Unmanned aerial vehicle scheduling method and device, computer equipment and storage medium

Abstract

The invention discloses an unmanned aerial vehicle scheduling method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining to-be-predicted data, and carrying out the feature extraction of the to-be-predicted data, so as to obtain to-be-predicted features; predicting user demand vectors of different demand positions based on the to-be-predicted features; based on the user demand vector, constructing a target function and a constraint condition of a preset equipment scheduling target; based on an objective function, calculating a first objective function value of each equipment site selection scheme meeting the constraint condition, so as to determine an optimal equipment site selection scheme based on the first objective function value; and calculating a second target function value of each path planning scheme meeting the constraint condition based on the target function and the optimal equipment site selection scheme, so as to determine an optimal path planning scheme based on the second target function value. Complex requirements in logistics distribution can be better met, and the resource utilization rate and the overall efficiency of the unmanned aerial vehicle are improved.

Description

Technical Field

[0001] The present application relates to the field of logistics technology, and in particular to a drone scheduling method, device, computer equipment and storage medium. Background Art

[0002] Drone logistics and delivery has developed rapidly in recent years, especially in areas with inconvenient transportation and emergency delivery scenarios, showing great potential. The efficiency and flexibility of drones have led to their increasing application in industries such as e-commerce, medical care, and emergency rescue. However, as demand grows, the site selection and path planning issues of drones in the delivery network have become increasingly complex, and how to reasonably allocate resources and optimize delivery routes has become a problem that needs to be solved urgently.

[0003] At present, traditional distribution planning methods rely more on static demand assumptions, fail to fully consider the dynamics and complexity of customer demand, and ignore the impact of market environment, seasonal factors and emergencies on demand. This static forecasting method cannot accurately respond to changes in demand, resulting in site selection and route planning being difficult to adapt to actual demand, resulting in waste or delays in distribution resources. This makes it difficult for the dispatching system to cope with high-frequency demand fluctuations and unable to fully tap the delivery potential of drones. Summary of the invention

[0004] Based on this, it is necessary to provide a drone scheduling method, device, computer equipment and storage medium to address the above technical problems, so as to solve at least one problem existing in the above-mentioned prior art.

[0005] In a first aspect, the embodiment of the present application is implemented as follows: a method for dispatching a drone is provided, comprising the following steps: Acquire data to be predicted, and perform feature extraction on the data to be predicted to obtain features to be predicted; Based on the features to be predicted, predict user demand vectors at different demand locations; Based on the user demand vector, construct an objective function and constraint conditions for a preset equipment scheduling target; Based on the objective function, calculating a first objective function value of each equipment site selection scheme that satisfies the constraint condition, so as to determine an optimal equipment site selection scheme based on the first objective function value; Based on the objective function and the optimal equipment site selection plan, a second objective function value of each path planning plan that meets the constraint condition is calculated to determine the optimal path planning plan based on the second objective function value.

[0006] In one embodiment, the step of calculating, based on the objective function, a first objective function value of each equipment site selection scheme satisfying the constraint condition, so as to determine an optimal equipment site selection scheme based on the first objective function value, includes: Constructing a set of equipment site selection solutions, wherein the set of equipment site selection solutions includes a plurality of initial equipment site selection solutions; Iteratively search each initial equipment site selection plan, and obtain a local optimal equipment site selection plan for this iteration process based on the first objective function value; The best solution is selected from all the local optimal equipment location solutions as the global optimal equipment location solution.

[0007] In one embodiment, the iterative search for each initial equipment site selection scheme and obtaining a local optimal equipment site selection scheme of this iteration process based on the first objective function value includes: Performing a neighborhood search on the initial equipment location scheme to obtain a neighborhood solution; Based on the objective function, calculating the objective function value of the neighborhood solution; The objective function value of the neighborhood solution is compared with the objective function value of the initial equipment site selection plan. When the neighborhood solution is better, the neighborhood solution is used as the local optimal equipment site selection plan for this iteration process. Otherwise, a simulated annealing strategy is used to determine whether to use the neighborhood solution as the local optimal equipment site selection plan for this iteration process.

[0008] In one embodiment, the selecting the best solution from all the local optimal equipment site selection solutions as the global optimal equipment site selection solution includes: Compare the local equipment location scheme obtained in each iteration process with the local optimal equipment location scheme obtained in the previous iteration process to obtain the current local optimal equipment location scheme, and save it to the optimal solution set, while deleting the local optimal equipment location scheme generated in the previous iteration process saved in the optimal solution set; When the iteration is completed, the local optimal equipment location solution retained in the optimal solution set is used as the global optimal equipment location solution.

[0009] In one embodiment, the calculating, based on the objective function and the optimal equipment site selection scheme, a second objective function value of each path planning scheme that satisfies the constraint condition, so as to determine the optimal path planning scheme based on the second objective function value, includes: Based on the optimal equipment site selection plan, determine an initial path planning plan; Iteratively search the initial path planning solution, and obtain a local optimal path planning solution of this iterative process based on the second objective function value; The best solution is selected from all the local optimal path planning solutions as the global optimal path planning solution.

[0010] In one embodiment, the predicting of user demand vectors at different demand locations based on the features to be predicted includes: Inputting the feature to be predicted into a pre-trained prediction model, wherein the prediction model includes a plurality of decision trees; Each decision tree predicts the feature to be predicted respectively to obtain multiple prediction results; All the prediction results are aggregated to obtain user demand vectors of the different demand locations.

[0011] In one embodiment, the trained prediction model is calculated in the following manner: Acquire historical order data, perform feature extraction on the historical order data, and establish a training feature set based on the extracted features, wherein the training feature set includes multiple training feature subsets; Recursively constructing a prediction model using Gini impurity or information gain as a splitting criterion, the prediction model comprising a plurality of decision trees; Each decision tree is independently iteratively trained on the training feature subset until a preset convergence condition is met, thereby obtaining the trained prediction model.

[0012] In a second aspect, a drone dispatching device is provided, comprising: A feature acquisition unit to be predicted, used to acquire data to be predicted, and perform feature extraction on the data to be predicted to obtain features to be predicted; A user demand vector prediction unit, configured to predict user demand vectors at different demand locations based on the features to be predicted; An objective function construction unit, used to construct an objective function and constraint conditions of a preset equipment scheduling objective based on the user demand vector; An optimal equipment site selection scheme determining unit, configured to calculate a first objective function value of each equipment site selection scheme satisfying the constraint condition based on the objective function, so as to determine an optimal equipment site selection scheme based on the first objective function value; The optimal path optimization scheme determination unit is used to calculate the second objective function value of each path planning scheme that meets the constraint conditions based on the objective function and the optimal equipment site selection scheme, so as to determine the optimal path planning scheme based on the second objective function value.

[0013] In a third aspect, a computer device is provided, comprising a memory, a processor, and computer-readable instructions stored in the memory and running on the processor, wherein the processor implements the drone scheduling method as described above when executing the computer-readable instructions.

[0014] In a fourth aspect, a readable storage medium is provided, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor, the drone scheduling method as described above is implemented.

[0015] The above-mentioned drone scheduling method, device, computer equipment and storage medium, the method implementation includes: obtaining the data to be predicted, extracting features of the data to be predicted to obtain the features to be predicted; based on the features to be predicted, predicting the user demand vectors of different demand locations; based on the user demand vectors, constructing the objective function and constraint conditions of the preset equipment scheduling target; based on the objective function, calculating the first objective function value of each equipment site selection scheme that meets the constraint conditions, so as to determine the optimal equipment site selection scheme based on the first objective function value; based on the objective function and the optimal equipment site selection scheme, calculating the second objective function value of each path planning scheme that meets the constraint conditions, so as to determine the optimal path planning scheme based on the second objective function value. In the embodiment of the present application, through accurate user demand prediction, distribution hotspots can be identified in advance, and drones and facility locations can be reasonably allocated, thereby improving the efficiency and service quality of logistics distribution. The prediction model can not only help cope with changes in demand, but also provide real-time updated data support for the scheduling system, thereby optimizing the distribution path, reducing empty flights, and improving the resource utilization of drones. Thereby, it can better cope with the complex needs in logistics distribution and improve the overall efficiency of drone logistics. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0017] Figure 1 This is a flow chart of a method for dispatching a drone in one embodiment of the present application; Figure 2 This is a structural schematic diagram of a drone dispatching device in one embodiment of the present application; Figure 3 It is a schematic diagram of a computer device in one embodiment of the present application. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] In one embodiment, if Figure 1 As shown, a drone scheduling method is provided, comprising the following steps: In step S110, the data to be predicted is obtained, and features are extracted from the data to be predicted to obtain features to be predicted; In the embodiment of the present application, the data to be predicted may include order data, which may include information such as demand location, order time, order quantity, etc. Through feature engineering, the data to be predicted may be subjected to feature extraction, such as time features, such as hours, days of the week, etc., and location features, such as longitude and latitude, area type, area name, etc. It should be noted that the acquired data to be predicted may be first subjected to data cleaning processing, such as removing null values, outliers, etc.

[0020] In step S120, based on the features to be predicted, user demand vectors of different demand locations are predicted; In an embodiment of the present application, a user demand prediction model can be constructed and trained using a training data set to obtain a trained user demand prediction model. The features to be predicted are input into the user demand prediction model for prediction processing to obtain user demand vectors at different demand locations. It should be noted that different demand locations refer to different business districts, regions, locations, etc. with service demands, such as demand locations for drone delivery demands. The user demand vector refers to all users with service demands within a certain range of different demand locations, such as drone delivery service demands.

[0021] Optionally, the user demand prediction model may be a random forest model, a neural network model, etc. Taking the random forest model as an example, it may include multiple decision trees, each of which may predict a result, so the user demand vector output by the prediction is It can be expressed as the aggregation of all decision tree prediction results. The user demand prediction model can accurately predict user demand at different demand locations, providing key input for equipment site selection and route planning, thereby optimizing the efficiency and response speed of distribution services.

[0022] In step S130, based on the user demand vector, an objective function and constraint conditions of a preset equipment scheduling target are constructed; For example, if there are 100 users in business district A, it is predicted that 70 of the 100 users have delivery needs, and the remaining 30 users have no delivery needs. By establishing a user demand prediction model to predict user demand, interference items can be removed, and drone site selection and path planning can be prioritized for locations with a large number of demand users and dense demand users, so as to improve service coverage and maximize revenue. Among them, the preset equipment scheduling goal can be to maximize total revenue or maximize service coverage.

[0023] The decision variables may be obtained first, the value range of each element in the decision variables may be determined, and the objective function and constraint conditions of the preset equipment scheduling target may be constructed based on the decision variables. For example, the decision variables may include , indicating that if the drone In the itinerary Zhongcong Facilities Service Request , if there is a service demand, it is 1, otherwise it is 0. , indicating that if the facility is established, 1; otherwise, 0. It means that if the drone Assigned to facility , then it is 1, otherwise it is 0.

[0024] The objective function can be specifically shown in the following formula (1), and the constraint condition can be shown in the following formulas (2)-(14):

[0025] Wherein, formula (1) is the objective function that maximizes total revenue. Constraint (2) ensures that each demand is served at most once. Constraint (3) ensures that each drone trip is served only once. The number of facilities is limited by constraint (4). Constraint (5) enforces the drone capacity constraint. The capacity of each facility and the total number of assigned drones are subject to constraints (6) and (7). Constraints (8) and (9) ensure that each drone is assigned to a facility at most. In constraint (10), a drone can only serve a demand in a trip assigned to a facility if the drone is assigned to the facility. Constraint (11) ensures that the demand must be satisfied before the latest delivery time. In constraints (12) and (13), consecutive deliveries need to start later and cannot overlap. Constraint (14) defines the variables.

[0026] in, : The set of all required locations. : The set of all potential facility locations. : The collection of all drones. : The set of itineraries for each drone. :Service demand location Revenue from customers. :From the required position To potential facilities travel distance. :From the required position To potential facilities travel time. : Maximum number of facilities. :need The latest delivery time. :facility The capacity (maximum number of drones that can be assigned). : Drone exist As required during the trip Time of service. : The travel distance limit of the drone.

[0027] Based on the above objective function, the maximum number of flights for each drone can be calculated as follows: ; in, The calculation formula is as follows: For each potential facility, sort it from largest to smallest according to the distance to the facility, and then accumulate the previous distances of customers until the sum of their distances is greater than ,therefore, .

[0028] By obtaining the maximum number of flights, the number of flights of the UAV can be limited when solving the objective function to reduce the solution space and improve the search efficiency and algorithm efficiency.

[0029] It should be noted that the above objective function can be solved by an adaptive large neighborhood search algorithm (ALNS), or by other heuristic or meta-heuristic algorithms to obtain an approximate optimal solution.

[0030] In step S140, based on the objective function, a first objective function value of each equipment site selection scheme that meets the constraint condition is calculated to determine an optimal equipment site selection scheme based on the first objective function value; In an embodiment of the present application, the objective function can be solved by an adaptive large neighborhood search algorithm (ALNS), or other heuristic or meta-heuristic algorithms can be used to solve the above objective function to obtain an approximate optimal solution. Taking the adaptive large neighborhood search algorithm (ALNS) as an example, the solution process is described. Exemplarily, a greedy algorithm can be used to construct an initial device location solution, or an initial device location solution can be constructed based on a heuristic method, and a neighborhood search is performed to obtain a neighborhood solution, and the objective function value of the neighborhood solution is calculated, and compared with the objective function value of the initial device location solution. If the neighborhood solution is better, it is used as the latest current solution. Otherwise, a simulated annealing method can be used to determine whether to accept it as the latest current solution, and the latest current solution obtained is used as the optimal solution for this iteration. Repeat the above process until the optimal device location solution is obtained.

[0031] In step S150, based on the objective function and the optimal equipment site selection plan, the second objective function value of each path planning plan that meets the constraint condition is calculated to determine the optimal path planning plan based on the second objective function value.

[0032] In an embodiment of the present application, according to the determined optimal equipment site selection scheme, all flight points of the drone can be determined, and then based on the flight points, path planning can be further performed. Specifically, it can be solved by an adaptive large neighborhood search algorithm (ALNS), or the above objective function can be solved by other heuristic or meta-heuristic algorithms to obtain an approximate optimal solution as a path planning solution. Taking the adaptive large neighborhood search algorithm (ALNS) as an example, the solution process is described. Exemplarily, a greedy algorithm can be used to construct an initial path planning solution, or all initial path planning solutions can be constructed based on a heuristic method, and a neighborhood search is performed to obtain a neighborhood solution, and the objective function value of the neighborhood solution is calculated, and compared with the objective function value of the initial path planning solution. If the neighborhood solution is better, it is used as the latest current solution. Otherwise, a simulated annealing method can be used to determine whether to accept it as the latest current solution, and the latest current solution obtained is used as the optimal solution for this iteration. Repeat the above process until the optimal path planning solution is obtained.

[0033] In an embodiment of the present application, the optimal equipment site selection scheme and the optimal path planning scheme can be displayed through the user interface to present to the user, realize user interaction, and improve user participation and experience. Among them, the user interface may specifically include a data view, which can be used to display historical order data and extracted feature sets. The prediction result view can display the user demand predicted for each demand location in the form of a chart. The real-time site selection view is used to display the optimized facility location and drone coverage on the map. The path planning view is used to display the optimal path and estimated delivery time of the drone. And through the control panel of the control center, the user can start the prediction, optimization and path planning process, and can monitor the system status in real time.

[0034] The above-mentioned drone scheduling method includes: obtaining data to be predicted, extracting features from the data to be predicted to obtain features to be predicted; based on the features to be predicted, predicting user demand vectors of different demand locations; based on the user demand vector, constructing the objective function and constraint conditions of the preset equipment scheduling target; based on the objective function, calculating the first objective function value of each equipment site selection scheme that meets the constraint conditions, so as to determine the optimal equipment site selection scheme based on the first objective function value; based on the objective function and the optimal equipment site selection scheme, calculating the second objective function value of each path planning scheme that meets the constraint conditions, so as to determine the optimal path planning scheme based on the second objective function value. In the embodiment of the present application, through accurate user demand prediction, distribution hotspots can be identified in advance, and drones and facility locations can be reasonably allocated, thereby improving the efficiency and service quality of logistics distribution. The prediction model can not only help cope with changes in demand, but also provide real-time updated data support for the scheduling system, thereby optimizing the distribution path, reducing empty flights, and improving the resource utilization of drones. Thereby, it can better cope with the complex needs in logistics distribution and improve the overall efficiency of drone logistics.

[0035] In an embodiment of the present application, the step of calculating, based on the objective function, a first objective function value of each device site selection scheme that satisfies the constraint condition, so as to determine an optimal device site selection scheme based on the first objective function value, includes: Construct initial equipment siting plan; Iteratively search the initial equipment site selection plan, and obtain a local optimal equipment site selection plan for this iteration process based on the first objective function value; The best solution is selected from all the local optimal equipment location solutions as the global optimal equipment location solution.

[0036] Optionally, a greedy algorithm can be used to construct an initial path planning solution, or all initial equipment location solutions can be constructed based on a heuristic approach. Define a neighborhood structure, for example, select a location and change its establishment state from 0 to 1, or from 1 to 0. The objective function can be solved by an adaptive large neighborhood search algorithm (ALNS), or by other heuristic or meta-heuristic algorithms. A local optimal solution can be obtained in each iteration process. Based on the objective function value of the local optimal solution, a global optimal solution can be selected. For example, the larger the objective function value, the higher the priority as the global optimal solution. The global optimal solution can be used as the global optimal equipment location solution.

[0037] It should be noted that for the global optimal equipment site selection plan, heuristic rules can be applied, such as giving priority to facility locations with high demand or good geographical locations, such as business districts, residential areas, and other locations where users are concentrated and there is more demand.

[0038] In an embodiment of the present application, the iterative search for each initial device site selection scheme and obtaining a local optimal device site selection scheme of this iteration process based on the first objective function value includes: Performing a neighborhood search on the initial equipment location scheme to obtain a neighborhood solution; Based on the objective function, calculating the objective function value of the neighborhood solution; The objective function value of the neighborhood solution is compared with the objective function value of the initial equipment site selection plan. When the neighborhood solution is better, the neighborhood solution is used as the local optimal equipment site selection plan for this iteration process. Otherwise, a simulated annealing strategy is used to determine whether to use the neighborhood solution as the local optimal equipment site selection plan for this iteration process.

[0039] Optionally, a neighborhood search can be performed on the initial device location solution to obtain a neighborhood solution. The objective function value of the neighborhood solution is calculated and compared with the objective function value of the initial device location solution. If the neighborhood solution is better, it is used as the latest current solution. Otherwise, a simulated annealing method can be used to determine whether to accept it as the latest current solution, and the latest current solution obtained is used as the optimal solution for this iteration. Then repeat the above process for the device location solution of the next iteration until the iteration is completed and the optimal device location solution can be obtained.

[0040] Among them, the specific approach of neighborhood search is: execute user greedy strategy search, that is, replace the user with the lowest benefit in the solution with the user with the highest benefit among the unserved users. Execute facility greedy strategy, that is, replace the facility with the lowest customer coverage income in the solution with the remaining unselected facilities that cover the most customers or have the highest income, and then reallocate customers. Adopt greedy-random strategy: sort users or facilities from low to high according to income, randomly select one from the first K (K=3, 4, 5, etc.), and then execute user greedy strategy and facility greedy strategy.

[0041] In one embodiment of the present application, the selecting the best solution from all the local optimal device site selection solutions as the global optimal device site selection solution includes: Compare the local equipment location scheme obtained in each iteration process with the local optimal equipment location scheme obtained in the previous iteration process to obtain the current local optimal equipment location scheme, and save it to the optimal solution set, while deleting the local optimal equipment location scheme generated in the previous iteration process saved in the optimal solution set; When the iteration is completed, the local optimal equipment location solution retained in the optimal solution set is used as the global optimal equipment location solution.

[0042] Optionally, the local device site selection plan of the local iteration obtained in each iteration process is compared with the local optimal device site selection plan obtained in the previous iteration process to obtain the current local optimal equipment site selection plan, and saved in the optimal solution set. At the same time, the local optimal equipment site selection plan generated in the previous iteration process saved in the optimal solution set is deleted, so that the optimal solution set only retains one current local optimal equipment site selection plan in each iteration process. When the iteration is completed, the local optimal equipment site selection plan retained in the optimal solution set is used as the global optimal equipment site selection plan.

[0043] In an embodiment of the present application, the calculating, based on the objective function and the optimal equipment site selection scheme, a second objective function value of each path planning scheme that satisfies the constraint condition, so as to determine the optimal path planning scheme based on the second objective function value, includes: Based on the optimal equipment site selection plan, determine an initial path planning plan; Iteratively search the initial path planning solution, and obtain a local optimal path planning solution of this iterative process based on the second objective function value; The best solution is selected from all the local optimal path planning solutions as the global optimal path planning solution.

[0044] Optionally, according to the determined optimal equipment site selection plan, all flight points of the UAV can be determined, and then based on the flight points, path planning can be further performed. Exemplarily, a greedy algorithm can be used to construct an initial path planning solution, or all initial path planning solutions can be constructed based on a heuristic approach, and a neighborhood search is performed to obtain a neighborhood solution, and the objective function value of the neighborhood solution is calculated and compared with the objective function value of the initial path planning solution. If the neighborhood solution is better, it is used as the latest current solution. Otherwise, a simulated annealing method can be used to determine whether to accept it as the latest current solution, and the latest current solution obtained is used as the optimal solution for this iteration. Repeat the above process until the optimal path planning solution is obtained.

[0045] In an embodiment of the present application, predicting user demand vectors of different demand locations based on the features to be predicted includes: Inputting the feature to be predicted into a pre-trained prediction model, wherein the prediction model includes a plurality of decision trees; Each decision tree predicts the feature to be predicted respectively to obtain multiple prediction results; All the prediction results are aggregated to obtain user demand vectors of the different demand locations.

[0046] Optionally, historical order data can be collected to establish a historical order data set, and feature extraction can be performed on the historical order data in the historical order data set, such as time features, regional features, etc., to establish feature set X, and use the prediction model Each tree in the required position The feature set Make predictions, calculate the aggregate value of the prediction results of each decision tree, and obtain the predicted user needs , aggregate the prediction results of all trees to obtain the final predicted customer demand vector .

[0047] In the embodiment of the present application, the trained prediction model is calculated in the following way: Acquire historical order data, perform feature extraction on the historical order data, and establish a training feature set based on the extracted features, wherein the training feature set includes multiple training feature subsets; Recursively constructing a prediction model using Gini impurity or information gain as a splitting criterion, the prediction model comprising a plurality of decision trees; Each decision tree is independently iteratively trained on the training feature subset until a preset convergence condition is met, thereby obtaining the trained prediction model.

[0048] Optionally, historical order data can be collected as training sample data, which may specifically include demand location, order time, order quantity, etc. Features, such as time features, location features, etc., are extracted from the historical order data through feature engineering to construct a training feature set for iterative training of the prediction model until the preset convergence conditions are met, such as the loss value is less than a preset threshold, or the number of iterations reaches a preset number, and a trained prediction model can be obtained.

[0049] It should be noted that the prediction model can be composed of T decision trees, and the Gini impurity or information gain can be used as the splitting criterion to recursively construct the decision tree. Then the training feature set can be divided into multiple subsets, and then each decision tree t can be independently trained on a randomly selected subset, and the loss function (such as mean square error MSE, determination coefficient R², etc.) can be used to calculate the loss value of the model's prediction result. If the loss value is greater than or equal to the preset threshold, iterative training is performed again until the loss value is less than the preset threshold, and a trained prediction model is obtained.

[0050] It can be understood that the decision tree may include multiple nodes, and the best splitting feature is found at each node. and split point , to minimize the impurity of the node. The best split feature and split point The search can be done as follows. For continuous features, all possible split points can be traversed and the point with the largest Gini impurity or information gain can be used as the split point. For discrete features, each feature value is considered as a potential split point, and the best split feature and split point are selected. The above process is recursively applied to the left and right child nodes formed after the split until the stopping condition is met, such as the number of samples in the node is less than the preset threshold or the maximum tree depth is reached. Then a decision tree can be obtained, and the decision tree is added to the random forest model Make predictions in .

[0051] For each demand location ∈ , using the model Make forecasts and get predicted customer demand , The prediction results It can be expressed as the aggregation of all decision tree predictions: ; in, It is Trees for required locations The predicted value of Represents the required location set The predicted customer demand vector on . :Model The number of decision trees in . Indicates the required location Forecast customer demand. Indicates Trees for required locations The predicted value of .

[0052] In the embodiment of the present application, through accurate user demand prediction, distribution hotspots can be identified in advance, and drone and facility locations can be reasonably allocated, thereby improving the efficiency and service quality of logistics distribution. The prediction model can not only help cope with changes in demand, but also provide real-time updated data support for the scheduling system, thereby optimizing the distribution path, reducing empty flights, and improving the resource utilization of drones. This can better cope with the complex needs in logistics distribution, and can find the optimal path for each drone, ensuring efficient distribution. Through this method, the facility layout and path planning of drone delivery services have been significantly optimized, improving service efficiency and customer satisfaction, and improving the overall effectiveness of drone logistics.

[0053] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0054] In one embodiment, a drone dispatching device is provided, which corresponds one-to-one to the drone dispatching method in the above embodiment. Figure 2 As shown, the UAV scheduling device includes a feature acquisition unit 10 to be predicted, a user demand vector prediction unit 20, an objective function construction unit 30, an optimal equipment location scheme determination unit 40 and an optimal path optimization scheme determination unit 50. Each functional module is described in detail as follows: The feature acquisition unit 10 is used to acquire the data to be predicted and perform feature extraction on the data to be predicted to obtain the feature to be predicted; A user demand vector prediction unit 20, configured to predict user demand vectors at different demand locations based on the features to be predicted; An objective function construction unit 30 is used to construct an objective function and constraint conditions of a preset equipment scheduling objective based on the user demand vector; The optimal equipment site selection scheme determining unit 40 is used to calculate the first objective function value of each equipment site selection scheme that meets the constraint condition based on the objective function, so as to determine the optimal equipment site selection scheme based on the first objective function value; The optimal path optimization solution determination unit 50 is used to calculate the second objective function value of each path planning solution that meets the constraint conditions based on the objective function and the optimal equipment site selection solution, so as to determine the optimal path planning solution based on the second objective function value.

[0055] In one embodiment of the present application, the optimal equipment site selection scheme determining unit 40 is further configured to: Construct initial equipment siting plan; Iteratively search the initial equipment site selection plan, and obtain a local optimal equipment site selection plan for this iteration process based on the first objective function value; The best solution is selected from all the local optimal equipment location solutions as the global optimal equipment location solution.

[0056] In one embodiment of the present application, the optimal equipment site selection scheme determining unit 40 is further configured to: Performing a neighborhood search on the initial equipment location scheme to obtain a neighborhood solution; Based on the objective function, calculating the objective function value of the neighborhood solution; The objective function value of the neighborhood solution is compared with the objective function value of the initial equipment site selection plan. When the neighborhood solution is better, the neighborhood solution is used as the local optimal equipment site selection plan for this iteration process. Otherwise, a simulated annealing strategy is used to determine whether to use the neighborhood solution as the local optimal equipment site selection plan for this iteration process.

[0057] In one embodiment of the present application, the optimal equipment site selection scheme determining unit 40 is further configured to: Compare the local equipment location scheme obtained in each iteration process with the local optimal equipment location scheme obtained in the previous iteration process to obtain the current local optimal equipment location scheme, and save it to the optimal solution set, while deleting the local optimal equipment location scheme generated in the previous iteration process saved in the optimal solution set; When the iteration is completed, the local optimal equipment location solution retained in the optimal solution set is used as the global optimal equipment location solution.

[0058] In one embodiment of the present application, the optimal path planning solution determining unit 50 is further configured to: Based on the optimal equipment site selection plan, determine an initial path planning plan; Iteratively search the initial path planning solution, and obtain a local optimal path planning solution of this iterative process based on the second objective function value; The best solution is selected from all the local optimal path planning solutions as the global optimal path planning solution.

[0059] In one embodiment of the present application, the user demand vector prediction unit 20 is further configured to: Inputting the feature to be predicted into a pre-trained prediction model, wherein the prediction model includes a plurality of decision trees; Each decision tree predicts the feature to be predicted respectively to obtain multiple prediction results; All the prediction results are aggregated to obtain user demand vectors of the different demand locations.

[0060] In one embodiment of the present application, the optimal path planning solution determining unit 50 is further configured to: Acquire historical order data, perform feature extraction on the historical order data, and establish a training feature set based on the extracted features, wherein the training feature set includes multiple training feature subsets; Recursively constructing a prediction model using Gini impurity or information gain as a splitting criterion, the prediction model comprising a plurality of decision trees; Each decision tree is independently iteratively trained on the training feature subset until a preset convergence condition is met, thereby obtaining the trained prediction model.

[0061] In the embodiment of the present application, through accurate user demand prediction, distribution hotspots can be identified in advance, and drone and facility locations can be reasonably allocated, thereby improving the efficiency and service quality of logistics distribution. The prediction model can not only help cope with changes in demand, but also provide real-time updated data support for the scheduling system, thereby optimizing the distribution path, reducing empty flights, and improving the resource utilization of drones. This can better cope with the complex needs in logistics distribution, and can find the optimal path for each drone, ensuring efficient distribution. Through this method, the facility layout and path planning of drone delivery services have been significantly optimized, improving service efficiency and customer satisfaction, and improving the overall effectiveness of drone logistics.

[0062] For the specific definition of the drone dispatching device, please refer to the definition of the drone dispatching method above, which will not be repeated here. Each module in the above-mentioned drone dispatching device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0063] In one embodiment, a computer device is provided. The computer device may be a terminal device, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a readable storage medium. The readable storage medium stores computer-readable instructions. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer-readable instructions are executed by the processor, a method for scheduling a drone is implemented. The readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0064] In an embodiment of the present application, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, the steps of the drone scheduling method described above are implemented.

[0065] In an embodiment of the application, a readable storage medium is provided, wherein the readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the drone scheduling method as described above are implemented.

[0066] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through computer-readable instructions, and the computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When the computer-readable instructions are executed, they may include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0067] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0068] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for dispatching a drone, characterized in that: The method comprises: Acquire data to be predicted, and perform feature extraction on the data to be predicted to obtain features to be predicted; Based on the features to be predicted, predict user demand vectors at different demand locations; Based on the user demand vector, construct an objective function and constraint conditions for a preset equipment scheduling target; Based on the objective function, calculating a first objective function value of each equipment site selection scheme that satisfies the constraint condition, so as to determine an optimal equipment site selection scheme based on the first objective function value; Based on the objective function and the optimal equipment site selection plan, a second objective function value of each path planning plan that meets the constraint condition is calculated to determine the optimal path planning plan based on the second objective function value.

2. The method for dispatching a drone according to claim 1, characterized in that: The step of calculating the first objective function value of each equipment site selection scheme satisfying the constraint condition based on the objective function, so as to determine the optimal equipment site selection scheme based on the first objective function value, comprises: Construct initial equipment siting plan; Iteratively search the initial equipment site selection plan, and obtain a local optimal equipment site selection plan for this iteration process based on the first objective function value; The best solution is selected from all the local optimal equipment location solutions as the global optimal equipment location solution.

3. The method for dispatching a drone according to claim 2, characterized in that: The iterative search of the initial equipment site selection scheme and obtaining a local optimal equipment site selection scheme of this iteration process based on the first objective function value includes: Performing a neighborhood search on the initial equipment location scheme to obtain a neighborhood solution; Based on the objective function, calculating the objective function value of the neighborhood solution; The objective function value of the neighborhood solution is compared with the objective function value of the initial equipment site selection plan. When the neighborhood solution is better, the neighborhood solution is used as the local optimal equipment site selection plan for this iteration process. Otherwise, a simulated annealing strategy is used to determine whether to use the neighborhood solution as the local optimal equipment site selection plan for this iteration process.

4. The method for dispatching a drone according to claim 2, wherein: The selecting the best solution from all the local optimal equipment location solutions as the global optimal equipment location solution includes: Compare the local equipment location scheme obtained in each iteration process with the local optimal equipment location scheme obtained in the previous iteration process to obtain the current local optimal equipment location scheme, and save it to the optimal solution set, while deleting the local optimal equipment location scheme generated in the previous iteration process saved in the optimal solution set; When the iteration is completed, the local optimal equipment location solution retained in the optimal solution set is used as the global optimal equipment location solution.

5. The method for dispatching a drone according to claim 1, wherein: The step of calculating the second objective function value of each path planning solution that satisfies the constraint condition based on the objective function and the optimal equipment site selection solution, so as to determine the optimal path planning solution based on the second objective function value, includes: Based on the optimal equipment site selection plan, determine an initial path planning plan; Iteratively search the initial path planning solution, and obtain a local optimal path planning solution of this iterative process based on the second objective function value; The best solution is selected from all the local optimal path planning solutions as the global optimal path planning solution.

6. The method for dispatching a drone according to claim 1, wherein: The step of predicting user demand vectors at different demand locations based on the features to be predicted includes: Inputting the feature to be predicted into a pre-trained prediction model, wherein the prediction model includes a plurality of decision trees; Each decision tree predicts the feature to be predicted respectively to obtain multiple prediction results; All the prediction results are aggregated to obtain user demand vectors of the different demand locations.

7. The method for dispatching a UAV according to claim 6, characterized in that: The pre-trained prediction model is calculated as follows: Acquire historical order data, perform feature extraction on the historical order data, and establish a training feature set based on the extracted features, wherein the training feature set includes multiple training feature subsets; Recursively constructing a prediction model using Gini impurity or information gain as a splitting criterion, the prediction model comprising a plurality of decision trees; Each decision tree is independently iteratively trained on the training feature subset until a preset convergence condition is met, thereby obtaining the trained prediction model.

8. A drone dispatching device, characterized in that: The device comprises: A feature acquisition unit to be predicted, used to acquire data to be predicted, and perform feature extraction on the data to be predicted to obtain features to be predicted; A user demand vector prediction unit, configured to predict user demand vectors at different demand locations based on the features to be predicted; An objective function construction unit, used to construct an objective function and constraint conditions of a preset equipment scheduling objective based on the user demand vector; An optimal equipment site selection scheme determining unit, configured to calculate a first objective function value of each equipment site selection scheme satisfying the constraint condition based on the objective function, so as to determine an optimal equipment site selection scheme based on the first objective function value; The optimal path optimization scheme determination unit is used to calculate the second objective function value of each path planning scheme that meets the constraint conditions based on the objective function and the optimal equipment site selection scheme, so as to determine the optimal path planning scheme based on the second objective function value.

9. A computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executed on the processor, characterized in that: When the processor executes the computer-readable instructions, the drone scheduling method according to any one of claims 1 to 7 is implemented.

10. A readable storage medium having computer readable instructions stored thereon, characterized in that: When the computer-readable instructions are executed by a processor, the drone scheduling method according to any one of claims 1 to 7 is implemented.

Images