step S3.3.5: Transport resources and carry out rescue operations according to the node and segment sequence of the rescue route. During the rescue process, whenever a node is reached, determine whether the current segment risk probability of the next segment exceeds an acceptable safety threshold. If so, it is considered a road segment interruption, triggering dynamic path adjustment. The current node is regarded as the output node, and the process returns to step S3.3.4 to replan the subsequent path until the node to be rescued is reached or no passable path can be found. If not, resource transportation continues according to the node and road segment order of the rescue path. Step S3.3.6: Check the risk probability of all nodes and road segments in the emergency rescue road network structure every 5 minutes. Based on step S3.3.2, select nodes to be rescued, check whether there are already rescue tasks being performed, and assess the risk probability of nodes to be rescued after the completion of existing rescue tasks. If the risk probability is less than the first threshold... If no new rescue mission is added, proceed to steps S3.3.3 to S3.3.

5. The risk probability check will continue until the simulation time limit is reached.

4. The adaptive risk-avoidance emergency rescue path planning method according to claim 1, characterized in that, Step S4 is as follows: After completing all Monte Carlo simulations, path statistics were performed based on all simulation results, including the final planned path after reaching the node to be rescued, the average path risk probability, and the variance of the path risk probability. Establish a pre-planned path library based on the final implemented planned path; The set of planned paths in the pre-planned path library is set as follows: For each path Based on the statistical results of Monte Carlo simulation analysis, the average path risk probability is calculated using the following formula. Path risk probability variance : （13） （14） in: Path for Monte Carlo Analysis The number of times it is executed. For the first Path in this simulation The probability of risk when Value too small order .

5. The adaptive risk-avoidance emergency rescue path planning method according to claim 1, characterized in that, In step S5, establishing the prior risk distribution of each planned path in the pre-planned path library specifically involves: For each path Based on the average path risk probability and path risk probability variance Paths are constructed using asymmetric Logit probability curve distribution. Prior risk distribution, path probability density function for: （15） ； ； in: For path The risk factor; For path The offset coefficient.

6. The adaptive risk-avoidance emergency rescue path planning method according to claim 5, characterized in that, In step S6, sequential Bayesian inference is used to dynamically correct the prior risk distribution of each planned path during the emergency rescue process, obtaining the mean posterior risk of each planned path, specifically: Step S6.1: Obtain the risk probability change value for each planned path: During emergency rescue operations, the risk probability change values ​​of all nodes and road segments in the emergency rescue road network structure are checked every minute. This updates the risk probability change value of each planned path, which is then used as the observation variable for Bayesian updates. ;when Indicate path The probability of risk is increasing, when Indicate path The probability of risk is decreasing, when Indicate path The probability of risk remains unchanged. It is calculated using the following formula: （16） in: For the first Minute path The change in the probability of risk, For the first Minute path The probability of risk, For the first Minute path The risk probability is the sum of the risk probabilities of all segments in the path; Step S6.2: Update the posterior risk distribution for each planned path: Based on the statistical distribution model of path risk, the posterior parameters of path risk are adjusted to update the prior risk distribution of the path to the posterior risk distribution; specifically, formula (17) is used to update the path... Risk coefficient as a prior parameter Updated to posterior parameters , and, the path The bias coefficient as a prior parameter Updated to posterior parameters : , （17） Step S6.3: Calculate the path using formula (18). mean posterior risk of the path : （18） path mean posterior risk of the path As a route in emergency rescue process The updated mean posterior risk of the path.

7. The adaptive risk-avoidance emergency rescue path planning method according to claim 1, characterized in that, In step S7, dynamic planning of emergency rescue paths is performed based on the mean posterior risk of each planned path after sequential Bayesian inference. The optimal rescue path is selected and the emergency rescue task is executed until the emergency rescue activity ends. Specifically: Step S7.1: Based on measured data, determine the risk probability of all nodes and road segments in the emergency rescue road network structure; Step S7.2: Based on the traffic data and node risk probability of all road segments, select the nodes to be rescued; Step S7.3: For each selected node to be rescued, check if a rescue mission is already in progress. If not, proceed to step S7.

4. If so, proceed to step S7.9; Step S7.4: For each node to be rescued, select the three nodes with the lowest risk probability from the emergency rescue road network structure as output nodes; Step S7.5: Between each output node and the node to be rescued, path planning is carried out based on the mean posterior risk of each planned path determined in step S6, and an optimal rescue path is selected as the rescue path. Step S7.6: Carry out the rescue according to the order of nodes and road segments of the rescue route; Step S7.7: During the rescue operation, whenever a node is reached, determine whether the current risk probability of the next road segment exceeds an acceptable safety threshold. If so, it is considered a road segment interruption, triggering dynamic path adjustment, treating the current node as an output node, and returning to step S7.5 to re-plan the subsequent path; if not, proceed to step S7.

8. Step S7.8: Determine whether the node to be rescued has been reached. If not, return to step S7.6; if yes, proceed to step S7.

9. Step S7.9: End the rescue mission.

8. The adaptive risk-avoidance emergency rescue path planning method according to claim 7, characterized in that, Also includes: Based on current traffic data, the risk probability of all nodes and road segments in the emergency rescue road network structure is checked every 5 minutes. Nodes awaiting rescue are selected, and it is checked whether any rescue missions are already in progress. The risk probability of nodes awaiting rescue is then assessed after the completion of any existing rescue missions. If the risk probability is less than a first threshold... If no new rescue task is added, then no new rescue task will be added; otherwise, return to step S7.4, add an output node, add a new rescue task, and repeat this process continuously. Step S7.5: Between each output node and the node to be rescued, path planning is performed based on the mean posterior risk of each planned path determined in Step S6, and an optimal rescue path is selected as the rescue path, specifically as follows: First, filter the set of feasible paths between the output node and the node to be rescued. For each path Based on its path posterior risk mean Choose the path with the lowest mean posterior risk. As the optimal rescue route, the route It is calculated using the following formula: （19） This allows us to determine the optimal rescue route.

9. A system for an adaptive risk-avoidance emergency rescue path planning method according to any one of claims 1-8, characterized in that, include: The system includes a BeiDou short message communication module, a GIS visualization and risk assessment module, a path planning and calculation module, and a rescue dispatch terminal. The BeiDou short message communication module is used for data transmission in disaster environments to ensure information exchange during rescue missions. Installed on rescue vehicles, drones, and fixed monitoring stations, the module is responsible for sending and receiving road status information, rescue mission allocation instructions, and real-time route adjustment plans. Uplink data includes the location information of rescue teams, the execution status of rescue missions, road conditions, and risk update information; downlink data includes optimal route instructions, mission adjustment commands, and detour suggestions. The GIS visualization and risk assessment module is used to model the location and road topology of the disaster area, convert it into an emergency rescue road network structure composed of nodes and road segments, and dynamically assess the risk probability of road segments and nodes; The GIS visualization and risk assessment module is installed in the rescue command center or on rescue vehicles. When there is a network environment, it connects to a remote server to obtain more accurate data. When there is no network environment, it runs independently by relying on Beidou short message data and historical disaster models. The road section risk probability is dynamically updated using sequential Bayesian inference code set in the module. The route planning and calculation module is used to dynamically plan rescue routes. It is installed on rescue vehicles and stores the route statistical analysis results obtained through Monte Carlo analysis. It can independently plan rescue routes based on the latest road network risk information sent by the GIS visualization and risk assessment module, and quickly adjust the rescue routes in the event of a sudden disaster to ensure that the rescue mission is not interrupted. The rescue dispatch terminal is installed on transportation vehicles, rescue personnel equipment, or emergency material distribution equipment to execute rescue missions and report mission progress. The rescue dispatch terminal has path display and navigation functions, can receive path planning instructions from the path planning calculation module, and guide rescue personnel to proceed according to the path calculated by the system. At the same time, the rescue dispatch terminal has status feedback capabilities, and can send mission completion status, road damage, and risk change information to the Beidou short message communication module to ensure that the GIS visualization and risk assessment module updates the road section risk assessment based on the latest data.

Images