A heterogeneous unmanned aerial vehicle transport scheduling method and device for urban and rural logistics

Abstract

The application discloses a heterogeneous unmanned aerial vehicle transport capacity scheduling method and equipment for urban and rural logistics, and belongs to the technical field of intelligent logistics and heterogeneous unmanned aerial vehicle transport capacity. The method comprises the following steps: determining an initial population based on multi-source data, wherein each individual in the initial population represents a heterogeneous unmanned aerial vehicle group transport capacity scheduling scheme; determining a physical constraint model of the heterogeneous unmanned aerial vehicle group based on a preset constraint rule and the multi-source data; constructing at least two types of operation scheduling target functions in combination with the multi-source data and taking the physical constraint model as a boundary constraint; integrating all the target functions through weighted aggregation to obtain a multi-objective optimization model; and performing iterative calculation on the individuals in the initial population by using a multi-objective optimization algorithm based on the multi-objective optimization model and the physical constraint model to obtain an optimal solution set of the unmanned aerial vehicle transport capacity scheduling. The method improves the flexible adjustment capability of the heterogeneous unmanned aerial vehicle diversified distribution transport capacity strategy for urban and rural logistics.

Description

Technical Field

[0001] This application relates to the fields of smart logistics and heterogeneous drone transportation capacity technology, and in particular to a method and equipment for scheduling heterogeneous drone transportation capacity for urban and rural logistics. Background Technology

[0002] With the rapid development of drone technology and the increasing demand for smart logistics, the application of drone delivery capacity in urban delivery scenarios is gradually increasing. However, existing drone dispatching technology still has certain limitations.

[0003] On the one hand, traditional drone scheduling methods are usually based on homogeneous drone models and often use fixed models to configure all delivery tasks. This method is difficult to adapt to the differences in tasks and the diversity of needs in urban and rural scenarios. As a result, in urban and rural logistics networks with significant differences in infrastructure conditions, varied terrain and environment, and uneven delivery distance and load requirements, drone performance is redundant or mission capabilities are insufficient, resulting in low resource utilization.

[0004] On the other hand, urban and rural logistics and distribution often need to balance multiple objectives such as cost sensitivity, timeliness requirements, airspace control, and environmental impact. Because conflicts can easily arise between these multi-objective optimization goals, and refined modeling is difficult, existing technologies mostly focus on optimizing a single objective, such as the shortest delivery time or lowest energy consumption, lacking the ability to collaboratively optimize multiple objectives. This may result in an inability to adapt to diverse delivery needs and limit the scope of application scenarios.

[0005] On the other hand, in urban and rural logistics and distribution scenarios, factors such as the number of packages, weather conditions, and airspace restrictions are highly dynamic. However, existing algorithms often rely on static constraint modeling, resulting in a lack of flexible adjustment capabilities for capacity strategies, making it difficult to adapt to the dynamic changes in actual scenarios. Summary of the Invention

[0006] This application provides a method and device for scheduling heterogeneous drones for urban and rural logistics, which solves the problems of inflexible drone scheduling and low resource utilization when performing urban and rural logistics delivery tasks.

[0007] The embodiments of this application adopt the following technical solutions:

[0008] On the one hand, embodiments of this application provide a method for scheduling heterogeneous unmanned aerial vehicle (UAV) capacity, the method comprising:

[0009] Based on multi-source data, an initial population is determined, and each individual in the initial population represents a heterogeneous UAV swarm capacity scheduling scheme.

[0010] Based on the preset constraint rules and the multi-source data, the physical constraint model of the heterogeneous UAV swarm is determined.

[0011] Using the physical constraint model as the boundary constraint and combining the multi-source data, at least two types of objective functions for runtime scheduling are constructed.

[0012] By integrating all objective functions, a multi-objective optimization model is obtained.

[0013] Based on the multi-objective optimization model and the physical constraint model, a multi-objective optimization algorithm is used to iteratively calculate the individuals in the initial population to obtain the optimal solution set for UAV capacity scheduling.

[0014] In one example, the multi-source data includes real-time multi-source data and historical multi-source data;

[0015] The real-time multi-source data includes: mission requirement data, status data of various UAV models, environmental perception data, airspace control data, and electricity price data;

[0016] The historical multi-source data includes: historical mission execution data, historical operation data of various UAV models, historical environmental impact data, and historical scheduling decision data.

[0017] In one example, determining the physical constraint model of a heterogeneous drone swarm based on preset constraint rules and multi-source data specifically includes:

[0018] By using the status data of various UAV models from the multi-source data and combining the safety redundancy rules in the preset constraint rules, a nonlinear load constraint model is established.

[0019] Based on the status data of various UAV models, environmental perception data, and historical environmental impact data from the multi-source data, and combined with the environmental impact correction rules in the preset constraint rules and the environmental compensation coefficients that match the environmental perception data, a range constraint model is constructed.

[0020] Based on the status data of various UAV models and airspace control data from the multi-source data, and combined with the airspace layering rules in the preset constraint rules, a layered speed constraint model is established.

[0021] The nonlinear load constraint model, the range constraint model, and the speed constraint model are integrated to obtain the physical constraint model.

[0022] In one example, the construction of at least two types of objective functions for runtime scheduling, using the physical constraint model as boundary constraints and combining the multi-source data, specifically includes:

[0023] Using the status data, task requirement data, historical operation data, airspace control data and environmental perception data of various types of UAVs from multiple sources as input, the physical constraint model as boundary constraint, and the latest time for the heterogeneous UAV swarm to complete all delivery tasks as the optimization index, a delivery time objective function is constructed.

[0024] Using the status data of various types of UAVs, mission requirement data, historical operation data of various types of UAVs, electricity price data, airspace control data and environmental perception data from multiple sources as inputs, the physical constraint model is used as the boundary constraint, and the minimum total delivery cost to complete the delivery task is used as the optimization index to construct the delivery cost objective function.

[0025] Based on the preset coupling rules of load and energy consumption, the energy consumption objective function is constructed by taking the status data, task requirement data and historical operation data of each type of UAV from multiple sources as input, the physical constraint model as boundary constraint, and the minimum total energy consumption for completing the delivery task as the optimization index.

[0026] In one example, the process of integrating all objective functions to obtain a multi-objective optimization model specifically includes:

[0027] Standardize all objective functions;

[0028] The standardized objective function is multiplied by its corresponding weight coefficient to obtain the weighted term of each objective function;

[0029] The multi-objective optimization model is obtained by summing all weighted terms.

[0030] In one example, the step of using a multi-objective optimization algorithm to iteratively calculate the individuals in the initial population based on the multi-objective optimization model and the physical constraint model to obtain the optimal solution set for UAV capacity scheduling specifically includes:

[0031] Based on the parameter values ​​corresponding to each individual in the initial population and the multi-objective optimization model, the multi-objective optimization algorithm is used to calculate the multi-dimensional objective value of each individual in the initial population;

[0032] The multidimensional objective values ​​of all individuals are hierarchically divided using a preset sorting method, and non-dominated solutions that satisfy the constraints of the physical constraint model are selected to form the latest solution set.

[0033] Continue multi-objective optimization on the latest solution set until the maximum number of iterations is reached;

[0034] The latest solution set is taken as the optimal solution set.

[0035] In one example, the method further includes:

[0036] The iteration process terminates when the change in the distribution entropy of the latest solution set is less than a preset threshold.

[0037] The latest solution set is taken as the optimal solution set.

[0038] In one example, before determining the physical constraint model of a heterogeneous UAV swarm based on preset constraint rules and multi-source data, the method further includes:

[0039] Outliers in the multi-source data are filtered using a statistical threshold method, and the filtered outliers are processed using interpolation or deletion methods.

[0040] A unique identifier is generated for each data record in the multi-source data after outlier cleaning based on a hash algorithm. By comparing the unique identifier with the database primary key stored in a preset database, duplicate and conflicting data in the multi-source data are eliminated.

[0041] Based on preset composite feature conditions, feature extraction processing is performed on the cleaned multi-source data to construct composite features, which are used to quantify task priorities.

[0042] The latitude and longitude coordinates in the multi-source data are discretized into grid codes to enhance spatial correlation;

[0043] The multi-source data, after feature extraction and encoding, is mapped to a unified numerical range using a normalization method.

[0044] In one example, the method further includes:

[0045] Acquire basic demand data, spatiotemporal correlation data, auxiliary verification data, geographic environment data, airspace condition data, and status data of various UAV models;

[0046] The aforementioned basic demand data, spatiotemporal correlation data, and auxiliary verification data are subjected to feature parsing processing to extract demand scale features, temporal concentration features, spatial dispersion features, and time sensitivity features.

[0047] By using time series analysis, the seasonal fluctuation pattern of agricultural product demand is identified based on the time concentration characteristics. Combined with the time sensitivity characteristics, the periodic change characteristics of demand are mined. Based on the seasonal fluctuation pattern and the periodic change characteristics, the dynamic requirements for delivery timeliness and transportation capacity are determined.

[0048] Based on the spatial dispersion characteristics, the distribution dispersion of industrial goods downstream demand in geographic space is quantified. Combined with the demand scale characteristics, the total demand in different regions is evaluated. By integrating the distribution dispersion and the total demand, the spatial distribution dispersion characteristics and regional accessibility of industrial goods downstream demand are determined. Combined with the temporal concentration characteristics, the overall density of regional logistics demand and the concentration of demand in different time periods are determined.

[0049] The geographic environment data and the airspace condition data are input into a preset three-dimensional spatial analysis model to determine the site selection spatial constraints.

[0050] The dynamic requirements for delivery timeliness and capacity, the overall density of regional logistics demand and the concentration of demand at different times, and the status data of each type of UAV are input into a preset multi-objective integer programming location model. The location space constraint is used as the boundary constraint. The preset hybrid heuristic algorithm is used to iteratively solve the multi-objective integer programming location model to obtain the optimal solution set for the delivery station location.

[0051] Based on the optimal solution set for the distribution station location, real-time multi-source data is determined from the multi-source data.

[0052] On the other hand, embodiments of this application provide a heterogeneous drone capacity scheduling device for urban and rural logistics, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute any of the above-described heterogeneous drone capacity scheduling methods for urban and rural logistics.

[0053] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects:

[0054] The method provided in this application solves the technical problem that traditional drone scheduling based on fixed configurations of homogeneous drone models cannot adapt to the differences in urban delivery tasks by obtaining an initial population that characterizes the capacity scheduling scheme of heterogeneous drone swarms. It enables heterogeneous drone swarms to flexibly adapt to different delivery tasks and improves resource utilization.

[0055] By constructing an objective function and deriving a multi-objective optimization model based on it, this approach addresses the shortcomings of existing technologies that focus on single-objective optimization, lack multi-objective collaborative optimization capabilities, and are unable to adapt to diverse delivery needs. It achieves multi-objective collaborative optimization, improving adaptability to diverse delivery demands and scenario coverage. Furthermore, by determining the physical constraint model of heterogeneous drone swarms based on preset constraint rules and multi-source data, and combining the multi-objective optimization model with the physical constraint model for iterative calculations, this approach solves the technical problems of existing algorithms relying on static constraint modeling, lacking flexible adjustment capabilities for capacity strategies, and being unable to adapt to dynamic changes in community delivery scenarios. It enables capacity scheduling schemes to adapt to dynamic factors such as package quantity, weather, and airspace restrictions, improving the flexibility of capacity strategy adjustments. Attached Figure Description

[0056] To more clearly illustrate the technical solution of this application, some embodiments of this application will be described in detail below with reference to the accompanying drawings, in which:

[0057] Figure 1 A flowchart illustrating a drone capacity scheduling method for urban and rural logistics provided in this application embodiment;

[0058] Figure 2 This is a structural schematic diagram of a heterogeneous unmanned aerial vehicle (UAV) capacity scheduling device for urban and rural logistics, provided as an embodiment of this application. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0060] Some embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0061] Figure 1 This is a flowchart illustrating a drone capacity scheduling method for urban and rural logistics, provided as an embodiment of this application. This method can be applied to different business domains. The process can be executed by computing devices specific to that domain, and certain input parameters or intermediate results can be manually adjusted to improve accuracy.

[0062] The analysis method involved in the embodiments of this application can be implemented by a terminal device or a server, and this application does not impose any special limitations on it. For ease of understanding and description, the following embodiments are all described in detail using a server as an example.

[0063] It should be noted that the server can be a single device or a system composed of multiple devices, i.e., a distributed server. This application does not make any specific limitations on this.

[0064] Figure 1 The process includes the following steps:

[0065] S101. Based on multi-source data, determine the initial population. Each individual in the initial population represents a heterogeneous UAV swarm capacity scheduling scheme.

[0066] The initial population is the starting solution set for the multi-objective optimization algorithm iterations, containing multiple scheduling schemes with basic feasibility. An individual is a single element in the initial population, representing the drone type matching rule, task allocation logic, and path planning sequence. The heterogeneous drone swarm is a cluster composed of various drone models. Different models differ in performance parameters such as payload, range, and speed, adapting to different delivery needs. The drone type matching rule is a logical rule that determines the appropriate drone model based on the delivery task, such as package characteristics (weight and volume), and drone performance, such as rated payload and range.

[0067] Task allocation logic refers to the allocation strategy that assigns multiple delivery tasks to different drones, enabling configurations of a single drone for a single task or a single drone for multiple tasks. Path planning sequence refers to the sequence of delivery points that a drone passes through in order as it completes its assigned tasks, determining the drone's flight path.

[0068] Multi-source data refers to a set of input data across all scenarios that supports multi-objective optimization of heterogeneous UAV capacity scheduling. In some embodiments of this application, multi-source data may include, for example, real-time multi-source data and historical multi-source data. Real-time multi-source data may include, for example, mission requirement data, status data of various UAV models, environmental perception data, airspace control data, and electricity price data. Historical multi-source data may include, for example, historical mission execution data, historical operation data of various UAV models, historical environmental impact data, and historical scheduling decision data.

[0069] Among them, real-time multi-source data refers to dynamic data collected in real time during the scheduling process, reflecting the real-time status of the current delivery scenario. Task requirement data refers to relevant data for packages to be delivered, such as package weight, volume, delivery distance, delivery point geographical coordinates, and delivery priority. Drone status data refers to the real-time operating status data of each drone model, such as remaining battery power, current payload, cumulative flight count, and battery degradation. Environmental perception data refers to environmental data for the current delivery area, such as real-time wind speed, precipitation, and visibility. Airspace control data refers to management restrictions in the current airspace, such as no-fly zones, speed limits at different altitudes, and speed-limited periods. Electricity price data refers to current electricity charges, used to calculate the electricity consumption cost of drone operation.

[0070] Historical multi-source data refers to statistical data accumulated during past delivery processes, providing historical reference for constraining models. Historical task execution data refers to the execution records of past delivery tasks, which may include, for example, package characteristics, delivery routes, and completion times. Historical operational data for each drone model refers to the past operational records of each drone model, which may include, for example, historical failure rates, maintenance costs, battery degradation patterns, and energy consumption data under different operating conditions. Historical environmental impact data refers to statistical data on the impact of different environmental conditions, such as different wind speeds and precipitation, on drone endurance and speed. Historical scheduling decision data refers to records of past scheduling schemes and their execution effects, which may include, for example, aircraft model matching results, task allocation strategies, and path planning schemes.

[0071] For example, the initial population size can be set to 500 individuals. Each individual uses a multi-dimensional parameter encoding to represent the scheduling scheme. The heterogeneous UAV swarm capacity scheduling scheme can include, for example, UAV type matching rules, task allocation logic, and path planning sequence.

[0072] Among them, the drone type matching rule encoding can be binary encoding, with each bit corresponding to a drone model. The encoding value of 1 indicates that the drone model participates in the current scheduling task. The encoding length is consistent with the number of available drone models. The matching logic is based on the composite feature of delivery urgency to initially screen suitable drone models.

[0073] The task allocation logic encoding can, for example, use integer encoding, where the encoded value corresponds to the delivery task number and the encoded position corresponds to the drone number. For example, if the encoded value of the i-th position is j, it means that the i-th drone is assigned to the j-th delivery task.

[0074] For example, real number encoding can be used for path planning sequence encoding. The encoding value can correspond to the nine-square grid encoding of the delivery point network map. The drone flight path is formed according to the encoding order to ensure that the path is adapted to the spatial correlation.

[0075] It should be noted that the initial population generation must meet basic constraints, such as the payload of a single flight not exceeding the threshold of the drone's maximum payload, and the total path distance not exceeding the threshold of the drone's current range, to avoid the initial solution being directly eliminated due to violation of core constraints.

[0076] Furthermore, to ensure the quality and consistency of multi-source data and eliminate dimensional differences between them, before determining the physical constraint model of the heterogeneous UAV swarm based on preset constraint rules and multi-source data, the method also includes:

[0077] Outliers in multi-source data are filtered using statistical thresholding, and interpolation or deletion methods are used to process them. A unique identifier is generated for each data record in the cleaned multi-source data using a hash algorithm. This unique identifier is then compared with the primary key stored in a pre-defined database to remove duplicate and / or conflicting data. Based on pre-defined composite feature conditions, feature extraction is performed on the cleaned multi-source data to construct composite features, which are used to quantify task priorities. The latitude and longitude coordinates in the multi-source data are discretized into grid codes to enhance spatial correlation. A normalization method is used to map the feature-extracted and encoded multi-source data to a unified numerical range.

[0078] For example, firstly, outliers are screened using the 3σ principle, which means exceeding the mean by plus or minus three standard deviations, combined with a statistical threshold method. For instance, oversized packages or unreasonable delivery distances can be marked as outliers. A second screening process using business logic directly deletes invalid outliers and corrects them, such as slightly off-center delivery distances. Missing weather influencing factors are filled into the daily weather station data using linear interpolation.

[0079] Secondly, for example, a unique identifier can be generated for each data record based on the SHA-256 hash algorithm, and compared with the preset database primary key to remove duplicate data, such as duplicate orders and conflicting data, such as contradictory drone status records. This ensures the integrity of data transmission and storage, and avoids data tampering that could lead to deviations in the constraint modeling.

[0080] Subsequently, composite features are constructed, such as delivery urgency = (package volume / drone rated payload) × (delivery distance / drone range), to quantify task priority.

[0081] Finally, the adaptive Min-Max normalization method is adopted. The adaptive Min-Max normalization formula can be expressed by, for example, the following formula (1).

[0082] (1)

[0083] in, This is used to avoid interference from extreme values. For long-tailed distribution data such as package weight, an additional Box-Cox transformation can be performed to improve the normality of the data distribution. The Box-Cox transformation formula can be expressed, for example, by the following formula (2).

[0084] (2)

[0085] in, Determined by the maximum likelihood estimation method.

[0086] S102. Based on preset constraint rules and multi-source data, determine the physical constraint model of the heterogeneous UAV swarm.

[0087] Among them, the preset constraint rules refer to a set of rules formulated based on the safety requirements for drone operation and the delivery business specifications. For example, they may include safety redundancy rules, environmental impact correction rules, airspace layering rules, etc.

[0088] A physical constraint model refers to a set of boundary conditions that limit the feasibility of a drone scheduling scheme. These conditions may include constraints such as payload, range, and speed, ensuring that the scheduling scheme meets the physical performance and operational safety requirements of the drone.

[0089] In some embodiments of this application, a physical constraint model for a heterogeneous UAV swarm is determined based on preset constraint rules and multi-source data, specifically including:

[0090] A nonlinear load constraint model is established using state data of various UAV models from multi-source datasets, combined with the safety redundancy rule in the preset constraint rules. A range constraint model is constructed based on state data of various UAV models from multi-source datasets, environmental perception data, and historical environmental impact data, combined with the environmental impact correction rule in the preset constraint rules and the environmental compensation coefficient matching the environmental perception data. A hierarchical speed constraint model is established based on state data of various UAV models from multi-source datasets and airspace control data, combined with the airspace layering rule in the preset constraint rules. The nonlinear load constraint model, range constraint model, and speed constraint model are integrated to obtain the physical constraint model.

[0091] Among them, the nonlinear load constraint model refers to a load limit model constructed based on nonlinear programming, which considers safety redundancy and avoids operational risks caused by exceeding load limits. The endurance constraint model refers to a constraint model that limits the flight distance of the UAV, which is dynamically adjusted in combination with environmental factors to ensure that the UAV can complete its mission and return safely. The hierarchical speed constraint model refers to a speed limit model constructed according to airspace hierarchical rules, with different altitude layers corresponding to different speed ranges, in compliance with airspace control requirements.

[0092] For example, based on the status data of various UAV models in multi-source data, such as rated load, and combined with preset safety redundancy rules, such as safety redundancy coefficient, a nonlinear programming model is established. The formula for the nonlinear programming model can be expressed by, for example, the following formula (3).

[0093] (3)

[0094] in, For rated load, Let be the weight of the i-th package, and n be the number of packages delivered in a single trip. The safety redundancy factor can be expressed as, for example, as... By reserving a 10% load redundancy, the instability of the drone flight is avoided due to overloading.

[0095] It should be noted that the total payload of a single drone flight It should be less than or equal to the rated load. .

[0096] For example, based on the status data of various UAV models in multi-source data, such as rated range, environmental perception data, such as real-time wind speed, and historical environmental impact data, combined with environmental impact correction rules and wind speed compensation coefficients, a range constraint model is constructed. The range constraint model formula can be expressed by, for example, the following formula (4).

[0097] (4)

[0098] in, For the rated driving range, For real-time wind speed, the wind speed compensation coefficient can be obtained, for example, through... The coefficient 0.05 is based on historical environmental impact data and reflects the proportion of range loss caused by air resistance for every 1 m / s increase in wind speed. The absolute value ensures that both positive and negative wind speeds (tailwind / headwind) have a loss effect.

[0099] Furthermore, due to the current battery capacity The actual driving range is affected by the energy consumption conversion relationship. Battery aging effect correction can also be introduced. The battery aging correction formula can be expressed by the following formula (5).

[0100] (5)

[0101] in, This is the initial energy of the battery. To accumulate the number of charge and discharge cycles, The maximum number of charge-discharge cycles represents the maximum battery life. Battery aging directly affects the maximum driving range. 0.9 is the degradation coefficient, indicating that after each 100% charge-discharge cycle, the battery capacity retains 90% of its initial value, which is consistent with the exponential degradation characteristics of lithium batteries.

[0102] Based on this, the actual driving range formula can be expressed, for example, by the following formula (6).

[0103] (6)

[0104] in, The rated flight range of the drone, This is the battery capacity degradation ratio, combined with the wind speed compensation coefficient, to comprehensively correct the impact of environmental and equipment losses on battery life.

[0105] For example, based on the status data of various drone models from multi-source data, such as rated speed range and airspace control data (e.g., altitude restrictions), combined with airspace stratification rules, layered speed constraints are established. For instance, airspace stratification rules can define altitudes below 100 meters as low-altitude, suitable for safety in densely populated urban areas; 100 to 300 meters as mid-altitude, balancing flight efficiency and operational safety; and above 300 meters as high-altitude, suitable for long-distance suburban delivery scenarios. The speed coefficients for different altitude levels are set based on airspace control data: lower speeds at low altitudes ensure personnel safety, while higher speeds at high altitudes optimize efficiency.

[0106] The establishment of hierarchical velocity constraints can be expressed, for example, by the following formulas (7), (8), and (9).

[0107] Lower atmosphere: (7)

[0108] Hollow layer: (8)

[0109] Upper atmosphere: (9)

[0110] in, , For the rated speed range of the drone, This refers to the actual flight speed.

[0111] The aforementioned nonlinear load constraint model, endurance constraint model, and hierarchical speed constraint model are logically integrated to form a unified physical constraint model, thereby clarifying the feasible solution boundary of the UAV scheduling scheme.

[0112] S103. Using physical constraint models as boundary constraints and combining multi-source data, construct at least two types of objective functions for runtime scheduling.

[0113] Boundary constraints refer to the boundary conditions that limit the feasible solutions of the objective function, ensuring that the optimization direction meets practical requirements. The objective function is a mathematical expression that quantifies the optimization objective and evaluates the performance of the scheduling scheme on a specific objective.

[0114] In some embodiments of this application, at least two types of objective functions for runtime scheduling are constructed using physical constraint models as boundary constraints and preprocessed multi-source data, specifically including:

[0115] Using multi-source data such as the status data of various UAV models, mission requirement data, historical operation data of various UAV models, airspace control data, and environmental perception data as inputs, and physical constraint models as boundary constraints, the latest time for the heterogeneous UAV swarm to complete all delivery tasks is used as the optimization index to construct a delivery time objective function.

[0116] For example, the objective function for delivery time can be expressed by the following formula (10).

[0117] (10)

[0118] in, The latest time to complete all tasks is determined by taking the maximum value of the completion times of all drone tasks to ensure overall delivery timeliness, rather than the average time, and to avoid some tasks exceeding the time limit. The flight time formula can be expressed, for example, by the following formula (11).

[0119] (11)

[0120] in, For flight distance, This refers to the actual flight speed. The maximum speed limit in the airspace is regulated; the speed must not exceed the upper limit stipulated by airspace control.

[0121] Furthermore, the actual flight speed of the drone can be dynamically adjusted based on the cargo weight w. The formula for correcting the speed under load can be expressed, for example, by the following formula (12).

[0122] (12)

[0123] in, This is the corrected actual flight speed. Where w is the maximum speed of the drone and w is the weight of the cargo. When w ≤ 3kg, there is no speed reduction. When w > 3kg, the speed decreases linearly with the increase of weight, and the maximum decrease is to a preset ratio to adapt to the effect of the payload on flight drag.

[0124] In addition, the delivery time objective function incorporates a dynamic adjustment mechanism. When sudden weather occurs in the delivery area, such as heavy rain or strong winds, the flight speed and waiting time are adjusted in real time based on environmental perception data to ensure the accuracy of delivery time prediction.

[0125] The loading and unloading time can be expressed, for example, by the following formula (13).

[0126] (13)

[0127] Here, 3 minutes is the preset fixed loading and unloading base time, and 0.5 minutes / kg is the load sensitivity coefficient, which can be calibrated based on historical task execution data to reflect the impact of cargo weight on loading and unloading efficiency. This refers to the weight of the goods.

[0128] This refers to the waiting time of the i-th drone during the delivery mission.

[0129] Using multi-source data such as the status data of various UAV models, mission requirement data, historical operation data of various UAV models, electricity price data, airspace control data, and environmental perception data as inputs, and with physical constraint models as boundary constraints, the objective function of delivery cost is constructed with the minimum total delivery cost to complete the delivery task as the optimization index.

[0130] For example, the objective function for delivery costs can be expressed by the following formula (14).

[0131] (14)

[0132] in, For the total delivery cost, For depreciation costs, For regular maintenance costs, For sudden repair costs, For electricity consumption costs, It covers airspace usage costs and the entire lifecycle costs of drone operation.

[0133] Depreciation cost For example, it can be expressed by the following formula (15),

[0134] (15)

[0135] in, The cost of purchasing a single drone, The total number of flights during the design life of a drone. This represents the actual number of flights.

[0136] Regular maintenance costs For example, it can be expressed by the following formula (16),

[0137] (16)

[0138] in, This refers to the duration of a single flight. Price is per unit of flight time. Routine maintenance costs are calculated cumulatively based on flight time.

[0139] Unexpected repair costs For example, it can be expressed by the following formula (17),

[0140] (17)

[0141] in, This is a correction factor for the number of historical failures. ≥1 indicates a higher number of failures. The larger the value, the higher the threshold is set based on the historical failure rate model. When the failure probability is less than 5%, the cost of sudden maintenance is ignored. When it is greater than 5%, a correction coefficient is added according to a fixed standard to balance the accuracy of cost estimation and the computational complexity.

[0142] Electricity consumption cost For example, it can be expressed by the following formula (18),

[0143] (18)

[0144] in, To consume electricity for operation, 0.05 is the basic energy consumption coefficient, which is related to distance; 0.02 is the load sensitivity coefficient. For flight distance, For load-bearing capacity, The energy consumption is linearly related to distance and payload, which is consistent with the actual energy consumption patterns of flight.

[0145] airspace usage costs The pricing is based on flight distance and altitude, and will not be elaborated further in this embodiment.

[0146] Based on the preset coupling rules of load and energy consumption, the energy consumption objective function is constructed by taking the status data, task requirement data and historical operation data of each type of UAV from multiple sources as input, the physical constraint model as boundary constraint, and the minimum total energy consumption for completing the delivery task as the optimization index.

[0147] For example, the energy consumption objective function can be expressed by the following formula (19),

[0148] (19)

[0149] in, The total energy consumption of the system. Let i be the flight distance of the i-th segment. Let i be the payload of the i-th segment of flight. Let be the flight speed of segment i. 0.05 is the basic cruise energy consumption coefficient, which is related to air resistance. 0.02 is the load sensitivity coefficient. 0.001 is the speed square term coefficient, which reflects the characteristic that aerodynamic drag increases with the square of wind speed, and comprehensively couples the effects of distance, load, and speed on energy consumption.

[0150] The upper limit of energy consumption for a single flight can be expressed, for example, by the following formula (20).

[0151] (20)

[0152] in, This represents the maximum energy consumption for flight i. This refers to the current battery energy. For example, a preset value of battery energy can be reserved as a safety redundancy to prevent the drone from losing connection due to excessive battery discharge.

[0153] S104. Integrate all objective functions to obtain a multi-objective optimization model.

[0154] In some embodiments of this application, all objective functions are integrated to obtain a multi-objective optimization model, specifically including:

[0155] Standardize all objective functions. Multiply each standardized objective function by its corresponding weight coefficient to obtain a weighted term for each objective function. Sum all weighted terms to obtain the multi-objective optimization model.

[0156] Standardization refers to the method of eliminating the differences in the dimensions of different objective functions, mapping the objective function values ​​to a unified interval, and ensuring the fairness of weight allocation. For example, the standardization of objective functions can be achieved through the following formula (21). The weight coefficients reflect the importance of each objective function; the larger the weight, the higher the priority. The sum of all weight coefficients is 1. The weighting term is the product of the standardized objective function and the corresponding weight coefficient, reflecting the contribution of the objective to the comprehensive optimization. The multi-objective optimization model comprehensively reflects the optimization needs of the three types of objectives.

[0157] S105. Based on the multi-objective optimization model and the physical constraint model, the multi-objective optimization algorithm is used to iteratively calculate the individuals in the initial population to obtain the optimal solution set for UAV capacity scheduling.

[0158] Among them, the multi-objective optimization algorithm is an algorithm for solving multi-objective optimization problems and is used to generate the optimal solution set. For example, it can be NSGA-II, NSGA-III, multi-objective particle swarm optimization, simulated annealing, tabu search, etc. The embodiments of this application do not limit this.

[0159] In some embodiments of this application, based on a multi-objective optimization model and a physical constraint model, a multi-objective optimization algorithm is used to iteratively calculate the individuals in the initial population to obtain the optimal solution set for UAV capacity scheduling, specifically including:

[0160] Based on the parameter values ​​of each individual in the initial population and the multi-objective optimization model, a multi-objective optimization algorithm is used to calculate the multi-dimensional objective value of each individual in the initial population. A pre-defined sorting method is used to hierarchically divide the multi-dimensional objective values ​​of all individuals, and non-dominated solutions that satisfy the physical constraint model are selected to form the current latest solution set. Multi-objective optimization continues on the current latest solution set until the maximum number of iterations is reached. The current latest solution set is then taken as the optimal solution set.

[0161] In this context, multidimensional objective values ​​refer to the values ​​of an individual on each single objective function, such as delivery time, delivery cost, and system energy consumption. The calculation of multidimensional objective values ​​requires the integration of real-time environmental perception data, such as real-time wind speed, dynamic correction time for precipitation, cost, and energy consumption, to ensure the calculation results accurately reflect the current scenario. The preset sorting method refers to the method for hierarchically dividing the multidimensional objective values ​​of individuals; for example, a fast non-dominated sorting method can be used. A non-dominated solution is a solution for which there is no better individual on the multidimensional objective value, meaning no other solution is superior to this solution on all objectives. The final non-dominated solution set obtained after the algorithm iterations terminate contains optimal scheduling schemes with multiple different objective trade-offs.

[0162] The method also includes terminating the iteration process when the change in the distribution entropy of the latest solution set is less than a preset threshold. The latest solution set is then taken as the optimal solution set.

[0163] Among them, distribution entropy is an indicator of the dispersion of the solution set. The larger the distribution entropy, the more dispersed the solution set is and the better its diversity. A preset threshold is a critical value for judging whether the solution set has converged. For example, it can be set to 0.01. When the change in distribution entropy is less than this value for 50 consecutive generations, it indicates that the solution set tends to be stable.

[0164] In a concrete example, for each individual in the initial population, the delivery time, delivery cost, and system energy consumption are calculated to verify whether the individual meets the physical constraint model. For example, the payload of a single flight is less than the maximum payload limit of the drone, the total path distance is less than the battery endurance limit, and the flight speed is within the safe flight range in the airspace. This ensures that the initial solution has basic feasibility, so as to screen feasible solutions and eliminate invalid solutions.

[0165] A fast non-dominated sorting method is used to divide feasible solutions into different non-dominated levels. The first level consists of non-dominated solutions, meaning that no other solution is superior to this solution on all objectives. The crowding distance of the non-dominated solutions is calculated. A larger crowding distance indicates a more dispersed distribution of solutions in the solution set; solutions with larger crowding distances are retained to ensure the diversity of the solution set. Non-dominated solutions that satisfy the physical constraints are selected to form the current latest solution set.

[0166] The algorithm performs selection, crossover, and mutation operations on the latest solution set to generate a new generation of population. For example, a roulette wheel selection algorithm can be used, with a crossover probability of 0.9 and a mutation probability of 0.1. Crossover combines superior genes from parent generations, and mutation introduces new genes, preventing the algorithm from getting trapped in local optima. An elite retention strategy is employed, retaining the top 30% of non-dominated solutions in each generation to avoid losing high-quality solutions. The multidimensional objective value calculation, non-dominated sorting, and solution set selection steps are repeated to update the latest solution set. Iteration terminates after 500 generations or when the entropy change in the solution set distribution is less than 0.01 for 50 consecutive generations. After iteration termination, the latest solution set is the Pareto optimal solution set, which may include 30-50 non-dominated solutions, each labeled with its objective trade-off ratio.

[0167] Furthermore, in order to improve the scenario adaptability of the scheduling scheme and increase the utilization rate of heterogeneous drone resources,

[0168] In some embodiments of this application, the unpacking filtering rules can be dynamically adjusted according to the package type, and weight coefficients can be assigned in real time through the business rule engine, such as the time weight for emergency medical supplies. Cost weight Energy consumption weight Cost weighting for ordinary express delivery Time weight Energy consumption weight According to the weighting formula Calculate the overall score and filter out solutions that do not meet the weight threshold.

[0169] Furthermore, it can handle no-fly zones by calling a preset airspace API to identify whether the solution path covers a no-fly zone and eliminate conflicting solutions. It can also force a correction for solutions requiring detours, demanding that the detour distance be less than or equal to a preset value for the original path.

[0170] Traffic flow speed adaptation processing is performed. During urban speed limit periods, such as from 9 pm to 7 am, the speed constraint is adjusted to v being less than or equal to the preset speed limit threshold, and the time target is recalculated.

[0171] The route is corrected. If the delivery time exceeds the standard after detouring, the sub-route is split. For example, a single delivery is split into relay stations to ensure that the delivery time meets the standard.

[0172] It can also perform capacity configuration generation and task allocation processing, such as aircraft model matching, allocating drone models according to payload requirements, for example, selecting a heavy-duty drone for a 5 kg package, and accurately matching the performance of heterogeneous drones with mission requirements. Another example is building... Matrix, where, For the number of drones, For each delivery point, the matrix element values ​​represent task priorities, clearly defining the correspondence between drones and delivery points. A drone capacity allocation table is generated, for example, using a preliminary configuration file in JSON format, including drone identification, latitude and longitude path sequence, and charging plan.

[0173] The method provided in this application achieves flexible adaptation of heterogeneous drone swarms to different delivery tasks and improves resource utilization by obtaining an initial population representing the capacity scheduling scheme of heterogeneous drone swarms. By constructing an objective function and obtaining a multi-objective optimization model based on the objective function, multi-objective collaborative optimization is achieved, improving adaptability to diverse delivery needs and scenario coverage.

[0174] By determining the physical constraint model of heterogeneous drone swarms based on preset constraint rules and multi-source data, and combining the technical features of multi-objective optimization model and physical constraint model for iterative calculation, the capacity scheduling scheme can be adapted to dynamic factors such as package quantity, weather, and airspace restrictions, thereby improving the flexibility of capacity strategy adjustment.

[0175] It should be noted that, although the embodiments in this application are based on... Figure 1 Steps S101 to S105 will be described sequentially, but this does not mean that steps S101 to S105 must be performed in a strict order. The reason this embodiment follows this order is... Figure 1 The order in which steps S101 to S105 are described is provided to facilitate understanding of the technical solutions of the embodiments of this application by those skilled in the art. In other words, in the embodiments of this application, the order of steps S101 to S105 can be appropriately adjusted according to actual needs.

[0176] based on Figure 1 In addition to the method described herein, this specification also provides some specific implementation schemes and extension schemes of this method, which will be further explained below.

[0177] Furthermore, to address the problem of existing site selection data collection being one-sided, focusing only on a single demand or environmental factor, leading to a disconnect between site selection schemes and actual scenarios, and to adapt to the dynamic changes in demand in urban and rural remote areas, improve the spatial coverage adaptability of drone delivery networks, and enhance the efficiency of heterogeneous drone capacity scheduling, the methods also include:

[0178] In some embodiments of this application, basic demand data, spatiotemporal correlation data, auxiliary verification data, geographic environment data, airspace condition data, and status data of various UAV models are obtained.

[0179] Demand-based data refers to fundamental information that directly reflects the scale and main actors of rural logistics demand. This data may include, for example, data on the distribution of the regional resident population (e.g., regional density, location of settlements), agricultural product output data (e.g., category, output, harvest cycle), and industrial product consumption data (e.g., the frequency and scale of purchases of daily consumer goods and agricultural inputs).

[0180] Spatiotemporal data refers to auxiliary information that is related to the temporal patterns and spatial distribution of demand. Examples of spatiotemporal data include agricultural product sales cycle data, such as seasonal market availability and shelf life; industrial product procurement time data, such as differences between busy and slack farming seasons; and geographical accessibility data, such as village distribution and road connectivity.

[0181] Auxiliary verification data refers to data used to verify the authenticity of demand data and supplement background information on the demand. Auxiliary verification data may include, for example, historical logistics and delivery records, such as transportation distance, timeliness requirements, and cargo volume fluctuations; and regional economic development data, such as per capita income and the degree of agricultural industrialization.

[0182] Geographic environmental data refers to information describing the natural environment of rural areas, such as topography and landforms. Examples of geographic environmental data include digital elevation model (DEM) data, used to build the foundation for three-dimensional terrain analysis and adapting to complex terrains such as mountainous areas.

[0183] Airspace condition data refers to airspace management information that affects drone flights. This data may include, for example, airspace altitude layer division rules, no-fly zones, and time limits for speed restrictions, such as nighttime noise reduction speed limits in urban areas.

[0184] Feature analysis is performed on the basic demand data, spatiotemporal correlation data, and auxiliary verification data to extract demand scale features, temporal concentration features, spatial dispersion features, and time sensitivity features.

[0185] Among these, demand scale characteristics can quantify the total logistics demand in remote rural areas, reflecting the overall size of the demand. For example, demand scale characteristics can be directly quantified using data such as annual agricultural output and annual industrial product procurement volume, reflecting the total regional demand.

[0186] Temporal concentration characteristics can describe, for example, the degree to which demand is concentrated in different time periods, reflecting the temporal distribution pattern of demand. For instance, temporal concentration characteristics can be generated by analyzing the proportion of demand during key periods such as the agricultural harvest season and the pre-harvest season, reflecting the temporal concentration pattern of demand.

[0187] Spatial dispersion characteristics can describe, for example, the degree of dispersion of demand in geographic space, reflecting the spatial distribution pattern of demand. For instance, spatial dispersion characteristics can be calculated from data such as the number of demand points, the distance between villages, and the area of ​​a region, reflecting the spatial dispersion characteristics of industrial product demand.

[0188] Time sensitivity features can be quantified as the degree to which demand requires delivery speed, reflecting the time urgency of the demand. For example, time sensitivity features can be quantified based on data such as the shelf life of agricultural products and the urgency of agricultural input usage; the shorter the shelf life and the higher the urgency, the stronger the time sensitivity.

[0189] By using time series analysis, we can identify the seasonal fluctuation patterns of demand for agricultural products based on time concentration characteristics, and explore the periodic changes in demand by combining time sensitivity characteristics. Based on the seasonal fluctuation patterns and periodic changes, we can determine the dynamic requirements for delivery timeliness and transportation capacity.

[0190] Seasonal fluctuations can refer to the cyclical ups and downs in agricultural product demand as the seasons change, such as a surge in demand during the harvest season and a slowdown in demand outside the harvest season. Periodic variations can refer to the recurring patterns of agricultural product demand within fixed time intervals, such as the demand for agricultural products available in specific months each year. Dynamic requirements for delivery timeliness and capacity can refer to the dynamically determined delivery speed standards and capacity allocation scale based on the temporal patterns of agricultural product demand.

[0191] Based on the spatial dispersion characteristics, the distribution dispersion of industrial goods downstream demand in geographic space is quantified. Combined with the demand scale characteristics, the total demand in different regions is assessed. By integrating the distribution dispersion and the total demand, the spatial distribution dispersion characteristics and regional accessibility of industrial goods downstream demand are determined. Combined with the temporal concentration characteristics, the overall density of regional logistics demand and the degree of concentration in different time periods are determined.

[0192] Among these, the degree of dispersion of distribution can refer to, for example, the degree of dispersion of industrial product demand in rural geographical space, reflecting the concentration or dispersion of demand points. Total demand can refer to, for example, the total procurement scale of industrial products within a specific rural area, reflecting the overall size of regional demand. Spatial distribution dispersion characteristics can refer to, for example, the specific features of the spatially dispersed distribution of industrial product demand, such as the range of dispersion and locally dense areas.

[0193] Regional accessibility can refer to, for example, the ease with which drones can reach various industrial product demand points from delivery stations, which is related to the distance to the demand point and terrain conditions. For instance, regional accessibility analysis can combine Geographic Information System (GIS) data with village distribution information to identify remote and poorly accessible demand points as key coverage targets during site selection. The overall density of regional logistics demand can refer to the total logistics demand within a unit of rural area, reflecting the density of regional demand. The concentration at different times can refer to the degree of aggregation of industrial product demand at different times, reflecting the temporal distribution characteristics of demand.

[0194] Geographic environment data and airspace condition data are input into a preset three-dimensional spatial analysis model to determine the spatial constraints of site selection.

[0195] The pre-defined three-dimensional spatial analysis model can be constructed based on digital elevation model (DEM) data and geographic information system (GIS) data. Traditional site selection models lack clear spatial constraints, resulting in an excessively large solution range and low computational efficiency. By constructing three-dimensional spatial constraints adapted to the complex terrain and airspace requirements of rural areas, the engineering feasibility and flight safety of the site selection scheme are ensured.

[0196] Site selection constraints can include, for example, terrain constraints, airspace constraints, and coverage constraints. Terrain constraints can exclude areas with excessively steep slopes or high altitudes, ensuring that the delivery station site meets the terrain requirements. Airspace constraints can avoid no-fly zones and comply with airspace altitude classification rules to ensure drone flight safety. Coverage constraints can define the spatial boundaries of the delivery station's service area based on village distribution and geographical accessibility.

[0197] The dynamic requirements for delivery timeliness and capacity, the overall density and concentration of regional logistics demand at different times, and the status data of various drone models are input into a pre-set multi-objective integer programming location selection model. Using location space constraints as boundary constraints, a pre-set hybrid heuristic algorithm iteratively solves the multi-objective integer programming location selection model to obtain the optimal solution set for delivery station location. Based on the optimal solution set for delivery station location, real-time multi-source data is determined from the multi-source data.

[0198] The pre-defined multi-objective integer programming location model refers to a pre-constructed mathematical model centered on multi-objective optimization, used to solve for the optimal location of rural drone delivery stations under constraints. The hybrid heuristic algorithm refers to an optimization algorithm that integrates the advantages of multiple heuristic algorithms, used to efficiently solve complex multi-objective optimization problems. The optimal solution set for delivery station location refers to a Pareto optimal solution set containing multiple non-dominated solutions, each corresponding to a location scheme with different objective trade-offs. The multi-objective integer programming location model can maximize service coverage, minimize total system cost, and improve operational safety and energy efficiency.

[0199] In addition to site selection constraints, model constraints may also include, for example, drone performance constraints and facility capacity constraints. The drone performance constraints are adapted to the range, payload, and other performance parameters of various drone models. The facility capacity constraints match the maximum service capacity of the distribution station with the total regional logistics demand. This application's embodiments do not impose restrictions on the model constraints.

[0200] Hybrid heuristic algorithms, for example, can be optimization algorithms that combine adaptive large neighborhood search algorithms with multi-objective genetic algorithms, and can effectively handle the trade-offs between multiple objectives.

[0201] Furthermore, in order to ensure the engineering feasibility and practical implementation of the drone capacity scheduling method provided in this application, the embodiments of this application perform multi-dimensional verification of the method provided in this application.

[0202] For example, battery degradation can be verified using a formula. Calculate the current battery capacity, where, This refers to the cumulative number of charge-discharge cycles for the battery. If the energy consumption per cycle is greater than 0.85 × This will trigger a replanning process.

[0203] It can also perform overload inspection, using the formula. If the actual load exceeds the rated load by 10%, the priority of the task will be automatically reduced and the machine model will be reassigned.

[0204] Dynamic simulation evaluation can also be performed, and simulation verification can be performed based on historical data. For example, the Monte Carlo method can be used for testing. For instance, 1,000 sets of historical orders can be sampled to verify that the timeliness compliance rate is greater than or equal to 95%.

[0205] It can also perform energy consumption deviation analysis, calculating the deviation between the simulated measured energy consumption and the model's predicted value, requiring... To ensure the accuracy of energy consumption calculations.

[0206] It can also perform timeliness and cost balance verification. For example, scatter plots can be drawn to analyze the Pareto solution set, ensuring that 90% of the solutions satisfy the timeliness and cost balance. .in, This refers to the total delivery cost of a certain scheduling plan. This refers to a preset reference cost benchmark, such as the historical average of the best delivery costs or the theoretical minimum cost ceiling.

[0207] Based on the same idea, some embodiments of this application also provide devices and non-volatile computer storage media corresponding to the above methods.

[0208] Figure 2 A schematic diagram of a heterogeneous unmanned aerial vehicle (UAV) capacity scheduling device for urban and rural logistics, provided as an embodiment of this application, includes:

[0209] At least one processor; and,

[0210] A memory communicatively connected to the at least one processor; wherein,

[0211] The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform any of the above-described heterogeneous UAV capacity scheduling methods for urban and rural logistics.

[0212] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0213] The devices and methods provided in this application are one-to-one correspondences. Therefore, the devices also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices will not be repeated here.

[0214] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0215] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0216] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0217] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0218] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0219] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0220] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0221] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0222] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the technical principles of this application should fall within the protection scope of this application.

Claims

1. A heterogeneous UAV capacity scheduling method for urban and rural logistics, characterized in that, The method includes: Based on multi-source data, an initial population is determined, and each individual in the initial population represents a heterogeneous UAV swarm capacity scheduling scheme. Based on the preset constraint rules and the multi-source data, the physical constraint model of the heterogeneous UAV swarm is determined. Using the physical constraint model as the boundary constraint and combining the multi-source data, at least two types of objective functions for runtime scheduling are constructed. By integrating all objective functions, a multi-objective optimization model is obtained. Based on the multi-objective optimization model and the physical constraint model, a multi-objective optimization algorithm is used to iteratively calculate the individuals in the initial population to obtain the optimal solution set for UAV capacity scheduling. The multi-source data includes real-time multi-source data and historical multi-source data; The real-time multi-source data includes: mission requirement data, status data of various UAV models, environmental perception data, airspace control data, and electricity price data; The historical multi-source data includes: historical mission execution data, historical operation data of various UAV models, historical environmental impact data, and historical scheduling decision data; The determination of the physical constraint model for the heterogeneous UAV swarm based on preset constraint rules and multi-source data specifically includes: By using the status data of various UAV models from the multi-source data and combining the safety redundancy rules in the preset constraint rules, a nonlinear load constraint model is established. Based on the status data of various UAV models, environmental perception data, and historical environmental impact data from the multi-source data, and combined with the environmental impact correction rules in the preset constraint rules and the environmental compensation coefficients that match the environmental perception data, a range constraint model is constructed. Based on the status data of various UAV models and airspace control data from the multi-source data, and combined with the airspace layering rules in the preset constraint rules, a layered speed constraint model is established. The nonlinear load constraint model, the range constraint model, and the speed constraint model are integrated to obtain the physical constraint model. The objective function for runtime scheduling is constructed using the physical constraint model as boundary constraint and the multi-source data, specifically including: Using the status data, task requirement data, historical operation data, airspace control data and environmental perception data of various types of UAVs from multiple sources as input, the physical constraint model as boundary constraint, and the latest time for the heterogeneous UAV swarm to complete all delivery tasks as the optimization index, a delivery time objective function is constructed. Using the status data of various types of UAVs, mission requirement data, historical operation data of various types of UAVs, electricity price data, airspace control data and environmental perception data from multiple sources as inputs, the physical constraint model is used as the boundary constraint, and the minimum total delivery cost to complete the delivery task is used as the optimization index to construct the delivery cost objective function. Based on the preset coupling rules of load and energy consumption, the system takes the status data, task requirement data and historical operation data of each type of UAV from multi-source data as input, the physical constraint model as boundary constraint, and the minimum total energy consumption to complete the delivery task as the optimization index to construct an energy consumption objective function. The method further includes: Acquire basic demand data, spatiotemporal correlation data, auxiliary verification data, geographic environment data, airspace condition data, and status data of various UAV models; The aforementioned basic demand data, spatiotemporal correlation data, and auxiliary verification data are subjected to feature parsing processing to extract demand scale features, temporal concentration features, spatial dispersion features, and time sensitivity features. By using time series analysis, the seasonal fluctuation pattern of agricultural product demand is identified based on the time concentration characteristics. Combined with the time sensitivity characteristics, the periodic change characteristics of demand are mined. Based on the seasonal fluctuation pattern and the periodic change characteristics, the dynamic requirements for delivery timeliness and transportation capacity are determined. Based on the spatial dispersion characteristics, the distribution dispersion of industrial goods downstream demand in geographic space is quantified. Combined with the demand scale characteristics, the total demand in different regions is evaluated. By integrating the distribution dispersion and the total demand, the spatial distribution dispersion characteristics and regional accessibility of industrial goods downstream demand are determined. Combined with the temporal concentration characteristics, the overall density of regional logistics demand and the concentration of demand in different time periods are determined. The geographic environment data and the airspace condition data are input into a preset three-dimensional spatial analysis model to determine the site selection spatial constraints. The dynamic requirements for delivery timeliness and capacity, the overall density of regional logistics demand and the concentration of demand at different times, and the status data of each type of UAV are input into a preset multi-objective integer programming location model. The location space constraint is used as the boundary constraint. The preset hybrid heuristic algorithm is used to iteratively solve the multi-objective integer programming location model to obtain the optimal solution set for the delivery station location. Based on the optimal solution set for the distribution station location, real-time multi-source data is determined from the multi-source data.

2. The method according to claim 1, characterized in that, The process of integrating all objective functions to obtain a multi-objective optimization model specifically includes: Standardize all objective functions; The standardized objective function is multiplied by its corresponding weight coefficient to obtain the weighted term of each objective function; The multi-objective optimization model is obtained by summing all weighted terms.

3. The method according to claim 1, characterized in that, The step of using the multi-objective optimization model and the physical constraint model to iteratively calculate the individuals in the initial population using a multi-objective optimization algorithm to obtain the optimal solution set for UAV capacity scheduling specifically includes: Based on the parameter values ​​corresponding to each individual in the initial population and the multi-objective optimization model, the multi-objective optimization algorithm is used to calculate the multi-dimensional objective value of each individual in the initial population; The multidimensional objective values ​​of all individuals are hierarchically divided using a preset sorting method, and non-dominated solutions that satisfy the constraints of the physical constraint model are selected to form the latest solution set. Continue multi-objective optimization on the latest solution set until the maximum number of iterations is reached; The latest solution set is taken as the optimal solution set.

4. The method according to claim 3, characterized in that, The method further includes: The iteration process terminates when the change in the distribution entropy of the latest solution set is less than a preset threshold. The latest solution set is taken as the optimal solution set.

5. The method according to claim 1, characterized in that, Before determining the physical constraint model of the heterogeneous UAV swarm based on preset constraint rules and multi-source data, the method further includes: Outliers in the multi-source data are filtered using a statistical threshold method, and the filtered outliers are processed using interpolation or deletion methods. A unique identifier is generated for each data record in the multi-source data after outlier cleaning based on a hash algorithm. By comparing the unique identifier with the database primary key stored in a preset database, duplicate and conflicting data in the multi-source data are eliminated. Based on preset composite feature conditions, feature extraction processing is performed on the cleaned multi-source data to construct composite features, which are used to quantify task priorities. The latitude and longitude coordinates in the multi-source data are discretized into grid codes to enhance spatial correlation; The multi-source data, after feature extraction and encoding, is mapped to a unified numerical range using a normalization method.

6. A heterogeneous unmanned aerial vehicle (UAV) capacity scheduling device for urban and rural logistics, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the heterogeneous UAV capacity scheduling method for urban and rural logistics as described in any one of claims 1-5.

Images