Urban logistics unmanned aerial vehicle scheduling method and device considering path risk

Abstract

The invention discloses an urban logistics unmanned aerial vehicle scheduling method and device considering path risks. The method comprises the following steps: constructing an unmanned aerial vehicle urban operation third-party risk assessment model and drawing a risk map; solving the urban logistics unmanned aerial vehicle path planning model to obtain a path planning result; based on the path planning result, solving the initial scheduling model of the urban logistics unmanned aerial vehicle to obtain an initial scheduling scheme; in response to the situation that distribution needs to be interrupted when the unmanned aerial vehicle breaks down in the process of flying to the customer, the standby landing path planning model is solved based on the standby landing point set, and a standby landing path planning result of the faulted unmanned aerial vehicle is obtained; and solving the rescheduling model to obtain a rescheduling result, and re-arranging the unmanned aerial vehicle and the distribution time for the remaining unstarted distribution tasks and the interrupted distribution tasks. Aiming at the problems of unmanned aerial vehicle path planning and distribution interruption in an urban environment, a flight path with balanced operation risk and cost and a rescheduling scheme with lower cost are established.

Description

Technical Field

[0001] The present invention belongs to the field of drone path planning and scheduling, and specifically relates to a method and device for scheduling urban logistics drones taking path risks into consideration. Background Art

[0002] In recent years, with the booming development of the low-altitude economy, the application of drone technology in urban environments has attracted widespread attention. In particular, there are many application scenarios in the field of urban logistics and distribution, such as terminal logistics distribution for e-commerce platforms and emergency medical supplies delivery, which provide new solutions for improving urban logistics efficiency. With the expansion of drone delivery, urban low-altitude airspace management faces new challenges.

[0003] Drones operate in complex low-altitude urban environments, involving issues such as coordination with other low-altitude aircraft, protection of residents' privacy, prevention and control of crash risks, noise pollution and environmental impacts. In addition, through a survey of takeout delivery scenarios, it was found that when a drone interrupts its flight due to route conflicts, sudden weather changes, etc., it will make an alternate landing and then the rider will complete the subsequent delivery. Although this method is currently a commonly used solution, it has problems such as high cost and long delivery time. Therefore, it is urgent to build a drone path planning model, scheduling and rescheduling model that comprehensively considers operational risks and operating costs, so as to effectively improve the efficiency and safety of drone delivery.

[0004] Focusing on the scenario where drones deliver one order at a time and multiple drones deliver goods simultaneously, the paper comprehensively considers the operation risks and operation costs, and constructs the drone path planning model and scheduling model by considering the constraints such as drone energy consumption, load, and delivery time window. Based on the initial scheduling plan, the paper considers the emergency situation that causes the drone to carry goods for emergency landing, constructs the drone emergency landing path planning model and rescheduling model, and selects the optimal redistribution method. Summary of the invention

[0005] The purpose of the present invention is to make up for the deficiencies in existing research and provide a method and device for urban logistics drone scheduling that takes path risks into consideration, integrating third-party risk assessment of drone urban operations, drone path planning and scheduling, and establishing a flight path that takes operational risks and costs into consideration, a scheduling that takes time window penalty costs into consideration, and a rescheduling plan after interruption, which can effectively balance the risks and costs of drone operations and reduce scheduling costs after distribution interruptions.

[0006] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0007] In a first aspect, the present invention provides a method for dispatching urban logistics drones taking into account path risks, comprising:

[0008] Obtain data on buildings, roads, trees and water bodies in the target city to build a three-dimensional city model and rasterize it, obtain population and vehicle density data, build a third-party risk assessment model for drone city operations and draw a risk map;

[0009] Based on the three-dimensional city model, risk map, customer coordinates, cargo weight and flight parameters of the drone, a pre-built urban logistics drone path planning model is solved using a multi-objective A* algorithm to obtain a path planning result; wherein the urban logistics drone path planning model is based on the third-party risk assessment model, with minimization of third-party risk value, minimization of operating costs and minimization of carbon emissions as multiple objectives;

[0010] Based on the path planning results, an improved genetic algorithm is used to solve the pre-constructed urban logistics drone initial scheduling model to obtain an initial scheduling plan; wherein the urban logistics drone initial scheduling model aims to minimize the penalty cost of the delivery time window;

[0011] In response to a drone malfunctioning on the way to a customer and requiring delivery to be interrupted, a pre-built alternate landing path planning model is solved based on an alternate landing point set to obtain an alternate landing path planning result for the faulty drone; wherein the alternate landing path planning model aims to minimize the third-party risk value and the shortest flight time;

[0012] The pre-built rescheduling model is solved to obtain the rescheduling result, and the drones and delivery times are rescheduled for the remaining unstarted delivery tasks and the interrupted delivery tasks; wherein the rescheduling model aims to minimize the penalty cost of the running time window.

[0013] In a second aspect, the present invention provides an urban logistics drone scheduling device that considers path risk, including a processor and a storage medium;

[0014] The storage medium is used to store instructions;

[0015] The processor is configured to operate according to the instructions to execute the method according to the first aspect.

[0016] In a third aspect, the present invention provides a storage medium having a computer program stored thereon, wherein the computer program implements the method described in the first aspect when executed by a processor.

[0017] In a fourth aspect, the present application provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the method of the first aspect when executing the computer program.

[0018] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0019] Compared with the prior art, the present invention has significant advantages: the present invention can address the problems of drone path planning and delivery interruption in urban environments. The method is based on third-party risk assessment of drone urban operations, and can establish a flight path that balances operational risks and costs. The lower-cost rescheduling solution can provide a reference for the safe and efficient operation of drone urban delivery in the future. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a schematic diagram of a process for dispatching urban logistics drones taking into account path risks provided by an embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of a building "privacy protection layer" according to an embodiment of the present invention;

[0022] Figure 3 is a schematic diagram of a UAV path flying to a customer point planned by the path planning method according to an embodiment of the present invention;

[0023] Figure 4 is a schematic diagram of a drone path flying back to a distribution center planned by the path planning method according to an embodiment of the present invention;

[0024] Figure 5 It is a schematic diagram of a UAV path flying to an alternate landing point planned by the path planning method described in an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The present invention will be further described below in conjunction with the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0026] Embodiment 1: This embodiment provides a method for dispatching urban logistics drones taking into account path risks, including:

[0027] Obtain data on buildings, roads, trees and water bodies in the target city to build a three-dimensional city model and rasterize it, obtain population and vehicle density data, build a third-party risk assessment model for drone city operations and draw a risk map;

[0028] Based on the three-dimensional city model, risk map, customer coordinates, cargo weight and flight parameters of the drone, a pre-built urban logistics drone path planning model is solved using a multi-objective A* algorithm to obtain a path planning result; wherein the urban logistics drone path planning model is based on the third-party risk assessment model, with minimization of third-party risk value, minimization of operating costs and minimization of carbon emissions as multiple objectives;

[0029] Based on the path planning results, an improved genetic algorithm is used to solve the pre-constructed urban logistics drone initial scheduling model to obtain an initial scheduling plan; wherein the urban logistics drone initial scheduling model aims to minimize the penalty cost of the delivery time window;

[0030] In response to a drone malfunctioning on the way to a customer and requiring delivery to be interrupted, a pre-built alternate landing path planning model is solved based on an alternate landing point set to obtain an alternate landing path planning result for the faulty drone; wherein the alternate landing path planning model aims to minimize the third-party risk value and the shortest flight time;

[0031] The pre-built rescheduling model is solved to obtain the rescheduling result, and the drones and delivery times are rescheduled for the remaining unstarted delivery tasks and the interrupted delivery tasks; wherein the rescheduling model aims to minimize the penalty cost of the running time window.

[0032] In some embodiments, Figure 1 As shown in FIG. 1 , a method for dispatching urban logistics drones considering path risk includes:

[0033] Step (1): Obtain and rasterize urban building, road, tree and water data from the geographic information database, obtain population and vehicle density data, build a third-party risk assessment model for drone urban operations and draw a risk map;

[0034] Step (2): Based on the three-dimensional city model, risk map, customer coordinates, cargo weight and flight parameters of the drone, a path planning model for urban logistics drones is constructed, with minimization of third-party risk value, minimization of operating costs and minimization of carbon emissions as multiple objectives, and range limitation, turning angle limitation, load limitation and avoiding no-fly zones as constraints;

[0035] Step (3): Use the multi-objective A* algorithm to solve the urban logistics drone path planning model and obtain the path planning result;

[0036] Step (4): Based on the path planning results, an initial scheduling model for urban logistics drones is constructed, with the goal of minimizing the penalty cost of the delivery time window, and the constraints of a single drone serving one customer at a time, safe interval determination and delay time setting at take-off time, the start time of the next task being later than the end time of the previous task, energy consumption restrictions, customer time window requirements, battery power remaining calculation, and battery replacement judgment;

[0037] Step (5): Use the improved genetic algorithm to solve the initial scheduling model of urban logistics drones and obtain the initial scheduling plan;

[0038] Step (6): In response to a failure of the drone on the way to the customer and the need to interrupt delivery, a pre-built alternate landing path planning model is solved based on the alternate landing point set to obtain an alternate landing path planning result for the failed drone; wherein the alternate landing path planning model aims to minimize the third-party risk value and the shortest flight time;

[0039] Step (7): Solve the pre-built rescheduling model to obtain the rescheduling result, and reschedule the drones and delivery times for the remaining unstarted delivery tasks and the interrupted delivery tasks; wherein the rescheduling model aims to minimize the penalty cost of the running time window.

[0040] In some embodiments, step (1) specifically includes:

[0041] Step (101): Obtaining building, road, tree and water data in the region from an open source global geographic information database OSM (Open Street Map); wherein the building data includes the longitude and latitude coordinates, outline range, height information and building type of the buildings in the region, the road data includes the longitude and latitude coordinates and road shape, the tree data includes the longitude and latitude coordinates, outline range and height information of the trees, and the water data includes the longitude and latitude coordinates and water shape;

[0042] Step (102): Use Python language to construct a three-dimensional city model containing buildings, roads, trees and water bodies based on the data of buildings, roads, trees and water bodies. For the three-dimensional city model and the airspace OABC-O′A′B′C′, in this embodiment, the grid size l is selected. r is 10m, and the airspace is divided into several three-dimensional grids; for any grid r, the information it contains is the coordinates (xr, yr, zr), where xr, yr, zr are the coordinates of grid r in the x, y, z directions;

[0043] Step (103): Draw an obstacle layer based on the building outline and height; draw a "privacy protection layer" (20m horizontally and 10m vertically) around each building. Figure 2 ), prohibit drones from flying in, and draw a no-fly zone layer together with other no-fly zones;

[0044] Step (104): The third-party risk assessment model includes a UAV-to-ground casualty risk model and a UAV noise impact model; a UAV-to-ground casualty risk model is constructed, using the number of ground personnel deaths N caused by UAV crashes per flight hour fat To measure casualties, N fat Including casualties caused by direct drone strikes Casualties caused by drone collisions with vehicles Two parts, and Expressed as:

[0045]

[0046] Among them, P cra is the probability of a drone crashing, S imp is the impact area of ​​the drone. is the population density of grid r, The fatality rate of drone strikes against pedestrians is is the vehicle density of grid r, is the fatality rate of traffic accidents;

[0047] Step (105): Derive the fatality rate of drone collisions with pedestrians

[0048]

[0049] Where E is the impact kinetic energy, f s is the shielding coefficient, α is f s = 0.5, the impact kinetic energy required to achieve a mortality rate of 50%, β is the impact kinetic energy required to achieve a mortality rate of 50% when f s The impact energy threshold required to cause death when it approaches 0, E, is solved by the following formula:

[0050]

[0051] Where m is the total weight of the drone and the cargo it carries, which can be solved by the following formula: m = m u +m fre , where m u is the weight of the drone and battery, m fre is the weight of the cargo; v is the speed of the drone when it hits the ground, g is the acceleration due to gravity, and f res is the air resistance coefficient, ρ air is the air density, h is the altitude of the UAV, and in this embodiment, there are 4 altitude layers: 30m, 60m, 90m and 120m;

[0052] Step (106): Use the gravity model to calculate the population density in grid r and vehicle density Make an estimate:

[0053]

[0054] in, and are the average population and vehicle density of the entire study area, respectively. The average population density is obtained through the WorldPop population dataset, and the average vehicle density is obtained through Amap. l is the influence radius of the attractiveness of consumer facilities.

[0055] Step (107): Construct a drone noise impact model and use the noise annoyance index NI to characterize the noise impact of the drone on the third party:

[0056]

[0057] PA = -0.107ln(h) + 0.7995;

[0058] Among them, PA is the degree of annoyance perceived by pedestrians to drone noise, A r is the area of ​​grid r;

[0059] Step (108): Calculate and draw a risk map, which is completed by calculating and superimposing a population and vehicle density layer, an obstacle layer, a shielding factor layer, and a no-fly zone layer. Each layer contains different parameters for calculating the risk value of a specific geographic area.

[0060] In some embodiments, step (2) specifically includes:

[0061] Step (201): Establish decision variables, assuming that there are N u A battery-swap multi-rotor vertical take-off and landing drone of the same model is responsible for N i (N i >N u ) customers’ package delivery, the drone set is U={1,2,...,N u}, the set of distribution centers and customer points is I = {0,1,2,...,N i}, 0 represents the distribution center. The drone only carries one customer's package at a time and returns to the distribution center after completing the delivery. The decision variable x for task allocation uij for:

[0062]

[0063] Among them, u represents any drone, i and j represent any two customers served by the same drone;

[0064] Step (202): Setting the objective function, minimizing the third-party risk Z1, minimizing the operating cost Z2, and minimizing the carbon emission Z3 is expressed as:

[0065]

[0066] in, The normalized number of ground fatalities caused by drone crashes per flight hour is solved by the following formula: is the normalized noise impact index, which is solved by the following formula:

[0067] minZ2=C fix +C var ;

[0068] Among them, C fix is the fixed cost, and is solved by the following formula: Among them, C u is the UAV operation unit price after the purchase cost of the empty aircraft is amortized, C batt is the battery operation unit price after the battery purchase cost is allocated, C cap is the background captain cost, C gro C is the cost of on-site ground staff; var is the variable cost, which can be solved by the following formula: Among them, C e is the energy consumption price per hour of the drone, is the energy consumption of the segmented flight of drone u carrying cargo to customer i, is the energy consumption of the segmented flight of UAV u returning empty-handed from customer i;

[0069]

[0070] Among them, ε is the weight of CO2 emitted per unit energy consumption of the drone;

[0071] Step (203): Calculate the energy consumption of each segment of the drone delivery round trip and

[0072]

[0073] Among them, v up is the speed of the drone during take-off, v hor is the speed of the UAV during the cruising phase, v down is the speed of the drone during the descent phase, h carr h is the flight altitude of the drone carrying cargo to the customer point, emp is the flight altitude of the drone returning to the distribution center without a load, z0 is the take-off and landing point altitude at the distribution center, and z i is the take-off and landing point height at the customer, is the UAV propeller power transmission efficiency, σ is the UAV flight lift-to-drag ratio;

[0074] Step (204): Establish path planning constraints, as follows:

[0075]

[0076] in, is the horizontal cruising distance from the distribution center to customer point i, is the horizontal cruising distance from customer i to the distribution center, d max is the maximum range of the drone, is the battery power of the drone before it takes off from the distribution center, B min is the minimum safe power of the battery; λ is the flight path point, and its position coordinate is (x λ ,y λ ,z λ ), Ω is the set of obstacles and no-fly zones.

[0077] In some embodiments, step (3) specifically includes:

[0078] Step (301): inputting the data required for path planning, including reading the obstacle map and risk map and converting the map data into NumPy arrays to store obstacle information and risk levels respectively; reading the customer coordinates and cargo weight, providing the coordinates of the target point for path planning and calculating the data of energy consumption and carbon emissions;

[0079] Step (302): setting parameters and defining energy consumption and carbon emission models, calculating the energy consumption of the horizontal cruise, ascent and descent phases based on the flight parameters of the UAV (such as flight speed, flight altitude, total weight, etc.), and further deriving the carbon emissions based on the energy consumption calculation results, providing a basis for multi-objective optimization;

[0080] Step (303): For each optimization target, the A* algorithm is used to solve the path, and Manhattan distance and Euclidean distance are used as heuristic functions; Manhattan distance is suitable for grid maps, while Euclidean distance is suitable for continuous space, which can be flexibly selected according to actual needs; using open lists and closed sets, nodes with lower objective functions are expanded first and the shortest path is finally returned; if the path exists, the relevant cost, carbon emissions, risk and other information are calculated; otherwise, a message that the path is not found is returned;

[0081] Step (304): Evaluate multiple optimization objectives of each path, and use the dominance relationship to determine whether a solution is better than another solution, that is, solution 1 is considered to dominate solution 2 if and only if solution 1 is not inferior to solution 2 in all objectives and is better than solution 2 in at least one objective; select the Pareto optimal solution through the dominance relationship to obtain a set of solutions, in which no other solution is better than these solutions in all objectives at the same time;

[0082] Step (305): Path conflict detection, analyzing the path segments in all paths (i.e., two consecutive points on the path) to determine whether multiple customers share the same path segment; if the paths of multiple customers overlap, it is considered that a conflict occurs; for the detected conflicting path segments, the conflicting customer IDs are recorded and the conflict information is exported for subsequent analysis or scheduling adjustments.

[0083] In some embodiments, step (4) specifically includes:

[0084] Step (401): Add a new decision variable. Assume that there are N u A battery-swap multi-rotor vertical take-off and landing drone of the same model is responsible for N i (N i >N u ) customers’ package delivery, the drone set is U={1,2,...,N u}, the set of distribution centers and customer points is I = {0,1,2,...,N i}, 0 represents the distribution center, the drone only carries one customer's package at a time, and returns to the distribution center after completing the delivery. The decision variable y for replacing the battery uij and, whether the time of departure from the distribution center meets the time interval decision variable and whether the time of arrival at the distribution center meets the time interval decision variable for:

[0085]

[0086] When drone u completes the delivery task for customer i and returns to the delivery center, it continues to perform the delivery task for customer j. At this time, battery y needs to be replaced. uij Take 1, otherwise take 0. is the battery power of the drone after it returns to the distribution center; when there is an identical path segment between the paths from the distribution center to customer i and customer i′, and the time interval between the two taking off from the distribution center does not meet the safety interval Take 1, otherwise take 0. is the moment when drone u starts delivering to customer i, T min is the safe interval for drone takeoff, Γ is the set of customers that have the same path segment when flying from the distribution center to the customer point; when there is the same path segment between the paths from customer i and customer j′ returning to the distribution center respectively, and the time interval between the two arriving at the distribution center does not meet the safe interval Take 1, otherwise take 0. is the time when drone u completes the delivery of customer i and arrives at the distribution center, and Π is the set of customers that have the same path segment when returning from the customer point to the distribution center;

[0087] Step (402): Set the objective function, time window penalty cost C pen The minimum representation is:

[0088]

[0089] Among them, γ ear is the penalty coefficient for the drone to arrive at the customer ahead of schedule, γ latis the penalty coefficient for the drone’s delayed arrival at the customer, [a i ,b i ] is the delivery time window required by customer i, The time when robot u arrives at customer i and completes unloading the goods;

[0090] Step (403): Establish the delivery task and load constraints, as follows:

[0091]

[0092] m fre ≤m max ;

[0093] Among them, m max The maximum load limit for the drone;

[0094] Step (404): Establish the delivery task time-related constraints, as follows:

[0095]

[0096] And customer i is the first customer delivered by drone u

[0097]

[0098]

[0099] in, is the time interval between customer i and customer i′ that needs to be delayed in bringing goods. is the idle arrival time interval that needs to be delayed between customer i and customer j′, is the decision variable for whether the time when customer j and customer i′ take off from the distribution center meets the time interval; is the time interval between customer j and customer i′ that needs to be delayed in bringing goods; t STA is the time when the distribution center starts to operate, T loa The time required to load the drone, T exc The time required to replace the battery of the drone, T unl The time required to unload the drone; The time it takes for drone u to carry cargo to customer i is calculated by the following formula: is the time it takes for UAV u to return from customer i without payload, calculated by the following formula:

[0100]

[0101] Step (405): Establish battery power related constraints, as follows:

[0102] And customer i is the first customer delivered by drone u;

[0103]

[0104] Among them, B max The maximum charge of the battery.

[0105] In some embodiments, step (5) specifically includes:

[0106] Step (501): Initialize the population, consider key parameters such as the number of drones, the number of customer points, battery energy consumption, and time window, randomly arrange the delivery order of customers, and use different arrangements as different individuals, that is, each individual represents a delivery plan, to generate an initial population;

[0107] Step (502): fitness function definition, which takes into account the time window penalty cost and finds the optimal solution by calculating the performance of each individual scheduling solution in actual application;

[0108] Step (503): Genetic algorithm evolution, using customized crossover and mutation operations to generate new individuals; the crossover operation introduces new gene combinations by exchanging task sequences between parent individuals, and the mutation operation introduces disturbances in the solution space by randomly adjusting the task sequence, thereby jumping out of the local optimal solution and expanding the solution space;

[0109] Step (504): Population update and elimination. After each generation of evolution, the algorithm selects individuals in the population according to their fitness values, eliminates individuals with poor fitness, and retains and replicates individuals with high fitness to form the next generation of population.

[0110] Step (505): Repeated path segment identification, after finding the preliminary optimal solution, based on the pre-imported path segment overlap information, identify customers with repeated path segments;

[0111] Step (506): Delivery time adjustment: for customers with duplicated route segments, delivery time is adjusted to ensure that the flight times of different drones on the same route are staggered, thereby avoiding air conflicts; the fitness of the adjusted scheduling scheme is recalculated, and the better scheme is retained;

[0112] Step (507): Conflict check and further adjustment. After the initial delivery time adjustment is completed, the adjusted scheduling plan is checked. If new conflicts are found during the check, adjustments will be made again to ensure that no new potential conflicts arise.

[0113] Step (508): Output of the initial scheduling plan. After all adjustments and optimizations are completed, the final initial scheduling plan is output. The plan includes the specific task sequence of each drone, the time nodes in the delivery, the battery usage, whether the battery needs to be replaced, and the operating cost of the plan.

[0114] In some embodiments, step (6) specifically includes:

[0115] Step (601): Establish decision variables. During the transport cruise, if the drone fails due to internal reasons and needs to interrupt the delivery, it will immediately select the best alternate landing point for maintenance. Then, it will determine whether to continue the mission based on the maintenance situation. The decision variable x for the failed drone to fly to the alternate landing point is uk for:

[0116]

[0117] in, Assemble for faulty drones, is any faulty UAV, K is the set of alternate landing points, K = {(x k ,y k ,z k )k∈{0,1,2,...,N k}}, 0 represents the distribution center, k is any alternate landing point;

[0118] Step (602): Setting the objective function, the third-party risk Z1′ is minimized and the flight time Z2′ is shortest, which is expressed as:

[0119]

[0120]

[0121] in, For malfunctioning drones The time required to reach the alternate landing point k is calculated by the following formula: in, For malfunctioning drones Horizontal cruising distance to alternate landing point k, h div The flight altitude at which the faulty drone will fly to the alternate landing point;

[0122] Step (603): Establish path planning constraints, as follows:

[0123]

[0124] in, The battery power when the drone is ready to fly to the alternate landing point, (x λ′ ,yλ′ ,z λ′ ) is the coordinate of the point λ′ that the faulty UAV passes through when it flies to the alternate landing point;

[0125] Step (604): Use the multi-objective A* algorithm in step (3) to solve the alternate landing path planning model to obtain the alternate landing path planning result of the faulty UAV.

[0126] In some embodiments, step (7) specifically includes:

[0127] Step (701): Add a new decision variable. Assuming that the faulty drone arrives at the alternate landing point, after inspection by the staff, it is determined that the drone cannot continue to perform the task, then it is necessary to reschedule and reschedule the drones and delivery times for the interrupted task and all other tasks that have not started delivery. The delivery paths of other tasks except the interrupted task remain unchanged; the interrupted task is rescued by other drones, that is, other drones are arranged to take off empty from the distribution center and arrive at the alternate landing point. The staff will hand over the goods carried by the faulty drone to the rescue drone, and the drone will deliver the goods to the customer point; according to whether there is a spare drone, the rescheduling strategy is divided into two types: the existing other drone rescheduling strategy (that is, using the remaining normal drones to complete the delivery of the remaining tasks and the interrupted tasks) and the adding spare drone rescheduling strategy (that is, using the remaining normal drones and spare drones to complete the delivery of the remaining tasks and the interrupted tasks). The decision variables for task reallocation for:

[0128]

[0129] Step (702): Set the objective function. The minimum penalty cost Z3′ of the overall running time window is expressed as:

[0130]

[0131] in, is the set of normal drones (hereinafter referred to as ordinary drones) remaining in the original drone set. A is the set of backup drones, A={1,2,...,N a}, The time when drone u arrives at customer i and unloads the goods after the delivery is interrupted;

[0132] Step (703): Establish a constraint that only one of the rescheduling strategies can be selected, as follows:

[0133]

[0134] Among them, DI is the set of customers who have not started delivery when the delivery is interrupted, DI = {1, 2, ..., n}, DI∈I;

[0135] Step (704): Establish the available time constraints of the drone, as follows:

[0136]

[0137] in, is the time when drone u starts delivering to customer i after the delivery is interrupted. is the time when drone u becomes available, i * For customers whose delivery tasks are interrupted, t * is the moment when the delivery task is interrupted, OI is the set of customers in transit when the delivery is interrupted, OI={1,2,...,m}OI∈I,I={0,CI,DI,OI};

[0138] Step (705): Establish the time constraint of the rescue interruption task, as follows:

[0139]

[0140] Among them, T che The time required to repair a faulty drone. is the time when drone u completes the delivery task for customer i and arrives at the delivery center after the delivery is interrupted; is the time it takes for drone u to fly from the distribution center to the alternate landing point k, calculated by the following formula: in, is the horizontal cruising distance from the distribution center to the alternate landing point k, z k The take-off and landing point altitude for the alternate landing point; For drone u to fly from alternate landing point k to customer i with cargo * The time is calculated by the following formula: in, To fly from alternate point k to customer i * Horizontal cruising distance;

[0141] Step (706): Establish the delivery task time-related constraints, as follows:

[0142]

[0143] And customer i is the first customer delivered by drone u after the delivery interruption

[0144]

[0145] in, The time it takes for drone u to carry cargo to customer i is calculated by the following formula: is the time it takes for UAV u to return from customer i without payload, calculated by the following formula:

[0146] Step (707): Establish battery power related constraints, as follows:

[0147]

[0148] in, The decision variables for replacing the drone’s battery after a delivery interruption, The decision variables for task reallocation, The battery power of drone u when it is ready to fly from the distribution center to customers i and j, The battery charge of the drone after it returns to the distribution center; is the energy consumption of the segmented flight of drone u carrying cargo to customer j, B is the energy consumption of the segmented flight of UAV u returning from customer j without any payload; min B is the minimum safe power of the battery; max The maximum charge of the battery.

[0149] Verification example: In order to verify the effectiveness of the urban logistics drone scheduling method considering path risks provided by the embodiment of the present invention, the example of the present invention selected a 5km range around Galaxy World in Shenzhen as the distribution environment, and used the multi-objective A* algorithm to solve the distribution paths at different flight altitudes for small-scale examples of three distributions, thereby determining the optimal drone flight altitude. The specific results are shown in Tables 1 and 2.

[0150] Table 1 Path planning results for flying to customer points at different flight altitudes

[0151]

[0152] Table 2 Path planning results for flying back to the distribution center at different flight altitudes

[0153]

[0154] From Table 1 and Table 2, we can see that: for the path planning to the customer point, the fixed cost of all distribution types is the same, and the variable cost increases slightly with the increase of flight altitude, with an average growth rate of 0.0004%; the time window penalty cost is not calculated during path planning, but this cost is closely related to the flight time of the drone. The flight time increases with the increase of flight altitude, and its average growth rate is 1.98%; carbon emissions also increase with the increase of flight altitude, with an average growth rate of 11.52%; the third-party risk value increases first and then decreases with the increase of flight altitude, with an average growth rate of 0.83%; the path planning results for flying back to the distribution center have a similar pattern to the path to the customer point; the total flight distance directly or indirectly The ground has a positive impact on the variable cost, time window penalty cost and carbon emissions in the target value, while the third-party risk value is mainly affected by the distribution of buildings, population, vehicles, etc. under different flight paths; the target values ​​for flying to or returning from each customer point at different altitudes are calculated and sorted, and the results are: 30m is better than 60m, better than 90m, better than 120m, that is, the round-trip route is optimal at the 30m altitude; but in order to reduce the air conflict between the outbound and return trips, it is stipulated that the round-trip flights are performed at different altitudes, so different altitudes are selected in the paths flying to the customer point and returning to the distribution center for pairing, and it is finally determined that the 30m altitude layer for the outbound trip and the 60m altitude layer for the return trip are the optimal combination according to the multiple target values.

[0155] In order to demonstrate the effect of the drone scheduling method considering path risk provided by the example of the present invention, the example of the present invention first performs round-trip path planning between the distribution center and the customer point at the 30m and 60m altitude layers, and takes 15 drones delivering 50 customers with mixed distribution as an example. The path planning results to the customer point are shown as follows: Figure 3 As shown in Figure 2, the path planning result of flying back to the distribution center is as follows: Figure 4 Some initial scheduling schemes are shown in Table 3.

[0156] Table 3 Initial scheduling results of medium-scale example (mixed distribution)

[0157]

[0158] From Table 3, we can see that: the customer points with path conflicts have been staggered to avoid possible mid-air collisions; in addition, since each drone performs fewer delivery tasks, there is no need to replace the battery; the time window penalty cost for small-scale customers is 2922.4 yuan in total; the fixed cost of this scheduling plan is 383.25 yuan, the variable cost is 0.0311 yuan, the carbon emissions are 0.0157 kg, and the risk value is 21835.062.

[0159] Take the example of UAV 2 failing at pixel coordinates (379, 261) at 8:17:39 when delivering to customer 5. The optimal alternate landing path planning results are shown as follows: Figure 5 As shown in Table 4, the partial rescheduling schemes using the existing other UAV rescheduling strategy and adding the backup UAV rescheduling strategy are shown in Table 4 and Table 5 respectively.

[0160] Table 4: Remaining task rescheduling results of medium-scale example (mixed distribution) (other existing drone rescheduling strategies)

[0161]

[0162] Table 5: Remaining task rescheduling results of medium-scale example (mixed distribution) (adding backup drone rescheduling strategy)

[0163]

[0164] It can be seen from Tables 4 and 5 that when adopting other existing drone rescheduling strategies, 25 customer tasks need to be assigned to 14 drones. The time window penalty cost in the rescheduling scheme under this strategy is 2631.4 yuan, which is 15.9% higher than the time window penalty cost for customers who have not started delivery and delivery interruption in the initial scheduling scheme; when adopting the rescheduling strategy of adding backup drones, 25 customer tasks need to be assigned to 14 ordinary drones and 1 backup drone. The time window penalty cost in the rescheduling scheme under this strategy is 1945.4 yuan, which is 14.31% lower than the initial scheme and 26.07% lower than the rescheduling scheme under other drone rescheduling strategies; by comparing the two drone rescheduling strategies A brief comparative analysis shows that the advantage of the first strategy "rescheduling strategy with other existing drones" is that it does not require additional equipment and can only rely on existing drone resources to complete the task. However, due to resource constraints, the time window penalty cost is high and cannot fully meet the timeliness requirements of the task. The second strategy "rescheduling strategy with backup drones" introduces an additional backup drone on the basis of the original 4 ordinary drones to improve the flexibility of task allocation. The first strategy is suitable for resource-constrained scenarios and completes tasks by optimizing the scheduling of existing drones, but the time window penalty cost is high. The second strategy shows obvious advantages in the flexibility and timeliness of task allocation by introducing backup drones, and significantly reduces the time window penalty cost.

[0165] In order to verify the effectiveness of the interruption scheduling method provided by the embodiment of the present invention, a backup drone rescheduling strategy with a lower time window cost is selected for comparison with the rider rescue strategy (i.e., the rider goes to the alternate landing point and carries the goods to the customer point, while other tasks that have not started delivery continue to be delivered by drones). The comparison results are shown in Table 6.

[0166] Table 6 Comparison between drone and rider rescue

[0167]

[0168] Table 6 shows that: from the perspective of travel distance, drones can choose more direct routes when flying in the air, and their flight distance is about 28.95% less than the distance ridden by riders, effectively shortening the transportation distance; from the perspective of operating costs, the operating costs of drones are 15.66% lower than those of riders, which is mainly due to the high labor costs and energy costs of riders during delivery; from the perspective of carbon emissions, drones emit only 0.0002kg of carbon, which is 99.83% less than the electric drones used by riders, mainly due to their lower energy consumption; from the perspective of timeliness, drones rescue and deliver to customers The time consumed by the UAV is 10.1 minutes, which is significantly better than the 25.5 minutes of the rider, a reduction of about 60.39%, indicating that the UAV is more efficient in quickly completing the rescue. In addition, although the difference between the UAV and the rider rescue in terms of the time window penalty cost is very small, this is because the rider rescue task does not participate in the overall rescheduling of the remaining delivery tasks, and the rider can immediately set out for rescue when the faulty UAV is judged to be unable to continue the mission. The use of UAV rescue is one of the overall rescheduling arrangements. In order to minimize the overall time window penalty cost, the departure time of the rescue UAV is about 15 minutes later than that of the rider.

[0169] In summary, the urban logistics drone scheduling method considering path risks provided by the present invention can make up for the shortcomings of existing methods, and can address the problems of drone path planning and delivery interruption in urban environments. Based on the third-party risk assessment of drone urban operations, a flight path that balances operation risks and costs can be established. A lower-cost rescheduling solution can provide a reference for the safe and efficient operation of drone urban delivery in the future.

[0170] Embodiment 2: Based on Embodiment 1, this embodiment provides an urban logistics drone scheduling device considering path risk, including a processor and a storage medium;

[0171] The storage medium is used to store instructions;

[0172] The processor is used to operate according to the instruction to execute the method according to embodiment 1.

[0173] Embodiment 3: Based on Embodiment 1, this embodiment provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the method described in Embodiment 1 is implemented.

[0174] Embodiment 4: Based on Embodiment 1, this embodiment provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the method described in Embodiment 1 when executing the computer program.

[0175] Embodiment 5: Based on Embodiment 1, this embodiment provides a computer program product, including a computer program, which implements the method described in Embodiment 1 when executed by a processor.

[0176] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0177] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0178] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0179] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0180] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for dispatching urban logistics drones considering path risk, characterized in that: include: Obtain data on buildings, roads, trees and water bodies in the target city to build a three-dimensional city model and rasterize it, obtain population and vehicle density data, build a third-party risk assessment model for drone city operations and draw a risk map; Based on the three-dimensional city model, risk map, customer coordinates, cargo weight and flight parameters of the drone, a pre-built urban logistics drone path planning model is solved using a multi-objective A* algorithm to obtain a path planning result; The urban logistics drone path planning model is based on the third-party risk assessment model, with minimization of third-party risk value, minimization of operating costs and minimization of carbon emissions as multiple objectives; Based on the path planning results, an improved genetic algorithm is used to solve the pre-built urban logistics drone initial scheduling model to obtain an initial scheduling plan; The initial scheduling model of urban logistics drones aims to minimize the penalty cost of the delivery time window; In response to a UAV failure on the way to the customer and the need to interrupt delivery, a pre-built alternate landing path planning model is solved based on the alternate landing point set to obtain an alternate landing path planning result for the faulty UAV; The alternate landing path planning model aims to minimize the third-party risk value and the shortest flight time; The pre-built rescheduling model is solved to obtain the rescheduling result, and the drones and delivery times are rescheduled for the remaining unstarted delivery tasks and the interrupted delivery tasks; wherein the rescheduling model aims to minimize the penalty cost of the running time window.

2. The urban logistics drone scheduling method considering path risk according to claim 1 is characterized in that: Obtain data on buildings, roads, trees and water bodies in the target city to build a three-dimensional city model and rasterize it, obtain population and vehicle density data, build a third-party risk assessment model for drone city operations and draw a risk map, including: Step (101): Acquire building, road, tree and water data; wherein the building data includes the longitude and latitude coordinates, outline range, height information and building type of the buildings in the area; the road data includes the longitude and latitude coordinates and road shape; the tree data includes the longitude and latitude coordinates, outline range and height information of the trees; and the water data includes the longitude and latitude coordinates and water shape; Step (102): construct a three-dimensional city model containing buildings, roads, trees and water bodies based on the data of buildings, roads, trees and water bodies. For the three-dimensional city model and airspace, select a grid size l r , the airspace is divided into several three-dimensional grids; for any grid r, the coordinates are expressed as (xr, yr, zr), where xr, yr, zr are the coordinates of grid r in the x, y, z directions; Step (103): Draw an obstacle layer based on the building outline and height; draw a "privacy protection layer" around each building to prohibit drones from flying in, and draw a no-fly zone layer together with other no-fly zones; Step (104): The third-party risk assessment model includes a UAV-to-ground casualty risk model and a UAV noise impact model; wherein the UAV-to-ground casualty risk model uses the number of ground personnel deaths N caused by UAV crashes per flight hour fat To measure casualties, N fat Including casualties caused by direct drone strikes Casualties caused by drone collisions with vehicles Two parts, and Expressed as: Among them, P cra is the probability of a drone crashing, S imp is the area where the drone crashes to the ground, is the population density of grid r, The fatality rate of drone strikes against pedestrians is is the vehicle density of grid r, is the fatality rate of traffic accidents; Step (105): The fatality rate of drone collision with pedestrians Where E is the impact kinetic energy, f s is the shielding coefficient, α is f s = 0.5, the impact kinetic energy required to achieve a mortality rate of 50%, β is the impact kinetic energy required to achieve a mortality rate of 50% when f s The impact energy threshold required to cause death when it approaches 0, the impact kinetic energy E is solved by the following formula: Where m is the total weight of the drone and the cargo it carries, which can be solved by the following formula: m = m u +m fre , where m u is the weight of the drone and battery, m fre is the weight of the cargo; v is the speed of the drone when it hits the ground, g is the acceleration due to gravity, and f res is the air resistance coefficient, ρ air is the air density, S imp is the impact area of ​​the drone, and h is the flying height of the drone; Step (106): Use the gravity model to calculate the population density in grid r and vehicle density Make an estimate: in, are the average population density and average vehicle density of the entire study area, respectively; l is the influence radius of the attractiveness of consumer facilities; Step (107): The drone noise impact model uses a noise annoyance index NI to characterize the noise impact of the drone on a third party: PA = -0.107ln(h) + 0.7995; Among them, PA is the degree of annoyance perceived by pedestrians to drone noise, A r is the area of ​​grid r, h is the flying height of the UAV; Step (108): Calculate and draw a risk map, wherein the risk map is completed by calculating and superimposing a population and vehicle density layer, an obstacle layer, a shielding coefficient layer, and a no-fly zone layer.

3. The urban logistics drone scheduling method considering path risk according to claim 1 is characterized in that: The method for constructing the urban logistics UAV path planning model includes: Step (201): Establish decision variables, assuming that there are N u A battery-swap multi-rotor vertical take-off and landing drone of the same model is responsible for N i Package delivery for N customers, i >N u , the drone set is U={1,2,...,N u },N u is the total number of drones, the set of distribution centers and customer points is I = {0,1,2,...,N i },N i is the total number of customers, 0 represents the distribution center, the drone only carries one customer's package at a time and returns to the distribution center after completing the delivery. The decision variable x for task allocation uij for: Among them, u represents any drone, i and j represent any two customers served by the same drone; Step (202): Setting the objective function of the urban logistics drone path planning model, the third-party risk Z1 is minimized, the operating cost Z2 is minimized, and the carbon emission Z3 is minimized, respectively, as follows: in, The normalized number of ground fatalities caused by drone crashes per flight hour is solved by the following formula: N fat The number of fatalities on the ground caused by drone crashes per flight hour; is the normalized noise impact index, which is solved by the following formula: NI is the noise annoyance index; minZ2=C fix +C var ; Among them, C fix is the fixed cost, and is solved by the following formula: Among them, C u is the UAV operation unit price after the purchase cost of the empty aircraft is amortized, C batt is the battery operation unit price after the battery purchase cost is allocated, C cap is the background captain cost, C gro C is the cost of on-site ground staff; var is the variable cost, which can be solved by the following formula: Among them, C e is the energy consumption price per hour of the drone, is the energy consumption of the segmented flight of drone u carrying cargo to customer i, is the energy consumption of the segmented flight of UAV u returning empty-handed from customer i; Among them, ε is the weight of CO2 emitted per unit energy consumption of the drone; Step (203): Calculate and Where m is the total weight of the drone and the cargo it carries, m u is the weight of the drone and the battery, g is the acceleration of gravity, v up is the speed of the drone during take-off phase, is the propeller power transmission efficiency of the UAV, σ is the lift-to-drag ratio of the UAV; h carr is the flight height of the drone carrying goods to the customer point, z0 is the take-off and landing point height at the distribution center, and v hor is the speed of the UAV during the cruising phase, is the horizontal cruising distance from the distribution center to customer point i, is the horizontal cruising distance from customer i to the distribution center, v down is the speed of the drone during the descent phase, z i h is the take-off and landing point height at the customer; emp is the flight altitude of the drone returning to the distribution center without any payload, and e is a natural constant; Step (204): Establish the constraint conditions of the urban logistics drone path planning model, as follows: Among them, d max is the maximum range of the drone, is the battery power of the drone before it takes off from the distribution center, B min is the minimum safe power of the battery; λ is the flight path point, and its position coordinates are (x λ ,y λ ,z λ ), Ω is the set of obstacles and no-fly zones.

4. The urban logistics drone scheduling method considering path risk according to claim 1 is characterized in that: The multi-objective A* algorithm is used to solve the urban logistics drone path planning model, including: Step (301): inputting the data required for path planning, including reading the obstacle map and risk map and converting the map data into an array to store obstacle information and risk level respectively; reading the customer coordinates and cargo weight, providing the coordinates of the target point for path planning and calculating the data of energy consumption and carbon emissions; Step (302): setting parameters and defining energy consumption and carbon emission models, calculating the energy consumption of the horizontal cruising, ascent and descent phases based on the flight parameters of the UAV, and deriving the carbon emission based on the energy consumption calculation results, thereby providing a basis for multi-objective optimization; wherein the flight parameters of the UAV include flight speed, flight altitude and total weight; Step (303): For each optimization target, the A* algorithm is used to solve the path, and Manhattan distance and Euclidean distance are used as heuristic functions; Manhattan distance is suitable for grid maps, while Euclidean distance is suitable for continuous space, which can be flexibly selected according to actual needs; using open lists and closed sets, nodes with lower objective functions are expanded first and the shortest path is finally returned; if the path exists, the relevant operating cost, carbon emission and third-party risk information are calculated; otherwise, a message that the path is not found is returned; Step (304): Evaluate multiple optimization objectives of each path, and use the dominance relationship to determine whether a solution is better than another solution, that is, solution 1 is considered to dominate solution 2 if and only if solution 1 is not inferior to solution 2 in all objectives and is better than solution 2 in at least one objective; select the Pareto optimal solution through the dominance relationship to obtain a set of solutions, in which no other solution is better than these solutions in all objectives at the same time; Step (305): Path conflict detection, analyzing the path segments in all paths to determine whether multiple customers share the same path segment; if the paths of multiple customers overlap, it is considered that a conflict occurs; for the detected conflicting path segments, the conflicting customer ID is recorded and the conflict information is exported for subsequent analysis or scheduling adjustment.

5. The urban logistics drone scheduling method considering path risk according to claim 1 is characterized in that: The method for constructing the initial scheduling model of urban logistics drones includes: Step (401): Add a new decision variable. Suppose there are N u A battery-swap multi-rotor vertical take-off and landing drone of the same model is responsible for N i Package delivery for N customers, i >N u , the drone set is U={1,2,...,N u }, the set of distribution centers and customer points is I = {0,1,2,...,N i }, 0 represents the distribution center, the drone only carries one customer's package at a time, and returns to the distribution center after completing the delivery. The decision variable y for replacing the battery uij , whether the time of departure from the distribution center meets the time interval decision variable and whether the time of arrival at the distribution center meets the time interval decision variable for: When drone u completes the delivery task for customer i and returns to the delivery center to continue the delivery task for customer j, it needs to replace battery y. uij Take 1, otherwise y uij Take 0, The battery charge of the drone after it returns to the distribution center; is the energy consumption of the segmented flight of drone u carrying cargo to customer j, B is the energy consumption of the segmented flight of UAV u returning from customer j without any payload; min is the minimum safe power of the battery; x uij is the decision variable for task allocation; when there is an identical path segment between the paths from the distribution center to customer i and customer i′, and the time interval between the two departures from the distribution center does not meet the safety interval Take 1, otherwise take 0. The moment when drone u starts delivering to customer i, is the time when drone u′ starts delivering to customer i′, T min is the safe interval for drone takeoff, Γ is the set of customers that have the same path segment when flying from the distribution center to the customer point; when there is the same path segment between the paths from customer i and customer j′ returning to the distribution center respectively, and the time interval between the two arriving at the distribution center does not meet the safe interval Take 1, otherwise take 0. The time when drone u completes the delivery for customer i and arrives at the distribution center, is the time when drone u′ completes the delivery of customer j′ and arrives at the distribution center, Π is the set of customers that have the same path segment when returning from the customer point to the distribution center; Step (402): Set the objective function of the initial scheduling model of urban logistics drones, the delivery time window penalty cost C pen The minimum representation is: minC pen ; Among them, γ ear is the penalty coefficient for the drone to arrive at the customer ahead of schedule, γ lat is the penalty coefficient for the drone’s delayed arrival at the customer, [a i ,b i ] is the delivery time window required by customer i, The time when robot u arrives at customer i and completes unloading the goods; Step (403): Establish the delivery task and load constraints, as follows: m fre ≤m max ; Among them, m fre is the weight of the cargo, m max The maximum load limit for the drone; Step (404): Establish the delivery task time-related constraints, as follows: And customer i is the first customer delivered by drone u in, is the time interval between customer i and customer i′ that needs to be delayed in bringing goods. is the idle arrival time interval that needs to be delayed between customer i and customer j′, is the decision variable for whether the time when customer j and customer i′ take off from the distribution center meets the time interval; is the time interval between customer j and customer i′ that needs to be delayed in bringing goods; t STA is the time when the distribution center starts to operate, T loa The time required to load the drone, T exc The time required to replace the battery of the drone, T unl The time required to unload the drone; The time it takes for drone u to carry cargo to customer i is calculated by the following formula: h carr is the flight height of the drone carrying goods to the customer point, z0 is the take-off and landing point height at the distribution center, and v up is the speed of the drone during take-off phase, is the horizontal cruising distance from the distribution center to customer point i, v hor is the speed of the UAV during the cruising phase, z i is the take-off and landing point height at the customer, v down is the speed of the drone during the descent phase; is the time it takes for UAV u to return from customer i without payload, calculated by the following formula: h emp The altitude at which the drone returns to the distribution center empty. is the horizontal cruising distance from customer i to the distribution center; Step (405): Establish battery power related constraints, as follows: And customer i is the first customer delivered by drone u; in, The battery power of drone u when it is ready to fly to customer i and customer j, B max is the maximum charge of the battery, is the energy consumption of the segmented flight of drone u carrying cargo to customer i, is the energy consumption of the segmented flight of UAV u returning empty-handed from customer i.

6. The urban logistics drone scheduling method considering path risk according to claim 1 is characterized in that: The improved genetic algorithm is used to solve the initial scheduling model of urban logistics drones, including: Step (501): Initialize the population, consider the number of drones, the number of customer points, battery energy consumption, and the key parameters of the time window, randomly arrange the delivery order of customers, and use different arrangements as different individuals, that is, each individual represents a delivery plan, to generate the initial population; Step (502): fitness function definition, which takes into account the time window penalty cost and finds the optimal solution by calculating the performance of each individual scheduling solution in actual application; Step (503): Genetic algorithm evolution, using customized crossover and mutation operations to generate new individuals; the crossover operation introduces new gene combinations by exchanging task sequences between parent individuals, and the mutation operation introduces disturbances in the solution space by randomly adjusting the task sequence, thereby jumping out of the local optimal solution and expanding the solution space; Step (504): Population update and elimination. After each generation of evolution, the algorithm selects individuals in the population according to their fitness values, eliminates individuals with poor fitness, and retains and replicates individuals with high fitness to form the next generation of population. Step (505): Repeated path segment identification, after finding the preliminary optimal solution, based on the pre-imported path segment overlap information, identify customers with repeated path segments; Step (506): Delivery time adjustment: for customers with duplicated route segments, delivery time is adjusted to ensure that the flight times of different drones on the same route are staggered, thereby avoiding air conflicts; the fitness of the adjusted scheduling scheme is recalculated, and the better scheme is retained; Step (507): Conflict check and further adjustment. After the initial delivery time adjustment is completed, the adjusted scheduling plan is checked. If new conflicts are found during the check, adjustments will be made again to ensure that no new potential conflicts arise. Step (508): Output of the initial scheduling plan. After all adjustments and optimizations are completed, the final initial scheduling plan is output. The plan includes the specific task sequence of each drone, the time nodes in the delivery, the battery usage, whether the battery needs to be replaced, and the operating cost of the plan.

7. The urban logistics drone scheduling method considering path risk according to claim 1 is characterized in that: The method for constructing the alternate landing path planning model includes: Step (601): Establish decision variables. During the transport cruise, if the drone fails due to internal reasons and needs to interrupt the delivery, it will immediately select the best alternate landing point for maintenance. Then, it will determine whether to continue the mission based on the maintenance situation. The decision variable x for the failed drone to fly to the alternate landing point is uk for: in, Assemble for faulty drones, For any faulty drone, is the total number of faulty drones, U is the set of drones, K is the set of alternate landing points, K={(x k ,y k ,z k )k∈{0,1,2,...,N k }},N k is the total number of alternate landing points, 0 represents the distribution center, k is any alternate landing point, x k ,y k ,z k is the coordinate of the alternate landing point k in the x, y, z directions; Step (602): Setting the objective function of the alternate path planning model, the third-party risk Z1′ is minimized and the flight time Z2′ is shortest, which is expressed as: in, is the normalized number of ground fatalities caused by drone crashes per flight hour, is the normalized noise impact index, For malfunctioning drones The time required to reach the alternate landing point k is calculated by the following formula: in, For malfunctioning drones Horizontal cruising distance to alternate landing point k, h div v is the flight altitude of the faulty drone flying to the alternate landing point, hor is the speed of the UAV during the cruising phase, v down is the speed of the drone during the descent phase; Step (603): Establish the constraint conditions of the alternate landing path planning model, as follows: Among them, d max is the maximum range of the drone, is the battery power of the drone when it is ready to fly to the alternate landing point, m is the total weight of the drone and the cargo it carries, g is the acceleration of gravity, is the propeller power transmission efficiency of the UAV, σ is the lift-to-drag ratio of the UAV; e is a natural constant; B min is the minimum safe power of the battery; (x λ′ ,y λ′ ,z λ′ ) is the coordinate of the point λ′ that the faulty UAV passes through when flying to the alternate landing point, and Ω is the set of obstacles and no-fly zones.

8. The urban logistics drone dispatching method considering path risk according to claim 1 is characterized in that: The method for constructing the rescheduling model includes: Step (701): Add a new decision variable. Assuming that the faulty drone arrives at the alternate landing point, after inspection by the staff, it is determined that the drone cannot continue to perform the task, and it is necessary to reschedule the drones and delivery times for the interrupted task and all other tasks that have not started delivery. The delivery paths of other tasks remain unchanged except for the interrupted task; the interrupted task is rescued by other drones, that is, other drones are arranged to take off empty from the distribution center and arrive at the alternate landing point. The staff will hand over the goods carried by the faulty drone to the rescue drone, and the drone will deliver the goods to the customer point; according to whether there is a spare drone, the rescheduling strategy is divided into two types: the rescheduling strategy of other existing drones and the rescheduling strategy of adding spare drones. The decision variables for task reallocation for: Step (702): Set the objective function of the rescheduling model. The minimum penalty cost Z3′ of the overall running time window is expressed as: in, is the set of normal drones (hereinafter referred to as ordinary drones) remaining in the original drone set. U is the drone collection, Assemble for faulty drones, is the total number of remaining normal drones, A is the set of spare drones, A={1,2,...,N a },N a is the total number of backup drones, I is the set of customer points; γ ear is the penalty coefficient for the drone to arrive at the customer ahead of schedule, γ lat is the penalty coefficient for the drone’s delayed arrival at the customer, [a i ,b i ] is the delivery time window required by customer i; The time when drone u arrives at customer i and unloads the goods after the delivery is interrupted; Step (703): Establish a constraint that only one of the rescheduling strategies can be selected, as follows: in, is the decision variable for the redistribution of backup drone tasks, DI is the set of customers who have not started delivery when the delivery is interrupted, n is the total number of customers who have not started delivery when the delivery is interrupted, DI={1,2,...,n},DI∈I; Step (704): Establish the available time constraints of the drone, as follows: in, is the time when drone u starts delivering to customer i after the delivery is interrupted. is the time when drone u becomes available, i * For customers whose delivery tasks are interrupted, t * is the moment when the delivery task is interrupted, OI is the set of customers in transit when the delivery is interrupted, m is the total number of customers in transit when the delivery is interrupted, OI={1,2,...,m}OI∈I,I={0,CI,DI,OI}, 0 represents the distribution center, CI is the set of customers who have completed the delivery when the delivery is interrupted; Step (705): Establish the time constraint of the rescue interruption task, as follows: in, After the delivery was interrupted, drone u began to serve customers i * The time of delivery, For malfunctioning drones The time required to reach the alternate landing point k, T che The time required to repair a faulty drone. After the delivery is interrupted, drone u completes customer i and customer i * The time when the delivery task arrives at the distribution center; is the time it takes for drone u to fly from the distribution center to the alternate landing point k, calculated by the following formula: Among them, h carr is the flight height of the drone carrying goods to the customer point, z0 is the take-off and landing point height at the distribution center, and v up is the speed of the drone during take-off phase, is the horizontal cruising distance from the distribution center to the alternate landing point k, v hor is the speed of the UAV during the cruising phase, z k is the take-off and landing point height of the alternate landing point, v down T is the speed of the drone during the descent phase; loa The time required to load the drone, For drone u to fly from alternate landing point k to customer i with cargo * The time is calculated by the following formula: Among them, z k The take-off and landing point altitude for the alternate landing point; To fly from alternate point k to customer i * Horizontal cruising distance; T unl The time required to unload the drone; For customers * and the decision variable of whether the time when customer j′ arrives at the distribution center meets the time interval; For customers * The time interval of no-load arrival that needs to be delayed between client j and client j′; For drone u from customer i * Time for no-load return; Step (706): Establish the constraints related to the delivery task time, as follows: whether the time of departure from the distribution center meets the decision variable of the time interval and whether the time of arrival at the distribution center meets the time interval decision variable for: And customer i is the first customer delivered by drone u after the delivery interruption in, is the time when drone u′ starts delivering to customer i′ after the delivery is interrupted, is the time when drone u′ completes the delivery task for customer j′ and arrives at the delivery center after the delivery is interrupted; T min Provide safe interval for drone takeoff; is the time interval between customer i and customer i′ that needs to be delayed in bringing goods. is the idle arrival time interval that needs to be delayed between customer i and customer j′; T loa The time required to load the drone, T exc The time required to replace the battery of the drone, T unl The time required to unload the drone; The time it takes for drone u to carry cargo to customer i is calculated by the following formula: z i is the take-off and landing point height at customer i; is the time it takes for UAV u to return from customer i without payload, calculated by the following formula: is the horizontal cruising distance from the distribution center to customer point i, is the horizontal cruising distance from customer i to the distribution center, h emp is the flight altitude of the drone returning to the distribution center without a load; Γ is the set of customers that have the same path segment when flying from the distribution center to the customer point; Π is the set of customers that have the same path segment when returning from the customer point to the distribution center; Step (707): Establish battery power related constraints, as follows: in, The decision variables for replacing the drone’s battery after a delivery interruption, The decision variables for task reallocation, The battery power of drone u when it is ready to fly from the distribution center to customers i and j, The battery charge of the drone after it returns to the distribution center; is the energy consumption of the segmented flight of drone u carrying cargo to customer j, B is the energy consumption of the segmented flight of UAV u returning from customer j without any payload; min B is the minimum safe power of the battery; max The maximum charge of the battery.

9. An urban logistics drone dispatching device considering path risk, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the method according to any one of claims 1 to 8.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images