Intelligent decision-making method and system for emergency allocation of first-aid materials

Abstract

The invention provides an intelligent decision-making method and system for emergency allocation of first-aid materials. According to the method, by integrating a graph neural network, deep learning, reinforcement learning, Bayesian optimization and a genetic algorithm, the demand information of multiple emergency response points is efficiently processed, and by receiving the demand information including the number, the type and the severity of injuries and patients, the demand urgency degree is dynamically calculated through an intelligent evaluation system, and a preliminary deployment plan is generated. And a traffic prediction model is combined with real-time and historical data to pre-estimate the road traffic capacity, various logistics distribution strategies are simulated to optimize the preliminary scheme, and an optimal path plan is determined. And then, the material distribution effect is verified in different scenes by means of an AI simulation drilling system, the configuration scheme is adjusted, and finally a comprehensive allocation instruction is formed to guide actual operation. According to the technical scheme of the invention, intelligent decision support for the first-aid materials can be realized, and the efficiency and accuracy of first-aid material allocation are remarkably improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence technology, and in particular to an intelligent decision-making method and system for emergency deployment of first aid supplies. Background Art

[0002] In emergency rescue scenarios, rapid and accurate deployment of emergency supplies is crucial to improving the survival rate and recovery quality of the injured.

[0003] Currently, the allocation of emergency supplies mainly relies on manual scheduling or automated systems based on preset rules. These systems usually combine geographic information systems to assist decision-making and may integrate some basic traffic prediction models to optimize delivery routes.

[0004] However, such methods often lack flexibility and adaptability when dealing with emergencies, and fail to fully consider uncertainties in actual situations, such as traffic accidents or temporary traffic controls. Summary of the invention

[0005] The embodiments of the present application provide an intelligent decision-making method and system for emergency allocation of first aid supplies, so as to solve the problems of low efficiency, lack of flexibility and adaptability in the prior art of first aid supply allocation.

[0006] In a first aspect, an embodiment of the present application provides an intelligent decision-making method for emergency deployment of emergency supplies, including:

[0007] Receiving demand information from multiple emergency response points, the demand information at least including the number of injured, the type of injured, the severity, the status of on-site medical resources, the priority identification, and the geographic spatial distribution;

[0008] Based on the graph neural network intelligent evaluation system, the urgency of the demand of each emergency response point is evaluated according to the demand information, and a preliminary deployment plan is generated. The preliminary deployment plan is determined according to the urgency of the demand of each emergency response point, and includes the type and quantity of first aid supplies required by each emergency response point, as well as the distribution order of the first aid supplies;

[0009] Using a deep learning prediction model, a prediction result of road capacity is generated based on historical traffic data and real-time traffic obtained from multiple sources, and a variety of logistics distribution strategies are simulated in combination with a reinforcement learning algorithm, the preliminary deployment plan is refined and adjusted, and a target logistics distribution strategy is determined; wherein, the prediction result of road capacity refers to the deep learning prediction model using historical traffic data, real-time traffic data and model algorithms to predict the traffic conditions on the road in the future period of time, and the logistics distribution strategy is formulated according to the distribution order of the emergency supplies described in the preliminary deployment plan; the target logistics distribution strategy can avoid the estimated congested sections and ensure timely arrival at the designated location;

[0010] Initiate an AI simulation exercise system based on Bayesian optimization and genetic algorithm, simulate the configuration schemes of the types and quantities of the first aid supplies in multiple scenarios according to the target logistics distribution strategy, and iteratively evolve the optimal configuration scheme of the types and quantities of the first aid supplies that can adapt to the actual situation from the configuration schemes in multiple scenarios through genetic algorithm, and generate a comprehensive deployment instruction, which includes a list of first aid supplies, an estimated delivery time and an optimal transportation route;

[0011] Execute the comprehensive deployment instruction to realize the emergency deployment of the emergency supplies.

[0012] Optionally, the deep learning prediction model is used to generate a prediction result of road capacity based on historical traffic data and real-time traffic acquired from multiple sources, including:

[0013] Integrate historical traffic data and real-time traffic data obtained from multiple sources to obtain a traffic data set, wherein the historical traffic data at least includes traffic flow, accident records and weather impacts in the same time period in the past, and the real-time traffic data at least includes current traffic camera images, vehicle location information and instant traffic conditions provided by sensors;

[0014] The deep learning prediction model is used to analyze and process the traffic data in the traffic data set to generate a prediction result of the road capacity in the future. The deep learning prediction model can identify and estimate the impact of different factors on the road capacity through training, and output the expected traffic efficiency of each road in a specific time period in the future to generate a prediction result of the road capacity.

[0015] Optionally, according to the predicted result of the road capacity, multiple logistics distribution strategies are simulated in combination with the reinforcement learning algorithm, the preliminary deployment plan is refined and adjusted, and a target logistics distribution strategy is determined, including:

[0016] Using the predicted results of the road capacity and combining with the reinforcement learning algorithm, the distribution order of the emergency supplies in the preliminary deployment plan is simulated to obtain a variety of logistics distribution strategies aimed at optimizing time cost and transportation efficiency;

[0017] Based on a variety of logistics distribution strategies, the preliminary deployment plan is refined and adjusted. The refinement and adjustment process includes optimizing the transportation routes of the emergency supplies under different logistics distribution strategies and adjusting the types and quantities of emergency supplies required for each emergency response point to adapt to the actual situation;

[0018] For each of the logistics distribution strategies, a performance evaluation is performed in terms of avoiding estimated congested road sections and ensuring timely arrival at designated locations, and the performance evaluation indicators at least include delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility, to obtain an evaluation result for each of the logistics distribution strategies;

[0019] Based on the evaluation results of all logistics distribution strategies, select the target logistics distribution strategy that can maximize transportation efficiency and minimize delays;

[0020] The target logistics distribution strategy is applied to the adjusted preliminary allocation plan for verification and confirmation to ensure that the target logistics distribution strategy can achieve the expected effect in actual operation.

[0021] Optionally, the step of selecting a target logistics distribution strategy that can maximize transportation efficiency and minimize delays based on the evaluation results of all logistics distribution strategy effects includes:

[0022] According to the evaluation results of the effects of all logistics distribution strategies, eligible logistics distribution strategies are screened out, and based on multi-dimensional evaluation, the screened logistics distribution strategies are comprehensively scored to obtain the score of each logistics distribution strategy, and one or several logistics distribution strategies with the highest score are selected as candidate logistics distribution strategies;

[0023] Analyze the selection of each route, expected traffic conditions and delay risk in each candidate logistics distribution strategy to determine the target logistics distribution strategy that can maximize transportation efficiency and minimize delays

[0024] Optionally, an AI simulation and exercise system based on Bayesian optimization and genetic algorithm is started to simulate the configuration schemes of the types and quantities of the first aid supplies in multiple scenarios according to the target logistics distribution strategy, and iteratively evolve the optimal configuration scheme of the types and quantities of the first aid supplies that can adapt to the actual situation from the configuration schemes in multiple scenarios through genetic algorithm, and generate comprehensive deployment instructions, including:

[0025] Start the AI ​​simulation and exercise system based on Bayesian optimization and genetic algorithm, load the target logistics distribution strategy as input data, and initialize the simulation environment;

[0026] Using Bayesian optimization technology, the model parameters of the AI ​​simulation and drill system are automatically adjusted to improve simulation accuracy and obtain optimized parameter configuration;

[0027] Based on the optimized parameter configuration, simulate the configuration scheme of the types and quantities of the emergency supplies in multiple predefined scenarios, and evaluate the distribution effect corresponding to the configuration scheme in each predefined scenario;

[0028] By using a genetic algorithm, the allocation effect corresponding to the configuration scheme in each predefined scenario is iterated and evolved to determine the optimal configuration scheme for the types and quantities of the first aid supplies that can adapt to various actual situations;

[0029] Verify the effectiveness of the optimal configuration plan and generate a comprehensive deployment instruction containing a detailed list of emergency supplies, estimated delivery time and optimal transportation route.

[0030] Optionally, based on the optimized parameter configuration, simulating the configuration scheme of the type and quantity of the emergency supplies in multiple predefined scenarios, and evaluating the distribution effect corresponding to the configuration scheme in each predefined scenario, includes:

[0031] In the AI ​​simulation and drill system, the optimized parameter configuration is loaded as the initial setting to initialize the simulation environment to ensure that the simulation environment can accurately reflect the actual situation;

[0032] Based on the optimized parameter configuration, in each predefined scenario, the distribution of the first aid supplies is simulated, and the simulation process includes path selection, time arrangement and the distribution of the first aid resources, so as to obtain a configuration scheme of the type and quantity of the first aid supplies in each predefined scenario;

[0033] The configuration scheme of the type and quantity of each of the first aid supplies is evaluated, and the evaluation indicators include at least delivery time, transportation cost, material utilization efficiency, rescue success rate and emergency response speed, to obtain the distribution effect corresponding to the configuration scheme.

[0034] Optionally, the graph neural network-based intelligent assessment system assesses the urgency of the demand for each emergency response point according to the demand information, and generates a preliminary deployment plan, including:

[0035] Utilize natural language processing algorithms to parse and process demand information from multiple emergency response points to obtain specific demand results for each emergency response point;

[0036] According to the specific demand results, a graph structure data model is constructed, wherein the nodes in the graph structure data model represent various emergency response points, and the edges represent the correlation and impact range between different emergency response points;

[0037] Based on the pre-trained model parameters, the graph neural network intelligent evaluation system is started to load and process the graph structure data model to ensure that the graph neural network intelligent evaluation system can identify and process different types of node and edge attributes, and the graph neural network intelligent evaluation system is used to evaluate the demand urgency of each emergency response point and calculate the demand urgency score of each node. The demand urgency score is determined according to the correlation and impact range between different emergency response points;

[0038] Based on the demand urgency score, a preliminary deployment plan is generated.

[0039] Optionally, a performance evaluation process is performed on each of the logistics distribution strategies in terms of avoiding estimated congested road sections and ensuring timely arrival at a designated location, and the performance evaluation indicators at least include delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility, and the evaluation results of each of the logistics distribution strategies are obtained, including:

[0040] For each of the logistics distribution strategies S i ,Performance evaluation is performed in terms of avoiding estimated congested sections and ensuring timely arrival at the designated location. The performance evaluation indicators include at least: delivery time T i , transportation cost C i , path reliability P i Traffic adaptability A i , Resource flexibility i ;

[0041] Evaluation Results i Calculated by the following formula:

[0042]

[0043] Among them, w 1 ,w 2 ,w 3 ,w 4 ,w 5 are the weight coefficients of delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility; a 1 is an exponential adjustment factor for delivery time, used to reflect the severity of delay; P max ,A max ,F max They are the maximum values ​​of path reliability, traffic adaptability, and resource flexibility, respectively, and are used for normalization so that the values ​​of each performance evaluation indicator are in the range of [0,1].

[0044] Optionally, the configuration scheme of the type and quantity of each first aid material is evaluated, and the evaluation index at least includes material utilization efficiency, rescue success rate and emergency response speed, and the distribution effect corresponding to the configuration scheme is obtained, including:

[0045] Based on the optimized parameters, P opt , in each predefined scenario S j The distribution of the emergency supplies is simulated, and the simulation process includes path selection R k , Time arrangement k and the allocation of emergency resources Q k , in order to obtain the configuration scheme A of the types and quantities of the emergency supplies in each predefined scenario j ;

[0046] Among them, each allocation scheme A j Expressed as:

[0047] A j = {R k ,T k ,Q k}

[0048] R k represents the kth transport route; T k represents the time schedule of the kth transport route; Q k represents the demand for the kth category of emergency supplies;

[0049] A configuration plan for the type and quantity of each of the first aid supplies j Carry out evaluation, and the evaluation index shall at least include material utilization efficiency E i , rescue success rate R i 、Emergency response speed V i ;

[0050] Distribution Effect V j Calculated by the following formula:

[0051]

[0052] w E ,w R ,w R are the weight coefficients of material utilization efficiency, rescue success rate, and emergency response speed; a E ,a R ,a V is the corresponding exponential adjustment factor; E k (Q k ) represents the allocation of emergency resources on the kth path Q kThe calculated material utilization efficiency represents the material utilization efficiency E i A specific instance under a specific path and resource allocation; R k (T k ) indicates that T is scheduled for a given time on the kth path k The rescue success rate represents the rescue success rate R i A specific instance with a specific path and schedule; min k T k Represents the shortest delivery time among all paths, used to measure the emergency response speed V i ; E max It is the maximum value of material utilization efficiency and is used for normalization processing.

[0053] In a second aspect, the embodiment of the present application provides an intelligent decision-making system for emergency deployment of emergency supplies, including:

[0054] A receiving module, used to receive demand information from multiple emergency response points, wherein the demand information at least includes the number of injured, the type of injured, the severity, the status of on-site medical resources, the priority identification and the geographical space distribution;

[0055] A generation module, for evaluating the urgency of the demand of each emergency response point based on the demand information and generating a preliminary deployment plan based on the graph neural network intelligent evaluation system, wherein the preliminary deployment plan is determined according to the urgency of the demand of each emergency response point and includes the type and quantity of first aid supplies required by each emergency response point, as well as the distribution order of the first aid supplies;

[0056] A determination module is used to generate a prediction result of road capacity based on historical traffic data and real-time traffic obtained from multiple sources using a deep learning prediction model, and simulate multiple logistics distribution strategies in combination with a reinforcement learning algorithm, refine and adjust the preliminary deployment plan, and determine a target logistics distribution strategy; wherein the prediction result of the road capacity refers to the deep learning prediction model using historical traffic data, real-time traffic data and model algorithms to predict the traffic conditions on the road in a certain period of time in the future, and the logistics distribution strategy is formulated according to the distribution order of the emergency supplies in the preliminary deployment plan; the target logistics distribution strategy can avoid the estimated congested sections and ensure timely arrival at the designated location;

[0057] The generation module is also used to start an AI simulation and exercise system based on Bayesian optimization and genetic algorithm, simulate the configuration schemes of the types and quantities of the first aid supplies in multiple scenarios according to the target logistics distribution strategy, and iteratively evolve the best configuration scheme of the types and quantities of the first aid supplies that can adapt to the actual situation from the configuration schemes in multiple scenarios through genetic algorithm, and generate a comprehensive deployment instruction, which includes a list of first aid supplies, an estimated delivery time and an optimal transportation route;

[0058] The execution module is used to execute the comprehensive deployment instruction to realize the emergency deployment of the emergency supplies.

[0059] In an embodiment of the present application, demand information is received from multiple emergency response points, and the demand information includes at least the number of injured, the type of injured, the severity, the status of on-site medical resources, the priority identification, and the geographic spatial distribution; based on the graph neural network intelligent evaluation system, the demand urgency of each emergency response point is evaluated according to the demand information, and a preliminary deployment plan is generated, and the preliminary deployment plan is determined according to the demand urgency of each emergency response point, including the type and quantity of first aid materials required by each emergency response point, and the distribution order of the first aid materials; using a deep learning prediction model, a road capacity prediction result is generated based on historical traffic data and real-time traffic obtained from multiple sources, and a variety of logistics distribution strategies are simulated in combination with a reinforcement learning algorithm, the preliminary deployment plan is refined and adjusted, and a target logistics distribution strategy is determined; wherein the road capacity prediction The result means that the deep learning prediction model uses historical traffic data, real-time traffic data and model algorithms to predict the traffic conditions on the road in the future. The logistics distribution strategy is formulated according to the distribution order of the first aid materials in the preliminary deployment plan; the target logistics distribution strategy can avoid the estimated congested sections and ensure timely arrival at the designated location; start the AI ​​simulation exercise system based on Bayesian optimization and genetic algorithm, simulate the configuration scheme of the types and quantities of the first aid materials in multiple scenarios according to the target logistics distribution strategy, and iteratively evolve the optimal configuration scheme of the types and quantities of the first aid materials that can adapt to the actual situation from the configuration schemes in multiple scenarios through genetic algorithms, and generate a comprehensive deployment instruction, which includes a list of first aid materials, estimated delivery time and optimal transportation route; execute the comprehensive deployment instruction to realize emergency deployment of the first aid materials.

[0060] The technical solution of this application has the following beneficial effects:

[0061] This application uses a graph neural network intelligent evaluation system to dynamically evaluate the urgency of the needs of each emergency response point and quickly generate a preliminary deployment plan. This method can determine the optimal resource allocation plan in the shortest time, significantly shorten the decision-making time, and improve the rescue efficiency. Based on demand information and on-site medical resource conditions, the types and quantities of emergency supplies required for each emergency response point are accurately calculated to ensure that the supplies will not be excessively wasted or insufficiently supplied, maximizing the efficiency of material utilization. The optimal configuration plan is iteratively evolved using Bayesian optimization and genetic algorithms, further improving the scientificity and rationality of material distribution. The deep learning prediction model combines historical and real-time traffic data to accurately estimate the road capacity in the future, providing a reliable basis for logistics distribution strategies. The reinforcement learning algorithm simulates a variety of logistics distribution strategies to ensure that the selected path not only avoids the estimated congested sections, but also saves transportation time and costs to the maximum extent, and improves the overall distribution efficiency. The AI ​​simulation and drill system automatically adjusts parameters through Bayesian optimization to improve the simulation accuracy; the genetic algorithm continuously adjusts and optimizes according to the actual situation, ensuring the high adaptability and flexibility of the plan. The comprehensive deployment instructions contain a detailed list of emergency supplies, estimated delivery time and optimal transportation routes, guide actual operations, and reduce the uncertainty caused by human factors. The entire system design fully considers the uncertain factors in emergencies, such as changes in traffic conditions and injuries, and enhances the ability to deal with complex scenarios. By simulating the distribution of emergency supplies in multiple predefined scenarios, the efficiency of material use, rescue success rate and emergency response speed are comprehensively evaluated to ensure efficient rescue in various situations, thereby improving the overall rescue success rate.

[0062] Furthermore, the embodiments of the present application also utilize road capacity prediction and reinforcement learning algorithms to optimize the distribution order of emergency supplies, and generate a variety of logistics distribution strategies aimed at reducing time costs and improving transportation efficiency; based on these strategies, the transportation routes are further optimized and the types and quantities of materials required for each emergency response point are adjusted to adapt to actual conditions; each strategy is performance evaluated, and the evaluation indicators cover delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility, and a target logistics distribution strategy that maximizes transportation efficiency and minimizes delays is selected; finally, the target logistics distribution strategy is applied to the adjusted preliminary deployment plan to verify its feasibility in actual operations.

[0063] Through the above method, not only can we dynamically adapt to changing road conditions, but we can also continuously optimize the logistics distribution strategy through reinforcement learning algorithms to ensure the optimization of material distribution order and transportation routes. The performance evaluation process comprehensively considers multiple key indicators to ensure that the selected path planning scheme is not only fast and efficient, but also has high reliability and adaptability. The target logistics distribution strategy finally selected has been verified and can achieve the expected results in actual operations, significantly improving the scientificity and accuracy of decision-making, enhancing the ability to respond to emergencies, and ensuring that emergency supplies can be efficiently delivered to the destination in the shortest time.

[0064] In summary, this method not only improves the speed and accuracy of emergency material allocation, but also optimizes resource allocation through intelligent technology and algorithms, enhances the adaptability and reliability of the system, and effectively improves emergency response capabilities and rescue success rate.

[0065] These and other aspects of the present application will become more clearly understood in the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0067] Figure 1 A flowchart of an intelligent decision-making method for emergency deployment of emergency supplies provided in an embodiment of the present application;

[0068] Figure 2 A schematic diagram of the structure of an intelligent decision-making system for emergency allocation of first aid supplies provided in an embodiment of the present application. DETAILED DESCRIPTION

[0069] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0070] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0071] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0072] Figure 1 A flowchart of an intelligent decision-making method for emergency allocation of emergency supplies is provided for the embodiment of the present application, such as Figure 1 As shown, the method includes:

[0073] 101. Receive demand information from multiple emergency response points.

[0074] The required information includes, but is not limited to, the number of injured, the type of injured (such as trauma, poisoning, etc.), the severity (mild, moderate, severe), the status of on-site medical resources (the number and capabilities of existing medicines, equipment, and medical staff), priority identification (emergency or non-emergency), and geographic spatial distribution (geographic location, accessibility). These data are used to construct a comprehensive description of the emergency scenario and are the basis for all subsequent decisions.

[0075] In actual operation, the system collects the above demand information through multiple channels, such as through telephone, mobile applications, IoT devices, etc. This information is uploaded to the central server in real time for processing and analysis, ensuring the timeliness and accuracy of the data, and providing solid data support for subsequent intelligent evaluation.

[0076] For example, in a city emergency system, when a traffic accident occurs, on-site rescuers use mobile devices to quickly input accident details, including the number of injured, injury classification, currently available medical resources, etc., and upload them to the system. At the same time, nearby hospitals and emergency stations also update their own resource status. After the information is aggregated, a complete description of the entire incident is formed, providing a basis for the next step of intelligent evaluation.

[0077] 102. Based on the graph neural network intelligent evaluation system, the urgency of the demand for each emergency response point is evaluated according to the demand information, and a preliminary deployment plan is generated.

[0078] Graph Neural Network (GNN) is a machine learning model that can process complex relational data and is suitable for analyzing the correlation and impact range between emergency response points. It models the relationship between response points and dynamically evaluates the urgency of each point, thereby providing scientific guidance for resource allocation.

[0079] The preliminary deployment plan is determined according to the urgency of the needs of each of the emergency response points, and includes the types and quantities of emergency supplies required by each of the emergency response points, as well as the distribution order of the emergency supplies.

[0080] Based on the demand information received, GNN takes into account factors such as the distance between different response points, traffic conditions, and sharing of medical resources, calculates the urgency of demand at each point, and generates a preliminary allocation plan including the type, quantity, and distribution order of materials.

[0081] For example, in the case of a traffic accident, GNN found that the two emergency stations A and B closest to the scene of the accident had relatively abundant resources at Station A and relatively tight resources at Station B. Therefore, the system decided to first deploy more resources from Station A to the accident site, and planned to consider mobilizing Station C, which is farther away but has more resources, if further support is needed, to ensure that the most needed resources can arrive as quickly as possible.

[0082] 103. Utilize a deep learning prediction model to generate predictions of road capacity based on historical traffic data and real-time traffic obtained from multiple sources, and combine with a reinforcement learning algorithm to simulate a variety of logistics distribution strategies, refine and adjust the preliminary deployment plan, and determine the target logistics distribution strategy.

[0083] The deep learning prediction model combines historical and real-time traffic data to estimate road capacity in the future and provide accurate road condition predictions for logistics distribution. The reinforcement learning algorithm simulates multiple distribution strategies to find the target logistics distribution strategy to ensure that the transportation of materials is fast and safe.

[0084] The prediction result of the road capacity refers to the deep learning prediction model using historical traffic data, real-time traffic data and model algorithms to estimate the traffic conditions on the road in the future. The logistics distribution strategy is formulated according to the distribution order of the emergency supplies in the preliminary deployment plan. The target logistics distribution strategy can avoid the estimated congested sections and ensure timely arrival at the designated location.

[0085] The model uses a variety of factors such as traffic flow, weather forecasts, and construction information to predict possible traffic bottlenecks and adjust the delivery routes in the preliminary dispatch plan accordingly. Through continuous iterative optimization, the best route to avoid congested sections is finally determined.

[0086] For example, in the aforementioned traffic accident, the system predicted that the main road leading to Station A would be severely congested due to another traffic accident in half an hour. Therefore, AI suggested changing to a smaller road that was farther away but expected to be unobstructed, ensuring that emergency supplies could arrive at the accident site smoothly within the specified time and avoiding delays in treatment.

[0087] 104. Initiate an AI simulation and exercise system based on Bayesian optimization and genetic algorithm, simulate the configuration schemes of the types and quantities of the first aid supplies in multiple scenarios according to the target logistics distribution strategy, and iteratively evolve the optimal configuration scheme of the types and quantities of the first aid supplies that can adapt to the actual situation from the configuration schemes in multiple scenarios through genetic algorithm, and generate comprehensive deployment instructions.

[0088] Bayesian optimization and genetic algorithms work together in the AI ​​simulation and exercise system. The former automatically adjusts parameters to improve simulation accuracy, while the latter finds the best configuration solution that adapts to the actual situation through iterative evolution. The system is designed to verify and optimize the preliminary deployment plan to ensure that it can be effectively executed under various conditions.

[0089] Based on the target logistics distribution strategy obtained in the previous step, the AI ​​simulation system simulates multiple possible scenarios, such as different time periods, weather changes, emergencies, etc., evaluates the efficiency of material use, rescue success rate and emergency response speed, and finally generates a comprehensive deployment instruction containing a detailed list, delivery time and route.

[0090] For example, taking the above traffic accident as an example, the AI ​​simulation system took into account the impact of various situations, such as poor visibility at night and slippery roads in rainy days, and finally confirmed the best material distribution plan. The plan not only specifies a specific list of materials and the estimated delivery time, but also plans an alternative route that can ensure efficient transportation even in severe weather conditions, ensuring that the rescue operation is foolproof.

[0091] It should be noted that although the target logistics distribution strategy determines the optimal route for material distribution, it does not fully take into account the various complex situations that may be encountered in the actual distribution process, such as resource utilization efficiency, treatment success rate, adaptability of material types and quantities in different scenarios, etc. Therefore, after generating the target logistics distribution strategy, it is necessary to further conduct multi-scenario simulations through the AI ​​simulation and exercise system to ensure that the final comprehensive deployment instructions are the most optimized and most adaptable to the actual situation.

[0092] 105. Execute the comprehensive deployment instruction to realize the emergency deployment of the emergency supplies.

[0093] Among them, the comprehensive deployment order is the final decision-making document after multiple rounds of evaluation and optimization, which contains a detailed list of emergency supplies, estimated delivery time and optimal transportation route. These orders directly guide on-site operations to ensure the efficiency and accuracy of resource deployment.

[0094] Once the comprehensive deployment order is generated, it is immediately communicated to relevant parties, such as emergency vehicle drivers and material management personnel, to ensure that everyone clearly understands the mission details. During the execution process, the system continuously monitors the progress and makes real-time adjustments when necessary to ensure the smooth completion of the rescue operation.

[0095] For example, in traffic accident rescue, comprehensive deployment instructions are quickly conveyed to all units involved in the rescue. Emergency vehicles set out along the scheduled route, keep in touch with the command center on the way, and report their locations and problems at any time. The system dynamically adjusts the deployment of other resources based on the latest feedback to ensure that the entire rescue operation is carried out in an orderly manner until all the injured are properly treated.

[0096] Through the implementation of steps 101 to 105, the intelligent decision-making method for emergency allocation of first aid materials realizes the intelligent management of the whole process from comprehensive collection of demand information, intelligent evaluation and generation of preliminary plans, traffic forecasting and optimization of logistics distribution strategies, AI simulation exercises to the final execution of comprehensive allocation instructions. First, the system receives detailed demand information from multiple emergency response points, ensuring the accuracy and real-time nature of the data, laying a solid foundation for subsequent decision-making; then, based on the graph neural network, the demand urgency of each response point is dynamically evaluated, and a scientific and reasonable preliminary allocation plan is generated, which improves the pertinence and timeliness of resource allocation; then, the deep learning prediction model is used to combine historical and real-time traffic data to estimate road traffic capacity, and a variety of logistics distribution strategies are simulated through reinforcement learning algorithms to determine the target logistics distribution strategy to ensure that materials can be delivered quickly and safely; further, the AI ​​simulation exercise system based on Bayesian optimization and genetic algorithms is started to simulate the material distribution effects under different scenarios, optimize and verify the preliminary allocation plan, and generate a comprehensive allocation instruction containing a detailed list, estimated delivery time and optimal transportation route; finally, these instructions are executed to achieve efficient allocation of first aid materials. The entire process not only significantly improved the speed and accuracy of emergency response, but also optimized resource allocation efficiency, enhanced the adaptability and reliability of the system, thereby greatly improving the rescue success rate and overall emergency handling capabilities.

[0097] In order to solve the prediction challenges brought by the diversity and complexity of traffic data and further improve the accuracy and real-time performance of road capacity prediction, in some embodiments, the deep learning prediction model described in step 103 is used to generate the prediction results of road capacity based on the historical traffic data and real-time traffic obtained from multiple sources, including:

[0098] Historical traffic data and real-time traffic data obtained from multiple sources are integrated to obtain a traffic data set, wherein the historical traffic data at least includes traffic flow, accident records and weather impacts in the same time period in the past, and the real-time traffic data at least includes current traffic camera images, vehicle location information and instant traffic conditions provided by sensors; the traffic data in the traffic data set is analyzed and processed using a deep learning prediction model to generate a prediction result of road capacity in the future. The deep learning prediction model can identify and estimate the impact of different factors on road capacity through training, and output the expected traffic efficiency of each road in a specific time period in the future to generate a prediction result of road capacity.

[0099] In this embodiment, historical traffic data at least includes information such as traffic flow, accident records, and weather impacts in the same time period in the past, which are used to identify long-term trends and periodic patterns. Real-time traffic data at least includes current traffic camera images, vehicle location information (such as from GPS or on-board sensors), and instant traffic conditions provided by sensors (such as traffic light status, road slipperiness, etc.), which are used to capture instantaneous changes and emergencies.

[0100] In the embodiment of the present application, first, the system cleans, aligns and standardizes data from different sources through multi-source data fusion technology to ensure data consistency and availability. Then, the integrated traffic data set is analyzed and processed using a deep learning prediction model. Through a large amount of training, the model can identify and estimate the impact of different factors (such as time, location, weather conditions, traffic accidents, holiday effects, etc.) on road capacity, and output the estimated traffic efficiency of each road in a specific time period in the future. The prediction model not only takes into account static road infrastructure information, but also dynamically adjusts to adapt to the ever-changing traffic environment.

[0101] Here is a specific example:

[0102] In a case of emergency material allocation in a large city, the system needs to formulate a target logistics distribution strategy based on the upcoming weekday evening rush hour. To achieve this goal, the system collects and integrates traffic flow data, accident records, and weather conditions during the same weekday evening rush hour in the past few months as historical data. At the same time, it obtains real-time traffic camera images, vehicle GPS location information, and instant traffic conditions fed back by roadside sensors. These data are input into a fully trained deep learning prediction model.

[0103] The model predicts the traffic efficiency of major roads and intersections in the next two hours based on historical and real-time data. For example, the model identifies that a major road leading to an emergency station may be temporarily congested in the next hour due to the dismissal of a nearby school, but will then return to normal. Therefore, the system recommends that emergency material transport vehicles choose a side road that is slightly farther but expected to be smoother as an alternative route, thereby ensuring that the materials can reach their destination in the shortest time.

[0104] In addition, the model also takes into account possible emergencies, such as temporary traffic control or unexpected accidents, and automatically adjusts the prediction results to provide alternative route options. This not only improves the accuracy of road capacity prediction, but also enhances the system's flexibility to ensure that the best logistics distribution decisions can be made in any situation.

[0105] In order to solve the problem of staticness of the preliminary deployment plan and its lack of adaptability to the actual situation, and to further improve the flexibility and efficiency of the logistics distribution strategy, in some embodiments, in step 103, according to the prediction result of the road capacity, a plurality of logistics distribution strategies are simulated in combination with a reinforcement learning algorithm, the preliminary deployment plan is refined and adjusted, and a target logistics distribution strategy is determined, including:

[0106] Using the predicted results of the road capacity and combining with the reinforcement learning algorithm, the distribution order of the emergency supplies in the preliminary allocation plan is simulated to obtain a variety of logistics distribution strategies aimed at optimizing time cost and transportation efficiency; based on a variety of logistics distribution strategies, the preliminary allocation plan is refined and adjusted, and the refinement and adjustment process includes optimizing the transportation routes of the emergency supplies under different logistics distribution strategies and adjusting the types and quantities of emergency supplies required for each emergency response point to adapt to the actual situation; for each of the logistics distribution strategies, a performance evaluation is performed in terms of avoiding estimated congested sections and ensuring timely arrival at the designated location. The performance evaluation indicators include at least delivery time, transportation cost, path reliability, traffic adaptability and resource flexibility, and the evaluation results of each of the logistics distribution strategies are obtained; based on the evaluation results of the effects of all logistics distribution strategies, a target logistics distribution strategy that can maximize transportation efficiency and minimize delays is selected; the target logistics distribution strategy is applied to the adjusted preliminary allocation plan for verification and confirmation to ensure that the target logistics distribution strategy can achieve the expected effect in actual operation.

[0107] In this embodiment, the prediction results of road capacity are combined with reinforcement learning algorithms to simulate the order of emergency material distribution in the preliminary deployment plan. Here, the prediction results of road capacity are generated based on historical and real-time traffic data, providing the estimated traffic efficiency of each road in the future. The reinforcement learning algorithm learns the best action sequence through trial and error, and evaluates the effects of different distribution strategies in a simulated environment, aiming to optimize time cost and transportation efficiency.

[0108] The detailed adjustment process includes optimizing the emergency material transportation routes under different logistics distribution strategies and adjusting the types and quantities of emergency materials required for each emergency response point. This not only takes into account the choice of routes, but also dynamically adjusts resource allocation based on the latest demand information to ensure efficient operation in a changing environment.

[0109] Performance evaluation indicators include at least delivery time, transportation cost, route reliability, traffic adaptability, and resource flexibility. These indicators are used to comprehensively measure the performance of each logistics distribution strategy to ensure that the selected solution is not only fast but also reliable and can flexibly respond to emergencies.

[0110] In the embodiment of the present application, the system first uses the road capacity prediction results and combines them with the reinforcement learning algorithm to simulate multiple possible logistics distribution strategies. In this process, the algorithm will try different allocation orders, path selection and resource adjustment combinations to find a strategy that can complete the distribution in the shortest time and at the lowest cost. Subsequently, the system will conduct a detailed performance evaluation of each strategy, comprehensively considering multiple dimensions such as delivery time, transportation cost, and path reliability, and finally determine one or more candidate target logistics distribution strategies.

[0111] In order to verify the actual feasibility of the selected solution, the system will apply it to the adjusted preliminary deployment plan and confirm its effect through simulation exercises or other means. If the simulation results show that the solution can meet the expected goals, the solution will be officially adopted; otherwise, the system will continue to iterate and optimize until the optimal solution is found.

[0112] Here is a specific example:

[0113] In a case study of emergency material distribution in a city, it is assumed that the initial distribution plan has specified a plan to deliver materials from three different warehouses to four emergency sites. However, as traffic conditions change (such as temporary congestion on certain roads), the original route may no longer be the optimal choice.

[0114] At this point, the system starts the reinforcement learning algorithm and combines the latest road capacity prediction results to simulate a variety of possible logistics distribution strategies. For example, one strategy may be to give priority to a side road that is slightly farther away but expected to be more unobstructed, while another strategy is to use the main road during non-peak hours. The system will also consider whether some delivery tasks can be merged to reduce the total driving distance and time.

[0115] Next, the system conducts a comprehensive evaluation of each strategy. For delivery time, the system checks whether all materials can arrive within the specified time; for transportation costs, it calculates expenses such as fuel consumption and vehicle wear and tear; for path reliability, it evaluates possible obstacles or risks on the path; for traffic adaptability, it examines the strategy's ability to adapt to different traffic conditions; and finally, for resource flexibility, it ensures that it can be flexibly adjusted even when demand changes.

[0116] After multiple rounds of simulation and evaluation, the system finally selected a strategy that can minimize delays while maximizing transportation efficiency as the target logistics distribution strategy. This solution not only avoids the estimated congested sections, but also reduces the total driving time and cost by optimizing the distribution sequence. In addition, the system also conducted a pre-operation verification to ensure that the solution can achieve the expected results in a real environment, thereby significantly improving the efficiency and reliability of the entire emergency material allocation process.

[0117] In order to solve the multi-objective optimization problem in the selection of logistics distribution strategies and further improve the scientificity and reliability of path planning, in some embodiments, the target logistics distribution strategy that can maximize transportation efficiency and minimize delays is selected based on the evaluation results of all logistics distribution strategy effects, including:

[0118] According to the evaluation results of all logistics distribution strategies, eligible logistics distribution strategies are screened out, and based on multi-dimensional evaluation, the screened logistics distribution strategies are comprehensively scored to obtain the score of each logistics distribution strategy, and one or several logistics distribution strategies with the highest score are selected as candidate logistics distribution strategies; the selection of each route segment, the expected traffic conditions and the risk of delay in each candidate logistics distribution strategy are analyzed to determine the target logistics distribution strategy that can maximize transportation efficiency and minimize delays

[0119] In this embodiment, qualified logistics distribution strategies are those that excel in delivery time, transportation cost, route reliability, traffic adaptability, resource flexibility, etc. Multi-dimensional evaluation involves multiple key performance indicators (KPIs), each with a corresponding weight and scoring criteria, which are used to comprehensively measure the effectiveness of each strategy.

[0120] The scoring mechanism ensures the comparability between different strategies, while selecting multiple high-scoring strategies as candidates provides more flexibility and selection space for subsequent analysis.

[0121] Analyze the path planning information contained in the candidate logistics distribution strategy, including the choice of each route, expected traffic conditions, possible delay risks, and integrate this path planning information into a specific path planning plan. This step deeply examines the specific details of each candidate strategy to ensure that the final selected path is not only excellent in theory, but also feasible in practice.

[0122] The path planning scheme that can maximize transportation efficiency and minimize delays is selected as the target logistics distribution strategy. Through detailed analysis and comparison, the system determines one or more target logistics distribution strategies that can achieve the best transportation efficiency and minimize delays while satisfying all constraints.

[0123] In the embodiment of the present application, first, the system preliminarily screens all logistics distribution strategies according to the pre-set multi-dimensional evaluation indicators, and removes the options that obviously do not meet the requirements. Then, the remaining strategies are quantitatively evaluated using a weighted scoring model, where the weight of each indicator reflects its importance in the overall evaluation. The scoring model may consider the following factors:

[0124] Delivery time: The shorter the better, and the higher the weight.

[0125] Shipping cost: The lower the cost, the better, but timeliness cannot be sacrificed too much.

[0126] Path reliability: Avoid high-risk sections and ensure the safe arrival of supplies.

[0127] Traffic adaptability: the ability to flexibly respond to different traffic conditions.

[0128] Resource flexibility: Ability to adjust quickly to changing needs.

[0129] Next, the system selects several high-scoring strategies as candidate logistics distribution strategies. For each candidate strategy, the system analyzes its path planning information in detail, including but not limited to:

[0130] Selection of each route segment: selection of specific roads and their characteristics.

[0131] Predicted traffic conditions: Estimation of future traffic flow based on predictive models.

[0132] Possible risk of delay: Identify potential risk points and assess their impact.

[0133] Finally, the system integrates the above information into a specific route planning plan and confirms its practical feasibility through simulation exercises or other verification methods. On this basis, it selects the target logistics distribution strategy that can maximize transportation efficiency and minimize delays.

[0134] In order to solve the uncertainty and complexity of the target logistics distribution strategy in actual operation and further improve the adaptability and reliability of the distribution of emergency supplies, in some embodiments, an AI simulation exercise system based on Bayesian optimization and genetic algorithm is started in step 104, and according to the target logistics distribution strategy, the configuration schemes of the types and quantities of the emergency supplies in multiple scenarios are simulated, and the optimal configuration scheme of the types and quantities of the emergency supplies that can adapt to the actual situation is iteratively evolved from the configuration schemes in multiple scenarios through genetic algorithms, and a comprehensive deployment instruction is generated, including:

[0135] The AI ​​simulation and exercise system based on Bayesian optimization and genetic algorithm is started, the target logistics distribution strategy is loaded as input data, and the simulation environment is initialized; the model parameters of the AI ​​simulation and exercise system are automatically adjusted using Bayesian optimization technology to improve the simulation accuracy and obtain the optimized parameter configuration; based on the optimized parameter configuration, the configuration schemes of the types and quantities of the first aid supplies are simulated in multiple predefined scenarios, and the allocation effect corresponding to the configuration scheme in each predefined scenario is evaluated; through the genetic algorithm, the allocation effect corresponding to the configuration scheme in each predefined scenario is iteratively evolved to determine the optimal configuration scheme of the types and quantities of the first aid supplies that can adapt to various actual situations; the effectiveness of the optimal configuration scheme is verified, and a comprehensive allocation instruction is generated including a detailed list of first aid supplies, an estimated delivery time, and an optimal transportation route.

[0136] In this embodiment, the target logistics distribution strategy is determined by the previous step (such as step 103), including detailed information such as the transportation route, the estimated delivery time, and the order of material distribution. The simulation environment is a virtual platform that can reproduce factors such as traffic conditions, weather changes, and emergencies in different scenarios, and is used to test and verify the actual effect of the target logistics distribution strategy.

[0137] Bayesian optimization is an efficient global optimization method that is particularly suitable for parameter tuning problems in high-dimensional spaces. It builds a proxy model to predict the relationship between parameters and performance, and dynamically adjusts parameters based on historical evaluation results to find the best configuration. This helps ensure that the output of the simulation system is as close to the actual situation as possible.

[0138] Multiple predefined scenarios can cover different time periods, weather conditions, traffic flows, and possible emergencies such as road closures or new emergencies. Each simulation generates a detailed report on the effectiveness of material distribution, including key indicators such as material utilization efficiency, rescue success rate, and emergency response speed.

[0139] Genetic algorithms imitate the natural selection process, continuously optimizing the solution set through operations such as selection, crossover, and mutation, and eventually evolving a configuration solution with strong adaptability and robustness. This step ensures that no matter what kind of emergency situation is encountered, the system can quickly adjust and provide the best resource allocation suggestions.

[0140] The system will conduct multiple rounds of verification on the selected optimal configuration plan to ensure that it can achieve the expected effect in various predefined scenarios. Once the verification is passed, the system will generate the final comprehensive deployment instructions to guide the actual operators to complete the material deployment task.

[0141] In the embodiment of the present application, first, the system loads the target logistics distribution strategy, initializes the simulation environment, and sets the initial parameter values. Then, the Bayesian optimization technology is used to automatically adjust the model parameters to gradually improve the simulation accuracy. Next, based on the optimized parameter configuration, the system simulates the distribution process of emergency supplies in multiple predefined scenarios and records the distribution effect in each scenario. Subsequently, these effects are iteratively evolved through genetic algorithms to find the optimal configuration solution. Finally, the system verifies the effectiveness of the optimal configuration solution and generates a comprehensive deployment instruction containing a detailed list of emergency supplies, estimated delivery time, and optimal transportation route.

[0142] In order to solve the difference between the simulation environment and the actual situation and further improve the accuracy and adaptability of the configuration scheme of the types and quantities of the first aid supplies, based on the above embodiments, another embodiment is provided, which simulates the configuration scheme of the types and quantities of the first aid supplies in multiple predefined scenarios based on the optimized parameter configuration, and evaluates the distribution effect corresponding to the configuration scheme in each predefined scenario, including:

[0143] In the AI ​​simulation and drill system, the optimized parameter configuration is loaded as the initial setting, and the simulation environment is initialized to ensure that the simulation environment can accurately reflect the actual situation; based on the optimized parameter configuration, the distribution of the first aid supplies is simulated in each predefined scenario, and the simulation process includes path selection, time arrangement and distribution of the first aid resources, so as to obtain the configuration scheme of the type and quantity of the first aid supplies in each predefined scenario; the configuration scheme of each type and quantity of the first aid supplies is evaluated, and the evaluation indicators include at least delivery time, transportation cost, material utilization efficiency, rescue success rate and emergency response speed, so as to obtain the distribution effect corresponding to the configuration scheme.

[0144] The optimized parameter configuration is the best model parameters adjusted by Bayesian optimization technology. These parameters are used to simulate key factors in the environment, such as vehicle speed, loading capacity, traffic flow, etc. Initializing the simulation environment means setting the initial state and conditions of the simulation system to make it as close to the actual operating environment as possible. This step is crucial to ensure the validity and reliability of subsequent simulation results.

[0145] Path selection involves determining the specific route from the warehouse to each emergency response point; time scheduling takes into account the time required for loading, transporting and unloading materials; and the allocation of emergency resources covers the types and quantities of materials required at different sites. By simulating these processes in detail, the system can generate specific allocation plans for each predefined scenario.

[0146] Evaluate the allocation scheme of the type and quantity of each of the first aid supplies, and the evaluation indicators include at least delivery time, transportation cost, material utilization efficiency, rescue success rate and emergency response speed, and obtain the allocation effect corresponding to the allocation scheme. These evaluation indicators are used to comprehensively measure the performance of each allocation scheme to ensure that the final selected scheme is not only fast but also efficient. For example, delivery time reflects whether the supplies can reach the destination on time; transportation cost takes into account factors such as fuel consumption and vehicle wear and tear; material utilization efficiency measures whether the supplies have been fully utilized; rescue success rate evaluates the impact of material distribution on the treatment of the wounded; and emergency response speed focuses on the system's ability to respond to emergencies.

[0147] In the embodiment of the present application, first, the system loads the optimal parameter configuration adjusted by Bayesian optimization technology and initializes the simulation environment. This step ensures that the basic conditions of the simulation environment (such as traffic conditions, weather forecasts, etc.) are as close to the actual situation as possible, thereby improving the credibility of the simulation results. Then, the system simulates the distribution of emergency supplies in multiple predefined scenarios according to the optimized parameter configuration. Each simulation includes detailed path selection, time arrangement, and distribution of emergency resources to generate a specific and feasible distribution plan.

[0148] Next, the system conducts a comprehensive evaluation of each generated allocation plan. During the evaluation process, the system will comprehensively consider multiple key performance indicators such as delivery time, transportation cost, material utilization efficiency, rescue success rate, and emergency response speed. Each indicator has a corresponding weight and scoring standard to ensure the objectivity and scientificity of the evaluation results. For example, if a certain allocation plan has the shortest delivery time, but the transportation cost is too high or the material utilization efficiency is inefficient, its overall score may be affected.

[0149] Finally, the system uses the evaluation results to determine the distribution effect of emergency supplies in each predefined scenario. This process not only helps identify the optimal distribution plan, but also provides valuable data support for subsequent decision-making, ensuring that the expected results can be achieved in actual operations.

[0150] Here is a specific example:

[0151] In a city emergency supplies allocation case, it is assumed that the system has adjusted the model parameters through Bayesian optimization technology, initialized the simulation environment, and set basic conditions such as traffic flow and weather conditions in the current period.

[0152] The system loads the optimized parameter configuration and sets the basic parameters of the simulation environment to ensure that the simulation environment can accurately reflect the actual situation. For example, considering that the main road is under construction during the current period, the system lowers the traffic efficiency of the road section to simulate the real traffic conditions.

[0153] The system simulates the distribution of emergency supplies in multiple predefined scenarios. For example, in one scenario, it simulates traffic congestion during rush hour during the day; in another scenario, it simulates unobstructed roads at night. Each simulation includes detailed path selection (such as choosing a main road or a side road), time arrangement (such as estimated departure time and arrival time), and allocation of emergency resources (such as the number of medicines and equipment required at each station). In this way, the system generates a specific allocation plan for each predefined scenario.

[0154] The system conducts a comprehensive evaluation of each generated allocation plan. Evaluation indicators include but are not limited to delivery time, transportation cost, material utilization efficiency, rescue success rate, and emergency response speed. For example, in one simulation, the system found that a certain path was farther but expected to be more unobstructed, so although the total driving distance increased, the delivery time was greatly shortened and the transportation cost was lower. At the same time, the system also evaluated the material utilization efficiency and rescue success rate on the path to ensure that the selected plan can achieve the best balance in all aspects.

[0155] Finally, the system uses the evaluation results to determine the distribution effect of emergency supplies in each predefined scenario. For example, in the simulation of the daytime rush hour, the system selected a small road that avoids the main road as the distribution route, ensuring that the supplies can arrive at the destination smoothly within the specified time while reducing transportation costs. In the night simulation, the system took advantage of the smooth flow of the main road to further optimize the distribution sequence and improve overall efficiency.

[0156] In this way, the system not only improves the accuracy and reliability of the simulation results, but also enhances its ability to adapt to various actual situations, ensuring that emergency supplies can be efficiently delivered to the destination in the shortest time, significantly improving the efficiency and reliability of the entire emergency supplies allocation process.

[0157] In order to solve the complexity and diversity of demand information analysis and further improve the accuracy and scientificity of emergency response point demand urgency assessment, in some embodiments, the graph neural network-based intelligent assessment system in step 102 assesses the demand urgency of each emergency response point according to the demand information and generates a preliminary deployment plan, including:

[0158] Utilize a natural language processing algorithm to parse demand information from multiple emergency response points and obtain specific demand results for each emergency response point; construct a graph structure data model based on the specific demand results, wherein the nodes in the graph structure data model represent the emergency response points, and the edges represent the correlation and impact range between different emergency response points; based on pre-trained model parameters, start a graph neural network intelligent evaluation system to load the graph structure data model to ensure that the graph neural network intelligent evaluation system can identify and process different types of node and edge attributes, and utilize the graph neural network intelligent evaluation system to evaluate the demand urgency of each emergency response point and calculate the demand urgency score of each node, wherein the demand urgency score is determined based on the correlation and impact range between different emergency response points; generate a preliminary deployment plan based on the demand urgency score.

[0159] In this embodiment, the demand information can be a text report, voice recording, or structured data form. NLP algorithms are used to extract and understand key content in this information, such as the number, type, severity, on-site medical resource status, priority identification, and geographic distribution. In this way, the system can convert unstructured input into structured specific demand results, providing a basis for subsequent evaluation.

[0160] The graph data model is a mathematical abstraction that represents complex real-world relationships in the form of nodes (emergency response points) and edges (relevance and scope of influence). Node attributes may include location, demand urgency score, etc.; edge attributes reflect factors such as the distance between different response points, traffic conditions, and resource sharing. This modeling method helps to intuitively display the interactions between response points and provides structured data support for intelligent evaluation.

[0161] Graph Neural Network (GNN) is a deep learning model specifically designed to process graph structured data. It can capture the complex relationships between nodes, thereby more accurately assessing demand urgency. The pre-trained model parameters enable the system to have good generalization capabilities when facing different types of data, and the demand urgency score is the result of comprehensive consideration of factors such as the correlation between different response points, the scope of influence, and the priority identification.

[0162] The preliminary deployment plan specifies the types and quantities of emergency supplies required at each emergency response point, as well as the priority of material distribution. This process not only takes into account the needs of a single response point, but also takes into account the optimal allocation of resources within the entire network to ensure that the most urgently needed resources can reach their destination as quickly as possible.

[0163] Here is a specific example:

[0164] In a case of emergency material allocation in a city, suppose the system receives demand information from multiple emergency response points, which includes text descriptions, voice recordings, table data, etc. In order to effectively process this information, the system first uses the NLP algorithm to parse and process it, and extract the specific demand results of each emergency response point, such as 5 seriously injured patients who need immediate surgery at a certain location, and 3 slightly injured patients waiting for bandages at another location.

[0165] The system uses NLP algorithms to analyze the above information and identify key factors such as the number, type, and severity of the specific injuries. For example, for the text description "5 seriously injured patients", the system will automatically interpret it as 5 seriously injured people; for the voice recording "need more bandages", the system will convert it into a demand for bandages.

[0166] The system builds a graph data model based on the specific demand results after analysis. In this model, each emergency response point is represented as a node, and the node attributes include location, demand urgency score, etc. The correlation and influence range between different response points are connected as edges, and the edge attributes reflect factors such as traffic conditions and resource sharing. For example, if there is a fast channel between two adjacent stations, the edge weight between them is higher.

[0167] The system loads a pre-trained graph neural network intelligent evaluation system to process the graph structure data model. The system is able to identify and process different types of node and edge attributes, evaluate the urgency of each emergency response point, and calculate the urgency of the need score. For example, a response point close to a hospital with convenient transportation may receive a higher score because it is easier to get additional support.

[0168] Based on the urgency score, the system generates a preliminary deployment plan, which clarifies the types and quantities of emergency supplies required at each emergency response point, as well as the priority of material distribution. For example, the system decides to first deploy more medicines and equipment to the highest-scoring response point to ensure that the most urgently needed resources can arrive as quickly as possible.

[0169] In this way, the system not only improves the efficiency and accuracy of demand information analysis, but also enhances the scientificity and reliability of the assessment of the urgency of emergency response points, ensuring that emergency supplies can be efficiently deployed in the shortest time, significantly improving the efficiency and effectiveness of the entire emergency supplies deployment process.

[0170] This application considers that the evaluation of logistics distribution strategies needs to comprehensively consider multiple key performance indicators (KPIs) to ensure that the selected path planning scheme can not only avoid estimated congested sections and arrive at the designated location in time, but also achieve the best in terms of transportation cost, path reliability, traffic adaptability and resource flexibility. The traditional single indicator evaluation method is difficult to fully reflect the complex and changing actual needs, so a new optional solution is proposed, which includes:

[0171] For each of the logistics distribution strategies, a performance evaluation process is performed in terms of avoiding estimated congested sections and ensuring timely arrival at the designated location. The performance evaluation indicators include at least delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility. The evaluation results of each of the logistics distribution strategies are obtained, including:

[0172] For each of the logistics distribution strategies S i ,Performance evaluation is performed in terms of avoiding estimated congested sections and ensuring timely arrival at the designated location. The performance evaluation indicators include at least: delivery time T i , transportation cost C i , path reliability P iTraffic adaptability A i , Resource flexibility i ;

[0173] Evaluation Results i Calculated by the following formula:

[0174]

[0175] Among them, w 1 ,w 2 ,w 3 ,w 4 ,w 5 are the weight coefficients of delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility; a 1 is an exponential adjustment factor for delivery time, used to reflect the severity of delay; P max ,A max ,F max They are the maximum values ​​of path reliability, traffic adaptability, and resource flexibility, respectively, and are used for normalization so that the values ​​of each performance evaluation indicator are in the range of [0,1].

[0176] The following is a detailed explanation of each parameter:

[0177] V i : The comprehensive evaluation score of the i-th logistics distribution strategy, which is used to measure the overall performance of the strategy. It is calculated by comprehensively considering multiple key performance indicators (such as delivery time, transportation cost, path reliability, etc.) and based on preset weight coefficients and index adjustment factors.

[0178] w 1 ,w 2 ,w 3 ,w 4 ,w 5 : are the weight coefficients of delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility. These weights reflect the relative importance of different indicators in the overall evaluation and can be adjusted according to actual conditions. They are determined by expert systems or historical data analysis. For example, initial weights can be set based on past successful cases or expert opinions, and dynamically adjusted through feedback from actual operations.

[0179] a 1 : An exponential adjustment factor for delivery time, used to reflect the severity of the delay. Emphasize the impact of long delays on the evaluation results. Determine through experimental data fitting or simulation. Usually, by analyzing the impact of different delay times on the rescue effect, the appropriate exponential adjustment factor value can be found. Obtain the actual driving time of each route through the GPS tracking system or simulation. It is also possible to combine real-time traffic data to predict the future delivery time.

[0180] T i : The delivery time (hours) of the i-th logistics distribution strategy, the shorter the better. Use an exponential function to amplify the impact of long delays and ensure the importance of fast delivery. Obtain the actual driving time of each route through the GPS tracking system or simulation. It is also possible to combine real-time traffic data to predict the future delivery time.

[0181] C i : The transportation cost (yuan) of the i-th logistics distribution strategy, the lower the better. It directly linearly affects the evaluation score to ensure cost-effectiveness. Obtain through the quotation list of logistics companies or the internal cost accounting system. It is also possible to consider factors such as fuel consumption and vehicle wear for a detailed cost analysis.

[0182] P i : The path reliability of the i-th logistics distribution strategy, with a value range of [0,1], the higher the better. After normalization, compare with the maximum value P max and calculate the insufficient part of the path reliability. Obtain through historical data statistics and real-time monitoring. For example, the success rate of past similar paths can be analyzed, or the reliability of the current path can be evaluated through real-time traffic conditions and weather forecasts (the calculation method can be designed according to requirements, such as calculating the path reliability by weighted summing the above information).

[0183] A i : The traffic adaptability of the i-th logistics distribution strategy, with a value range of [0,1], the higher the better. After normalization, compare with the maximum value A max and calculate the insufficient part of the traffic adaptability. Obtain through the traffic flow prediction model and real-time traffic data. For example, the adaptability of the path to different traffic conditions can be evaluated using real-time road condition information provided by traffic cameras, sensor data, or navigation software (the calculation method can be designed according to requirements, such as calculating the traffic adaptability by weighted summing each real-time road condition information).

[0184] F i : The resource flexibility of the i-th logistics distribution strategy, with a value range of [0,1], the higher the better. After normalization, compare with the maximum value F maxCompare and calculate the deficiencies of resource flexibility. Obtained through the material management system and emergency response database. For example, the types, quantities and distribution of existing resources can be analyzed to evaluate the flexibility and response speed of resource allocation (the calculation method can be designed according to demand, such as calculating resource flexibility by weighted summation of the above information).

[0185] The following is an introduction to the design reasons of each sub-item:

[0186] Delivery time is one of the key factors, especially in the allocation of emergency supplies, time is of the essence. Use exponential function The impact of long delays can be magnified, ensuring the importance of fast delivery. 1 Used to adjust the sensitivity of time delay.

[0187] w 2 ·C i : Transportation costs directly affect the economics of the entire delivery solution. They are directly added to the evaluation score in a linear manner to ensure cost-effectiveness. Lower costs help improve the overall score.

[0188] Path reliability reflects the safety and stability of the path. After normalization, the insufficient part of the path reliability is calculated to ensure that the selected path is safe and reliable enough. Higher path reliability means less risk and higher success rate.

[0189] Traffic adaptability takes into account the adaptability of the path to different traffic conditions. After normalization, the insufficient part of traffic adaptability is calculated to ensure that the selected path can flexibly respond to various traffic changes and reduce the occurrence of unexpected situations.

[0190] Resource flexibility reflects the flexibility and response speed of the system in resource allocation. After normalization, the insufficient part of resource flexibility is calculated to ensure that resources can be quickly deployed when needed, thus improving the overall response capability of the system.

[0191] The purpose of adding up the sub-items is to comprehensively consider multiple key performance indicators (KPIs) to ensure that the evaluation results are comprehensive and scientific. Each sub-item represents an important evaluation dimension, and different weight coefficients w are assigned to each sub-item. 1 ,w 2 ,w 3 ,w 4 ,w 5, the importance of different indicators can be reflected in the overall evaluation. Specifically: by adding up the various sub-items, the performance of each logistics distribution strategy in terms of delivery time, transportation cost, route reliability, traffic adaptability and resource flexibility can be comprehensively evaluated.

[0192] Weight adjustment: The weight coefficient allows the importance of different indicators to be adjusted according to actual needs. For example, in an emergency, delivery time may be more critical, which can be adjusted by increasing w 1 The final score V i It comprehensively reflects the performance of all key performance indicators to ensure that the selected solution is not only excellent in one aspect, but also achieves the best balance in multiple aspects.

[0193] This comprehensive evaluation method ensures that the selection of logistics distribution strategies is more scientific and reasonable, improves the accuracy and reliability of decision-making, and can better cope with various challenges, especially in complex and changing actual environments.

[0194] Here is a specific example:

[0195] In a case study of emergency supplies allocation in a city, suppose the system generates three different logistics distribution strategies S 1 ,S 2 ,S 3 ,The specific data are shown in Table 1 below:

[0196] Table 1

[0197]

[0198] Assume that the maximum values ​​of each performance evaluation index are:

[0199] P max =1.0

[0200] A max =1.0

[0201] F max =1.0

[0202] And the weight coefficient and exponential adjustment factor are set as:

[0203] w 1 =0.4

[0204] w 2 =0.2

[0205] w 3 =0.15

[0206] w 4 =0.15

[0207] w5 =0.1

[0208] a 1 =0.5

[0209] Calculation process:

[0210] For S 1 :

[0211]

[0212] V 1 =0.4·e 1.25 +0.2·800+0.15·0.1+0.15·0.15+0.1·0.3

[0213] V 1 =0.4·3.4903+160+0.015+0.0225+0.03

[0214] V 1 =1.39612+160+0.0675

[0215] V 1 =161.46362

[0216] For S 2 :

[0217]

[0218] V 2 =0.4·e 1.5 +0.2·700+0.15·0.15+0.15·0.1+0.1·0.2

[0219] V 2 =0.4·4.48169+140+0.0225+0.015+0.02

[0220] V 2 =1.792676+140+0.0575

[0221] V 2 =141.850176

[0222] For S 3 :

[0223]

[0224] V 3 =0.4·e 1.0 +0.2·900+0.15·0.05+0.15·0.2+0.1·0.35

[0225] V 3 = 0.4·2.71828 + 180 + 0.0075 + 0.03 + 0.035

[0226] V 3 = 1.087312 + 180 + 0.0725

[0227] V 3 = 181.159812

[0228] Conclusion Explanation:

[0229] Through the above calculations, the evaluation results of the three logistics distribution strategies are respectively:

[0230] V 1 = 161.46362

[0231] V 2 = 141.850176

[0232] V 3 = 181.159812

[0233] From the calculation results, it can be seen that although the evaluation score of S 3 is the highest, due to its relatively high transportation cost (900 yuan) and low resource flexibility (0.65), it may not be suitable for all situations in actual operation. In contrast, the scores of S 1 and S 2 are relatively close, but the delivery time of S 1 is shorter (2.5 hours vs 3.0 hours), and both the path reliability and traffic adaptability are better. Therefore, S 1 may be a more ideal choice.

[0234] Finally, the system selected S 1 as the target logistics distribution strategy, ensuring the best transportation efficiency and minimized delays while meeting all the constraints. This not only improves the scientificity and accuracy of decision-making but also enhances the adaptability to emergencies, ensuring that first-aid supplies can be delivered to the destination efficiently in the shortest possible time.

[0235] This application considers that the evaluation of the configuration plan of the types and quantities of the first-aid supplies needs to comprehensively consider multiple key performance indicators (KPIs) to ensure that the selected path planning plan can not only efficiently use resources, improve the rescue success rate, but also respond in the shortest possible time. The traditional single-index evaluation method is difficult to fully reflect the complex and changeable actual needs. Therefore, a new alternative solution is proposed, which includes:

[0236] The configuration scheme for evaluating the type and quantity of each first aid material, wherein the evaluation index at least includes material utilization efficiency, rescue success rate, and emergency response speed, and the distribution effect corresponding to the configuration scheme is obtained, including:

[0237] Based on the optimized parameters, P opt , in each predefined scenario S j The distribution of the emergency supplies is simulated, and the simulation process includes path selection R k , Time arrangement k and the allocation of emergency resources Q k , in order to obtain the configuration scheme A of the types and quantities of the emergency supplies in each predefined scenario j ;

[0238] Among them, each allocation scheme A j Expressed as:

[0239] A j = {R k ,T k ,Q k}

[0240] R k represents the kth transport route; T k represents the time schedule of the kth transport route; Q k represents the demand for the kth category of emergency supplies;

[0241] A configuration plan for the type and quantity of each of the first aid supplies j Carry out evaluation, and the evaluation index shall at least include material utilization efficiency E i , rescue success rate R i 、Emergency response speed V i ;

[0242] Distribution Effect V j Calculated by the following formula:

[0243]

[0244] w E ,w R ,w R are the weight coefficients of material utilization efficiency, rescue success rate, and emergency response speed; a E ,a R ,a V is the corresponding exponential adjustment factor; E k (Q k ) represents the allocation of emergency resources on the kth path Q k The calculated material utilization efficiency represents the material utilization efficiency Ei A specific instance under a specific path and resource allocation; R k (T k ) indicates that T is scheduled for a given time on the kth path k The rescue success rate represents the rescue success rate R i A specific instance with a specific path and schedule; min k T k Represents the shortest delivery time among all paths, used to measure the emergency response speed V i ; E max It is the maximum value of material utilization efficiency and is used for normalization processing.

[0245] The following is a detailed explanation of each parameter:

[0246] V j : The comprehensive evaluation score of the configuration scheme of the types and quantities of the first aid materials in the jth predefined scenario, which is used to measure the overall effect of the scheme. It is calculated by comprehensively considering multiple key performance indicators (such as material utilization efficiency, rescue success rate, emergency response speed, etc.) and based on the preset weight coefficient and index adjustment factor.

[0247] P opt : Optimized parameter configuration, including key parameters such as the speed and loading capacity of the transport vehicle. Determined through historical data statistics, simulation and expert system. For example, initial parameters can be set based on past successful delivery cases or expert opinions, and dynamically adjusted through feedback from actual operations.

[0248] S j : The jth predefined scenario simulates different emergency response conditions (such as weather, traffic conditions, etc.). It is generated by analyzing historical emergency response records and real-time environmental data. For example, multiple possible emergency response scenarios can be constructed based on past meteorological data, traffic flow information, and emergency reports.

[0249] A j : Each allocation scheme is represented by A j = {R k ,T k ,O k}, where R k : The kth transport route. T k : The time schedule of the kth transport route. Q k : The demand for the kth type of emergency supplies. Generated by path planning algorithm and time scheduling model. For example, the shortest path algorithm (such as Dijkstra algorithm), time window constraint optimization model and demand forecasting model can be used to generate detailed path selection and time scheduling.

[0250] wE ,w R ,w V : are the weight coefficients of material utilization efficiency, rescue success rate, and emergency response speed. These weights reflect the relative importance of different indicators in the overall evaluation and are determined through expert systems or historical data analysis. For example, initial weights can be set based on past successful cases or expert opinions, and dynamically adjusted through feedback from actual operations.

[0251] a E ,a R ,a V : The corresponding index adjustment factor is used to adjust the influence of each sub-item on the final score. It is determined by experimental data fitting or simulation. Usually, the appropriate index adjustment factor value can be found by analyzing the influence of different delay times, success rate changes and other factors on the rescue effect.

[0252] E k (Q k ): Allocation of emergency resources on the kth path Q k The calculated material utilization efficiency represents the material utilization efficiency E i Specific examples under specific paths and resource allocation. Obtained through the material management system and emergency response database. For example, the types, quantities and distribution of existing resources can be analyzed to evaluate the flexibility and response speed of resource allocation.

[0253] R k (T k ): Schedule T for a given time on the kth path k The rescue success rate, representing the rescue success rate R i Specific instances under specific routes and time arrangements. Obtained through historical data statistics and real-time monitoring. For example, the success rate of similar routes in the past can be analyzed, or the reliability of the current route can be evaluated through real-time traffic conditions and weather forecasts.

[0254] min k T k : The shortest delivery time among all paths, used to measure the emergency response speed V i The actual travel time for each route can be obtained through GPS tracking system or simulation. It can also be combined with real-time traffic data to predict future delivery times.

[0255] E max : The maximum value of material utilization efficiency is used for normalization processing to ensure that the values ​​of each performance indicator are in the interval [0,1]. It is determined through historical data statistics and best practice cases. For example, based on past successful material allocation cases, the highest material utilization efficiency can be found as a benchmark value.

[0256] The following is a brief introduction to the design reasons of each sub-item:

[0257] Material utilization efficiency E i It is a key indicator to measure whether resources are fully utilized. k (Q k ) and divided by the maximum value E max Normalize it to ensure that its value is in the range [0,1]. Exponential adjustment factor a E It is used to emphasize the importance of efficiency and ensure that solutions that use resources efficiently receive higher scores.

[0258] w R ·log(a R ∏ k R k (T k )+1): rescue success rate R i It reflects the impact of the distribution plan on the actual rescue effect. k (T k ) and then applying the logarithmic function log can smooth out large-scale changes and ensure that the score is not unbalanced due to too high or too low success rates of individual paths. The multiplication operation ensures that the success rates of all paths are taken into account, while the logarithmic function prevents too large or too small values ​​from affecting the overall score. R Used to adjust the impact on the rescue success rate.

[0259] Emergency response speed V i It is a crucial factor in emergency situations. By taking the minimum delivery time min among all the paths k T k , and apply the exponential function The importance of a fast response can be magnified. Exponential adjustment factor a V It is used to adjust the impact of response speed to ensure that solutions that reach the destination quickly receive higher scores.

[0260] This formula adds up the sub-items to comprehensively consider multiple key performance indicators (KPIs) to ensure that the evaluation results are comprehensive and scientific. Each sub-item represents an important evaluation dimension, and different weight coefficients w are assigned to each sub-item. E ,w R ,w V, the importance of different indicators can be reflected in the overall evaluation. Specifically: by adding up the various sub-items, the performance of each type and quantity configuration scheme of the first aid materials can be comprehensively evaluated in terms of material utilization efficiency, rescue success rate, emergency response speed, etc. The weight coefficient allows the importance of different indicators to be adjusted according to actual needs. For example, in an emergency, the emergency response speed may be more critical, and this can be increased by adding w V The final score V j It comprehensively reflects the performance of all key performance indicators to ensure that the selected solution is not only excellent in one aspect, but also achieves the best balance in multiple aspects.

[0261] This comprehensive evaluation method ensures that the selection of the configuration plan for the types and quantities of the first aid supplies is more scientific and reasonable, and improves the accuracy and reliability of decision-making. Especially in complex and changeable actual environments, it can better respond to various challenges and achieve the most efficient allocation of supplies.

[0262] Here is a specific example:

[0263] Assume we have three predefined scenarios S 1 ,S 2 ,S 3 , and set the following parameters:

[0264] Weight coefficient:

[0265] w E =0.4

[0266] w R =0.3

[0267] w V =0.3

[0268] Index adjustment factor:

[0269] a E =0.5

[0270] a R =0.8

[0271] a V =0.6

[0272] Maximum value:

[0273] E max =1.0

[0274] For each scenario, the following Table 2 shows the specific values:

[0275] Table 2

[0276]

[0277] Calculation process:

[0278] Calculate S 1 The distribution effect V 1 :

[0279]

[0280] V 1 =0.4·(2.55) 0.5 +0.3·log(0.8·0.72675+1)+0.3·e 1.2

[0281] V 1 =0.4·1.6+0.3·log(1.5814)+0.3·3.32

[0282] V 1 =0.64+0.3·0.204+0.996

[0283] V 1 =0.64+0.0612+0.996

[0284] V 1 =1.6972

[0285] Calculate S 2 The distribution effect V 2 :

[0286]

[0287] V 2 =0.4·(2.4) 0.5 +0.3·log(0.8·0.612+1)+0.3·e 1.5

[0288] V 2 =0.4·1.55+0.3·log(1.4896)+0.3·4.48

[0289] V 2 =0.62+0.3·0.176+1.344

[0290] V 2 =0.62+0.0528+1.344

[0291] V 2 =2.0168

[0292] Calculate S 3 The distribution effect V 3 :

[0293]

[0294] V 3 =0.4·(2.55) 0.5 +0.3·log(0.8·0.72675+1)+0.3·e 0.9

[0295] V 3 =0.4·1.6+0.3·log(1.5814)+0.3·2.46

[0296] V 3 =0.64+0.3·0.204+0.738

[0297] V 3 =0.64+0.0612+0.738

[0298] V 3 =1.4392

[0299] Through the above calculations, the evaluation results of the three predefined scenarios are obtained:

[0300] V 1 =1.6972

[0301] V 2 =2.0168

[0302] V 3 =1.4392

[0303] From the calculation results, it can be seen that although S 2 The evaluation score of is the highest, but its shortest delivery time is 2.5 hours, which is relatively long, which may lead to insufficient response speed. 3 The evaluation score of S is lower, but its shortest delivery time is 1.5 hours, which means it can respond to emergencies faster, and its material utilization efficiency and rescue success rate are also higher. Therefore, considering all factors, S 3 It is a more ideal choice.

[0304] Finally, the system selected S 3 As the target logistics distribution strategy, it ensures the best transportation efficiency and minimized delays under the premise of meeting all constraints. This not only improves the scientificity and accuracy of decision-making, but also enhances the ability to adapt to emergencies, ensuring that emergency supplies can be efficiently delivered to the destination in the shortest time.

[0305] This method not only improves the accuracy of decision-making, but also enhances the ability to adapt to emergencies, ensuring that the best logistics distribution decisions can be made in complex and changing actual environments.

[0306] Figure 2 A schematic diagram of the structure of an intelligent decision-making system for emergency allocation of emergency supplies is provided for the embodiment of the present application, such as Figure 2 As shown, the system includes:

[0307] A receiving module 21 is used to receive demand information from multiple emergency response points, where the demand information at least includes the number of injured, the type of injured, the severity, the status of on-site medical resources, the priority identification, and the geographic space distribution;

[0308] A generating module 22 is used to evaluate the urgency of the demand of each emergency response point based on the demand information based on the graph neural network intelligent evaluation system, and generate a preliminary deployment plan, wherein the preliminary deployment plan is determined according to the urgency of the demand of each emergency response point, and includes the type and quantity of first aid supplies required by each emergency response point, and the distribution order of the first aid supplies;

[0309] The determination module 23 is used to generate the prediction results of road capacity based on the historical traffic data and real-time traffic obtained from multiple sources by using the deep learning prediction model, and simulate multiple logistics distribution strategies in combination with the reinforcement learning algorithm, refine and adjust the preliminary deployment plan, and determine the target logistics distribution strategy; wherein, the prediction results of the road capacity refer to the deep learning prediction model using the historical traffic data, real-time traffic data and model algorithm to predict the traffic conditions on the road in the future period of time, and the logistics distribution strategy is formulated according to the distribution order of the emergency supplies in the preliminary deployment plan; the target logistics distribution strategy can avoid the estimated congested sections and ensure timely arrival at the designated location;

[0310] The generation module 22 is also used to start the AI ​​simulation and exercise system based on Bayesian optimization and genetic algorithm, simulate the configuration schemes of the types and quantities of the first aid supplies in multiple scenarios according to the target logistics distribution strategy, and iteratively evolve the best configuration scheme of the types and quantities of the first aid supplies that can adapt to the actual situation from the configuration schemes in multiple scenarios through genetic algorithm, and generate a comprehensive deployment instruction, which includes a list of first aid supplies, an estimated delivery time and an optimal transportation route;

[0311] The execution module 24 is used to execute the comprehensive allocation instruction to realize the emergency allocation of the emergency supplies.

[0312] Figure 2 The intelligent decision-making system for emergency allocation of emergency supplies can execute Figure 1The implementation principle and technical effect of the intelligent decision-making method for emergency allocation of first aid materials described in the illustrated embodiment will not be described in detail. The specific manner in which each module and unit performs operations in the intelligent decision-making system for emergency allocation of first aid materials in the above embodiment has been described in detail in the embodiment of the method, and will not be elaborated here.

[0313] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0314] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent decision-making method for emergency deployment of emergency supplies, characterized in that: include: Receiving demand information from multiple emergency response points, the demand information at least including the number of injured, the type of injured, the severity, the status of on-site medical resources, the priority identification, and the geographic spatial distribution; Based on the graph neural network intelligent evaluation system, the urgency of the demand of each emergency response point is evaluated according to the demand information, and a preliminary deployment plan is generated. The preliminary deployment plan is determined according to the urgency of the demand of each emergency response point, and includes the type and quantity of first aid supplies required by each emergency response point, as well as the distribution order of the first aid supplies; Using a deep learning prediction model, a prediction result of road capacity is generated based on historical traffic data and real-time traffic obtained from multiple sources, and a variety of logistics distribution strategies are simulated in combination with a reinforcement learning algorithm, the preliminary deployment plan is refined and adjusted, and a target logistics distribution strategy is determined; wherein, the prediction result of road capacity refers to the deep learning prediction model using historical traffic data, real-time traffic data and model algorithms to predict the traffic conditions on the road in the future period of time, and the logistics distribution strategy is formulated according to the distribution order of the emergency supplies described in the preliminary deployment plan; the target logistics distribution strategy can avoid the estimated congested sections and ensure timely arrival at the designated location; Initiate an AI simulation exercise system based on Bayesian optimization and genetic algorithm, simulate the configuration schemes of the types and quantities of the first aid supplies in multiple scenarios according to the target logistics distribution strategy, and iteratively evolve the optimal configuration scheme of the types and quantities of the first aid supplies that can adapt to the actual situation from the configuration schemes in multiple scenarios through genetic algorithm, and generate a comprehensive deployment instruction, which includes a list of first aid supplies, an estimated delivery time and an optimal transportation route; Execute the comprehensive deployment instruction to realize the emergency deployment of the emergency supplies.

2. The method according to claim 1, characterized in that The deep learning prediction model is used to generate road capacity prediction results based on historical traffic data and real-time traffic obtained from multiple sources, including: Integrate historical traffic data and real-time traffic data obtained from multiple sources to obtain a traffic data set, wherein the historical traffic data at least includes traffic flow, accident records and weather impacts in the same time period in the past, and the real-time traffic data at least includes current traffic camera images, vehicle location information and instant traffic conditions provided by sensors; The deep learning prediction model is used to analyze and process the traffic data in the traffic data set to generate a prediction result of the road capacity in the future. The deep learning prediction model can identify and estimate the impact of different factors on the road capacity through training, and output the expected traffic efficiency of each road in a specific time period in the future to generate a prediction result of the road capacity.

3. The method according to claim 2, characterized in that According to the predicted results of the road capacity, a variety of logistics distribution strategies are simulated in combination with the reinforcement learning algorithm, the preliminary deployment plan is refined and adjusted, and the target logistics distribution strategy is determined, including: Using the predicted results of the road capacity and combining with the reinforcement learning algorithm, the distribution order of the emergency supplies in the preliminary deployment plan is simulated to obtain a variety of logistics distribution strategies aimed at optimizing time cost and transportation efficiency; Based on a variety of logistics distribution strategies, the preliminary deployment plan is refined and adjusted. The refinement and adjustment process includes optimizing the transportation routes of the emergency supplies under different logistics distribution strategies and adjusting the types and quantities of emergency supplies required for each emergency response point to adapt to the actual situation; For each of the logistics distribution strategies, a performance evaluation is performed in terms of avoiding estimated congested road sections and ensuring timely arrival at designated locations, and the performance evaluation indicators at least include delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility, to obtain an evaluation result for each of the logistics distribution strategies; Based on the evaluation results of all logistics distribution strategies, select the target logistics distribution strategy that can maximize transportation efficiency and minimize delays; The target logistics distribution strategy is applied to the adjusted preliminary allocation plan for verification and confirmation to ensure that the target logistics distribution strategy can achieve the expected effect in actual operation.

4. The method according to claim 3, characterized in that According to the evaluation results of all logistics distribution strategies, the target logistics distribution strategy that can maximize transportation efficiency and minimize delays is selected, including: According to the evaluation results of the effects of all logistics distribution strategies, eligible logistics distribution strategies are screened out, and based on multi-dimensional evaluation, the screened logistics distribution strategies are comprehensively scored to obtain the score of each logistics distribution strategy, and one or several logistics distribution strategies with the highest score are selected as candidate logistics distribution strategies; The selection of each route segment in each candidate logistics distribution strategy, the expected traffic conditions and the delay risk are analyzed to determine the target logistics distribution strategy that can maximize transportation efficiency and minimize delays.

5. The method according to claim 1, characterized in that The AI ​​simulation and exercise system based on Bayesian optimization and genetic algorithm is started to simulate the configuration schemes of the types and quantities of the first aid materials in multiple scenarios according to the target logistics distribution strategy, and the optimal configuration schemes of the types and quantities of the first aid materials that can adapt to the actual situation are iteratively evolved from the configuration schemes in multiple scenarios through genetic algorithms, and a comprehensive deployment instruction is generated, including: Start the AI ​​simulation exercise system based on Bayesian optimization and genetic algorithm, load the target logistics distribution strategy as input data, and initialize the simulation environment; Using Bayesian optimization technology, the model parameters of the AI ​​simulation and drill system are automatically adjusted to improve simulation accuracy and obtain optimized parameter configuration; Based on the optimized parameter configuration, simulate the configuration scheme of the types and quantities of the emergency supplies in multiple predefined scenarios, and evaluate the distribution effect corresponding to the configuration scheme in each predefined scenario; By using a genetic algorithm, the allocation effect corresponding to the configuration scheme in each predefined scenario is iterated and evolved to determine the optimal configuration scheme for the types and quantities of the first aid supplies that can adapt to various actual situations; Verify the effectiveness of the optimal configuration plan and generate a comprehensive deployment instruction containing a detailed list of emergency supplies, estimated delivery time and optimal transportation route.

6. The method according to claim 5, characterized in that Based on the optimized parameter configuration, simulating the configuration scheme of the type and quantity of the emergency supplies in multiple predefined scenarios, and evaluating the distribution effect corresponding to the configuration scheme in each predefined scenario, includes: In the AI ​​simulation and drill system, the optimized parameter configuration is loaded as the initial setting to initialize the simulation environment to ensure that the simulation environment can accurately reflect the actual situation; Based on the optimized parameter configuration, in each predefined scenario, the distribution of the first aid supplies is simulated, and the simulation process includes path selection, time arrangement and the distribution of the first aid resources, so as to obtain a configuration scheme of the type and quantity of the first aid supplies in each predefined scenario; The configuration scheme of the type and quantity of each of the first aid supplies is evaluated, and the evaluation indicators include at least delivery time, transportation cost, material utilization efficiency, rescue success rate and emergency response speed, to obtain the distribution effect corresponding to the configuration scheme.

7. The method according to claim 1, characterized in that The graph neural network-based intelligent assessment system evaluates the urgency of the demand for each emergency response point according to the demand information and generates a preliminary deployment plan, including: Utilize natural language processing algorithms to parse and process demand information from multiple emergency response points to obtain specific demand results for each emergency response point; According to the specific demand results, a graph structure data model is constructed, wherein the nodes in the graph structure data model represent various emergency response points, and the edges represent the correlation and impact range between different emergency response points; Based on the pre-trained model parameters, the graph neural network intelligent evaluation system is started to load and process the graph structure data model to ensure that the graph neural network intelligent evaluation system can identify and process different types of node and edge attributes, and the graph neural network intelligent evaluation system is used to evaluate the demand urgency of each emergency response point and calculate the demand urgency score of each node. The demand urgency score is determined according to the correlation and impact range between different emergency response points. Based on the demand urgency score, a preliminary deployment plan is generated.

8. The method according to claim 3, characterized in that The performance evaluation of each logistics distribution strategy is performed in terms of avoiding estimated congested road sections and ensuring timely arrival at designated locations. The performance evaluation indicators include at least delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility. The evaluation results of each logistics distribution strategy are obtained, including: For each of the logistics distribution strategies S i , to perform performance evaluation in terms of avoiding estimated congested sections and ensuring timely arrival at the designated location. The performance evaluation indicators include at least: delivery time T i , transportation cost C i , path reliability P i Traffic adaptability A i , Resource flexibility i ; Evaluation Results i Calculated by the following formula: Among them, w1, w2, w3, w4, and w5 are the weight coefficients of delivery time, transportation cost, path reliability, traffic adaptability, and resource flexibility respectively; a1 is the exponential adjustment factor of delivery time, which is used to reflect the severity of delay; P max , A max , F max They are the maximum values ​​of path reliability, traffic adaptability, and resource flexibility, respectively, and are used for normalization so that the values ​​of each performance evaluation indicator are in the interval [0, 1].

9. The method according to claim 6, characterized in that The configuration scheme for evaluating the type and quantity of each first aid material, wherein the evaluation index at least includes material utilization efficiency, rescue success rate, and emergency response speed, and the distribution effect corresponding to the configuration scheme is obtained, including: Based on the optimized parameters, P opt , in each predefined scenario S j Under this condition, the distribution of the emergency supplies is simulated, and the simulation process includes path selection R k , Time arrangement k and the allocation of emergency resources Q k , in order to obtain the configuration scheme A of the types and quantities of the emergency supplies in each predefined scenario j ; Among them, each allocation scheme A is expressed as: A j ={R k ,T k ,Q k } R k represents the kth transport route; T k represents the time schedule of the kth transport route; Q k represents the demand for the kth type of emergency supplies; A configuration plan for the type and quantity of each of the first aid supplies j Carry out evaluation, and the evaluation index shall at least include material utilization efficiency E i , rescue success rate R i 、Emergency response speed V i ; Distribution Effect V j Calculated by the following formula: w E , w R , w R are the weight coefficients of material utilization efficiency, rescue success rate, and emergency response speed; a E , a R , a V is the corresponding exponential adjustment factor; E k (Q k ) represents the allocation of emergency resources on the kth path Q k The calculated material utilization efficiency represents the material utilization efficiency E i A specific instance under a specific path and resource allocation; R k (T k ) indicates that T is scheduled for a given time on the kth path k The rescue success rate represents the rescue success rate R i A specific instance with a specific path and schedule; min k T k Represents the shortest delivery time among all paths, used to measure the emergency response speed V i ; E max It is the maximum value of material utilization efficiency and is used for normalization processing.

10. An intelligent decision-making system for emergency deployment of emergency supplies, characterized in that: include: A receiving module, used to receive demand information from multiple emergency response points, wherein the demand information at least includes the number of injured, the type of injured, the severity, the status of on-site medical resources, the priority identification and the geographical space distribution; A generation module, for evaluating the urgency of the demand of each emergency response point based on the demand information and generating a preliminary deployment plan based on the graph neural network intelligent evaluation system, wherein the preliminary deployment plan is determined according to the urgency of the demand of each emergency response point and includes the type and quantity of first aid supplies required by each emergency response point, as well as the distribution order of the first aid supplies; A determination module is used to generate a prediction result of road capacity based on historical traffic data and real-time traffic obtained from multiple sources using a deep learning prediction model, and simulate multiple logistics distribution strategies in combination with a reinforcement learning algorithm, refine and adjust the preliminary deployment plan, and determine a target logistics distribution strategy; wherein the prediction result of the road capacity refers to the deep learning prediction model using historical traffic data, real-time traffic data and model algorithms to predict the traffic conditions on the road in a certain period of time in the future, and the logistics distribution strategy is formulated according to the distribution order of the emergency supplies in the preliminary deployment plan; the target logistics distribution strategy can avoid the estimated congested sections and ensure timely arrival at the designated location; The generation module is also used to start an AI simulation and exercise system based on Bayesian optimization and genetic algorithm, simulate the configuration schemes of the types and quantities of the first aid supplies in multiple scenarios according to the target logistics distribution strategy, and iteratively evolve the best configuration scheme of the types and quantities of the first aid supplies that can adapt to the actual situation from the configuration schemes in multiple scenarios through genetic algorithm, and generate a comprehensive deployment instruction, which includes a list of first aid supplies, an estimated delivery time and an optimal transportation route; The execution module is used to execute the comprehensive deployment instruction to realize the emergency deployment of the emergency supplies.

Images