First-aid platform work order management method based on digital twinborn and artificial intelligence

Abstract

The invention provides a first-aid platform work order management method based on digital twinning and artificial intelligence, and belongs to the technical field of information technology application innovation, and the method comprises the steps: collecting historical first-aid data and real-time data of a first-aid scene, and constructing a task scheduling system based on a knowledge graph; decomposing the task into a plurality of sub-tasks according to the work order data, and allocating an optimal first-aid resource combination for the sub-tasks; an optimization strategy is set to carry out collaborative allocation control on each resource, and task allocation is optimized; the optimized task allocation is monitored in real time, and meanwhile, in combination with real-time data flow monitored in real time, it is ensured that the change of the dynamic environment is responded at any time in the task execution process; and re-inputting the optimized task allocation and resource scheduling strategy into the knowledge graph, and carrying out closed-loop optimization. According to the invention, the problems of inaccurate task scheduling, weak real-time response capability, low resource utilization rate and the like in the existing work order management system of the emergency platform are effectively overcome, and the overall efficiency and the automation level of the emergency platform are greatly improved.

Description

Technical Field

[0001] The present invention belongs to the field of information technology application innovation technology, and in particular to a first aid platform work order management method based on digital twins and artificial intelligence. Background Art

[0002] With the development of information technology, emergency platforms play an increasingly important role in the modern medical system. Traditional emergency platform work order management systems usually rely on manual scheduling and fixed rule algorithms for resource allocation and task scheduling. However, the complexity, variability and urgency of emergency tasks make these traditional methods have many problems in practical applications, especially when faced with multiple tasks, frequent emergencies, resource shortages or unpredictable factors (such as traffic conditions, equipment failures, etc.), traditional systems are difficult to ensure efficient and accurate task allocation.

[0003] Most existing emergency work order management systems use a scheduling mechanism based on fixed priorities, which cannot respond to changes in task urgency and resource status in real time, often resulting in uneven or untimely resource allocation. For example, when multiple emergency tasks occur at the same time, the system often fails to respond to some tasks in a timely manner due to the inability to dynamically adjust priorities and optimize scheduling strategies, affecting the overall emergency efficiency.

[0004] In addition, resource scheduling and task allocation mainly rely on manual dispatchers, who need to manually judge and allocate each task based on experience, which not only increases the burden on dispatchers, but also may lead to human errors, especially in emergency task processing under high pressure. There is a high degree of correlation between resources such as personnel, equipment and vehicles in emergency tasks. For example, different emergency personnel have different skills, and different equipment has specific functions. The current system cannot fully model and understand these complex resource relationships, resulting in low utilization of emergency resources.

[0005] In addition, another limitation of existing technologies is the lack of effective processing capabilities for real-time dynamic data. With the advancement of urbanization, unpredictable factors such as traffic congestion and weather changes often have a significant impact on the timeliness of emergency tasks, while traditional systems rely only on historical data and static rules and cannot be combined with real-time data for dynamic optimization. This causes emergency vehicles to encounter unexpected delays during the execution of their missions, and the system cannot respond in a timely manner and cannot dynamically adjust the work order allocation or task path. Furthermore, traditional work order management systems lack the ability to perform fine-grained decomposition of the complexity of different tasks, and cannot automatically match resources according to the specific needs of each subtask (such as personnel skills, equipment configuration, etc.), resulting in poor coordination during the execution of emergency tasks, and often waste of resources or unreasonable allocation. Summary of the invention

[0006] The purpose of the present invention is to design a work order management method for an emergency platform based on digital twins and artificial intelligence, use the application of information technology (information technology application innovation) in the emergency platform, and improve the intelligence level and emergency response capability of the emergency platform through autonomous and controllable technologies and solutions. It effectively overcomes the problems of inaccurate task scheduling, weak real-time response capability, and low resource utilization in the existing emergency platform work order management system, and greatly improves the overall efficiency and automation level of the emergency platform.

[0007] In order to achieve the above object, the present invention provides a first aid platform work order management method based on digital twins and artificial intelligence, and the method comprises the following steps:

[0008] Step 1: Collect historical emergency data and real-time data of emergency scenarios to build a knowledge graph, and build a task scheduling system based on the knowledge graph, including:

[0009] S11, obtaining historical emergency data and real-time data to form total data;

[0010] S12. The emergency resources and tasks in the total data are used as nodes in the knowledge graph. The nodes are connected by relationship edges to represent the potential matching relationship between resources and tasks. Each relationship expresses the degree of association between nodes through a dynamic weight.

[0011] S13. For each task, calculate the comprehensive matching degree between each task and multiple emergency resources to select the optimal resource allocation. At the same time, design an urgency adjustment coefficient for each task to ensure that resources are allocated to high-urgency tasks first;

[0012] S14, calculating work order data including optimal resource allocation and optimal task priority;

[0013] Step 2: Decompose the task into several subtasks according to the work order data and assign the best combination of emergency resources to each subtask, including:

[0014] S21, automatically generate work orders based on the reasoning and matching results of the knowledge graph;

[0015] S22, decompose each emergency task into several subtasks, and then allocate the optimal resource combination to each subtask according to the comprehensive matching degree and urgency adjustment coefficient;

[0016] S23. After the work order of each subtask is generated, all tasks are sorted according to the comprehensive matching degree of the subtasks and the urgency adjustment coefficient to ensure that the most urgent tasks are responded to first and the final work order is output;

[0017] Step 3: For emergency scenarios that change at any time, set up a multi-agent-based optimization strategy to coordinate the allocation and control of each resource and optimize task allocation, including:

[0018] S31. Define the multi-agents to include: vehicle agent, medical staff agent and equipment agent;

[0019] S32. Design the state space and reward mechanism of multiple agents according to the real-time scheduling requirements of the emergency platform to control multiple agents. At the same time, design a resource coordination mechanism to optimize task allocation through game and negotiation mechanisms between resources. Design a resource conflict resolution mechanism for conflicts between emergency tasks to ensure maximum resource allocation efficiency.

[0020] S33. After multiple rounds of agent learning and collaborative game, all resource allocation schemes will eventually converge to the optimal state and be output;

[0021] Step 4: Monitor the optimized task allocation in real time, and combine the real-time data stream of real-time monitoring to ensure that the dynamic environment changes can be responded to at any time during the task execution process, including:

[0022] S41, collecting real-time data streams, and converting the real-time data streams into time series for modeling, wherein the real-time data streams contain multiple events occurring in different time periods, and designing different impact functions for each event on the task to measure the impact of the event, and then dynamically adjusting the priority of the task according to the impact of each event on the task;

[0023] S42. Design dynamic adjustment rules based on real-time data streams to ensure that task execution and resource allocation are always optimal;

[0024] S43. Design an event-task association mechanism to evaluate the actual impact of events on tasks and trigger corresponding adjustments. When the event impact index exceeds a certain threshold, the adjustment mechanism will be automatically triggered;

[0025] S45. Output the optimized and adjusted work order;

[0026] Step 5: Re-enter the optimized task allocation and resource scheduling strategy into the knowledge graph for closed-loop optimization, including:

[0027] Build a feedback model, take task execution feedback as input, and adjust future work order scheduling strategies in real time by comparing the actual task execution with the plan.

[0028] Furthermore, the structure of the knowledge graph includes:

[0029] The nodes include:

[0030] The vehicle node V(i) represents the i-th emergency vehicle;

[0031] The medical staff node P(j) represents the jth medical staff;

[0032] The equipment node E(k) represents the kth emergency equipment;

[0033] Task node T(l) represents the lth emergency task;

[0034] Relationships include:

[0035] Define the relationship between various resources and tasks, and use R(V,T), R(P,T), and R(E,T) to represent the matching relationship between each resource and task.

[0036] Furthermore, the S13 specifically includes:

[0037] Each task T(l) needs to be matched with multiple resources. The comprehensive matching degree between resources and tasks is calculated. To select the optimal resource allocation, it is expressed as follows:

[0038]

[0039] in, Represents task T(l) and resource R i The matching degree; Indicates resource R i The degree of match with a specific attribute of task T(l); w k represents the weight of attribute k, indicating the importance of different attributes; λ·ΔD(T l ,R i ) represents the real-time distance regularization term introduced innovatively, and represents the resource R i The difference in distance from task T(l) in geographic space, ΔD(T l ,R i ) is the geographical distance between the task and the resource, λ is the regularization coefficient; -γ·C(R i ) represents the resource load penalty term, C(R i ) is the current resource load level;

[0040] In the matching calculation of emergency tasks, the urgency of the tasks needs to be taken into consideration. Therefore, a dynamic adjustment item is designed for each task to ensure that resources are allocated to high-urgency tasks first, as shown below:

[0041]

[0042] Among them, μ l Indicates the dynamic priority adjustment coefficient of the lth task. l It represents the urgency of the lth task, and the value of U is determined by the task emergency assessment model; Indicates the sum of the urgency of all current tasks.

[0043] Furthermore, the optimal resource combination is allocated to each subtask according to the comprehensive matching degree and the urgency adjustment coefficient, which is expressed as follows:

[0044]

[0045] Among them, R(T(l) k ) represents subtask T(l) k the set of resources allocated; Indicates resource R i With subtask T(l) k The matching degree of l represents the urgency weight of task T(l);

[0046] In S22, in order to ensure reasonable resource allocation and avoid resource conflicts or duplication, task allocation needs to follow the following rules:

[0047] Each resource can be assigned to only one task or subtask, unless the resource can be used simultaneously;

[0048] Automatically check the current load of resources to avoid over-allocation of resources. If a resource has too high a load, its priority will be lowered through the load penalty item in the matching degree.

[0049] When there is a resource conflict, urgent tasks are given priority and low-priority tasks are delayed.

[0050] Furthermore, the final work order is a task-resource allocation table, including the subtasks of each task, the allocated resources, and the priority order of each task. The final work order structure is expressed as follows:

[0051] Task T(l): Task type, urgency μ l , Geographical location;

[0052] Subtask T(l) k :Detailed description of each subtask; corresponding allocation resources R(T(l) k );

[0053] Task priority sorting: According to the task priority calculated by the priority formula, tasks will be processed in order from high to low.

[0054] Furthermore, the S32 specifically includes:

[0055] The state space and reward mechanism of the intelligent agent are designed to solve the urgency and complexity of task scheduling. The state space is expressed as follows:

[0056]

[0057] Among them, L i Indicates resource R i The current load of U i Indicates resource R i availability, indicating whether the resource is idle; μ l represents the urgency adjustment coefficient of task T(l); D pos (T(l),R i ) represents the task T(l) and the resource R i geographical distance; Indicates that the resource R i Collaboration status with other resources;

[0058] The reward mechanism is as follows:

[0059]

[0060] Among them, α·μ l Indicates the emphasis on the urgency of the task; α controls the priority increase of urgent tasks; represents the matching weight between resources and tasks, which is used to reward resource agents that can complete tasks efficiently; δ·C collab (R i ) represents the collaborative work reward item, which is used to consider the positive feedback of cooperation between resources; γ·C(R i ) represents the resource load penalty term, which prevents resources from being overused or participating in tasks for a long time;

[0061] The collaboration mechanism is expressed as follows:

[0062] CollaborativeReward(R i ,R j )=C collab (R i ,R j )=λ·Efficiency(R i ,R j )-θ·D pos (R i ,R j )

[0063] Among them, λ represents the weight of collaborative efficiency, which indicates the efficiency of cooperation between resources; θ represents the distance penalty term. If the distance between two resources is far, the collaborative benefit will be reduced; Efficiency (R i ,R j ) indicates resource R i and R j The efficiency of cooperation on tasks is calculated using historical data;

[0064] The resource conflict resolution mechanism is expressed as follows:

[0065]

[0066] in, and Represents resource A i and A j With task T(l) k The matching degree of l Represents the urgency adjustment coefficient of task T(l).

[0067] Furthermore, all resource allocation schemes will eventually converge to the optimal state and be output, where the output content includes:

[0068] The final resources assigned to each task, along with task urgency and resource coordination, as well as the specific resource execution path, ensure that tasks can be completed according to priority and time sensitivity.

[0069] Furthermore, the priority of the task is dynamically adjusted according to the impact of each event on the task, as shown below:

[0070]

[0071] Among them, μ′ l Indicates the task priority adjusted after real-time event stream processing; μ l Represents the original task priority; Δμ l (e i (t)) represents each event e i (t) The increment to the task priority.

[0072] Furthermore, the design of dynamic adjustment rules according to the real-time data stream includes:

[0073] When the traffic situation changes, the event E traffic (t) Recalculate the path to ensure that resources can arrive at the task location on time. The path adjustment formula is as follows:

[0074] P′(T(l))=P(T(l))+ΔD(E traffic (t))

[0075] Where P(T(l)) represents the original path of task T(l); ΔD(E traffic (t)) represents the traffic event E traffic (t) is the path or time increment triggered; P′(T(l)) represents the optimal path after dynamic adjustment;

[0076] When the device status event E deviceWhen (t) occurs, the affected device is dynamically replaced. The device replacement formula is as follows:

[0077] R′(T(l) k )=R(T(l) k )-E fault +E backup

[0078] Among them, R(T(l) k ) represents the current subtask T(l) k Resource allocation; E fault Indicates a faulty device; E backup Represents a spare device that automatically replaces the affected resource.

[0079] In an emergency scenario, patients may be transferred to different locations, and the nearest resources need to be reallocated. The resource reallocation formula is as follows:

[0080]

[0081] in, Represents the task T(l) and the resource R i The matching degree of D pos (T(l),R i ) represents the new location of task T(l) and resource R i The geographical distance; R′(T(l)) represents the resource allocation after dynamic adjustment;

[0082] The association mechanism between the events and tasks is as follows:

[0083]

[0084] Where I(T(l),E(t)) represents the overall impact index of event E(t) on task T(l); w i represents the weight, based on the urgency of the event and the relevance of the task; δ(T(l),e i (t)) indicates the specific impact of the event on the task.

[0085] Furthermore, the feedback model is expressed as follows:

[0086] ΔT′(l)=λ·(P actual (T(l))-P′(T(l)))+γ·F(t)

[0087] Among them, ΔT′(l) represents the improvement of task execution effect, which is used to adjust future task orders; λ represents the path optimization coefficient, which weighs the difference between the planned path and the actual path; P actual(T(l)) represents the actual execution path of the task; P′(T(l)) represents the optimized planned path; γ represents the feedback weight coefficient, which measures the impact of feedback data on task adjustment; F(t) represents the execution feedback data.

[0088] The beneficial technical effects of the present invention are at least as follows:

[0089] In order to overcome the above shortcomings of the existing first aid platform work order management system, the present invention proposes a first aid work order dynamic generation and intelligent allocation system based on knowledge graph, multi-agent system (MAS) and real-time event stream processing technology, aiming to solve the problems of inaccurate first aid resource scheduling, poor real-time performance, and high manual dependence in the prior art. The system models the complex correlation between first aid resources and tasks by introducing knowledge graphs, and realizes the automated processing of task decomposition and resource matching. At the same time, the multi-agent system uses reinforcement learning technology to conduct games and learning in a digital twin environment to achieve collaborative optimization and scheduling of resources; and real-time event stream processing ensures that the system can dynamically adjust the work order and task priority in combination with real-time data (such as emergency calls, traffic conditions, equipment status, etc.) to ensure that first aid tasks are processed in a timely manner.

[0090] First, the present invention constructs a knowledge graph of the emergency platform to semantically express and reason about the complex relationship between emergency tasks and resources (such as vehicles, medical staff, equipment, etc.). Compared with the traditional fixed rule system, the present invention can automatically understand and match the specific requirements of the task (such as the medical staff skills and equipment types required for different tasks), achieve more accurate work order generation and task decomposition, and greatly improve the rationality and coordination of resource allocation. This can not only reduce the need for human intervention and reduce the burden on dispatchers, but also effectively avoid the waste of resources and unreasonable allocation.

[0091] Secondly, the present invention introduces a multi-agent system, which enables emergency resources to be coordinated and optimized when multiple tasks are concurrent. Each emergency resource (such as vehicles, medical staff) acts as an independent agent, and continuously adjusts its own scheduling decisions through a reinforcement learning algorithm to ensure that in an emergency, the system can dynamically generate or adjust work orders according to the priority of the task and automatically allocate the optimal resource combination. In addition, through the game mechanism of reinforcement learning, the system can optimize the action strategies of each agent in real time to ensure that resource scheduling is optimal overall.

[0092] Finally, the present invention solves the problem that the existing system cannot respond to environmental changes in a timely manner through real-time event stream processing technology. During the execution of emergency tasks, the system can analyze dynamic data from different sources (such as traffic conditions, ambulance driving status, weather changes, etc.) in real time, and dynamically adjust the generation of work orders and resource allocation according to the changes in these data. This mechanism ensures that the generation of work orders and resource scheduling for emergency tasks can not only adapt to changes in emergencies, but also continuously optimize paths and resource allocation during task execution, thereby improving the response speed and processing efficiency of emergency tasks.

[0093] In summary, through the application of information technology (artificial intelligence, digital twin), the first aid platform of the present invention can overcome the limitations of the traditional work order management system and achieve more efficient, intelligent and reliable first aid task scheduling and resource management. Technical means such as intelligent scheduling algorithms, real-time data processing, resource modeling and optimization, automated task management and visual decision support not only improve the response speed and processing efficiency of first aid tasks, but also reduce human errors and waste of resources, providing strong support for the modern medical system. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0095] Figure 1 This is a flow chart of a method for managing work orders on an emergency platform based on digital twins and artificial intelligence according to an embodiment of the present invention. DETAILED DESCRIPTION

[0096] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0097] In one or more embodiments, Figure 1 As shown, the present invention discloses a method for managing emergency platform work orders based on digital twins and artificial intelligence, and the method includes steps one to five, including:

[0098] Step 1: Collect historical emergency data and real-time data of emergency scenarios to build a knowledge graph, and build a task scheduling system based on the knowledge graph.

[0099] Specifically, in the scenario of the emergency platform, the present invention has a large amount of historical emergency data and real-time data, which include vehicle location, equipment status, medical staff skills, task urgency, etc. These data inputs will become the basis for building a knowledge graph.

[0100] Among them, the input data of the knowledge graph is:

[0101] Emergency vehicle data: vehicle type, real-time location, load capacity, equipment status, etc.

[0102] Medical staff data: Each medical staff member’s skill level, previous experience, availability, location, etc.

[0103] Equipment status data: equipment availability, model, remaining life, etc.

[0104] Task attributes: task urgency, task type (such as cardiac arrest, traffic accident), task location, etc.

[0105] These inputs will serve as the data source for node attributes in the knowledge graph, based on which semantic modeling will be performed.

[0106] Furthermore, the present invention abstracts emergency resources (vehicles, medical staff, equipment, etc.) and tasks (task type, urgency, geographic location, etc.) into nodes in the knowledge graph, and the nodes are connected by relationship edges to represent the potential matching relationship between resources and tasks. Each relationship expresses the degree of association between nodes through dynamic weights, which are calculated by a specific algorithm.

[0107] node:

[0108] Vehicle node V(i): represents the i-th emergency vehicle, and its attributes include real-time location, load capacity, equipment status, etc.

[0109] Medical staff node P(j): the jth medical staff, with attributes including skill level, availability, experience, etc.

[0110] Equipment node E(k): the kth emergency equipment, including attributes such as equipment model, remaining life, availability, etc.

[0111] Task node T(l): the lth emergency task, which contains attributes such as task urgency, task type, geographic location, etc.

[0112] Relationship: Define the relationship between various resources and tasks, such as whether a vehicle can respond to a task, whether the equipment meets the task requirements, whether the medical staff's skills are suitable for the current task, etc. Use R(V,T), R(P,T), and R(E,T) to represent the matching relationship between various resources and tasks.

[0113] Furthermore, for each task, the present invention calculates the matching degree of all resources (vehicles, medical staff, equipment) related to it. The present invention needs to design a special calculation formula for each match, which not only considers the static attributes of the resources (such as skills, equipment models), but also combines real-time data (such as equipment status, task location, etc.) for dynamic weighting. New regularization terms and innovation terms are introduced here to improve the rationality of the matching degree.

[0114] Furthermore, each task T(l) needs to be matched with multiple resources, and the comprehensive matching degree between resources and tasks is calculated. To select the optimal resource allocation. The present invention proposes the following formula to calculate the matching degree:

[0115]

[0116] in, Represents task T(l) and resource R i (such as vehicles, medical staff, equipment). Indicates resource R i The degree of matching with a specific attribute of the task T(l). For example, the degree of matching between the skills of medical staff and the skills required by the task, the degree of matching between the equipment model and the task requirements, etc. k The weight of attribute k indicates the importance of different attributes. This weight can be obtained through regression model training based on historical data, or it can be manually set based on expert experience. l ,R i ) represents the real-time distance regularization term introduced innovatively, and represents the resource R i The difference in distance from task T(l) in geographic space, ΔD(T l ,R i ) is the geographical distance between the task and the resource, and λ is the regularization coefficient. The closer the distance, the higher the matching degree, and the resource is allocated first. i ) represents the resource load penalty term, C(R i ) is the current resource load, such as whether the ambulance is fully loaded, the workload of medical staff, etc., γ is the penalty coefficient. The higher the load, the lower the matching degree.

[0117] Furthermore, the urgency of the task must be considered in the calculation of the matching degree of the emergency task. The present invention designs a dynamic adjustment item for each task to ensure that high-urgency tasks are allocated resources first.

[0118]

[0119] Among them, μ l Indicates the dynamic priority adjustment coefficient of the lth task. lIt indicates the urgency of the lth task. The value of U is determined by the task emergency assessment model (such as 0-1 score). Indicates the sum of the urgency of all current tasks.

[0120] This formula ensures the relative importance of the urgency of the task in the entire task queue. For more urgent tasks, their resource allocation priority will be dynamically increased.

[0121] Furthermore, through the above-mentioned matching degree calculation and dynamic adjustment of task priority, the present invention can finally obtain the optimal matching solution between each task and resource. After sorting according to the matching degree results, the system will output the work order data including resource allocation and task priority. These work order data are the basis for subsequent task allocation and scheduling optimization.

[0122] Output work order: including task T(l) and its assigned vehicle V(i), medical staff P(j), equipment E(k) and its assigned time, execution route and other detailed information. Each work order is generated based on the above matching degree and priority calculation.

[0123] This step is one of the core steps of the emergency rescue platform. It builds a dynamic resource allocation and task priority management system based on knowledge graph. By introducing the real-time distance regularization term and resource load penalty term, it effectively solves the problem of ignoring real-time and resource load in the traditional matching model. Through the dynamic adjustment term of the urgency, it ensures that the response of the emergency rescue task can maximize the adaptation to the changes in the emergency situation.

[0124] Step 2: Decompose the task into several subtasks according to the work order data and assign the optimal combination of emergency resources to each subtask.

[0125] Specifically, the input of this step comes directly from the output of step 1. Specifically, it includes:

[0126] Task node T(l): Each task T(l) is generated by the knowledge graph in step 1, and contains attributes such as urgency, task type, and task location.

[0127] Resource Node R i : Resources include vehicles V(i), medical staff P(j), and equipment E(k). These nodes come from the knowledge graph and have real-time status attributes (such as load, skill level, geographic location, etc.).

[0128] Match This is the task T(l) and resource R calculated from step 1 based on real-time distance, resource load, task requirements, etc. i The comprehensive matching degree.

[0129] Task urgency adjustment coefficient μ l: The urgency weight of task T(l), determined by the dynamic adjustment in step 1.

[0130] These inputs will be used to generate emergency work orders, broken down into subtasks, and resources will be assigned.

[0131] Furthermore, the work order generation process is based on the reasoning and matching results of the knowledge graph. At this point, the tasks have been associated according to the matching and priority information. The present invention decomposes the tasks into subtasks and allocates the most suitable resources to each subtask.

[0132] Furthermore, each emergency task T(l) needs to be decomposed into several subtasks. The task decomposition strategy depends on the type and requirements of the task. Common task decomposition methods include:

[0133] Rescue mission: usually includes on-site first aid, patient transfer and equipment use.

[0134] Furthermore, the representation of subtasks is:

[0135] T(l)={T(l)1,T(l)2,…,T(l) n}

[0136] Among them, T(l) k Represents the kth subtask. For example, for task T(1) (cardiac arrest), it can be decomposed into: T(1)1: on-site first aid. T(1)2: patient transfer.

[0137] Furthermore, according to the matching degree calculated in step 1 and the urgency adjustment factor μ l , the system automatically allocates the best resource combination for each subtask. The present invention uses the following formula for each subtask T(l) k Choose the best resource:

[0138]

[0139] Among them, R(T(l) k ) represents subtask T(l) k The collection of resources to which it is allocated. Indicates resource R i With subtask T(l) k The matching degree of μ is directly inherited from the output of step 1. l It represents the urgency weight of task T(l). A task with higher urgency will be allocated resources first.

[0140] Example: For subtask T(1)1 (on-site first aid) of task T(1), the present invention adjusts the allocation to the following resources based on the maximum matching degree and urgency:

[0141] R(T(1)1)={V(2),P(4)}, i.e., ambulance V(2) and medical staff P(4).

[0142] Furthermore, in order to ensure reasonable allocation of resources and avoid resource conflicts or duplication, the system needs to follow the following rules:

[0143] Resource uniqueness: Each resource can only be assigned to one task or subtask, unless the resource can be used simultaneously.

[0144] Load check: The system automatically checks the current load of resources (such as whether the vehicle is idle, the task load of medical staff) to avoid over-allocation of resources. If a resource has too high a load, the system will reduce its priority through the load penalty item in the matching degree.

[0145] Prioritize task urgency: When resources conflict, the system will prioritize the needs of urgent tasks, and lower priority tasks may be delayed.

[0146] Furthermore, after the work order is generated, the system adjusts the coefficient μ according to the urgency of the task. l and matching Sort all tasks to ensure that the most urgent tasks are responded to first. The priority of each task is calculated as:

[0147]

[0148] in, Indicates the matching degree between tasks and resources obtained from step 1. l Represents the urgency adjustment coefficient of task T(l). Tasks with higher priority will be allocated resources first to ensure that tasks are processed efficiently according to their urgency.

[0149] Furthermore, the final output of the work order generation is a detailed task-resource allocation table, including the subtasks of each task, the allocated resources, and the priority order of each task. The work order output structure is:

[0150] Task T(l):

[0151] Task type, urgency l , geographical location.

[0152] Subtask T(l) k :

[0153] A detailed description of each subtask, such as first aid scene processing, patient transfer, etc.

[0154] The corresponding allocated resources R(T(l) k ), such as ambulance V(i), medical staff P(j), and equipment E(k).

[0155] Task Prioritization:

[0156] According to the task priority calculated by the priority formula, tasks will be processed in order from high to low.

[0157] Example ticket:

[0158] Task T(1): Cardiac arrest, urgency μ1 = 0.95

[0159] Subtask T(1)1: On-site first aid

[0160] Allocated resources: ambulance V (2), first aid personnel P (4)

[0161] Subtask T(1)2: Patient transfer

[0162] Allocate resources: Ambulance V(3)

[0163] Task T(2): Traffic accident, urgency μ2 = 0.85

[0164] Subtask T(2)1: First aid at the accident scene

[0165] Allocated resources: Ambulance V (5), First Aid Personnel P (7)

[0166] Step 3: In response to the ever-changing emergency scenarios, a multi-agent-based optimization strategy is set up to coordinate the allocation and control of each resource and optimize task allocation.

[0167] Specifically, in the scenario of an emergency rescue platform, resources (such as ambulances, medical staff, and equipment) must respond to changing emergency rescue tasks at any time, and the requirements of each task will change dynamically with time and resource status. Therefore, traditional multi-agent scheduling solutions often cannot handle the suddenness of tasks, resource allocation conflicts, and dynamic task priorities.

[0168] To this end, the present invention has made the following targeted designs:

[0169] Input data structure:

[0170] Work order T(l): Work order information is passed from step 2, including the task urgency μ l , task type, geographical location, initial resource allocation R(T(l) k ).

[0171] Match Task T(l) and resource R i The initial matching degree.

[0172] Dynamic state of resources: Not only the static attributes of resources (skills, location, etc.) are considered, but also dynamic loads (such as the current number of tasks of the ambulance, fatigue of medical staff, etc.) are introduced, which are not fully considered in steps 1 and 2.

[0173] Agent definition:

[0174] Vehicle Agent V(i): Each ambulance acts as an independent agent and can make decisions based on task matching and vehicle status.

[0175] Medical staff agent P(j): Each medical staff acts as an agent and selects tasks based on skill matching and load conditions.

[0176] Equipment agent E(k): Each emergency equipment acts as an agent to respond to mission requirements.

[0177] The states of these agents not only rely on the preliminary allocated data, but also take into account the dynamic changes of each resource, ensuring that the agents can optimize their decisions based on the latest conditions.

[0178] Specifically, in order to better match the real-time scheduling requirements of the emergency platform, the present invention specially designs the state space and reward mechanism of the intelligent agent to address the urgency and complexity of task scheduling.

[0179]

[0180] Among them, L i Indicates resource R i The current load of U. This state variable updates the resource usage in real time and helps the agent decide whether to participate in new tasks. i Indicates resource R i The availability of μ indicates whether the resource is idle. l It represents the urgency adjustment coefficient of task T(l), which is particularly important in emergency mission scenarios and directly affects the priority of resource scheduling. pos (T(l),R i ) represents the task T(l) and the resource R i Considering the time sensitivity in emergency scenarios, resources farther away will have lower priority. Indicates that the resource R i Collaboration status with other resources. For example, an ambulance and medical personnel may need to work together to complete a task.

[0181] Specifically, in order to make the multi-agent system better serve the emergency task allocation, the present invention adds some special items to the reward function, especially adjusting it for urgency and collaborative tasks.

[0182]

[0183] Among them, α·μ l Indicates that the urgency of the task is emphasized. α controls the priority increase of urgent tasks. In the emergency scenario, the present invention must ensure that high-urgency tasks can obtain resources first. Represents the matching weight between resources and tasks, and is used to reward resource agents that can complete tasks efficiently. collab (R i ) represents the collaborative work reward, which is a positive feedback specifically designed to consider cooperation between resources. For example, when a task requires the collaboration of an ambulance and equipment, agents V(i) and E(k) will receive additional rewards to encourage them to respond to the task together. i ) represents the resource load penalty item, which prevents resources from being overused or involved in tasks for a long time. It is suitable for fatigue management of medical staff and equipment life management, etc.

[0184] Specifically, in emergency scenarios, tasks often require the collaborative completion of multiple resources, such as medical staff, emergency equipment, and ambulances working together to handle the same task. To this end, the present invention designs a collaborative mechanism that optimizes task allocation between resources through game and negotiation mechanisms.

[0185] CollaborativeReward(R i ,R j )=C collab (R i ,R j )=λ·Efficiency(R i ,R j )-θ·D pos (R i ,R j )

[0186] Among them, λ represents the weight of collaborative efficiency, which indicates the efficiency of cooperation between resources. For example, when ambulances and medical staff work together, they can complete tasks faster. θ represents the distance penalty term. If the distance between two resources is far, the collaborative efficiency will be reduced. i ,R j ) indicates resource R i and R j The efficiency of cooperation on tasks is calculated using historical data.

[0187] Specifically, in multi-agent systems, task conflicts and resource competition are common challenges. Especially in emergency platforms, different tasks have different urgency, but resources are limited. The present invention designs a resource conflict resolution mechanism to ensure maximum resource allocation efficiency.

[0188]

[0189] in, and Represents resource A i and A j With task T(l) k The matching degree of μ l Represents the urgency adjustment coefficient of task T(l).

[0190] Specifically, after multiple rounds of agent learning and collaborative games, all resource allocation schemes will eventually converge to the optimal state and output the following:

[0191] The final resources allocated to each task T(l), including vehicles, medical staff, equipment, etc., with task urgency and resource coordination.

[0192] Specific resource execution paths ensure tasks are completed according to priority and time sensitivity.

[0193] Step 4: Monitor the optimized task allocation in real time, and combine it with the real-time data stream of real-time monitoring to ensure that the dynamic environment changes can be responded to at any time during the task execution process.

[0194] Specifically, the input of step 4 must be the output from step 3 and combined with real-time event stream data to ensure that the task execution process can respond to dynamic environmental changes at any time.

[0195] Further, according to the output of step 3:

[0196] Task allocation table T(l): including task priority μ l , resource allocation R(T(l)), path planning P(T(l)), etc.

[0197] Resource Status i : Dynamic load and availability status of resources generated by the agent system.

[0198] New real-time data:

[0199] Real-time event stream E(t): includes real-time changing data such as traffic, equipment status, and personnel location.

[0200] Furthermore, in order to dynamically adjust the work order to adapt to the real-time changes in the emergency scenario, the present invention designs an event stream processing mechanism based on the application of information technology innovation, which associates the event stream with the task work order, so that the system can automatically adjust the task execution and resource scheduling under emergencies.

[0201] Furthermore, the present invention models the real-time event stream E(t) as a time series, which represents multiple events occurring in different time periods.

[0202] E(t)={e1(t),e2(t),...,e n (t)}

[0203] Each event i (t) contains its metadata (event type, location, timestamp, etc.) These events will affect different aspects of task execution.

[0204] Furthermore, each event e i (t) has different effects on task T(l). The present invention measures the impact of events on task priority, path, resource allocation, etc. by designing an impact function:

[0205] ΔT l (e i (t)) = w i ·δ(T l ,e i (t))

[0206] Among them, w i Indicates event e i The weight of (t) depends on the event type and task requirements (for example, traffic congestion has a greater impact on the path, while equipment failure has a greater impact on resource allocation). l ,e i (t)) represents the event e i (t) The specific impact on task T(l) (such as delay, equipment replacement, etc.).

[0207] Furthermore, according to the impact of real-time events on tasks, the system will dynamically adjust the priority of tasks. The present invention introduces a priority adjustment formula based on the application of information technology innovation to ensure that the priority of tasks can be flexibly changed according to external events.

[0208]

[0209] Among them, μ′ l Indicates the task priority after real-time event stream processing. μ l represents the original task priority output from step 3. Δμ l (e i (t)) represents each event e i (t) The increment to the task priority.

[0210] Furthermore, according to the real-time event stream, the present invention designs a set of dynamic adjustment rules based on the application innovation of information technology to ensure that task execution and resource allocation are always optimized.

[0211] Specifically, when traffic conditions change (such as traffic jam or road closure), the system will use event E traffic (t) Recalculate the path to ensure that resources can arrive at the task location on time. Path adjustment formula:

[0212] P′(T(l))=P(T(l))+ΔD(E traffic (t))

[0213] Where P(T(l)) represents the original path of task T(l). ΔD(E traffic (t)) represents the traffic event E traffic (t) is the path or time increment triggered. P′(T(l)) represents the optimal path after dynamic adjustment.

[0214] When the device status (such as device failure) event E device When (t) occurs, the system will dynamically replace the affected device. Device replacement formula:

[0215] R′(T(l) k )=R(T(l) k )-E fault +E backup

[0216] Among them, R(T(l) k ) represents the current subtask T(l) k Resource allocation. fault Indicates a failed device. E backup Represents a spare device that automatically replaces the affected resource.

[0217] In emergency scenarios, patients may be transferred to different locations, and the system needs to quickly reallocate the nearest resources. The resource reallocation formula is as follows:

[0218]

[0219] in, Represents the task T(l) and the resource R i The matching degree. pos (T(l),R i ) represents the new location of task T(l) and resource R i R′(T(l)) represents the dynamically adjusted resource configuration.

[0220] Furthermore, in order to ensure that the system can quickly respond to the triggering of each event, the present invention designs an association mechanism between events and tasks to evaluate the actual impact of events on tasks and trigger corresponding adjustments.

[0221] Furthermore, the present invention uses an event impact measurement formula to measure the overall impact of each event on the task:

[0222]

[0223] Where I(T(l),E(t)) represents the overall impact index of event E(t) on task T(l). i represents the weight, based on the urgency of the event and the relevance of the task. i (t)) indicates the specific impact of the event on the task.

[0224] When the event impact index exceeds a certain threshold, the system will automatically trigger the adjustment mechanism.

[0225] Output: Dynamically adjusted work order

[0226] Furthermore, the work order after dynamic event stream processing will contain the following:

[0227] Task priority updates: Task priorities are updated in real time, based on the impact of events on tasks.

[0228] Resource allocation adjustment: Dynamically replace or adjust the resource configuration of tasks to ensure optimal utilization of resources.

[0229] Path adjustment: Dynamically adjust path planning based on traffic changes to reduce task delays.

[0230] Step 5: Re-enter the optimized task allocation and resource scheduling strategy into the knowledge graph for closed-loop optimization, including:

[0231] Build a feedback model, take task execution feedback as input, and adjust future work order scheduling strategies in real time by comparing the actual task execution with the plan.

[0232] Specifically, the input of step 5 comes from the output of step 4, and combined with the feedback data during the task execution process, the system will use this data to form a closed-loop feedback and continuously optimize the work order execution effect. The input data includes:

[0233] Dynamically adjusted work order T′(l): Task priority μ′ generated after real-time event processing l , path planning P′(T(l)), resource allocation R′(T(l)) and other data.

[0234] Execution feedback F(t): feedback information collected during task execution, including task completion time, task delay, resource utilization efficiency, task response time, etc.

[0235] Furthermore, in order to optimize the work order execution effect, the system forms a closed-loop feedback system based on the task execution feedback, and continuously adjusts and improves the task allocation and resource scheduling strategy. The present invention introduces a feedback model and uses the task execution feedback F(t) as the core input for optimizing the work order execution effect. The core of the closed-loop feedback model is to adjust the future work order scheduling strategy in real time by comparing the actual situation of task execution with the plan. Feedback model formula:

[0236] ΔT′(l)=λ·(P actual (T(l))-P′(T(l)))+γ·F(t)

[0237] Among them, ΔT′(l) represents the improvement of task execution effect, which is used to adjust future task orders. λ represents the path optimization coefficient, which weighs the difference between the planned path and the actual path. actual (T(l)) represents the actual execution path of the task. P′(T(l)) represents the optimized planned path, which is output from step 4. γ represents the feedback weight coefficient, which measures the impact of feedback data on task adjustment. F(t) represents the execution feedback data, including task response time, device status, etc.

[0238] Furthermore, by comparing the actual execution path P of the task actual (T(l)) and planned path P′(T(l)), the system can continuously optimize the path selection for future tasks. After each task is executed, the system will adjust the path planning strategy according to the deviation to ensure a more accurate path for the next task assignment. The core formula for path optimization is:

[0239] P″(T(l))=P′(T(l))+ΔP(T(l))

[0240] Where P″(T(l)) represents the optimized path at the next execution. ΔP(T(l)) represents the path deviation, which corrects the planned path based on the execution feedback. The path deviation ΔP(T(l)) is calculated from the path feedback data, including road congestion, real-time traffic data, and differences in resource response time.

[0241] Furthermore, the closed-loop feedback system optimizes future resource allocation strategies by analyzing the resource utilization efficiency of each task. The system adjusts the resource allocation weights based on feedback data, such as resource utilization and task execution time, to ensure that efficient resources are prioritized for urgent tasks. Resource allocation optimization formula:

[0242] R″(T(l))=R′(T(l))+δ·F(t)

[0243] Among them, R″(T(l)) represents the resource allocation result of the next task. δ represents the resource allocation weight coefficient. F(t) represents resource usage feedback, including device status, resource efficiency in task execution, etc. Through this formula, the system can dynamically adjust the resource allocation strategy to make the use of resources in future tasks more efficient and avoid repeated use of overloaded resources.

[0244] Furthermore, based on closed-loop feedback, the system can dynamically adjust task priorities, especially when resources are tight or tasks are delayed. Priority adjustment not only considers the original urgency of the task, but also l , and also combines the actual feedback data of task execution to ensure more accurate priority allocation. Priority optimization formula:

[0245] μ″ l =μ′ l +θ·F(t)

[0246] Among them, μ″ l represents the optimized task priority, which is used for the next round of task scheduling. θ represents the priority adjustment coefficient, which adjusts the priority based on the task feedback data. F(t) represents the feedback data of task execution, including information such as task delay and resource conflict. This optimization formula ensures that the system can dynamically respond to unexpected problems in task execution and avoid delays in subsequent tasks due to the same problem.

[0247] Furthermore, after closed-loop feedback optimization, the system will generate an optimized work order for the next task scheduling, including:

[0248] Optimized task priority μ″ l : Dynamic adjustment based on feedback ensures more accurate priority allocation.

[0249] Optimized resource allocation R″(T(l)): Resource allocation adjusted by feedback data to ensure efficient use of resources.

[0250] Optimized path planning P″(T(l)): A more accurate path planning for future task execution after optimization based on the actual execution path data.

[0251] In summary, the present invention provides a more intelligent and dynamic solution that can better integrate multi-source information, automatically allocate and coordinate resources, and respond to complex emergency mission requirements in real time to improve the overall efficiency and response speed of the emergency system. This requires not only the introduction of more advanced technical tools to manage complex resource networks, but also the development of algorithms that can be optimized and adjusted dynamically in real time to address the inherent defects of traditional systems.

[0252] These are just some preferred embodiments of the present invention, which certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of implementing the above embodiments and making equivalent changes according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A work order management method for an emergency platform based on digital twins and artificial intelligence, characterized in that: The method comprises the following steps: Step 1: Collect historical emergency data and real-time data of emergency scenarios to build a knowledge graph, and build a task scheduling system based on the knowledge graph, including: S11, obtaining historical emergency data and real-time data to form total data; S12. The emergency resources and tasks in the total data are used as nodes in the knowledge graph. The nodes are connected by relationship edges to represent the potential matching relationship between resources and tasks. Each relationship expresses the degree of association between nodes through a dynamic weight. S13. For each task, calculate the comprehensive matching degree between each task and multiple emergency resources to select the optimal resource allocation. At the same time, design an urgency adjustment coefficient for each task to ensure that resources are allocated to high-urgency tasks first; S14, calculating work order data including optimal resource allocation and optimal task priority; Step 2: Decompose the task into several subtasks according to the work order data and assign the best combination of emergency resources to each subtask, including: S21, automatically generate work orders based on the reasoning and matching results of the knowledge graph; S22, decompose each emergency task into several subtasks, and then allocate the optimal resource combination to each subtask according to the comprehensive matching degree and urgency adjustment coefficient; S23. After the work order of each subtask is generated, all tasks are sorted according to the comprehensive matching degree of the subtasks and the urgency adjustment coefficient to ensure that the most urgent tasks are responded to first and the final work order is output; Step 3: For emergency scenarios that change at any time, set up a multi-agent-based optimization strategy to coordinate the allocation and control of each resource and optimize task allocation, including: S31. Define the multi-agents to include: vehicle agent, medical staff agent and equipment agent; S32. Design the state space and reward mechanism of multiple agents according to the real-time scheduling requirements of the emergency platform to control multiple agents. At the same time, design a resource coordination mechanism to optimize task allocation through game and negotiation mechanisms between resources. Design a resource conflict resolution mechanism for conflicts between emergency tasks to ensure maximum resource allocation efficiency. S33. After multiple rounds of agent learning and collaborative game, all resource allocation schemes will eventually converge to the optimal state and be output; Step 4: Monitor the optimized task allocation in real time, and combine the real-time data stream of real-time monitoring to ensure that the dynamic environment changes can be responded to at any time during the task execution process, including: S41, collecting real-time data streams, and converting the real-time data streams into time series for modeling, wherein the real-time data streams contain multiple events occurring in different time periods, and designing different impact functions for each event on the task to measure the impact of the event, and then dynamically adjusting the priority of the task according to the impact of each event on the task; S42. Design dynamic adjustment rules based on real-time data streams to ensure that task execution and resource allocation are always optimal; S43. Design an event-task association mechanism to evaluate the actual impact of events on tasks and trigger corresponding adjustments. When the event impact index exceeds a certain threshold, the adjustment mechanism will be automatically triggered; S45. Output the optimized and adjusted work order; Step 5: Re-enter the optimized task allocation and resource scheduling strategy into the knowledge graph for closed-loop optimization, including: Build a feedback model, take task execution feedback as input, and adjust future work order scheduling strategies in real time by comparing the actual task execution with the plan.

2. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 1, characterized in that: The structure of the knowledge graph includes: The nodes include: The vehicle node V(i) represents the i-th emergency vehicle; The medical staff node P(j) represents the jth medical staff; The equipment node E(k) represents the kth emergency equipment; Task node T(l) represents the lth emergency task; Relationships include: Define the relationship between various resources and tasks, and use R(V,T), R(P,T), and R(E,T) to represent the matching relationship between each resource and task.

3. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 2 is characterized in that: The S13 specifically includes: Each task T(l) needs to be matched with multiple resources. The comprehensive matching degree between resources and tasks is calculated. To select the optimal resource allocation, it is expressed as follows: in, Represents task T(l) and resource R i The matching degree; Indicates resource R i The degree of match with a specific attribute of task T(l); w k represents the weight of attribute k, indicating the importance of different attributes; λ·ΔD(T l ,R i ) represents the real-time distance regularization term introduced innovatively, and represents the resource R i The difference in distance from task T(l) in geographic space, ΔD(T l ,R i ) is the geographical distance between the task and the resource, λ is the regularization coefficient; -γ·C(R i ) represents the resource load penalty term, C(R i ) is the current resource load level; In the matching calculation of emergency tasks, the urgency of the tasks needs to be taken into consideration. Therefore, a dynamic adjustment item is designed for each task to ensure that resources are allocated to high-urgency tasks first, as shown below: Among them, μ l Indicates the dynamic priority adjustment coefficient of the lth task. l It represents the urgency of the lth task, and the value of U is determined by the task emergency assessment model; Indicates the sum of the urgency of all current tasks.

4. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 1, characterized in that: According to the comprehensive matching degree and the urgency adjustment coefficient, the optimal resource combination is allocated to each subtask, as shown below: Among them, R(T(l) k ) represents subtask T(l) k the set of resources allocated; Indicates resource R i With subtask T(l) k The matching degree of l represents the urgency weight of task T(l); In S22, in order to ensure reasonable resource allocation and avoid resource conflicts or duplication, task allocation needs to follow the following rules: Each resource can be assigned to only one task or subtask, unless the resource can be used simultaneously; Automatically check the current load of resources to avoid over-allocation of resources. If a resource has too high a load, its priority will be lowered through the load penalty item in the matching degree. When there is a resource conflict, urgent tasks are given priority and low-priority tasks are delayed.

5. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 4 is characterized in that: The final work order is a task-resource allocation table, including the subtasks of each task, the allocated resources, and the priority order of each task. The final work order structure is shown as follows: Task T(l): Task type, urgency μ l , Geographical location; Subtask T(l) k :Detailed description of each subtask; corresponding allocation resources R(T(l) k ); Task priority sorting: According to the task priority calculated by the priority formula, tasks will be processed in order from high to low.

6. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 1, characterized in that: The S32 specifically includes: The state space and reward mechanism of the intelligent agent are designed to solve the urgency and complexity of task scheduling. The state space is expressed as follows: Among them, L i Indicates resource R i The current load of U i Indicates resource R i availability, indicating whether the resource is idle; μ l represents the urgency adjustment coefficient of task T(l); D pos (T(l),R i ) represents the task T(l) and the resource R i geographical distance; Indicates that the resource R i Collaboration status with other resources; The reward mechanism is as follows: Among them, α·μ l Indicates the emphasis on the urgency of the task; α controls the priority increase of urgent tasks; represents the matching weight between resources and tasks, which is used to reward resource agents that can complete tasks efficiently; δ·C collab (R i ) represents the collaborative work reward item, which is used to consider the positive feedback of cooperation between resources; γ·C(R i ) represents the resource load penalty term, which prevents resources from being overused or participating in tasks for a long time; The collaboration mechanism is expressed as follows: CollaborativeReward(R i ,R j )=C collab (R i ,R j )=λ·Efficiency(R i ,R j )-θ·D pos (R i ,R j ) Among them, λ represents the weight of collaborative efficiency, which indicates the efficiency of cooperation between resources; θ represents the distance penalty term. If the distance between two resources is far, the collaborative benefit will be reduced; Efficiency (R i ,R j ) indicates resource R i and R j The efficiency of cooperation on tasks is calculated using historical data; The resource conflict resolution mechanism is expressed as follows: in, and Represents resource A i and A j With task T(l) k The matching degree of l Represents the urgency adjustment coefficient of task T(l).

7. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 6 is characterized in that: Eventually, all resource allocation schemes will converge to the optimal state and be output, where the output content includes: The final resources assigned to each task, along with task urgency and resource coordination, as well as the specific resource execution path, ensure that tasks can be completed according to priority and time sensitivity.

8. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 1, characterized in that: The priority of the task is dynamically adjusted according to the impact of each event on the task, as shown below: Among them, μ′ l Indicates the task priority adjusted after real-time event stream processing; μ l Represents the original task priority; Δμ l (e i (t)) represents each event e i (t) The increment to the task priority.

9. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 1, characterized in that: The design of dynamic adjustment rules based on real-time data streams includes: When the traffic situation changes, through event E traffic (t) Recalculate the path to ensure that resources can arrive at the task location on time. The path adjustment formula is as follows: P′(T(l))=P(T(l))+ΔD(E traffic (t)) Where P(T(l)) represents the original path of task T(l); ΔD(E traffic (t)) represents the traffic event E traffic (t) is the path or time increment triggered; P′(T(l)) represents the optimal path after dynamic adjustment; When the device status event E device When (t) occurs, the affected device is dynamically replaced. The device replacement formula is as follows: R′(T(l) k )=R(T(l) k )-E fault +E backup Among them, R(T(l) k ) represents the current subtask T(l) k Resource allocation; E fault Indicates a faulty device; E backup Represents a spare device that automatically replaces the affected resource. In an emergency scenario, patients may be transferred to different locations, and the nearest resources need to be reallocated. The resource reallocation formula is as follows: in, Represents the task T(l) and the resource R i The matching degree of D pos (T(l),R i ) represents the new location of task T(l) and resource R i The geographical distance; R′(T(l)) represents the resource allocation after dynamic adjustment; The association mechanism between the events and tasks is as follows: Where I(T(l),E(t)) represents the overall impact index of event E(t) on task T(l); w i represents the weight, based on the urgency of the event and the relevance of the task; δ(T(l),e i (t)) indicates the specific impact of the event on the task.

10. The method for managing emergency platform work orders based on digital twins and artificial intelligence according to claim 1, characterized in that: The feedback model is expressed as follows: ΔT′(l)=λ·(P actual (T(l))-P′(T(l)))+γ·F(t) Among them, ΔT′(l) represents the improvement of task execution effect, which is used to adjust future task orders; λ represents the path optimization coefficient, which weighs the difference between the planned path and the actual path; P actual (T(l)) represents the actual execution path of the task; P′(T(l)) represents the optimized planned path; γ represents the feedback weight coefficient, which measures the impact of feedback data on task adjustment; F(t) represents the execution feedback data.

Images