Emergency resource matching and rescue decision method and device based on injury situation

Abstract

This application relates to a method and apparatus for emergency resource matching and rescue decision-making based on injury situation. The method includes: acquiring real-time data from the disaster site; constructing a dynamic situational awareness model based on the real-time data and generating a multi-objective, multi-constraint rescue task model; solving the rescue task model using an intelligent matching algorithm and a hybrid multi-objective evolutionary algorithm to generate an initial casualty-resource matching scheme and a rescue route planning scheme; continuously monitoring situational indicators during the execution of the initial scheme, and performing a replanning operation to adjust resource matching and rescue routes when any indicator triggers a preset dynamic grading threshold. This method can significantly shorten the waiting time for critically injured patients, improve resource utilization, and reduce the expected mortality rate. It achieves efficient, robust, and adaptive rescue decision-making in large-scale emergency scenarios.

Description

Technical Field

[0001] This application relates to the field of emergency management technology, and in particular to a method and device for emergency resource matching and rescue decision-making based on injury status. Background Technology

[0002] Following major natural disasters such as earthquakes, floods, storms, and forest fires, the number of injured typically increases dramatically within a short period, leading to a surge in demand for medical care. The efficient and rational allocation of critical emergency resources, including medical personnel, ambulances, and medical equipment, directly determines the success or failure of rescue operations. Emergency resource allocation refers to the decision-making and management process by which an emergency dispatch center scientifically coordinates and optimizes the allocation of emergency resources based on the response needs of a sudden event. This aims to improve the timeliness and coordination of rescue operations, maximize resource utilization efficiency, and provide a solid guarantee for the successful achievement of emergency rescue objectives. However, in large-scale emergencies with limited resources and highly uncertain environments, traditional emergency resource allocation methods often have limitations. Most existing methods are based on static statistics of the number of injured or distance-priority principles, mixing injured individuals with different injury levels and ignoring the dynamic, heterogeneous, and time-sensitive nature of injuries. Within the "golden rescue window," the condition of the injured may continue to deteriorate, easily leading to delays in treating critically injured patients, resource misallocation, and low rescue efficiency. Therefore, how to achieve intelligent matching of emergency resources and dynamic decision-making for the rescue of the wounded under the dynamic evolution of the injury situation has become the key to improving the effectiveness of emergency rescue and reducing the mortality rate, and it is also a current research hotspot in the field of international emergency management.

[0003] Numerous challenges remain in actual emergency rescue operations, particularly in emergency resource allocation and casualty evacuation, where several issues urgently need to be addressed. After a disaster, rescue needs are highly uncertain in terms of time, location, type, and scale. Infrastructure such as roads and communications may be damaged, significantly hindering rescue efforts and potentially leading to mission failure. Regarding casualty evacuation, disaster sites often lack a unified mechanism for collecting and dynamically updating injury information. Information is often delayed and fragmented, the matching of resources and casualties lacks dynamic coordination, evacuation route planning is not updated in a timely manner, and there is a lack of effective balancing mechanisms among multiple objectives. This not only wastes emergency resources but also causes delays or incomplete treatment, severely impacting the effectiveness of emergency care. Therefore, based on the severity of injuries, comprehensive planning of emergency resource allocation and the formulation of reasonable evacuation decisions are crucial for protecting public safety, minimizing economic losses, and maintaining social stability. Summary of the Invention

[0004] Therefore, it is necessary to provide an emergency resource matching and rescue decision-making method and device based on the injury situation to address the above-mentioned technical problems, which can shorten the waiting time of critically injured patients, improve resource utilization, and reduce the expected mortality rate.

[0005] An emergency resource matching and rescue decision-making method based on injury status, the method comprising: Acquire real-time data from the disaster site, including data on the injuries of the injured, medical resources, and road network status.

[0006] Based on real-time data, a dynamic situational awareness model is constructed that includes time constraints, resource constraints, spatial constraints, and priority constraints, and a multi-objective, multi-constraint rescue mission model is generated.

[0007] The intelligent matching algorithm and the hybrid multi-objective evolutionary algorithm are used to solve the rescue mission model, and the initial casualty-resource matching scheme and rescue route planning scheme are generated.

[0008] During the execution of the initial plan, situation indicators are continuously monitored. When any indicator triggers a preset dynamic grading threshold, a replanning operation is performed to adjust resource matching and rescue routes.

[0009] Output and execute the adjusted resource scheduling instructions and rescue navigation instructions.

[0010] An emergency resource matching and rescue decision-making device based on injury status, the device comprising: The data acquisition module is used to acquire real-time data from the disaster site, including data on the injuries of the injured, medical resources, and road network status.

[0011] The model building module is used to construct a dynamic situational awareness model based on real-time data, which includes time constraints, resource constraints, spatial constraints, and priority constraints, and to generate a multi-objective, multi-constraint rescue mission model.

[0012] The scheme generation module is used to solve the rescue mission model using intelligent matching algorithm and hybrid multi-objective evolutionary algorithm to generate initial casualty-resource matching scheme and rescue route planning scheme.

[0013] The replanning module is used to continuously monitor situation indicators during the execution of the initial plan. When any indicator triggers a preset dynamic grading threshold, a replanning operation is performed to adjust resource matching and rescue routes.

[0014] The instruction output module is used to output and execute the adjusted resource scheduling instructions and rescue navigation instructions.

[0015] The aforementioned emergency resource matching and rescue decision-making method and device based on injury situation firstly acquires real-time data from multiple sources, including the injury status of the injured, medical resources, and road network status, and constructs a dynamic situational representation model that integrates time, resource, spatial, and priority constraints. This overcomes the shortcomings of traditional methods that treat the needs of the injured as static parameters and ignore the dynamic deterioration of injuries, enabling decisions to truly reflect the urgency and resource heterogeneity at the disaster site, laying a data foundation for subsequent accurate matching. Secondly, it uses intelligent matching algorithms and hybrid multi-objective evolutionary algorithms to solve the multi-objective, multi-constraint rescue task model. This allows for the generation of initial matching and path planning schemes that take into account the survival probability of the injured, resource utilization, and rescue costs in a short time, effectively solving the problems of resource misallocation and chaotic transportation order in large-scale emergency scenarios. In particular, it can still stably obtain high-quality Pareto front solutions under the complex constraints of hundreds of injured people and dozens of resource nodes. Secondly, during execution, core situation indicators are continuously monitored, and a replanning trigger mechanism based on dynamic hierarchical thresholds is introduced: when indicators such as the average waiting time of critically injured patients, regional resource saturation, and rate of injury deterioration exceed preset three-level thresholds, local or global replanning is automatically executed. This achieves a leap from fixed-cycle adjustment to event-driven adaptive adjustment, ensuring rapid response to critical changes in injury conditions while avoiding the waste of computational resources caused by frequent recalculations. Finally, the output and execution of adjusted resource scheduling and rescue navigation instructions can significantly shorten the waiting time of critically injured patients at disaster sites, improve the utilization efficiency of scarce resources such as ambulances, medical personnel, and beds, and reduce the mortality and permanent disability rates caused by rescue delays. This maximizes the overall treatment benefits within the "golden rescue window" and provides scientific, real-time, and executable decision support for emergency command departments. Attached Figure Description

[0016] Figure 1 This is a flowchart of an emergency resource matching and rescue decision-making method based on injury status in one embodiment; Figure 2 This is a schematic diagram of the emergency resource scheduling process in one embodiment; Figure 3 This is a schematic diagram of emergency medical services operation in one embodiment; Figure 4 This is a schematic diagram of the overall research framework of the project in one embodiment; Figure 5 Here is a flowchart of the NSGA-II algorithm in one embodiment; Figure 6 This is a schematic diagram of the parent-child merging process in NSGA-II in one embodiment; Figure 7 This is a schematic diagram illustrating the removal and insertion operations of the ALNS algorithm in one embodiment, wherein... Figure 7 (a) is the initial solution. Figure 7 (b) is the solution after removing the operator. Figure 7 (c) is the solution after re-inserting the operator; Figure 8 This is a schematic diagram of individual crowding in one embodiment; Figure 9 This is a structural block diagram of an emergency resource matching and rescue decision-making device based on injury status in one embodiment. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0018] In one embodiment, such as Figure 1 As shown, an emergency resource matching and rescue decision-making method based on injury status is provided, including the following steps: Step 102: Obtain real-time data from the disaster site. Real-time data includes data on the injuries of the injured, medical resources, and road network status.

[0019] Specifically, the emergency command center collects three types of data in real time through portable triage terminals deployed at disaster sites, ambulance GPS devices, hospital information system interfaces, and traffic management department road condition monitoring platforms: Injury data: Rescue personnel used the START triage system to conduct initial triage of each injured person and generate injury levels. (1 = critically injured, 2 = seriously injured, 3 = slightly injured), and record the geographical coordinates of the injured. Discovery time and initial vital signs parameters.

[0020] Medical resource data: including the current location of each ambulance. ,type (Normal / Negative Pressure), Supine Volume Walking wounded capacity Current occupancy status; dynamic bed capacity of each hospital. Operating room availability status; battery power of each drone platform. Maximum load capacity Maximum flight speed .

[0021] Road network status data: Loading disaster area road network map based on GIS system Each road section The initial travel time is calculated based on free-flow speed, while real-time traffic flow data is integrated to detect congestion and mark road sections interrupted by earthquakes (setting the travel time for these road sections). ).

[0022] Step 104: Based on real-time data, construct a dynamic situational awareness model that includes time constraints, resource constraints, spatial constraints, and priority constraints, and generate a multi-objective, multi-constraint rescue mission model.

[0023] Specifically, based on the initial injury level of the injured person. Assign a survival probability function to each wounded soldier: (Formula 1) in, The initial survival probability, This is a half-life parameter (unit: minutes). This is the degradation rate parameter. The values ​​for this embodiment are shown in Table 1 below: Table 1 Parameter Values

[0024] At the same time, a deprivation cost function is introduced to quantify the psychological and physiological losses suffered by wounded soldiers while waiting for medical treatment: (Formula 2) Among them, the deprivation cost rate ,coefficient Positively correlated with the severity of injury. In this example: critically injured... Serious injury Minor injury .

[0025] Furthermore, establish unified constraint expressions for the three types of core resources: Ambulance capacity constraints: (Formula 3) in Indicates the wounded Whether by vehicle transport, For vehicles Total capacity (lying position + walking). In this embodiment, the capacity of a standard ambulance is... Negative pressure ambulances are only used for patients with respiratory infectious diseases.

[0026] Hospital bed time-varying constraints: (Formula 4) in, Indicates the wounded Should we send them to the hospital? , For the hospital At any moment The number of available beds is dynamically updated as patients are admitted.

[0027] Constraints on the capabilities of healthcare personnel: (Formula 5) in Indicates the wounded Whether by medical staff Treatment This represents the number of injured persons that can be treated per unit of time.

[0028] Furthermore, the road network is modeled as a time-varying directed graph. Road section The travel time is represented as a time-dependent function: (Formula 6) in, For the length of the road segment, For a moment The passage speed. This function must satisfy the first-in-first-out property: (Formula 7) For road sections that are interrupted, They are directly eliminated in the path planning.

[0029] Furthermore, define the dynamic priority function: (Formula 8) in, This is the initial priority determined according to the START triage criteria (the higher the value, the higher the priority). This represents the change in the severity of the injury (positive for worsening and negative for improvement). The maximum tolerable waiting time corresponding to the injury level is as follows: critical injury ≤ 60 minutes, serious injury ≤ 120 minutes, minor injury ≤ 240 minutes. As an adjustable weighting coefficient, this embodiment takes... .

[0030] Furthermore, based on the above constraints, a mixed-integer nonlinear programming (MINLP) model is constructed, which includes three conflicting sub-objectives: Objective 1: Minimize the sum of weighted rescue time and deprivation cost. (Formula 9) in, Weighting of the injured (critical injury = 5, serious injury = 3, minor injury = 1). For the waiting time, For transit time, Unit operating cost for vehicles / drones.

[0031] Objective 2: Minimize expected mortality rate: (Formula 10) in, For the wounded The probability of survival upon arrival at the hospital is calculated using formula (1).

[0032] Objective 3: Maximize resource utilization: (Formula 11) in, For resources Actual usage Its total available quantity.

[0033] The three objectives mentioned above are solved jointly using the Pareto optimization method.

[0034] Step 106: Solve the rescue mission model using the intelligent matching algorithm and the hybrid multi-objective evolutionary algorithm to generate the initial casualty-resource matching scheme and rescue route planning scheme.

[0035] Specifically, the first step is to construct feature vectors of wounded soldier needs and resource service capabilities, and then normalize them.

[0036] Multi-attribute weighted matching degree function: (Formula 12) in, Injury severity mapping values ​​(critical injury = 1.0, serious injury = 0.6, minor injury = 0.3); : The reciprocal normalized value of the Euclidean distance The degree to which remaining resource capacity matches the needs of the wounded. The degree of matching between resource functionality and injury needs (e.g., negative pressure ambulance suitable for patients with respiratory infectious diseases = 1.0, otherwise = 0.2), weighting coefficient. The analytic hierarchy process (AHP) was used in conjunction with expert scoring to determine the following: .

[0037] The iterative formal description of the multi-round auction algorithm is as follows: (Formula 13) The specific process is as follows: In each round, unmatched wounded soldiers bid on all available resources (the bid value is the matching degree). Each resource selects the wounded soldier with the highest matching degree as the temporary winner. If a wounded soldier is selected by multiple resources at the same time, the resource with the highest matching degree is selected. The matched wounded soldiers and resources are removed from the candidate set. The process is iterated until all wounded soldiers are matched or resources are exhausted.

[0038] Furthermore, for multi-objective optimization steps in large-scale scheduling: Adaptive crossover and mutation operator: Crossover probability and mutation probability Based on population diversity indicators Dynamic adjustment: (Formula 14) (Formula 15) The diversity index is defined as the variance of the objective function values ​​of individuals in the population: (Formula 16) Let the objective function vector be... This represents the average target vector of the population. When the population diversity is high, the mutation probability is increased to enhance the exploration ability; when the population tends to converge, the crossover probability is increased to accelerate convergence.

[0039] Furthermore, the ALNS operator weights are adaptively updated: the operator set is disrupted. and repair operator set Weight update formula: (Formula 17) in The cumulative score for the operator (based on its contribution to the improvement of the objective function). For the number of times it is used, For the learning rate, this embodiment takes... .

[0040] Furthermore, non-dominated ordering and crowding distance: Congestion distance calculation formula: (Formula 18) in The number of objective functions (in this embodiment) ), On the same Pareto front and Two adjacent solutions.

[0041] Furthermore, nodes arrive shortest time Satisfying the Bellman optimal equation: (Formula 19) in For nodes The set of neighboring nodes. This embodiment uses an improved Dijkstra algorithm as the basic path search method, combined with real-time traffic data to solve for the time-varying shortest path. For sudden events such as road interruptions, a local replanning strategy is adopted, only taking the two ends of the affected road segment as the new start and end points, and re-searching the path in the subgraph to avoid the high overhead of global recalculation.

[0042] Step 108: During the execution of the initial plan, continuously monitor the situation indicators. When any indicator triggers a preset dynamic grading threshold, perform a replanning operation to adjust resource matching and rescue routes.

[0043] Specifically, during the execution of the initial rescue plan, the emergency command platform continuously calculates the following core situation indicators every 30 seconds: Average waiting time for critically injured patients: (Formula 20) Regional resource saturation: (Formula 21) in, These represent drones, ambulances, and truck platforms, respectively.

[0044] Injury deterioration rate index: (Formula 22) Percentage of injuries exceeding the threshold: (Formula 23) Critical platform failure rate: (Formula 24) Task saturation on different platforms: (Formula 25) Dynamic hierarchical threshold triggering mechanism: Furthermore, three threshold levels are set for each indicator. These correspond to yellow, orange, and red alerts, respectively. The comprehensive trigger level is defined as: (Formula 26)

[0045] The threshold settings in this embodiment are shown in Table 2 below: Table 2 Threshold Setting Parameters

[0046] when Time-triggered replanning: Level 1 triggers local matching adjustment (adjusting only affected casualty-resource pairs), Level 2 triggers partial path replanning (replanning the paths of affected vehicles), Level 3 triggers global rescheduling (resolving the entire optimization model).

[0047] Robust optimization module: A robust optimization module is embedded in the reprogramming solution framework, and box-type uncertainty sets are used to describe the uncertain parameters. Demand uncertainty: (Formula 27) Uncertainty regarding travel time: (Formula 28)

[0048] Platform availability uncertainty: (Formula 29)

[0049] Through robust equivalence transformation, uncertain constraints are converted into deterministic constraints. Taking demand constraints as an example, the original constraints... The robust correspondence is: (Formula 30) in, For uncertain budget parameters (in this embodiment, we take...) This controls the degree of conservatism in the solution. By adjusting... This allows for a balance between the robustness and optimality of the solution.

[0050] Step 110: Output and execute the adjusted resource scheduling instructions and rescue navigation instructions.

[0051] Specifically, after the replanning is completed, the decision-making system outputs two types of instructions in a visual format: Resource dispatch instructions: Send the patient transport sequence, mission priority, target hospital and estimated arrival time to the vehicle-mounted terminal or flight control system of each rescue platform (ambulance, drone, truck).

[0052] Rescue navigation instructions: Generate curve-by-curve navigation routes for each vehicle and drone, and update the ETA display in real time on the command center's large screen.

[0053] All instructions are transmitted to various terminals via 4G / 5G networks, and on-site rescue personnel or autonomous driving modules execute them directly. Simultaneously, the system synchronizes the adjusted plan to the hospital's information system, allowing the hospital to prepare beds and operating rooms in advance. The system continuously receives feedback data during execution, forming a closed-loop decision-making process of "matching-planning-replanning."

[0054] The aforementioned emergency resource matching and rescue decision-making method based on injury situation firstly acquires real-time data from multiple sources, including the injuries of the wounded, medical resources, and road network status, and constructs a dynamic situational awareness model that integrates time, resource, spatial, and priority constraints. This overcomes the shortcomings of traditional methods that treat the needs of the wounded as static parameters and ignore the dynamic deterioration of injuries, enabling decisions to truly reflect the urgency and resource heterogeneity at the disaster site, laying a data foundation for subsequent accurate matching. Secondly, it uses intelligent matching algorithms and hybrid multi-objective evolutionary algorithms to solve the multi-objective, multi-constraint rescue task model. This can generate initial matching schemes and path planning schemes that take into account the survival probability of the wounded, resource utilization, and rescue costs in a short time, effectively solving the problems of resource misallocation and chaotic transportation order in large-scale emergency scenarios. In particular, it can still stably obtain high-quality Pareto front solutions under the complex constraints of hundreds of wounded and dozens of resource nodes. Secondly, during execution, core situation indicators are continuously monitored, and a replanning trigger mechanism based on dynamic hierarchical thresholds is introduced: when indicators such as the average waiting time of critically injured patients, regional resource saturation, and rate of injury deterioration exceed preset three-level thresholds, local or global replanning is automatically executed. This achieves a leap from fixed-cycle adjustment to event-driven adaptive adjustment, ensuring rapid response to critical changes in injury conditions while avoiding the waste of computational resources caused by frequent recalculations. Finally, the output and execution of adjusted resource scheduling and rescue navigation instructions can significantly shorten the waiting time of critically injured patients at disaster sites, improve the utilization efficiency of scarce resources such as ambulances, medical personnel, and beds, and reduce the mortality and permanent disability rates caused by rescue delays. This maximizes the overall treatment benefits within the "golden rescue window" and provides scientific, real-time, and executable decision support for emergency command departments.

[0055] In one embodiment, the initial injury level and corresponding survival probability function parameters of each wounded soldier are obtained. The survival probability function characterizes the dynamic decay of the wounded soldier's survival probability with waiting time. Using the survival probability function, the real-time survival probability of each wounded soldier at any waiting time is calculated. Based on the real-time survival probability, a time window constraint is generated for each wounded soldier, wherein the upper limit of the time window constraint is inversely proportional to the real-time survival probability.

[0056] In one embodiment, a first objective function is defined as minimizing the sum of weighted rescue time and deprivation cost, wherein the deprivation cost is constituted by the integral of the deprivation cost rate function over the waiting time, and the deprivation cost rate function grows exponentially with the waiting time. A second objective function is defined as minimizing the expected mortality rate based on a survival probability function. A third objective function is defined as maximizing resource utilization. The first, second, and third objective functions are jointly optimized to form the multi-objective, multi-constraint rescue mission model.

[0057] In one embodiment, a feature vector representing the needs of the wounded and a feature vector representing the resource service capacity are constructed, and a multi-attribute weighted matching degree is calculated, incorporating injury priority, spatial distance, capacity suitability, and functional compatibility. A multi-round auction algorithm is used to globally optimize the pairing of wounded and resources based on the multi-attribute weighted matching degree, generating the wounded-resource matching scheme. Matching pairs from the wounded-resource matching scheme are treated as elite individuals and injected into the initial population of a hybrid multi-objective evolutionary algorithm. The hybrid multi-objective evolutionary algorithm is executed, iteratively optimizing through non-dominated sorting and crowding distance calculation to generate a Pareto optimal solution set, from which a rescue route planning scheme is selected.

[0058] In one embodiment, a hybrid strategy combining elite initialization and random replenishment is used to generate the initial population, where elite individuals are derived from the wounded-resource matching scheme. The crossover and mutation probabilities are dynamically adjusted based on the population diversity index; the mutation probability increases when population diversity is above a threshold, and the crossover probability increases when population diversity is below a threshold. A set of destruction and repair operators is invoked to perform a local search on the current solution. The destruction operators include random removal, worst-case removal, and relevant removal; the repair operators include greedy insertion, regret insertion, and 2-opt optimization. Based on the contribution of each operator to the objective function, the weights of the destruction and repair operators are adaptively updated until a termination condition is met, outputting the optimal rescue route planning scheme.

[0059] In one embodiment, such as Figure 2 As shown, an emergency resource dispatch process is provided. The emergency management center first receives requests for the rescue of injured persons from the disaster area. Combining real-time collected data on the injury situation, medical resource availability, and road network status, it conducts a demand assessment and formulates a plan. Then, it initiates a resource matching and task allocation process to generate a preliminary injured-resource matching plan. Next, it calculates the optimal rescue route and departure sequence based on a dynamic path planning algorithm. Finally, it issues dispatch instructions to various rescue platforms (ambulances, drones, trucks, etc.) to execute the injured transfer and medical resource delivery tasks. During execution, the system continuously monitors changes in situation indicators. When a preset dynamic grading threshold is triggered, it automatically initiates a replanning process, forming a closed-loop decision-making mechanism of "matching-planning-execution-monitoring-replanning".

[0060] In one embodiment, such as Figure 3The diagram illustrates an emergency medical service operation process, showcasing the collaborative relationship between the disaster site, ambulances, hospitals, and the emergency dispatch center. At the disaster site, injured personnel information, after triage, is uploaded to the emergency dispatch center in real time via mobile terminals. The dispatch center performs intelligent matching and route planning based on constraints such as the severity of injuries, ambulance location and capacity, available hospital beds, and staffing levels. The planning results are issued to ambulances as task instructions, which then proceed to the scene to pick up the injured and transport them to designated hospitals. Upon receiving the injured, the hospitals update their bed occupancy status and staff availability, which is fed back to the dispatch center in real time for subsequent decision-making. The entire operation embodies the collaborative flow of information, resources, and injured personnel.

[0061] In one embodiment, such as Figure 4 As shown, a general research framework for the project is provided. This framework is driven by injury situation awareness and is divided into three research layers from top to bottom: The first layer is the quantification of rescue constraints and task modeling. The system analyzes time constraints (including survival probability function and deprivation cost function), resource constraints (ambulance capacity, hospital bed capacity, medical staff capacity), spatial constraints (time-varying road network travel time, road interruption handling), and priority constraints (dynamic priority function), and constructs a multi-objective, multi-constraint mixed integer nonlinear programming model; the second layer is the intelligent matching and optimization algorithm for emergency resources, including an intelligent matching module based on multi-attribute weighted matching degree and multi-round auction algorithm, and a hybrid multi-objective evolutionary algorithm module that integrates improved NSGA-II and ALNS, combined with a time-varying shortest path planning algorithm; the third layer is the quantification of rescue constraints and task modeling, which adopts model predictive control and dynamic hierarchical threshold triggering mechanism to realize online adjustment and replanning of rescue strategy. The three layers form a closed-loop feedback through data flow and decision flow.

[0062] In one embodiment, such as Figure 5 As shown, an NSGA-II algorithm flow is provided, which specifically includes the following steps: Step 1, randomly generate an initial population according to the encoding rules, and perform a non-dominated sorting operation on the initial population, followed by genetic operations such as selection, crossover, and mutation to generate a progeny population; Step 2, merge the parent population and the progeny population into a joint population of size 2N, perform a fast non-dominated sorting on the joint population, divide the individuals into multiple Pareto fronts, and calculate the crowding distance of individuals within each front; Step 3, select N individuals from the joint population based on the non-dominated level (prioritizing individuals with lower levels) and crowding distance (prioritizing individuals with larger distances) to form a new parent population; Step 4, determine whether the preset maximum number of iterations has been reached. If not, return to Step 1 to continue iterating; if reached, output the Pareto optimal solution set for the decision-maker to select the final rescue plan.

[0063] In one embodiment, such as Figure 6 As shown, a parent-child merging process in NSGA-II is provided, assuming the first generation... The parent generation population is (Of size N), after genetic manipulation, a progeny population is generated. (Scale is also N). Compared to traditional genetic algorithms that directly use... As the next generation of parent populations differs, this embodiment will and Merge into a population of size 2N. Subsequently, on Perform a quick non-dominated sort, dividing the area into several front surfaces. ,in This is the optimal Pareto front. In constructing the next generation of the parent population... At that time, in sequence Individuals joining Until a certain frontier is added The population size then exceeds N. At this point, for Individuals are sorted in descending order of crowding distance, and only the top few are selected. Individuals join This approach maintains population diversity while preserving elite solutions. The merging and selection strategy ensures that the algorithm does not lose any excellent solutions found during iteration.

[0064] In one embodiment, such as Figure 7 As shown, a removal and insertion operation of the ALNS algorithm is provided. Figure 7 (a) to Figure 7 (b) Demonstrate removal operation: from the current complete rescue plan (e.g.) Figure 7 (a) shows multiple vehicle / drone paths and corresponding casualty evacuation tasks. Several task nodes are removed using destruction operators. The set of destruction operators used in this embodiment includes random removal (randomly selecting several casualty tasks), worst-case removal (removing the casualty task that contributes the least to the objective function), and related removal (removing a group of casualty tasks that are geographically close). Figure 7 (b) Independent nodes without connecting lines represent tasks that have been removed. Figure 7 (b) to Figure 7 (c) Demonstrating the insertion operation: Based on the partial solution formed after removing tasks, the removed tasks are re-inserted into the solution using repair operators. The set of repair operators used in this embodiment includes greedy insertion (inserting the task at the position with the minimum cost), regret insertion (considering both current and future insertion opportunities), and... Optimization (local path rearrangement). Through iterative destruction and repair, the ALNS algorithm can effectively escape local optima during the search process, improving the quality of the solution.

[0065] In one embodiment, such as Figure 8 As shown, a schematic diagram of individual crowding is provided, with two objective functions in the diagram. and For example, the distribution of individuals on the same Pareto front is illustrated. The crowding distance of each individual (represented by a dot) is defined as the sum of the side lengths of the rectangles formed by the two adjacent individuals in the target space. Specifically, for an individual... The individual to its left is The adjacent individual on the right is The formula for calculating congestion distance is:

[0066] Individuals are represented by dashed rectangles in the diagram. The crowding distance covers the region of the individual. A larger crowding distance indicates a sparser number of solutions around the individual, resulting in higher retention value during population selection; conversely, an individual with a small crowding distance represents an overly dense concentration of solutions in the region, leading to lower priority for retention. This mechanism effectively maintains the uniform distribution of the Pareto front, preventing the algorithm from prematurely converging to local optima.

[0067] It should be understood that, although Figures 1-3 , Figure 5 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figures 1-3 , Figure 5 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0068] In one embodiment, such as Figure 9 As shown, an emergency resource matching and rescue decision-making device based on injury situation is provided, including: a data acquisition module 902, a model building module 904, a plan generation module 906, a replanning module 908, and an instruction output module 910, wherein: The data acquisition module 902 is used to acquire real-time data from the disaster site, including data on the injuries of the injured, medical resources, and road network status.

[0069] The model building module 904 is used to build a dynamic situational representation model based on real-time data, which includes time constraints, resource constraints, spatial constraints and priority constraints, and generate a multi-objective, multi-constraint rescue mission model.

[0070] The scheme generation module 906 is used to solve the rescue mission model using intelligent matching algorithm and hybrid multi-objective evolution algorithm to generate initial casualty-resource matching scheme and rescue route planning scheme.

[0071] The replanning module 908 is used to continuously monitor situation indicators during the execution of the initial plan. When any indicator triggers a preset dynamic grading threshold, a replanning operation is performed to adjust resource matching and rescue routes.

[0072] The instruction output module 910 is used to output and execute the adjusted resource scheduling instructions and rescue navigation instructions.

[0073] Specific limitations regarding the injury situation-based emergency resource matching and rescue decision-making device can be found in the limitations of the injury situation-based emergency resource matching and rescue decision-making method described above, and will not be repeated here. Each module in the aforementioned injury situation-based emergency resource matching and rescue decision-making device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0074] Those skilled in the art will understand that Figure 4 , Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0075] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchlink, DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0076] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0077] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for emergency resource matching and rescue decision-making based on injury status, characterized in that, The method includes: Acquire real-time data from the disaster site, including data on the injuries of the injured, medical resources, and road network status. Based on the real-time data, a dynamic situational awareness model is constructed that includes time constraints, resource constraints, spatial constraints, and priority constraints, and a multi-objective, multi-constraint rescue mission model is generated. The rescue mission model is solved using an intelligent matching algorithm and a hybrid multi-objective evolutionary algorithm to generate an initial casualty-resource matching scheme and a rescue route planning scheme. During the execution of the initial plan, situation indicators are continuously monitored. When any indicator triggers a preset dynamic grading threshold, a replanning operation is performed to adjust resource matching and rescue routes. Output and execute the adjusted resource scheduling instructions and rescue navigation instructions.

2. The method according to claim 1, characterized in that, Based on the real-time data, a dynamic situational awareness model is constructed, incorporating time constraints, resource constraints, spatial constraints, and priority constraints, including: The initial injury level and corresponding survival probability function parameters of each wounded soldier are obtained. The survival probability function is used to characterize the dynamic decay law of the survival probability of the wounded soldier with the waiting time. Using the survival probability function, calculate the real-time survival probability of each wounded person at any waiting time; Based on the real-time survival probability, a time window constraint is generated for each wounded soldier, wherein the upper limit of the time window constraint is inversely proportional to the real-time survival probability.

3. The method according to claim 1, characterized in that, Generate a multi-objective, multi-constraint rescue mission model, including: The first objective function is defined as minimizing the sum of weighted rescue time and deprivation cost, wherein the deprivation cost is composed of the integral of the deprivation cost rate function over the waiting time, and the deprivation cost rate function grows exponentially with the waiting time; The second objective function is defined as minimizing the expected mortality rate based on the survival probability function; The third objective function is defined as maximizing resource utilization. The first, second, and third objective functions are jointly optimized to form the multi-objective, multi-constraint rescue mission model.

4. The method according to any one of claims 1 to 3, characterized in that, The rescue mission model is solved using an intelligent matching algorithm and a hybrid multi-objective evolutionary algorithm to generate an initial casualty-resource matching scheme and a rescue route planning scheme, including: Construct feature vectors of wounded patient needs and resource service capabilities, and calculate a multi-attribute weighted matching degree that integrates injury priority, spatial distance, capacity adaptability, and functional matching degree; A multi-round auction algorithm is used to globally optimize the pairing of wounded soldiers and resources based on the multi-attribute weighted matching degree, thereby generating the wounded soldier-resource matching scheme. The matched pairs in the wounded soldier-resource matching scheme are treated as elite individuals and injected into the initial population of the hybrid multi-objective evolutionary algorithm; The hybrid multi-objective evolutionary algorithm is executed to generate a Pareto optimal solution set through non-dominated sorting and crowding distance calculation, and the rescue route planning scheme is selected from iterative optimization.

5. The method according to claim 4, characterized in that, Executing the hybrid multi-objective evolutionary algorithm includes: An initial population is generated using a hybrid strategy that combines elite initialization with random replenishment, wherein elite individuals are derived from the wounded-resource matching scheme; The crossover and mutation probabilities are dynamically adjusted based on the population diversity index. When the population diversity is higher than the threshold, the mutation probability is increased, and when the population diversity is lower than the threshold, the crossover probability is increased. The set of destruction operators and the set of repair operators are invoked to perform a local search on the current solution. The destruction operators include random removal, worst removal and related removal, and the repair operators include greedy insertion, regret insertion and 2-opt optimization. Based on the contribution of each operator to the objective function, the weights of the destruction and repair operators are adaptively updated until the termination condition is met, and the optimal rescue path planning scheme is output.

6. An emergency resource matching and rescue decision-making device based on injury situation, characterized in that, The device includes: The data acquisition module is used to acquire real-time data from the disaster site, including data on the injuries of the injured, medical resources, and road network status. The model building module is used to build a dynamic situational representation model that includes time constraints, resource constraints, spatial constraints and priority constraints based on the real-time data, and generate a multi-objective and multi-constraint rescue mission model. The scheme generation module is used to solve the rescue mission model using an intelligent matching algorithm and a hybrid multi-objective evolutionary algorithm to generate an initial casualty-resource matching scheme and a rescue route planning scheme. The replanning module is used to continuously monitor situation indicators during the execution of the initial plan. When any indicator triggers a preset dynamic grading threshold, a replanning operation is performed to adjust resource matching and rescue routes. The instruction output module is used to output and execute the adjusted resource scheduling instructions and rescue navigation instructions.

Images