Medical distribution method and system based on unmanned aerial vehicle

Abstract

The invention discloses a medical distribution method and system based on an unmanned aerial vehicle, belongs to the technical field of medical distribution, and is used for solving the problems that an existing medical distribution technology is poor in scene adaptability, low in cooperative scheduling efficiency, insufficient in algorithm solving capability, lack of real-time state feedback and iterative optimization capability, incapable of adjusting model parameters according to historical distribution data, poor in reliability and the like. And the distribution scheme is separated from the actual scene. The method comprises the steps of determining a demand priority and extracting a distribution demand point based on demand information of a medical distribution demand; constructing a corresponding multi-target optimization model for the medical distribution tasks with different demand priorities; solving an optimal distribution path of the current distribution task through a multi-objective optimization model; monitoring transportation tool state data and environment data in a task execution process in real time, and extracting a dynamic disturbance factor; and carrying out local solution recalculation on the multi-objective optimization model based on the dynamic disturbance factor, and adjusting a distribution scheme.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical delivery, in particular to a medical delivery method and system based on a UAV. BACKGROUND

[0002] In the field of medical services, efficient transfer of medical samples such as blood samples, body fluid samples, tissue samples, and emergency medical supplies is a key link to ensure diagnosis and treatment efficiency, improve patient prognosis, and respond to public health emergencies and natural disasters. Traditional medical delivery methods rely heavily on manually driven vehicles, which face problems such as slow response due to traffic congestion, high labor costs, and biological safety risks of samples in daily scenarios. In emergency scenarios, road disruptions and complex terrain further exacerbate the limitations of vehicle delivery, making it difficult to meet the timeliness and reliability requirements of medical supplies.

[0003] Unmanned aerial vehicle technology, with its advantages of autonomous flight and avoidance of ground obstacles, provides a good solution to the "last mile" problem of medical delivery, but its endurance is limited and cannot independently undertake large-scale and long-distance delivery tasks. Although existing vehicle-UAV collaborative delivery research has been carried out, there are still significant shortcomings: On the one hand, most models are not specifically designed for medical scenarios, do not consider the time window differences (such as emergency blood samples that need to be delivered within 1 hour) and vulnerability (such as samples that need to be shock-proof and temperature-controlled) of medical samples / supplies, and do not establish a demand priority classification mechanism. On the other hand, traditional optimization algorithms (such as genetic algorithms and basic artificial bee colony algorithms) have slow convergence speed and are prone to local optimization, making it difficult to respond to dynamically changing demands and road conditions in medical delivery. When the demand scale expands or the scenario switches, the resource utilization rate and delivery success rate decrease significantly.

[0004] In addition, in medical delivery, the functions of the intelligent management and control center are limited to basic task allocation, and do not integrate real-time road conditions, UAV / vehicle state monitoring, and dynamic adjustment capabilities. The constraint conditions of existing collaborative models do not cover key scenarios such as "vehicle-UAV energy replenishment collaboration", leading to frequent problems such as insufficient endurance of UAVs, gaps in the connection between vehicles and UAVs, and inability to meet the high requirements of medical delivery for stability and safety. SUMMARY

[0005] The embodiments of the present application provide a medical delivery method and system based on a UAV, which solves the following technical problems: the existing medical delivery technology has poor scene adaptability, low collaborative scheduling efficiency, insufficient algorithm solving ability, and lacks real-time state feedback and iterative optimization capability, which cannot adjust model parameters according to historical delivery data, resulting in a mismatch between the delivery scheme and the actual scene.

[0006] The embodiments of the present application adopt the following technical solutions: In one aspect, the embodiment of the present application provides a medical delivery method based on a UAV, which comprises the following steps: receiving a medical delivery demand, and determining a demand priority based on demand information of the medical delivery demand; extracting a delivery demand point based on a current set of medical delivery demands to be processed and corresponding demand priorities, and performing delivery task scheduling on idle transportation tools; wherein the idle transportation tools comprise idle vehicles and idle UAVs; constructing a corresponding multi-objective optimization model for medical delivery tasks with different demand priorities, and dynamically configuring the weight of each objective according to different delivery scenarios; wherein the delivery scenarios at least include daily scenarios and emergency scenarios; solving an optimal delivery path of a current delivery task through the multi-objective optimization model, and performing the task through a transportation tool allocated by the current delivery task; extracting a dynamic disturbance factor by monitoring the state data and environment data of the transportation tool in the task execution process in real time; and recalculating a local solution of the multi-objective optimization model based on the dynamic disturbance factor, and adjusting the delivery scheme; generating an evaluation report based on actual delivery data after the delivery task is completed, and optimizing the parameters of the multi-objective optimization model.

[0007] In one possible implementation, the medical delivery demand is received, and the demand priority is determined based on the demand information of the medical delivery demand, specifically comprising the following steps: receiving the medical delivery demand and extracting corresponding demand information; wherein the demand information at least includes: delivery weight, delivery volume, time window and destination information; dividing the medical delivery demand into corresponding demand priorities based on the demand information, comprising: dividing the medical delivery demand with a time window less than or equal to a first preset threshold into an emergency level, corresponding to a UAV priority direct flight and vehicle escort supplement strategy; dividing the medical delivery demand with a time window greater than the first preset threshold and less than or equal to a second preset threshold into a regular level, corresponding to a vehicle and UAV collaborative delivery strategy; dividing the medical delivery demand with a time window greater than the second preset threshold into a reserve level, corresponding to a vehicle dominant transportation and UAV remote point supplement strategy; wherein the demand priority is sorted as: emergency level > regular level > reserve level.

[0008] In one possible implementation, the medical delivery demand point is extracted based on a current set of medical delivery demands to be processed and corresponding demand priorities, and the delivery task scheduling is performed on the idle transportation tools, specifically comprising the following steps: reading all medical delivery demands to be processed by an intelligent management and control scheduling center to form the set of medical delivery demands; According to the destination information in each medical delivery demand, a corresponding delivery demand point set is extracted; The reachable demand point coordinates, maximum range, and load data of each idle transportation tool are input into a DBSCAN clustering algorithm, and the demand priority of each delivery demand point is used as a clustering weight to perform density clustering on the delivery demand point set, thereby obtaining a plurality of delivery clusters; According to the resource state of each idle transportation tool and the preset combination strategy of the transportation tools corresponding to different demand priorities, a corresponding transportation tool combination is assigned to each delivery cluster; If the current idle transportation tool cannot cover all the delivery clusters, the delivery cluster with a higher demand priority is preferentially assigned.

[0009] In a feasible implementation, for medical delivery tasks of different demand priorities, a corresponding multi-objective optimization model is constructed, specifically including: For a medical delivery task with an emergency level demand priority, a first multi-objective optimization model is constructed; wherein the optimization objectives of the first multi-objective optimization model are total delivery time, satisfaction, and task completion rate; For a medical delivery task with a regular level demand priority, a second multi-objective optimization model is constructed; wherein the optimization objectives of the second multi-objective optimization model are total delivery time, delivery cost, satisfaction, and task completion rate; For a medical delivery task with a reserve level demand priority, a third multi-objective optimization model is constructed; wherein the optimization objectives of the third multi-objective optimization model are delivery cost and task completion rate.

[0010] In a feasible implementation, the weights of each objective are dynamically configured according to different delivery scenarios, specifically including: When the delivery scenario is a daily scenario, a first weight combination is configured for each optimization objective of the multi-objective optimization model corresponding to different demand priorities; When the delivery scenario is an emergency scenario, a second weight combination is configured for each optimization objective of the multi-objective optimization model corresponding to different demand priorities; According to the emergency level under the emergency scenario, the weight ratio in the second weight combination is dynamically adjusted.

[0011] In a feasible implementation, the optimal delivery path of the current delivery task is solved through the multi-objective optimization model, specifically including: According to the demand information of the current delivery task, real-time delivery environment data is retrieved; wherein the real-time delivery environment data includes real-time traffic data and real-time disaster data between the delivery starting point and the destination; According to the real-time delivery environment data and a preset scene division condition, a current delivery scene is determined; Based on the demand priority of the current delivery task and the current delivery scene, a corresponding multi-objective optimization model is called, and the demand information is input into the multi-objective optimization model for solving; Based on an adaptive double-algorithm switching strategy, the multi-objective optimization model is solved, including: When the number of demand points is less than or equal to a third preset threshold, an artificial bee colony algorithm fused with a simulated annealing strategy is called for solving; when the number of demand points is greater than the third preset threshold, a genetic algorithm fused with a simulated annealing strategy is called for solving.

[0012] In a feasible implementation, the state data and environment data of the transportation tool in the task execution process are monitored in real time, and a dynamic disturbance factor is extracted, specifically including: The state data and environment data returned by the transportation tool in the task execution process are received in real time; Based on the state data and environment data, a disturbance event is identified; wherein the judgment indexes of the disturbance event at least include: road state, unmanned aerial vehicle power, and demand point number change; When the disturbance event is identified, a disturbance parameter is calculated according to the state data and environment data; wherein the disturbance parameter at least includes event emergency degree, event influence range, and remaining processing time; the remaining processing time is the ratio of the remaining time limit window and the estimated adjustment time; According to the disturbance parameter, the dynamic disturbance factor is calculated.

[0013] In a feasible implementation, based on the dynamic disturbance factor, the local solution of the multi-objective optimization model is recalculated, and the delivery scheme is adjusted, specifically including: When the dynamic disturbance factor is greater than a fourth preset threshold, based on the occurrence position of the disturbance event, the event influence area boundary is determined through a spatial topology algorithm; For the delivery demand points within the event influence area boundary, the local solution is recalculated through the multi-objective optimization model, and the local recalculated path within the event influence area is obtained; Based on the local recalculated path, instructions are sent to the transportation tool to adjust the delivery scheme in real time to cope with the disturbance event.

[0014] In a feasible implementation, after the delivery task is completed, an evaluation report is generated based on the actual delivery data, and the parameters of the multi-objective optimization model are optimized, specifically including: After the delivery task is completed, actual delivery data is obtained; wherein the actual delivery data at least includes: actual delivery time, actual delivery satisfaction, and actual delivery cost; According to the actual delivery data and the expected delivery data, a delivery data deviation is calculated, and a preset template is called to generate an evaluation report; Based on a preset period, the evaluation reports are summarized, and data items to be optimized whose delivery data deviation is higher than an expected threshold are extracted; In a corresponding multi-objective optimization model, an optimization target corresponding to the data item to be optimized is adjusted in target weight and optimized in parameter, and the multi-objective optimization model is updated.

[0015] In another aspect, the embodiments of the present application also provide a medical delivery system based on a UAV, which comprises: A task scheduling module is configured to receive a medical delivery demand, determine a demand priority based on demand information of the medical delivery demand, extract a delivery demand point based on a current set of medical delivery demands to be processed and a corresponding demand priority, and schedule a delivery task for an idle transportation tool; wherein the idle transportation tool comprises an idle vehicle and an idle UAV. A path planning module is configured to construct a corresponding multi-objective optimization model for a medical delivery task of different demand priorities, and dynamically configure each target weight according to different delivery scenarios; wherein the delivery scenarios at least include a daily scenario and an emergency scenario; the optimal delivery path of the current delivery task is solved through the multi-objective optimization model, and the transportation tool allocated by the current delivery task is used to perform the task. A task delivery module is configured to monitor transportation tool state data and environment data in a task execution process in real time, extract a dynamic disturbance factor, re-calculate a local solution of the multi-objective optimization model based on the dynamic disturbance factor, adjust a delivery scheme, generate an evaluation report based on actual delivery data after the delivery task is completed, and optimize parameters of the multi-objective optimization model.

[0016] Compared with the prior art, the medical delivery method and system based on a UAV provided by the embodiments of the present application have the following beneficial effects: 1. Improved scene adaptability: by establishing a three-level demand priority system, a multi-objective optimization model containing different optimization targets is constructed for delivery demands of different priorities, and further, optimization target weights of the multi-objective optimization model are configured for daily and emergency scenarios respectively, so that the medical delivery is scheduled on demand, and the on-time rate of emergency-grade materials is improved, meeting the timeliness requirement of medical scenarios.

[0017] 2. Synergistic efficiency optimization: By establishing a vehicle and unmanned aerial vehicle cooperative energy supplement mechanism, the improved DBSCAN clustering algorithm reduces the unmanned aerial vehicle empty flight mileage, while shortening the vehicle detour distance, fully combining the respective transportation advantages of vehicles and unmanned aerial vehicles, and avoiding the respective limitations of the two. For example, the short endurance of unmanned aerial vehicles and the easy lengthening of the driving route of vehicles due to road restrictions, etc. Through vehicle and unmanned aerial vehicle cooperation, the transportation efficiency is greatly improved, and the total transportation time is shortened.

[0018] 3. Enhanced algorithm solving capability: The present application proposes an adaptive double-algorithm switching strategy, which dynamically switches between the improved artificial bee colony algorithm combined with simulated annealing and the genetic algorithm combined with simulated annealing according to the number of distribution demand points. A dynamic disturbance factor is added, which greatly shortens the local solution recalculation time compared to the original time when the road condition suddenly changes, meeting the real-time decision-making needs.

[0019] 4. Improved reliability of management and control: A closed-loop system of "demand access - scheduling solution - execution feedback - iterative optimization" is constructed, and the intelligent management and control center monitors the unmanned aerial vehicle / vehicle state in real time, optimizes the model parameters every month combined with historical data, and continuously optimizes the multi-objective optimization model, reduces the failure rate of the distribution scheme, and improves the satisfaction of medical service objects. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, a brief introduction will be given below to the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art without creative labor. In the drawings: Figure 1 A medical distribution method based on unmanned aerial vehicles is provided for the embodiments of the present application; Figure 2 A medical distribution scheduling model design scheme for the embodiments of the present application is provided; Figure 3 A medical distribution system based on unmanned aerial vehicles is provided for the embodiments of the present application. DETAILED DESCRIPTION

[0021] In order to make those skilled in the art better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0022] The embodiment of the present application provides a medical distribution method based on a UAV, as shown in the figure, the medical distribution method based on the UAV specifically comprises steps S101-S106: Figure 1 S101, receiving a medical distribution demand, and determining a demand priority based on demand information of the medical distribution demand.

[0023] Specifically, to solve the problems of low calculation efficiency, poor robustness or difficulty in guaranteeing the timeliness of key tasks of the existing medical scheduling method when dealing with complex scenes of high dynamics, strong constraints and multi-objective optimization, the present application proposes a medical distribution scheduling model based on a UAV under random task demand, Figure 2 A medical distribution scheduling model based on a UAV provided by the embodiment of the present application is shown in the figure, as shown in the figure, the medical distribution scheduling model based on the UAV mainly comprises the following entity elements: an intelligent management and control scheduling center, a hospital, a UAV, a distribution vehicle and a medical sample package (blood sample, body fluid sample, tissue sample and other samples). Figure 2

[0024] The specific distribution method is as follows: First, the intelligent management and control scheduling center receives a medical distribution demand and extracts corresponding demand information. The demand information at least includes: distribution weight, distribution volume, time window and destination information.

[0025] Further, the intelligent management and control scheduling center divides the medical distribution demand into corresponding demand priorities based on the demand information, including: The medical distribution demand with a time window less than or equal to a first preset threshold is divided into an emergency level, corresponding to a UAV priority direct flight and vehicle escort supplement strategy. The medical distribution demand with a time window greater than the first preset threshold and less than or equal to a second preset threshold is divided into a regular level, corresponding to a vehicle and UAV collaborative distribution strategy. The medical distribution demand with a time window greater than the second preset threshold is divided into a reserve level, corresponding to a vehicle dominant transportation and UAV remote point supplement strategy. The sorting of the demand priority is: emergency level> regular level> reserve level.

[0026] As a feasible implementation manner, the API interface of the intelligent management and control scheduling center is connected with the hospital HIS system and the emergency material management platform, and the distribution weight, distribution volume, time window and destination position of each medical distribution demand are automatically extracted. For missing information, such as unknown volume, the missing information is filled according to the preset coefficient of similar materials, such as the drug volume coefficient of 0.002 m 3 / kg, to ensure that the demand information completeness rate is greater than or equal to 99%.

[0027] In one embodiment, the demand priority division rule is: ​​The first preset threshold is set to 1 hour, and the second preset threshold is set to 4 hours.

[0028] Emergency level determination: time window ≤ 1 hour, or the material type is an emergency type, such as a blood sample, a thrombolytic drug, etc. The corresponding strategy is "drone priority direct flight + vehicle escort energy supplement", the vehicle arrives at the drone landing point 5 minutes in advance, and prepares a 600W fast charging device, which charges 50% in 10 minutes.

[0029] Routine level determination: 1 hour < time window ≤ 4 hours, or the material type is a routine type, such as a body fluid sample, a general detection reagent, etc. The corresponding strategy is "vehicle + drone cooperation", the vehicle is transported to the cluster edge take-off and landing point, and the drone is responsible for the end 3km distribution.

[0030] Reserve level determination: time window > 4 hours, or the material type is a reserve type, such as medical consumables, masks, etc. The corresponding strategy is "vehicle dominant transportation + drone remote point supplement", the vehicle is transported to the township distribution point, and the drone is distributed to the remote demand point more than 5km.

[0031] Priority sorting strictly follows "emergency level > routine level > reserve level", and high-priority demand can preempt low-priority resources, such as spare drones being preferentially allocated to emergency level distribution demand.

[0032] S102, based on the current set of medical distribution demands to be processed and the corresponding demand priorities, extract the distribution demand points, and schedule the distribution tasks for the idle transportation tools; wherein the idle transportation tools include idle vehicles and idle drones.

[0033] Specifically, read all the medical distribution demands currently to be processed by the intelligent management and control scheduling center to form a medical distribution demand set. According to the destination information in each medical distribution demand, the corresponding distribution demand point set is extracted.

[0034] Further, the reachable demand point coordinates, maximum range, and load data of each idle transportation tool are input into the DBSCAN clustering algorithm, and the demand priority of each distribution demand point is used as the clustering weight. The distribution demand point set is density clustered to obtain a plurality of distribution clusters.

[0035] Further, according to the resource state of each idle transportation tool and the preset combination strategy of transportation tools corresponding to different demand priorities, a corresponding transportation tool combination is allocated to each distribution cluster. If the current idle transportation tool cannot cover all the distribution clusters, the distribution cluster with a higher demand priority is preferentially allocated.

[0036] As a feasible implementation, the intelligent management and control scheduling center reads the current to-be-processed demand, filters the completed / cancelled demand, and constitutes a demand set. Then, the demand point set is generated by extracting the delivery demand point according to the longitude and latitude of the destination of each demand. The format can be: demand point ID-longitude-latitude-demand priority. And remove the repeated points, and the longitude and latitude deviation < 10 meters are regarded as repeated points.

[0037] The DBSCAN clustering algorithm implementation is as follows: Input data: demand point set, maximum range H of idle unmanned aerial vehicle, maximum load G, demand priority label; Parameter setting: search radius R = 0.8 x H, minimum number of points in the cluster MinPts is adjusted according to the priority (emergency level MinPts = 3, regular level / reserve level MinPts = 2), and the clustering weight is “priority x 0.6 + distance x 0.4”; Clustering process: traverse the unvisited demand points, if the number of neighbor points within the search radius R of a point ≥ MinPts and the proportion of emergency level demand ≥ 20%, mark it as a core point and expand it into a cluster. Finally, a number of delivery clusters are obtained.

[0038] The idle transportation tool scheduling and allocation process is as follows: Firstly, read the idle unmanned aerial vehicle state, including power, location, remaining load, etc.; read the idle vehicle state, including load rate, location, driving mileage, etc.

[0039] Then, sort according to “cluster priority-resource matching degree”, emergency level cluster is preferentially allocated unmanned aerial vehicle (power ≥ 70%, load ≥ total demand weight), regular level cluster is allocated “vehicle + unmanned aerial vehicle” combination, and reserve level cluster is preferentially allocated vehicle. If the resources are insufficient, high-priority clusters are preferentially covered, such as reserve level clusters can be delayed allocation, and ensure that emergency level clusters are 100% covered.

[0040] Finally, generate a “cluster-transportation tool” corresponding table, including vehicle route, unmanned aerial vehicle landing point and task order.

[0041] S103, for different demand priority medical delivery tasks, a corresponding multi-objective optimization model is constructed, and different distribution scenarios are dynamically configured according to different distribution scenarios. The distribution scenarios at least include daily scenarios and emergency scenarios.

[0042] Specifically, for the medical delivery task with emergency level demand priority, a first multi-objective optimization model is constructed; wherein the optimization objectives of the first multi-objective optimization model are: total delivery time, satisfaction and task completion rate.

[0043] Further, for the medical delivery task with the demand priority of the regular level, a second multi-objective optimization model is constructed; wherein, the optimization objectives of the second multi-objective optimization model are: total delivery time, delivery cost, satisfaction and task completion rate.

[0044] Further, for the medical delivery task with the demand priority of the reserve level, a third multi-objective optimization model is constructed; wherein, the optimization objectives of the third multi-objective optimization model are: delivery cost and task completion rate.

[0045] As a feasible implementation, the first multi-objective optimization model corresponds to the emergency diagnosis level delivery demand: The optimization objectives are: total delivery time (minimization), satisfaction (maximization), and task completion rate (100%); The objective function is: Min T_total, Max S, Max C; wherein, T_total is the maximum value of the delivery time of all emergency diagnosis level demands, S is the hospital feedback satisfaction, and C is the task completion rate; The constraint conditions are: the single flight mileage of the unmanned aerial vehicle is less than or equal to 0.8 times the maximum range, the remaining power is greater than or equal to 15% (when the power is supplemented), and the delivery time is less than or equal to the time limit for delivery.

[0046] The second multi-objective optimization model corresponds to the regular level delivery demand: The optimization objectives are: total delivery time (minimization), delivery cost (minimization), satisfaction (maximization), and task completion rate (100%); The objective function is: Min T_total, Min Cost, Max S, Max C; wherein, Cost contains vehicle fuel consumption and unmanned aerial vehicle power consumption; The constraint conditions are: the vehicle load is less than or equal to the maximum capacity, the unmanned aerial vehicle load is less than or equal to the maximum load, and the delivery time is less than or equal to the time limit for delivery. The third multi-objective optimization model corresponds to the reserve level delivery demand: The optimization objectives are: delivery cost (minimization) and task completion rate (100%); The objective function is: Min Cost, Max C; The constraint conditions are: the vehicle driving mileage is less than or equal to the preset upper limit, and the storage requirements of the materials have no special time limit, such as normal temperature transportation.

[0047] By analyzing the urgency of the delivery target corresponding to the delivery demand of different emergency degree, it is concluded that not all delivery demands need to achieve the optimization of each target. For example, very urgent delivery demand has very high requirements for delivery time efficiency, in such emergency situation, the delivery cost is not too high, and even completely without considering the cost, the shortest delivery time, the highest delivery success rate and the highest customer satisfaction can be used as the standard. For the regular level of delivery task, the time requirement is not so high, at this time, various factors can be considered comprehensively to achieve the perfect balance of delivery time, delivery cost, customer satisfaction and delivery success rate. For the reserve level of delivery task, the delivery time and customer satisfaction are not too high, as long as the cost is successfully delivered to the destination, therefore, only the delivery cost and the delivery success rate can be considered. Based on the actual existing demand difference, the multi-objective optimization model of different targets is constructed for different priority delivery demands, which can adapt to various demand delivery orders.

[0048] Further, when the delivery scene is a daily scene, a first weight combination is configured for each optimization target of the multi-objective optimization model corresponding to different demand priority. When the delivery scene is an emergency scene, a second weight combination is configured for each optimization target of the multi-objective optimization model corresponding to different demand priority. According to the emergency level under the emergency scene, the weight ratio in the second weight combination is dynamically adjusted.

[0049] In one embodiment, the delivery scene includes a daily scene and an emergency scene. Under the daily scene, the weight of the second multi-objective optimization model is: delivery cost weight 35%, total delivery time weight 30%, satisfaction weight 25%, and task completion rate weight 10%. Under the emergency scene, the weight of the second multi-objective optimization model is: delivery cost weight 10%, total delivery time weight 40%, task completion rate weight 30%, and satisfaction weight 20%. It can be seen that under the emergency scene, the target weights of delivery cost and satisfaction are appropriately lowered, and the delivery efficiency and task completion rate under the disaster situation and emergency scene are preferentially met.

[0050] S104, the optimal delivery path of the current delivery task is solved by the multi-objective optimization model, and the task is executed by the transportation tool allocated by the current delivery task.

[0051] Specifically, according to the demand information of the current delivery task, real-time delivery environment data is retrieved; wherein the real-time delivery environment data includes real-time road condition data and real-time disaster data between the delivery starting point and the destination.

[0052] Further, according to the real-time delivery environment data and the preset scene division condition, it is determined whether the current delivery scene is a daily scene or an emergency scene.

[0053] Further, based on the demand priority of the current delivery task and the current delivery scenario, a corresponding multi-objective optimization model is called, and the demand information is input into the multi-objective optimization model for solving.

[0054] Wherein, based on the adaptive double algorithm switching strategy, the multi-objective optimization model is solved, including: When the number of demand points is less than or equal to the third preset threshold, the artificial bee colony algorithm fused with the simulated annealing strategy is called for solving; when the number of demand points is greater than the third preset threshold, the genetic algorithm fused with the simulated annealing strategy is called for solving.

[0055] As a feasible implementation manner, the intelligent management and control scheduling center interfaces the following data through an interface: Real-time traffic data: obtained from the Gaode / Baidu map API, including road smoothness / semi-blockage / complete blockage state; real-time disaster data: obtained from the emergency management department platform, including disaster impact range, demand point addition / cancellation; transportation tool environment data: unmanned aerial vehicle aerial image, vehicle millimeter wave radar data, used for identifying road obstacles. If there is no warning from the emergency management department and the proportion of emergency level demand is <10%, it is a daily scenario; if there is a warning and the proportion of emergency level is ≥30%, it is an emergency scenario.

[0056] Multi-objective optimization model calling and input: based on the demand priority, the corresponding model is called, and the demand information (weight, volume, time window), transportation tool parameters (unmanned aerial vehicle range / load, vehicle capacity), and real-time environment data (traffic / disaster) are input.

[0057] The third preset threshold is set to 50, when the number of demand points ≤50, the artificial bee colony algorithm fused with the simulated annealing is adopted, and when the number of demand points >50, the genetic algorithm fused with the simulated annealing is adopted.

[0058] S105, real-time monitoring of transportation tool state data and environment data in the task execution process, extracting dynamic disturbance factors; based on the dynamic disturbance factors, the local solution of the multi-objective optimization model is recalculated, and the delivery scheme is adjusted.

[0059] Specifically, the state data and environment data returned by the transportation tool in the task execution process are received in real time. Based on the state data and environment data, disturbance events are identified; wherein, the judgment indexes of the disturbance events at least include: road state, unmanned aerial vehicle power, and demand point number change.

[0060] Further, when the disturbance event is identified, the disturbance parameters are calculated according to the state data and environment data; wherein, the disturbance parameters at least include event emergency degree, event influence range and remaining processing time; the remaining processing time is the ratio of the remaining time window and the estimated adjustment time. According to the disturbance parameters, the dynamic disturbance factors are calculated.

[0061] Further, when the dynamic disturbance factor is greater than a fourth preset threshold, based on the occurrence position of the disturbance event, the event influence area boundary is determined by a spatial topology algorithm. For the distribution demand points within the event influence area boundary, a multi-objective optimization model is used for local solution recalculation to obtain the local recalculation path within the event influence area.

[0062] Further, based on the local recalculation path, an instruction is sent to the transportation tool to adjust the distribution scheme in real time to cope with the disturbance event.

[0063] As a feasible implementation, the unmanned aerial vehicle returns the "position, remaining power, flight attitude" every 10 seconds through the 4G / 5G private network; the vehicle returns the "driving speed, remaining oil / power, load rate, road condition image" every 30 seconds; the unmanned aerial vehicle returns the regional road condition image once every 5 minutes; and the third-party platform pushes the road state update once every 1 minute.

[0064] Based on the preset threshold, three types of disturbance events are identified: road state disturbance: the road changes from smooth to semi-blocked or from semi-blocked to completely blocked; unmanned aerial vehicle state disturbance: the remaining power of the unmanned aerial vehicle is less than 20% and / or the acceleration change rate is greater than 5g / s and / or the navigation signal loss is greater than 10 seconds; demand point disturbance: new emergency level demand, demand point cancellation, and demand point position change.

[0065] Disturbance parameter calculation: Event urgency : emergency level demand = 1.0, regular level demand = 0.6, and reserve level demand = 0.3; Event influence range : number of affected demand points / total number of demand points (e.g. 5 points affected, total demand 20, = 0.25); Remaining processing time : (remaining time window - estimated adjustment time) / remaining time window (e.g. 30 minutes remaining, 5 minutes required for adjustment, = 25 / 30 ≈ 0.83); Dynamic disturbance factor calculation: the formula is λ = α × 0.5 + β × 0.3 + γ × 0.2 (the weight is based on the priority setting of the emergency scene), and the result range is 0-1.

[0066] In a further implementation, the fourth preset threshold is set to 0.5, and when λ ≥ 0.5, local solution recalculation is triggered, and when λ < 0.5, only a warning is given without adjustment.

[0067] Event influence area boundary determination: using a spatial topology algorithm, taking the disturbance event occurrence position as the center: Road blockage / UAV failure (primary disturbance, λ≥0.7): draw an impact area with a radius of 5 km, including the demand points in the area that have not completed delivery; demand point addition / power warning (secondary disturbance, 0.5≤λ<0.7): draw an impact area with a radius of 3 km. Finally, output the list of demand points in the impact area (including ID, priority, and time window remaining time).

[0068] Local solution recalculation of multi-objective optimization model: Input data: demand point information in the impact area, current idle transportation tools, and real-time environmental data; Solving algorithm: a lightweight greedy algorithm is used, with the following steps: sort the demand points in the impact area by priority; match the nearest available transportation tool for high-priority demand; check constraints, and replace the transportation tool if not satisfied. Solution time control: ≤3 minutes (primary disturbance), ≤5 minutes (secondary disturbance).

[0069] Delivery scheme adjustment and instruction issuance: After generating the local recalculated path, new instructions are issued to the affected transportation tools through the MQTT protocol, including new routes, takeoff and landing points, and task order. "Scheme update notification" (APP / SMS) is pushed to the demand point hospital / health center to ensure that the recipient is informed. Real-time monitoring of the progress of the adjusted task is performed every 30 seconds, and the status is updated until the task is completed.

[0070] S106, after the completion of the delivery task, an evaluation report is generated based on the actual delivery data, and the parameters of the multi-objective optimization model are optimized.

[0071] Specifically, after the completion of the delivery task, actual delivery data is obtained; wherein the actual delivery data at least includes: actual delivery time, actual delivery satisfaction and actual delivery cost; According to the actual delivery data and the expected delivery data, the delivery data deviation is calculated, and the preset template is called to generate the evaluation report; Based on the preset period, the evaluation report is summarized, and the to-be-optimized data items with delivery data deviation higher than the expected threshold are extracted; In the corresponding multi-objective optimization model, the optimization target corresponding to the to-be-optimized data item is adjusted and the parameters are optimized, and the multi-objective optimization model is updated.

[0072] In addition, the embodiment of the present application also provides a medical delivery system based on a UAV, as shown in Figure 3 As shown in the figure, the medical delivery system based on the UAV 300 specifically comprises: The task scheduling module 310 is configured to receive a medical distribution demand, determine a demand priority based on demand information of the medical distribution demand, extract a distribution demand point based on a current set of medical distribution demands to be processed and the corresponding demand priorities, and perform distribution task scheduling on an idle transportation tool; the idle transportation tool includes an idle vehicle and an idle unmanned aerial vehicle; The path planning module 320 is configured to construct a corresponding multi-objective optimization model for a medical distribution task of different demand priorities, and dynamically configure the weight of each target according to different distribution scenarios; the distribution scenarios include at least a daily scenario and an emergency scenario; the optimal distribution path of the current distribution task is solved through the multi-objective optimization model, and the transportation tool allocated by the current distribution task is used to perform the task. The task distribution module 330 is configured to monitor the transportation tool state data and the environment data in the task execution process in real time, extract a dynamic disturbance factor, perform local solution recalculation of the multi-objective optimization model based on the dynamic disturbance factor, adjust the distribution scheme, generate an evaluation report based on the actual distribution data after the distribution task is completed, and perform parameter optimization on the multi-objective optimization model.

[0073] Each of the embodiments in the present application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment mainly describes the difference from other embodiments. Especially, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the related parts can be referred to the part of the method embodiment.

[0074] The above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0075] The above only describes the embodiments of the present application and is not used to limit the present application. The embodiments of the present application can be variously changed and modified by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of the present application shall be included in the protection scope of the present application.

Claims

1. A method for medical delivery based on unmanned aerial vehicles (UAVs), characterized in that, The method includes: Receive medical delivery requests and determine request priorities based on the request information; Based on the current set of pending medical delivery demands and their corresponding priorities, delivery demand points are extracted, and delivery tasks are scheduled for available transportation vehicles; wherein, the available transportation vehicles include available vehicles and available drones; For medical delivery tasks with different demand priorities, a corresponding multi-objective optimization model is constructed, and the weights of each objective are dynamically configured according to different delivery scenarios; wherein, the delivery scenarios include at least routine scenarios and emergency scenarios; The optimal delivery route for the current delivery task is solved using the multi-objective optimization model, and the task is executed using the transportation vehicles allocated for the current delivery task. Real-time monitoring of vehicle status and environmental data during task execution; extraction of dynamic disturbance factors; recalculation of local solutions for a multi-objective optimization model based on the dynamic disturbance factors; and adjustment of the delivery plan. After the delivery task is completed, an evaluation report is generated based on the actual delivery data, and the parameters of the multi-objective optimization model are optimized.

2. The medical delivery method based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, Receive medical delivery requests and, based on the request information, determine the priority of the requests, specifically including: Receive medical delivery requests and extract corresponding request information; wherein, the request information includes at least: delivery weight, delivery volume, delivery time window, and destination information; Based on the demand information, the medical delivery demand is assigned to corresponding demand priorities, including: Medical delivery needs with a time window less than or equal to the first preset threshold are classified as emergency level, corresponding to a drone priority direct flight and vehicle escort and refueling strategy. Medical delivery needs with a time window greater than the first preset threshold and less than or equal to the second preset threshold are classified as regular, corresponding to a vehicle and drone collaborative delivery strategy. Medical delivery needs with a time window greater than the second preset threshold are classified as reserve level, corresponding to vehicle-led transportation and drone supplementation strategies for remote locations. The priority order of demand is: emergency level > routine level > reserve level.

3. The medical delivery method based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, Based on the current set of pending medical delivery demands and their corresponding priorities, delivery demand points are extracted, and delivery tasks are scheduled for available transportation vehicles. Specifically, this includes: Read all medical delivery requests currently pending processing from the intelligent management and dispatch center to form the medical delivery request set; Based on the destination information in each medical delivery request, extract the corresponding set of delivery request points; The coordinates of the reachable demand points, maximum range, and load data of each available transportation vehicle are input into the DBSCAN clustering algorithm, and the demand priority of each delivery demand point is used as the clustering weight to perform density clustering on the set of delivery demand points to obtain several delivery clusters. Based on the resource status of each idle transportation vehicle and the preset combination strategy of transportation vehicles corresponding to different demand priorities, a corresponding combination of transportation vehicles is assigned to each delivery cluster. If the current available transportation vehicles cannot cover all delivery clusters, then delivery clusters with higher demand priority will be allocated priority.

4. A medical delivery method based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, For medical delivery tasks with different demand priorities, corresponding multi-objective optimization models are constructed, specifically including: For medical delivery tasks with an emergency-level demand priority, a first multi-objective optimization model is constructed; wherein, the optimization objectives of the first multi-objective optimization model are: total delivery time, satisfaction, and task completion rate; For medical delivery tasks with a demand priority of routine level, a second multi-objective optimization model is constructed; wherein, the optimization objectives of the second multi-objective optimization model are: total delivery time, delivery cost, satisfaction, and task completion rate; For medical delivery tasks with a demand priority of reserve level, a third multi-objective optimization model is constructed; wherein, the optimization objectives of the third multi-objective optimization model are: delivery cost and task completion rate.

5. A method for medical delivery based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The weights of each target are dynamically configured according to different delivery scenarios, specifically including: When the delivery scenario is a routine scenario, configure a first weight combination for each optimization objective of the multi-objective optimization model corresponding to different demand priorities; When the delivery scenario is an emergency scenario, a second weight combination is configured for each optimization objective of the multi-objective optimization model corresponding to different demand priorities; The weighting ratio in the second weighting combination is dynamically adjusted according to the emergency level in the emergency scenario.

6. A medical delivery method based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The optimal delivery route for the current delivery task is solved using the multi-objective optimization model, specifically including: Based on the current delivery task requirements, real-time delivery environment data is retrieved; wherein, the real-time delivery environment data includes real-time road condition data and real-time disaster data between the delivery origin and destination; The current delivery scenario is determined based on the real-time delivery environment data and preset scenario classification conditions; Based on the priority of the current delivery task and the current delivery scenario, the corresponding multi-objective optimization model is invoked, and the requirement information is input into the multi-objective optimization model for solving. The multi-objective optimization model is solved based on an adaptive dual-algorithm switching strategy, including: When the number of demand points is less than or equal to the third preset threshold, the artificial bee colony algorithm incorporating simulated annealing strategy is invoked to solve the problem; when the number of demand points is greater than the third preset threshold, the genetic algorithm incorporating simulated annealing strategy is invoked to solve the problem.

7. A method for medical delivery based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, Real-time monitoring of vehicle status and environmental data during task execution, and extraction of dynamic disturbance factors, specifically including: Receive real-time status and environmental data returned by the transportation vehicle during task execution; Based on the aforementioned status and environmental data, disturbance events are identified; wherein, the indicators for judging disturbance events include at least: road conditions, drone battery level, and changes in the number of demand points; When a disturbance event is identified, disturbance parameters are calculated based on the state data and environmental data; wherein, the disturbance parameters include at least the urgency of the event, the scope of the event's impact, and the remaining processing time; the remaining processing time is the ratio of the remaining time window to the estimated adjustment time; The dynamic disturbance factor is calculated based on the disturbance parameters.

8. A method for medical delivery based on unmanned aerial vehicles according to claim 7, characterized in that, Based on the aforementioned dynamic disturbance factor, the local solution of the multi-objective optimization model is recalculated, and the delivery plan is adjusted, specifically including: When the dynamic disturbance factor is greater than the fourth preset threshold, the boundary of the event-affected area is determined by a spatial topology algorithm based on the location of the disturbance event. For delivery demand points within the boundary of the event-affected area, the local recalculation path within the event-affected area is obtained by recalculating the local solution through the multi-objective optimization model. Based on the locally recalculated path, instructions are sent to the transportation vehicles to adjust the delivery plan in real time in response to the disturbance events.

9. A method for medical delivery based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, After the delivery task is completed, an evaluation report is generated based on the actual delivery data, and the parameters of the multi-objective optimization model are optimized, specifically including: After the delivery task is completed, the actual delivery data is obtained; wherein, the actual delivery data includes at least: actual delivery time, actual delivery satisfaction, and actual delivery cost; Based on the actual delivery data and the expected delivery data, the delivery data deviation is calculated, and an evaluation report is generated by calling a preset template; Based on a preset cycle, summarize the evaluation report and extract the data items that need to be optimized if the deviation of the delivery data is higher than the expected threshold; In the corresponding multi-objective optimization model, the target weights and parameters of the optimization objectives corresponding to the data items to be optimized are adjusted, and the multi-objective optimization model is updated.

10. A drone-based medical delivery system, characterized in that, The system includes: The task scheduling module is used to receive medical delivery requests and determine the priority of the requests based on the request information; extract delivery request points based on the current set of pending medical delivery requests and their corresponding priorities, and schedule delivery tasks for idle transportation vehicles; wherein, the idle transportation vehicles include idle vehicles and idle drones. The route planning module is used to construct corresponding multi-objective optimization models for medical delivery tasks with different demand priorities, and dynamically configure the weights of each objective according to different delivery scenarios; wherein, the delivery scenarios include at least daily scenarios and emergency scenarios; through the multi-objective optimization model, the optimal delivery route for the current delivery task is solved, and the task is executed using the transportation tools allocated for the current delivery task; The task delivery module is used to monitor the status data of transportation vehicles and environmental data in real time during the task execution process, and extract dynamic disturbance factors; based on the dynamic disturbance factors, it performs local solution recalculation of the multi-objective optimization model and adjusts the delivery plan; after the delivery task is completed, it generates an evaluation report based on the actual delivery data and optimizes the parameters of the multi-objective optimization model.

Images