Collaborative delivery path planning method and related device based on trucks and drones

Through the collaborative distribution path planning method based on trucks and drones, the problem of untimely logistics distribution in the existing technology is solved. By recursively determining the target distribution time and minimizing delay risks, the collaborative distribution path is optimized, and the logistics distribution is improved Timeliness and efficiency.

CN118586572BActive Publication Date: 2024-11-05SHENZHEN RES INST OF BIG DATA
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411061549.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-05
Estimated Expiration
2044-08-05

AI Technical Summary

Technical Problem

While improving distribution efficiency and reducing costs, the existing collaborative distribution system is difficult to ensure the timeliness of logistics distribution, especially affected by weather conditions and uncertainty in delivery time, resulting in service delays and reduced delivery levels.

Method used

A collaborative delivery path planning method based on trucks and drones is proposed. By obtaining the location information and time windows of multiple delivery nodes, a candidate path is generated, and the target delivery time is determined through recursive means, the service delay risk index is calculated, and the service delay risk index is minimized. Delivery delay risk to optimize delivery paths.

Benefits of technology

The timeliness of logistics distribution and user satisfaction are improved, the delay risk of candidate paths is measured by summarizing the service delay risk index, and the coordinated distribution paths are optimized, which enhances the stability and efficiency of the distribution paths.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118586572B_ABST
    Figure CN118586572B_ABST
Patent Text Reader

Abstract

The embodiment of the present application provides a collaborative delivery path planning method based on trucks and drones and related devices. The method includes: obtaining multiple node location information corresponding to multiple delivery nodes that need to be collaboratively delivered, and generating multiple candidate delivery paths; each delivery node is delivered by at least one of a truck and a drone; obtaining the delivery time window of each delivery node, and determining the target delivery time of the delivery node from the first delivery time of the truck, the second delivery time of the drone, and the delivery time window in turn; determining the service delay risk index of the candidate delivery path based on the target delivery time and the corresponding delivery time window; inputting the service delay risk index into the target model to obtain the model to be solved; solving the minimized delivery delay risk of each candidate delivery path in the model to be solved, and obtaining the target delivery path of collaborative delivery. In this way, the timeliness of logistics distribution can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent logistics scheduling technology, and in particular to a collaborative delivery path planning method based on trucks and drones and related devices. Background Art

[0002] Collaborative distribution is a logistics model in which two or more parties work together, integrate resources, share information, and work together in the logistics distribution process. Collaborative distribution can achieve the optimal configuration and efficient use of logistics resources by integrating the transportation resources, distribution networks, and warehousing capabilities of different enterprises, aiming to improve distribution efficiency, reduce logistics costs, and optimize customer service levels.

[0003] With the rapid development of e-commerce and the improvement of people's pace of life, users have higher and higher requirements for the convenience and real-time nature of the purchasing experience. In logistics and distribution, completing delivery activities within the time window is becoming more and more important. In related technologies, the goals of collaborative distribution systems are mostly focused on reducing the overall delivery time, or striving to minimize the path and other costs, which will lead to the problem of untimely logistics and distribution. Summary of the invention

[0004] The main purpose of the embodiments of the present application is to propose a collaborative delivery path planning method based on trucks and drones and related devices, which can improve the timeliness of logistics delivery by planning the collaborative delivery path.

[0005] To achieve the above objectives, a first aspect of an embodiment of the present application proposes a collaborative delivery path planning method based on trucks and drones, the method comprising:

[0006] Acquire multiple node location information corresponding to multiple delivery nodes that need to be coordinated for delivery, and generate multiple candidate delivery paths based on the multiple node location information; wherein each of the delivery nodes in each of the candidate delivery paths is delivered by at least one of a truck and a drone;

[0007] Obtaining the delivery time window of each delivery node, and for each candidate delivery path, recursively determining the target delivery time of each delivery node from the first delivery time when the truck arrives at the delivery node, the second delivery time when the drone arrives at the delivery node, and the delivery time window;

[0008] Determine the service delay risk index of the corresponding candidate delivery path based on the target delivery time of the plurality of delivery nodes and the corresponding plurality of delivery time windows;

[0009] Inputting the service delay risk index into the target model to obtain a model to be solved;

[0010] The model to be solved is used to solve the delivery delay risk minimization of each candidate delivery path to obtain the target delivery path for collaborative delivery.

[0011] Accordingly, a second aspect of an embodiment of the present application proposes a collaborative delivery path planning device based on trucks and drones, the device comprising:

[0012] A generation module, used to obtain multiple node location information corresponding to multiple delivery nodes that need to be coordinated for delivery, and generate multiple candidate delivery paths based on the multiple node location information; wherein each of the delivery nodes in each of the candidate delivery paths is delivered by at least one of a truck and a drone;

[0013] an acquisition module, configured to acquire a delivery time window of each delivery node, and for each candidate delivery route, determine a target delivery time of each delivery node from a first delivery time when a truck arrives at the delivery node, a second delivery time when a drone arrives at the delivery node, and the delivery time window in a recursive manner;

[0014] A determination module, configured to determine a service delay risk index of a corresponding candidate delivery path based on target delivery times of a plurality of delivery nodes and a corresponding plurality of delivery time windows;

[0015] An input module, used for inputting the service delay risk index into a target model to obtain a model to be solved;

[0016] The solution module is used to solve the minimized delivery delay risk of each candidate delivery path for the model to be solved, and obtain the target delivery path of collaborative delivery.

[0017] In some embodiments, the input module is further used to:

[0018] When the number of historical samples of path planning exceeds a first threshold, the target model is a first planning model constructed based on empirical distribution; then the service delay risk index is input into the target model to obtain a model to be solved, including: inputting the service delay risk index into the first planning model to obtain a first model to be solved;

[0019] When the number of historical samples of the path planning does not exceed the first threshold, the target model is a second planning model constructed based on an uncertain set; then the service delay risk index is input into the target model to obtain a model to be solved, including: inputting the service delay risk index into the second planning model to obtain a second model to be solved.

[0020] In some embodiments, the collaborative delivery path planning device based on trucks and drones further includes a first building module for:

[0021] Obtaining a plurality of historical samples, and determining a plurality of historical delivery times of the plurality of historical samples;

[0022] Building a first initial model based on the average of multiple historical delivery times and the multiple candidate delivery routes;

[0023] Constructing a first constraint model based on the historical delivery device corresponding to any historical delivery node of each of the historical samples, the delivery sequence between the historical delivery nodes, and the preset value range of the historical delivery time of each of the historical delivery nodes;

[0024] A first planning model is constructed based on the first constraint model and the first initial model.

[0025] In some implementations, the model to be solved is a first model to be solved; and the building module is further used to:

[0026] Determining an initial first delivery path from the plurality of candidate delivery paths;

[0027] Decomposing a first linear programming model corresponding to a service delay risk index from the first model to be solved based on the initial first delivery path;

[0028] Obtaining a first dual model of the first linear programming model, and solving the first dual model to obtain a second delivery path and at least one first constraint for path planning;

[0029] Repeatingly determining a new first delivery path from the multiple candidate delivery paths based on at least one of the first constraints, decomposing a first linear programming model corresponding to a service delay risk index from the first model to be solved based on the first delivery path, obtaining a first dual model of the first linear programming model, and solving the first dual model to obtain a new second delivery path and at least one new first constraint for path planning, and comparing the new first delivery path with the new second delivery path until the comparison results of the new first delivery path and the new second delivery path are consistent, stopping determining the new first delivery path, and determining the current new first delivery path as the target delivery path.

[0030] In some embodiments, the collaborative delivery path planning device based on trucks and drones further includes a second building module for:

[0031] Acquire the time fluctuation range of each of the delivery nodes, and construct a time fluctuation function based on the time fluctuation ranges of the plurality of delivery nodes;

[0032] Building a second initial model based on the time fluctuation function and multiple candidate delivery routes;

[0033] Constructing a second constraint model based on the delivery order between the delivery nodes in each of the candidate delivery paths and the preset value range of the delivery time of each of the delivery nodes;

[0034] A second planning model is constructed according to the second initial model and the second constraint model.

[0035] In some implementations, the model to be solved is a second model to be solved; and the second building module is further used to:

[0036] Performing a dual transformation on the second model to be solved to obtain a second dual model;

[0037] Determining an initial third delivery path from the plurality of candidate delivery paths;

[0038] Decomposing a second linear programming model corresponding to a service delay risk index from the second dual model based on the initial third delivery path, and solving the second linear programming model to obtain a fourth delivery path and at least one second constraint for path planning;

[0039] Repeatedly determine a new third delivery path from the multiple candidate delivery paths based on at least one of the second constraints, decompose a second linear programming model corresponding to a service delay risk index from the second model to be solved based on the third delivery path, obtain a second linear programming model of the second linear programming model, solve the second linear programming model to obtain a new fourth delivery path and at least one new second constraint for path planning, compare the new third delivery path with the new fourth delivery path until the comparison results of the new third delivery path and the new fourth delivery path are consistent, stop determining the new third delivery path, and determine the current new third delivery path as the target delivery path.

[0040] In some embodiments, the second building block is further used to:

[0041] Obtaining the nominal value and the maximum deviation value of the truck and the drone at each delivery node respectively;

[0042] The time fluctuation range of each of the delivery nodes is determined based on the nominal value and the maximized deviation value.

[0043] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes a computer device, which includes a memory and a processor, and the memory stores a computer program. When the processor executes the computer program, it implements the collaborative delivery path planning method based on trucks and drones as described in any one of the embodiments of the first aspect of the present application.

[0044] To achieve the above-mentioned objectives, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the collaborative delivery path planning method based on trucks and drones as described in any one of the embodiments of the first aspect of the present application.

[0045] The embodiment of the present application obtains multiple node location information corresponding to multiple distribution nodes that need to be coordinated and generates multiple candidate distribution paths based on the multiple node location information; wherein each distribution node in each candidate distribution path is distributed by at least one of a truck and a drone; obtains the distribution time window of each distribution node, and for each candidate distribution path, determines the target distribution time of each distribution node in a recursive manner from the first distribution time when the truck arrives at the distribution node, the second distribution time when the drone arrives at the distribution node, and the distribution time window; determines the service delay risk index of the corresponding candidate distribution path based on the target distribution time of the multiple distribution nodes and the corresponding multiple distribution time windows; inputs the service delay risk index into the target model to obtain the model to be solved; solves the minimized distribution delay risk of each candidate distribution path for the model to be solved, and obtains the target distribution path for coordinated distribution. In this way, the distribution delay risk of each candidate distribution path can be measured by summarizing the service delay risk index of all distribution nodes, and the coordinated distribution path is optimized on this basis, thereby improving the timeliness of logistics distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Schematic diagram of the architecture of a collaborative delivery path planning system based on trucks and drones provided in an embodiment of the present application;

[0047] Figure 2 is a flow chart of a collaborative delivery path planning method based on trucks and drones provided in an embodiment of the present application;

[0048] Figure 3 is an example diagram of a target delivery path provided in an embodiment of the present application;

[0049] Figure 4 It is a schematic diagram of functional modules of a collaborative delivery path planning device based on trucks and drones provided in an embodiment of the present application;

[0050] Figure 5It is a schematic diagram of the hardware structure of the computer device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0054] Collaborative distribution is a logistics model in which two or more parties work together, integrate resources, share information, and work together in the logistics distribution process. Collaborative distribution can achieve the optimal configuration and efficient use of logistics resources by integrating the transportation resources, distribution networks, and warehousing capabilities of different enterprises, aiming to improve distribution efficiency, reduce logistics costs, and optimize customer service levels.

[0055] With the rapid development of e-commerce and the improvement of people's pace of life, users have higher and higher requirements for the convenience and real-time nature of the purchasing experience. In logistics and distribution, it is becoming more and more important to complete delivery activities within the time window. In related technologies, the goals of collaborative distribution systems are mostly focused on reducing the overall delivery time, or striving to minimize the path and other costs. However, the stability of collaborative distribution is affected by various uncertainties, such as weather conditions and uncertainty in delivery time, which leads to service delays, reduces the delivery level, and causes problems with untimely logistics and distribution.

[0056] Based on this, an embodiment of the present application provides a collaborative delivery path planning method and related devices based on trucks and drones, which improves the timeliness of logistics delivery by planning the collaborative delivery path.

[0057] The collaborative delivery path planning method based on trucks and drones and related devices provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the collaborative delivery path planning system based on trucks and drones in the embodiments of the present application is described.

[0058] Please refer to Figure 1 In some embodiments, an embodiment of the present application provides a collaborative delivery path planning system based on trucks and drones, including a terminal 11 and a server 12.

[0059] Exemplarily, the terminal 11 may be a mobile terminal on a delivery vehicle, a control terminal on a drone, or a user device such as a smart phone or tablet computer. The terminal 11 may be used to provide a user interaction interface such as placing an order, tracking the delivery status, and receiving notifications. The terminal 11 is integrated with or connected to a sensor to collect information about the user's location, order requirements, etc. Furthermore, the terminal 11 may also include a machine learning model to process local data and make preliminary decisions, such as recommending specific vehicles and drone service equipment, and sending relevant instructions.

[0060] The terminal 11 can receive the path planning instructions and real-time scheduling information sent by the server 12, and provide real-time feedback on the delivery progress and abnormal situations during the collaborative delivery process.

[0061] Exemplarily, the server 12 may be a high-performance server or cloud computing platform in a data center, including the main logic for executing data preprocessing, path planning, and optimization algorithms, as well as an interface for communicating with the terminal 11. The server 12 may be responsible for processing requests from the terminal 11, managing data, and performing complex calculations, such as solving the minimum delivery delay risk for each candidate delivery path for the model to be solved, and determining the optimal target delivery path. The server 12 may also include one or more databases for storing data such as user information, order history, and delivery routes.

[0062] The collaborative delivery path planning method based on trucks and drones in the embodiments of the present application can be illustrated by the following embodiments.

[0063] It should be noted that in each specific implementation of the present application, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first. Moreover, the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or separate consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or separate consent, the necessary user-related data for enabling the normal operation of the embodiment of the present application will be obtained.

[0064] In the embodiment of the present application, the collaborative delivery path planning device based on trucks and drones will be described from the perspective of the collaborative delivery path planning device based on trucks and drones. The collaborative delivery path planning device based on trucks and drones can be integrated into a computer device. Figure 2 , Figure 2 This is a flowchart of the steps of the collaborative delivery path planning method based on trucks and drones provided in an embodiment of the present application. In the embodiment of the present application, a collaborative delivery path planning device based on trucks and drones is specifically integrated on a terminal or a server as an example. When the processor on the terminal or server executes the program instructions corresponding to the collaborative delivery path planning method based on trucks and drones, the specific process is as follows:

[0065] Step 101, obtaining multiple node location information corresponding to multiple distribution nodes that need to be coordinated for distribution, and generating multiple candidate distribution paths based on the multiple node location information; wherein each distribution node in each candidate distribution path is distributed by at least one of a truck and a drone.

[0066] In some implementations, in order to optimize resource allocation in the collaborative delivery process, multiple feasible candidate delivery paths may be generated by acquiring the location information of each delivery node, so as to subsequently evaluate the delay risk of each candidate delivery path and select the optimal target delivery path.

[0067] Among them, the distribution node can be a customer node, that is, a specific location where logistics distribution is required, and a distribution node can correspond to a specific customer node. Each distribution node can be delivered by a vehicle, or a drone, or a vehicle and a drone in coordination.

[0068] The node location information may be the specific geographic coordinates of the customer node, such as longitude and latitude information, etc. The node location information may help the distribution device calculate candidate distribution routes.

[0069] Among them, the candidate delivery path can be a possible delivery route from the distribution center to all delivery nodes, and each candidate delivery path should include all delivery nodes.

[0070] Among them, trucks can be loadable vehicles. In addition to trucks in the general sense, trucks can also be delivery vehicles, smart delivery vehicles, etc. Trucks are equipment that can deliver over long distances and with large loads. Trucks have energy reserves that can at least deliver all delivery nodes.

[0071] Among them, drones can be unmanned aerial vehicles, which are suitable for short-distance, lightweight delivery tasks.

[0072] For example, in the process of generating candidate delivery paths, all delivery nodes that need to be delivered can be first determined based on node location information, and an initial delivery path can be generated. The initial delivery path can take into account the limitations of different delivery capabilities (such as maximum delivery distance and maximum load, etc.) of trucks (such as trucks) and drones (such as drones).

[0073] Furthermore, since trucks and drones are collaborative delivery, and the constraints imposed by the system on both need to be considered, for example, trucks and drones must comply with the law of conservation of energy during delivery, ensure that each delivery node is visited by at least one of the trucks and drones, and the distance limit between trucks and drones, etc. After evaluating or adjusting the initial delivery path according to the preset constraints, a candidate delivery path can be obtained.

[0074] In some implementations, in order to optimize the use of resources and ensure the feasibility of the system, the preset constraints can be set in the following manner. First, the definition and representation of the variables involved in the path planning process of this application are introduced. It should be noted that the specific definitions and representations can be different without violating the concept of this application, and can be adjusted according to the actual situation.

[0075] For example, the truck-drone collaborative delivery problem considering the window period can be defined as a completely directed graph The set of delivery nodes is defined as .make The set of delivery nodes in is represented by The set of arcs is defined as Each distribution node has and The delivery time window represents the earliest time and the latest time to provide delivery service for the delivery node. represents the delivery time of the truck, represents the delivery time of the drone, and it is assumed that the service time of each customer is also included in the delivery time. ,make represents the outgoing arc set of i, express The arc set of .

[0076] For the multi-item commodity flow of the truck-drone collaborative delivery system, there are the following expressions of preset constraints:

[0077]

[0078] In formula (1), the specific symbols have the following meanings:

[0079] For any arc , if it is only traversed by trucks, that is, only trucks work on this delivery path, this arc is called a truck arc, and To represent; similarly, if for any arc When only drones are working on this delivery path, it is called a drone arc, which is composed of and In addition, for any customer delivery node ,If the delivery node is only visited by trucks, then In other cases, ; When the delivery node is only visited by drones, In other cases, ; When the delivery node When being visited by a truck and a drone at the same time, it is recorded as .

[0080] Furthermore, the meaning of each preset constraint in formula (1) is explained in turn:

[0081] Specifically, constraints (1a) and (1b) ensure that the truck and the drone can leave and return to the warehouse exactly once; constraints (1c) and (1d) satisfy the flow conservation law that the truck and the drone should obey respectively; constraints (1e) and (1f) set the variable , Same variable , , are linked; the constraint (1g) ensures that the variable , and variables , , The logical relationship between them; constraint (1h) ensures that each node can be visited by a truck alone, a drone alone, or a truck and a drone simultaneously; constraint (1i) ensures that the drone only serves one customer (i.e., delivery node) in each departure-return cycle; constraint (1j) indicates that the drone cannot return to the warehouse immediately after serving a single customer; constraints (1k) and (1l) describe the range of the variables.

[0082] Based on this, the candidate delivery paths can be characterized as follows:

[0083]

[0084] For any , all have unique Make ,therefore and There is a one-to-one correspondence between them.

[0085] By characterizing the path with preset constraints, we can ensure the continuity, independence and rationality of each candidate delivery path, and ensure the coordination between trucks and drones during the delivery process, so as to facilitate the subsequent determination of the delivery time for the candidate delivery paths and then select the target delivery path.

[0086] Step 102, obtain the delivery time window of each delivery node, and for each candidate delivery path, determine the target delivery time of each delivery node in a recursive manner from the first delivery time when the truck arrives at the delivery node, the second delivery time when the drone arrives at the delivery node, and the delivery time window.

[0087] In some implementations, in order to ensure the timeliness of the delivery task under the time window constraint, a target delivery time, that is, a service start time, may be determined for each candidate delivery path, so as to facilitate the subsequent determination of a service delay risk index for each candidate delivery path.

[0088] The delivery time window can be the time within a certain time range that the customer can accept the delivery of logistics. The delivery time window can be composed of the earliest delivery time and the latest delivery time selected by the customer. The earliest delivery time refers to the earliest time that logistics can be delivered, and the latest delivery time refers to the latest time that logistics must be delivered. For example, if customer A requires [9:00-11:00] for delivery, then [9:00-11:00] is the delivery time window.

[0089] The first delivery time may be the time when the truck departs from the previous delivery node and arrives at the target delivery node.

[0090] The second delivery time may be the time it takes for the drone to depart from the previous delivery node and arrive at the target delivery node.

[0091] The target delivery time may be the time when the delivery service is provided to the customer, that is, the time when the goods are actually delivered. The customer may receive the goods at the target delivery time.

[0092] For example, suppose that the target delivery time of delivery node a needs to be determined, and the preceding delivery node of delivery node a is delivery node b, that is, the truck and / or drone first passes through delivery node b and then reaches delivery node a. Then, the first delivery time can be the time it takes for the truck to travel from delivery node b to delivery node a; and the second delivery time can be the time it takes for the drone to travel from delivery node b to delivery node a.

[0093] Exemplarily, for each candidate delivery route, in order to ensure that no customer service time is missed, for each candidate delivery route, among all arcs along the candidate delivery route, that is, in the arcs connecting any two delivery nodes, the maximum value can be determined as the target delivery time from the first delivery time, the second delivery time, and the earliest time when the delivery node provides delivery service. In this way, it can be ensured that during the delivery process, regardless of the actual situation, the delivery needs of any delivery node can be met. In other words, it can be ensured that there is enough time to start service when the truck and / or drone arrives at the delivery node. Specifically, it can be made is a deterministic tuple of trucks and drones. Assume that trucks and drones are Leave the warehouse at the same time, for a given candidate delivery path , each distribution node The target delivery time can be determined recursively as:

[0094]

[0095] In formula (3), represents the target delivery time of delivery node l, represents the target delivery time of the truck / drone at the predecessor node i of the delivery node l, represents the first delivery time of the truck from the previous node i to the delivery node l, that is, the time to arrive at the delivery node i, represents the second delivery time of the drone from the previous node i to the delivery node l, It is understandable that since there are delivery nodes in the candidate delivery paths that are only delivered by trucks or drones, or The value may be 0 or 1.

[0096] According to formula (3), for each delivery node, the maximum value among the first delivery time, the second delivery time and the earliest time to provide delivery service for the delivery node can be determined as the target delivery time of the delivery node. In this way, the target delivery time of each delivery node in each candidate delivery path can be recursively determined.

[0097] Through the above method, the delivery time window of the delivery node can be obtained, and the target delivery time of each delivery node can be recursively determined, so as to optimize the delivery route selection and improve the overall delivery efficiency. At the same time, it is convenient to determine the service delay risk index according to the target delivery time in the future.

[0098] Step 103: determining a service delay risk index of a corresponding candidate delivery path based on target delivery times of multiple delivery nodes and corresponding multiple delivery time windows.

[0099] In some implementations, in order to optimize the delivery process and reduce potential delay costs, the service delay risk index of each delivery node may be calculated to facilitate subsequent measurement of the total delay magnitude of each candidate delivery path.

[0100] The service delay risk index may be a quantitative indicator for evaluating the service delay risk during the delivery process. In some embodiments, before determining the service delay risk index, the service delay time of each delivery node may be determined based on the target delivery time of the multiple delivery nodes and the corresponding multiple delivery time windows.

[0101] For example, given a candidate delivery path solution and confirmed delivery time , at each node The service delay time can be characterized by the following delay function:

[0102]

[0103] in, , if and only if When the distribution node The service time is delayed relative to the delivery time window, which can be used To represent the delay vector, the service level can be expressed by to measure.

[0104] Furthermore, you can use As the service delay risk index corresponding to each candidate target path, the service delay time of each delivery node can be accumulated to measure the service delay risk index. Since the target delivery time of each delivery node can be recursively determined, the service delay risk index This can be obtained by solving the following linear programming:

[0105]

[0106] in, is a sufficiently large positive integer, i and j both represent distribution nodes, i is the preceding distribution node of j, represents the target delivery time of delivery node l, represents the earliest time to provide delivery service for delivery node l, represents the latest time to provide delivery service for delivery node l, represents the target delivery time of delivery node i, represents the target delivery time of delivery node j, represents the delivery time of a truck from delivery node i to delivery node j, represents the delivery time of the drone from delivery node i to delivery node j, N represents the set of delivery nodes, represents the arc generated by the truck between any two delivery nodes x and j, It can be 0 or 1. represents the arc generated by the drone between any two delivery nodes x and j, Can be 0 or 1.

[0107] By characterizing the service delay risk index corresponding to the candidate delivery path, we can quantify and evaluate the service delay risk that may occur during the delivery process, so that the service delay risk can be input into the target model in the future, the service delay risk can be minimized, and the target delivery path that can deliver to each delivery node in a timely manner can be obtained as much as possible.

[0108] Step 104: input the service delay risk index into the target model to obtain a model to be solved.

[0109] In some embodiments, in order to minimize the service delay risk associated with the delivery window, the service delay risk can be input into the target model to obtain a model to be solved, so as to subsequently solve the optimal path that minimizes the service delay risk based on the model to be solved.

[0110] Among them, the target model can be a model for optimizing collaborative delivery path planning generated based on candidate delivery paths and a preset value range. The target model can receive a service delay risk index as input to obtain a model to be solved.

[0111] The model to be solved may be a mathematical model. By solving the model to be solved, the risk of delivery delay for each candidate delivery path can be minimized while satisfying the constraints of the delivery time window.

[0112] In some embodiments, the target model may include a first planning model and a second planning model. Specifically, when there are abundant historical samples, the service delay risk of the current candidate delivery path can be estimated based on the empirical distribution of the historical samples. Specifically, the sample average approximation method can be used to construct the first planning model; when the number of historical samples is limited or the distribution of delivery time is unknown, a robust optimization framework can be used to handle uncertainty, and the uncertainty of delivery time can be characterized by defining an uncertainty set to find a path plan with good performance in all possible uncertain situations, so as to protect the risks brought by insufficient historical samples or uncertainty.

[0113] By inputting the service delay risk index into the target model, the model to be solved is obtained, which can facilitate the subsequent solution of the model to be solved, obtain the target delivery path that minimizes the service delay risk, and improve user satisfaction.

[0114] In some implementations, in order to effectively optimize and solve the uncertainty of delivery time under different circumstances, the specific model construction method can be determined based on the number of historical samples to improve the optimization and solving ability, efficiency and reliability of the system, while also enhancing the risk management ability. For example, step 104 can include the following two situations:

[0115] (A) When the number of historical samples of path planning exceeds a first threshold, the target model is a first planning model constructed based on the empirical distribution; the service delay risk index is input into the target model to obtain a model to be solved, including: inputting the service delay risk index into the first planning model to obtain a first model to be solved;

[0116] (B) When the number of historical samples of path planning does not exceed the first threshold, the target model is a second planning model constructed based on the uncertain set; then the service delay risk index is input into the target model to obtain the model to be solved, including: inputting the service delay risk index into the second planning model to obtain the second model to be solved.

[0117] Among them, the historical sample can be the actual delivery time data collected during the historical delivery process. The historical sample can include delivery time data, service delay data, delivery path data, delivery equipment and all other data related to historical delivery.

[0118] Among them, the first threshold can be a critical value used to distinguish whether the number of historical samples is sufficient. When the number of historical samples exceeds the first threshold, it indicates that the number of historical samples is sufficient, and the first planning model based on empirical distribution can be used; when the number of historical samples does not exceed the first threshold, it indicates that the number of historical samples is insufficient, and it is difficult to construct an accurate probability distribution based on the existing distribution of the number of historical samples. At this time, the second planning model based on the uncertain set can be used to protect the system from the impact of bad path decisions.

[0119] Among them, the first planning model can be a path planning model constructed based on empirical distribution, that is, a model constructed based on the sample average approximation method, which is suitable for situations where there are sufficient historical samples.

[0120] Among them, the second planning model can be a path planning model constructed based on an uncertain set, that is, a model constructed based on a robust optimization method, which is suitable for situations where the number of historical samples is insufficient.

[0121] In some embodiments, when the number of historical samples is sufficient, the actual delivery time distribution can be approximated by the empirical distribution of historical samples. In this case, a model based on Sample Average Approximation (SAA) (the first planning model) can be used for path planning, namely, the STDRP-TW model. When the number of samples is insufficient, statistical information cannot be effectively extracted. In this case, a model based on Robust Optimization (RO) (the second planning model) is used for path planning, namely, the RTDRP-TW model, to cope with the uncertainty of delivery time.

[0122] By utilizing the number of historical samples, we can build an adaptable path planning model for different historical sample numbers, improve the system's optimization and risk management capabilities, and find solutions to minimize risks under any circumstances, thereby improving user satisfaction.

[0123] In some implementations, in order to better handle the uncertainty of delivery time and improve the efficiency and accuracy of path planning, when there are a large number of historical samples, the historical samples can be used to construct an empirical distribution to approach the actual delivery time distribution. For example, the first planning model in (A) is constructed in the following way:

[0124] (A.1) obtaining a plurality of historical samples and determining a plurality of historical delivery times of the plurality of historical samples;

[0125] (A.2) constructing a first initial model based on the mean of multiple historical delivery times and multiple candidate delivery routes;

[0126] (A.3) constructing a first constraint model based on the historical delivery device corresponding to any historical delivery node of each historical sample, the delivery sequence between the historical delivery nodes, and the preset value range of the historical delivery time of each historical delivery node;

[0127] (A.4) Constructing a first planning model based on the first constraint model and the first initial model.

[0128] The historical delivery time may be the completion time of each historical target delivery route actually recorded in the past delivery process.

[0129] The first initial model may be a model including an objective function and its related variables, and the first initial model may include multiple objective optimization indicators, and the objective optimization indicators may include a delivery path and a delay risk.

[0130] Among them, the first constraint model can include various preset value ranges in the path planning process to ensure the feasibility of the planning and meet the actual delivery requirements.

[0131] The first planning model can be constructed by combining the first initial model and the first constraint model.

[0132] For example, when there are sufficient historical samples, the sample average approximation method can be used to construct the first initial model. Specifically, assuming there are K historical samples , assigning equal weight to each historical sample, and constructing an empirical distribution based on historical samples :

[0133]

[0134] in, yes Dirac distribution at , K is the number of the specific historical sample.

[0135] In some implementations, different weights may be assigned to historical samples in different time periods and then weighted averaged. Specifically, higher weights may be assigned to recent historical samples, that is, different weights may be set according to the time distance between historical samples and current samples, so as to better reflect recent environmental changes and trends.

[0136] Furthermore, since the present application aims to determine the empirical distribution in the process of constructing the first planning model Finding a feasible path solution that minimizes the risk of service delay is to solve the following stochastic planning problem:

[0137]

[0138] Therefore, by reconstructing equation (5), we can get the first initial model as follows:

[0139]

[0140] in, is an auxiliary variable used to replace the original objective function , N represents the set of delivery nodes for each candidate delivery path, and l represents the delivery node.

[0141] The first constraint model is:

[0142]

[0143] Where M is a sufficiently large positive integer, I, j and l all represent distribution nodes, i is the preceding distribution node of j, represents the target delivery time of delivery node l in historical sample k. Similarly, represents the target delivery time of delivery node i in historical sample k, represents the target delivery time of delivery node j in historical sample k, represents the delivery time of a truck from i to j in historical sample k, represents the delivery time of the drone from i to j in the historical sample k, represents the earliest time to provide delivery service for delivery node l, represents the latest time to provide delivery service for delivery node l, is an auxiliary variable used to replace the original objective function , is the set of all candidate delivery routes.

[0144] Combining the first initial model with the first constraint model, we can get the following first planning model:

[0145]

[0146] Where M is a sufficiently large positive integer, I, j and l all represent distribution nodes, i is the preceding distribution node of j, represents the target delivery time of delivery node l in historical sample k. Similarly, represents the target delivery time of delivery node i in historical sample k, represents the target delivery time of delivery node j in historical sample k, represents the delivery time of a truck from i to j in historical sample k, represents the delivery time of the drone from i to j in the historical sample k, represents the earliest time to provide delivery service for delivery node l, represents the latest time to provide delivery service for delivery node l, is an auxiliary variable used to replace the original objective function , is the set of all candidate delivery routes.

[0147] Specifically, the first planning model, that is, the stochastic planning model based on sample average approximation, requires that for each historical sample K, the relevant time constraints satisfy the preset value ranges in the first constraint model.

[0148] According to formula (8) For example, if a truck passes through distribution node i and arrives at distribution node j (i.e. ), the service time at point j should not be earlier than the service time at point i plus the travel time between (i, j).

[0149] By constructing the first planning model, historical data can be effectively utilized. By setting the first constraint model, a mathematical optimization model can be constructed that can minimize the risk of service delays while seeking efficient delivery paths. The steps for solving the first planning model will be expanded below. Please refer to the following for specific solution steps.

[0150] In some embodiments, when the number of historical samples is small, the empirical distribution may deviate significantly from the actual distribution of delivery time, which will affect the performance of the constructed model. Therefore, in order to effectively deal with uncertainty and change and avoid the degradation of the performance of the constructed model, in the scenario with only a small number of historical samples, a model with high robustness and reliability can be constructed by using a robust optimization method, so as to better deal with uncertainty and change and achieve efficient and accurate delivery management. For example, the second planning model in (B) is constructed in the following way:

[0151] (B.1) Obtain the time fluctuation range of each delivery node, and construct a time fluctuation function based on the time fluctuation ranges of multiple delivery nodes;

[0152] (B.2) Constructing a second initial model based on the time fluctuation function and multiple candidate delivery routes;

[0153] (B.3) constructing a second constraint model based on the delivery sequence between the delivery nodes in each candidate delivery path and the preset value range of the delivery time of each delivery node;

[0154] (B.4) Constructing a second planning model based on the second initial model and the second constraint model.

[0155] The time fluctuation range may be the possible variation range of the delivery time of each delivery node. The variation range includes a variety of uncertain factors, such as traffic factors, weather factors, etc., which may cause delivery time deviations.

[0156] Among them, the time fluctuation function can be a function constructed based on the time fluctuation range of multiple delivery nodes, which is used to describe the uncertainty of delivery time.

[0157] The second initial model may be a model including an objective function and its related variables, and the first initial model may include multiple objective optimization indicators, which may include delivery paths, delay risks, and time fluctuation functions.

[0158] The second constraint model may include various preset value ranges in the path planning process to ensure the feasibility of the planning and meet the actual delivery requirements.

[0159] The second planning model can be constructed by the second initial model and the second constraint model.

[0160] Specifically, when there are few historical samples, a robust optimization framework can be used to handle uncertainty, and it is assumed that the delivery time obeys a specific uncertainty set. In the uncertainty set, the nominal values ​​of the delivery time spent by trucks and drones between any two delivery nodes are given, that is, the nominal delivery time, and the maximum deviation value of the truck and drone from the nominal delivery time is given. Then, based on the nominal delivery time and the maximum deviation value, the time fluctuation range of the uncertainty level time of each delivery node is obtained, and based on the value range of multiple delivery nodes, the uncertainty set is constructed. Furthermore, the time fluctuation function of the candidate delivery path can be constructed based on the uncertainty set.

[0161] Specifically, uncertain sets can be expressed in the following forms:

[0162]

[0163] in , is a given budget parameter, for any arc between two delivery nodes , The nominal delivery time is the time it takes a truck to deliver on the arc, The nominal delivery time representing the delivery time of the drone on the arc; and Respectively represent the maximum deviation of truck and drone from the nominal delivery time; interval and Represents the uncertainty level time The time fluctuation range of It is used to control the total deviation of all delivery times from the nominal delivery time. In this uncertainty set, all parameters can be estimated through samples.

[0164] In the case of a small number of historical samples, we can find a feasible candidate delivery path solution so that the value of the service delay risk on the uncertain set is minimized in the worst case. On this basis, we can solve the problem to obtain the delivery solution with the minimum delay. Its essence is to solve the following robust optimization problem:

[0165]

[0166] in, represents the candidate delivery path, and Z represents the uncertain set.

[0167] Further, we can get is the second initial model.

[0168] Furthermore, a second constraint model can be constructed based on the order of delivery of each delivery node in each candidate delivery path and the preset value range of the delivery time of each delivery node. Specifically, the second constraint model is:

[0169]

[0170] Where M is a sufficiently large positive integer, I, j and l all represent distribution nodes, i is the preceding distribution node of j, represents the target delivery time of delivery node l. Similarly, represents the target delivery time of delivery node i, represents the target delivery time of delivery node j, represents the delivery time of the truck from i to j, represents the delivery time of the drone from i to j, represents the arc generated by the truck between any two delivery nodes i and j, It can be 0 or 1. represents the arc generated by the drone between any two delivery nodes i and j, Can be 0 or 1; represents the earliest time to provide delivery service for delivery node l, represents the latest time to provide delivery service for delivery node l, is an auxiliary variable used to replace the original objective function , is the set of all candidate delivery routes.

[0171] Thus, through the second initial model and the second constraint model, the second planning model (RTDRP-TW model) can be constructed as follows:

[0172]

[0173] Where M is a sufficiently large positive integer, I, j and l all represent distribution nodes, i is the preceding distribution node of j, represents the target delivery time of delivery node l. Similarly, represents the target delivery time of delivery node i, represents the target delivery time of delivery node j, represents the delivery time of the truck from i to j, represents the delivery time of the drone from i to j, represents the arc generated by the truck between any two delivery nodes i and j, It can be 0 or 1. represents the arc generated by the drone between any two delivery nodes i and j, Can be 0 or 1; represents the arc set of any two delivery nodes on the corresponding candidate delivery path, represents the node set of any two delivery nodes on the corresponding candidate delivery path; represents the earliest time to provide delivery service for delivery node l, represents the latest time to provide delivery service for delivery node l, is an auxiliary variable used to replace the original objective function , is the set of all candidate delivery routes.

[0174] Through the above method, when there are fewer historical samples, a more reliable and robust second model to be solved can be established to facilitate the subsequent solution of the second model to be solved and improve the efficiency and accuracy of distribution management.

[0175] In some embodiments, in order to provide a more robust and reliable path planning solution when facing the uncertainty in the actual situation when there are only a small number of historical samples, the time fluctuation range of each delivery node can be determined, so that the uncertainty set can be constructed based on the time fluctuation ranges of multiple delivery nodes in the future, so that the second model to be solved can still provide a better solution when facing various possible actual situations, avoiding performance degradation due to unforeseen uncertainties. For example, "obtaining the time fluctuation range of each delivery node" in (B.1) can include:

[0176] (B.1.1) Obtain the nominal value and maximum deviation value of the truck and drone at each delivery node;

[0177] (B.1.2) Determine the time fluctuation range of each delivery node based on the nominal value and the maximum deviation value.

[0178] The nominal value may be a nominal delivery time, that is, a target delivery time for each delivery node.

[0179] Among them, the maximum deviation value can be the maximum range that each delivery node may deviate from its nominal delivery time in actual operation. Specifically, for trucks (such as trucks) and drones (such as drones), the maximum deviation value represents the maximum amount of time that the node delivery time may exceed the nominal value in the worst case.

[0180] In some embodiments, the maximum deviation value of each delivery node may be the same, in which case a unified maximum deviation value may be directly set. The maximum deviation value of each delivery node may also be different. For example, the maximum deviation value may be determined by simulating different delivery scenarios. For example, for users who need to go up the stairs for delivery, the maximum deviation value may be appropriately increased. Different maximum deviation values ​​may also be determined for different areas. For example, some areas are densely populated and need to be avoided. In this case, the maximum deviation value may be appropriately increased, and so on.

[0181] Specifically, when there are few historical samples, a robust optimization framework can be used to handle uncertainty, and it is assumed that the delivery time obeys a specific uncertainty set. In the uncertainty set, the nominal values ​​of the delivery time spent by trucks and drones between any two delivery nodes are given, that is, the nominal delivery time, and the maximum deviation value of the truck and drone from the nominal delivery time is given. Then, based on the nominal delivery time and the maximum deviation value, the time fluctuation range of the uncertainty level time of each delivery node is obtained, and based on the value range of multiple delivery nodes, the uncertainty set is constructed. Furthermore, the time fluctuation function of the candidate delivery path can be constructed based on the uncertainty set.

[0182] By determining the time fluctuation range of each delivery node, the robustness and reliability of the path planning model can be significantly improved. In this way, uncertainty can be effectively handled, so that the second model to be solved has the ability to dynamically adjust the deviation value.

[0183] Step 105, solving the delivery delay risk minimization of each candidate delivery path for the model to be solved, and obtaining the target delivery path for collaborative delivery.

[0184] In some embodiments, since the first model to be solved and the second model to be solved belong to a two-stage robust optimization problem with a high complexity in solving, in order to improve the efficiency of the solution, the first model to be solved and the second model to be solved can be decomposed into smaller and easier to solve sub-problems through the Benders decomposition algorithm, thereby improving the efficiency and accuracy of path planning.

[0185] The target delivery path may be the final target delivery path of the truck and the drone obtained after solving the first model to be solved or the second model to be solved.

[0186] In some embodiments, when the model to be solved is the first model to be solved, in order to improve the efficiency and accuracy of the solution, the model to be solved can be decomposed into a main problem and sub-problems, and the optimal solution, that is, the target delivery path, is finally obtained through iterative solution and interaction. Exemplarily, step 105 may include:

[0187] (105.A1) Determine an initial first delivery path from a plurality of candidate delivery paths;

[0188] (105.A2) Decomposing a first linear programming model corresponding to a service delay risk index from the first model to be solved based on the initial first delivery path;

[0189] (105.A3) Obtaining a first dual model of the first linear programming model, and solving the first dual model to obtain a second delivery path and at least one first constraint for path planning;

[0190] (105.A4) Repeatedly determine a new first delivery path from multiple candidate delivery paths based on at least one first constraint, decompose a first linear programming model corresponding to a service delay risk index from the first model to be solved based on the first delivery path, obtain a first dual model of the first linear programming model, and solve the first dual model to obtain a new second delivery path and at least one new first constraint for path planning, and compare the new first delivery path with the new second delivery path until the comparison results of the new first delivery path and the new second delivery path are consistent, stop determining the new first delivery path, and determine the current new first delivery path as the target delivery path.

[0191] The initial first delivery path can be a path selected from all possible candidate delivery paths by a simplified method (such as a heuristic algorithm or a greedy strategy) when solving a complex delivery path optimization problem, for example, in the framework of the Benders decomposition algorithm. The initial first delivery path is the starting point of the solution process and is used to construct the initial restricted master problem, with the goal of quickly obtaining a feasible solution. It should be noted that the initial first delivery path cannot be directly regarded as the final optimal solution to the problem.

[0192] The first linear programming model may be a linear programming model corresponding to a service delay risk index decomposed from the first model to be solved, and the goal of the first linear programming model is to minimize the total delay risk on the path.

[0193] The first dual model may be a new optimization model formed by dual transformation of the first linear programming model. Solving the first dual model can improve the efficiency of solving the first model to be solved.

[0194] Among them, the first constraint can be an additional linear constraint condition obtained according to the optimal solution of the subproblem (first dual model) after solving the first dual model in order to limit the solution space of the main problem for solving the first delivery path. The first constraint can be used to optimize the delivery path and gradually approach the global optimal solution. Specifically, in each iteration process, it can be calculated based on the current solution (i.e., the second delivery path) and added to the main problem to exclude non-optimal solution paths, thereby gradually tightening the feasible solution space and guiding the solution process towards a better solution.

[0195] The first delivery path can be a path that is continuously updated based on the first constraint obtained from the subproblem (the solution of the first dual model), and each update is to better approximate the goal of minimizing the total service delay risk based on the first constraint. When the new first delivery path is found to be consistent with the second delivery path solved based on the path through comparison, it means that an optimal path that satisfies all current constraints has been found. At this time, the new first delivery path is determined as the final target delivery path, that is, the best delivery solution under given conditions.

[0196] The second delivery path may be a path obtained by solving the dual model of the first linear programming model.

[0197] In some embodiments, the Benders decomposition method can be used to solve the collaborative distribution path planning problem, that is, the collaborative distribution path planning problem in which each distribution node has a distribution time window required for distribution, and the minimization of the service delay risk index is focused on during the planning process. Exemplarily, the first model to be solved can be decomposed into a main problem and a sub-problem, the main problem is used for path planning, and the service delay risk is calculated through the sub-problem. Specifically, the sub-problem is solved first. If the optimal objective value of the sub-problem is unbounded, a feasibility cut is added; if the optimal objective value of the sub-problem is bounded, an optimality cut is added. Both the feasibility cut and the optimality cut can be called the first constraint. When the upper and lower bounds of the problem reach the convergence condition, the algorithm is stopped, and the current solution is output as the target distribution path, and an optimality cut or a feasibility cut is generated for iterative optimization. The algorithm continuously iterates, updates the solutions to the main problem and the sub-problem in each iteration, and gradually approaches the global optimal solution.

[0198] Exemplarily, an optimization cut can be generated by the solution result of the first dual model, and the optimization cut can be incorporated into the solution constraints of the first delivery path, that is, incorporated into the main problem, so as to gradually tighten the solution space and obtain a new first delivery path. The new first delivery path can be used as the lower bound of the main problem.

[0199] Furthermore, the updated main problem can be solved iteratively to obtain a new first distribution path, which is the lower bound of the main problem, until the lower bound of the main problem and the upper bound of the subproblem converge to the same value, and the distribution path corresponding to the main problem in the current iteration is determined as the target distribution path for the final collaborative distribution. The target distribution path ensures the optimal distribution strategy under all constraints. Through Benders decomposition, the efficiency of solving the first model to be solved can be greatly improved.

[0200] For example, for the first model to be solved (see Formula 8), the first model to be solved can be decomposed into a sub-problem and a main problem to greatly speed up the solution. For Formula (8), the first model to be solved (STDRP-TW model) can be re-expressed as follows:

[0201]

[0202] For example, the first model to be solved can be decomposed into a main problem and a sub-problem. The main problem mainly considers path selection and overall planning, ignoring the risk of service delay; the sub-problem calculates the service delay risk of each delivery node under the condition of the first delivery path. Specifically, a first delivery path can be selected from multiple candidate delivery paths. Given the first delivery path of the main problem, the first linear programming model can be decomposed from the first model to be solved. , Represents the distribution node The service delay risk based on historical samples can be expressed as follows:

[0203]

[0204] Given a path planning solution (i.e. the first delivery path) Under the condition of , the first linear programming model in equation (13) is defined as a subproblem. In order to solve the subproblem conveniently, the first dual model of the first linear programming model can be obtained:

[0205]

[0206] In formula (14), K represents the historical sample, and the coefficient The service delay risk index represented by can be calculated by finite-step recursion of formula (3), and the remaining parameters are obtained by dual transformation of formula (13). Among them, the optimal target value of the first dual model is always bounded. Solving the first dual model can obtain the dual solution (that is, the second delivery path) , and provides an optimality cut for equation (13), which is:

[0207]

[0208] use To represent the subproblem at the delivery node The obtained dual solution set. And add the optimality cut to the main problem, that is, add it to the first linear programming model, define the restricted main problem of formula (8), update the constraints of the main problem in formula (8), and obtain:

[0209]

[0210] Furthermore, formula (16) can be reconstructed into the following form:

[0211]

[0212] Where M is a sufficiently large positive integer, I, j and l all represent distribution nodes, i is the preceding distribution node of j, represents the target delivery time of delivery node l in the historical sample. Similarly, represents the target delivery time of delivery node i in the historical sample, represents the delivery time of trucks from i to j in the historical samples, represents the delivery time of drones from i to j in the historical samples, represents the arc generated by the truck between any two delivery nodes i and j, It can be 0 or 1. represents the arc generated by the drone between any two delivery nodes i and j, It can be 0 or 1; A represents the arc set of any two delivery nodes on the corresponding candidate delivery path, and N represents the arc set of any two delivery nodes on the corresponding candidate delivery path; represents the latest time to provide delivery service for delivery node l, is an auxiliary variable used to replace the original objective function , is the set of all candidate delivery routes.

[0213] Specifically, by resolving the updated main problem, a new first delivery path can be obtained, and the optimality cut (that is, the new constraint) is repeatedly added to the main problem to obtain a new first delivery path, until the first delivery path of the main problem and the second delivery path of the subproblem are equal, or the difference is within an acceptable range, the first model to be solved converges, and the new first delivery path (which is also the new second delivery path) is used as the target delivery path.

[0214] In some embodiments, in the process of performing Benders solution on the first model to be solved, not all delivery nodes will affect the total service delay risk. Therefore, the restricted main problem (17) does not consider all delivery nodes. A feasible path solution (the first delivery path) is first obtained. Given the first delivery condition, the delay risk of each delivery node is solved using the sub-problem. The new delivery nodes to be added are determined based on the results of the sub-problems, and further iterations are performed until the optimal solution is obtained.

[0215] In some embodiments, in the main problem of Benders decomposition, one of the candidate delivery paths can be selected as the initial first delivery path. Then, based on the first delivery path, the main problem of each delivery node, that is, the service delay risk, is calculated, and the optimality cut of the subproblem (first dual model) is added to the main problem (in the first linear programming model), and the solution of the main problem is adjusted so that the new first delivery path is closer to the global optimal solution. Repeat the above steps, update the path solution of the main problem, that is, update the first delivery path, and re-solve the subproblem until the convergence condition is met, that is, no new cut needs to be added, or the improvement of the solution of the main problem becomes very small, and the comparison results of the new first delivery path and the second delivery path are consistent, that is, the new first delivery path is determined as the target delivery path, and the target delivery path is output.

[0216] In general, when solving the first model to be solved, the problem is initialized first, and the initial upper and lower bounds are defined: LBD=−∞,UBD=∞, and the cut set is initialized. and F are empty sets respectively. Then, the first step of the solution is to solve the main problem (17) and obtain the optimal solution of the main problem The second step is to solve subproblem (14) and obtain the solution of the subproblem , and the main problem cut set Add the optimality cut (15), that is, add a new first constraint. The third step is to observe whether there is , that is, observe whether the first delivery path is equal to the second delivery path. If the condition is met, stop the iteration and output As the target delivery path; if the condition is not met, then return to the first step and iterate again. In the Benders decomposition of the first model to be solved (STDRP-TW model), the core idea of ​​the decomposition is that not all delivery nodes will affect the total service delay risk. Therefore, the restricted main problem (17) does not consider all delivery nodes, and first obtains a path feasible solution; under the condition of a given path feasible solution, the sub-problem is used to solve the service delay risk of each delivery node, and the results of the sub-problem are used to decide which new delivery nodes to add. Further iteration is performed until the optimal solution is obtained, that is, the target delivery path is obtained, thereby improving the efficiency and accuracy of the solution.

[0217] By using the Benders algorithm to solve the complex collaborative delivery path optimization problem, the first model to be solved is effectively decomposed into more manageable main problems and sub-problems, so that the optimal delivery path can be efficiently determined while considering the risk of service delay. By iteratively solving the main problem and sub-problems, the quality of the solution is gradually improved, and finally converges to the global optimal or approximate optimal solution to generate the target delivery path. In this way, the computational complexity can be effectively reduced and the efficiency of the solution can be improved.

[0218] In some implementations, when the model to be solved is the second model to be solved, in order to efficiently and accurately solve the second model to be solved, the second model to be solved may be decomposed into sub-problems that are easier to solve, and the optimal solution may be gradually approached by solving the problem in stages, so as to effectively solve the complex path planning problem. For example, step 105 may also include:

[0219] (105.B1) Perform a dual transformation on the second model to be solved to obtain a second dual model;

[0220] (105.B2) determining an initial third delivery route from a plurality of candidate delivery routes;

[0221] (105.B3) Decomposing a second linear programming model corresponding to the service delay risk index from the second dual model based on the initial third delivery path, and solving the second linear programming model to obtain a fourth delivery path and at least one second constraint for path planning;

[0222] (105.B4) Repeatedly determine a new third delivery path from multiple candidate delivery paths based on at least one second constraint, decompose a second linear programming model corresponding to the service delay risk index from the second model to be solved based on the third delivery path, obtain the second linear programming model of the second linear programming model, and solve the second linear programming model to obtain a new fourth delivery path and at least one new second constraint for path planning, and compare the new third delivery path with the new fourth delivery path until the comparison results of the new third delivery path and the new fourth delivery path are consistent, stop determining the new third delivery path, and determine the current new third delivery path as the target delivery path.

[0223] The second dual model can be a model obtained by performing a dual transformation on the second model to be solved. The purpose of the dual transformation is to transform the complex nonlinear constraints and objective functions in the original problem into linear constraints and objective functions on the dual variables, thereby simplifying the solution process.

[0224] The initial third delivery path can be a path selected from all possible candidate delivery paths by a simplified method (such as a heuristic algorithm or a greedy strategy) when solving a complex delivery path optimization problem, for example, in the framework of the Benders decomposition algorithm. The initial third delivery path is the starting point of the solution process and is used to construct the initial restricted master problem, with the goal of quickly obtaining a feasible solution. It should be noted that the initial third delivery path cannot be directly regarded as the final optimal solution to the problem.

[0225] The second linear programming model can be a linear programming model decomposed from the second dual model based on the third delivery path. The second linear programming model is used to solve the service delay risk index under the third delivery path. By solving the second linear programming model, the optimal solution for the service delay risk can be obtained, that is, the fourth delivery path.

[0226] Among them, the second constraint can be an additional linear constraint condition obtained according to the optimal solution of the sub-problem (the second linear programming model) in order to limit the solution space of the main problem for solving the third delivery path after solving the second linear programming model. The second constraint can be used to optimize the delivery path and gradually approach the global optimal solution. Specifically, in each iteration process, it can be calculated based on the current solution (i.e., the third delivery path) and added to the main problem to exclude non-optimal solution paths, thereby gradually tightening the feasible solution space and guiding the solution process towards a better solution.

[0227] Among them, the third delivery path can be a path that is continuously updated according to the second constraint obtained from the subproblem (the solution of the second linear programming model), and each update is to better approach the goal of minimizing the total service delay risk based on the second constraint. When it is found through comparison that the new third delivery path is consistent with the fourth delivery path solved based on the path, it means that an optimal path that meets all current constraints has been found. At this time, the new third delivery path is determined as the final target delivery path, that is, the best delivery solution under given conditions.

[0228] The fourth delivery path may be based on the service delay risk minimization result under the current third delivery path, and the fourth delivery path is an optimized path obtained by solving the second linear programming model.

[0229] Specifically, the second model to be solved can be solved by combining the Benders decomposition method. First, in order to simplify the solution process, the second model to be solved can be dually transformed to generate a second dual model. Furthermore, a third delivery path can be selected from multiple candidate delivery paths through a heuristic algorithm or other preliminary optimization strategies.

[0230] Furthermore, based on the third delivery path, the second linear programming model corresponding to the service delay risk index can be further decomposed from the second dual model, and the second linear programming model can be solved to obtain the fourth delivery path. Exemplarily, when the second model to be solved is decomposed by the Benders decomposition method, after obtaining the second dual model, the second dual model can be divided into a main problem and a sub-problem. The main problem is used to determine the third delivery path, and the sub-problem corresponds to the second linear programming model. In this way, the service delay risk in the worst case can be determined under the premise of a given path planning (the third delivery path). According to the solution of the sub-problem, an optimality cut or a feasibility cut, that is, a second constraint, is generated to gradually approach the global optimal solution. Furthermore, through continuous iteration, the solution to the main problem (the third delivery path) can be used as the lower bound of the problem, and the cut generated by the sub-problem helps to tighten the upper bound of the main problem. When the upper and lower bounds are close to the preset convergence threshold, the algorithm stops. At this time, the lower bound of the main problem is the optimal solution to the problem, that is, the third delivery path is used as the target delivery path. In other words, when the lower bound of the main problem and the upper bound of the main problem converge to the same value (the third delivery path and the fourth delivery path are the same), or when the difference is very small, the third delivery path corresponding to the main problem in the current iteration is used as the target delivery path. In this way, the main problem can be used to handle path planning, and the sub-problem focuses on service delay risk assessment. The interaction between the two promotes the gradual optimization of the problem through the Benders cut set mechanism until the global optimal target delivery path is found.

[0231] Exemplarily, for the second model to be solved, by performing a dual transformation on the second model to be solved, it can be transformed into the following two-stage robust optimization problem:

[0232]

[0233] in, is a dual variable, and its inner problem is a bilinear optimization problem, which is equivalent to the following second linear programming model:

[0234]

[0235] in, >0, is a given budget parameter; is a dual variable; I, j and l all represent delivery nodes, i is the preceding delivery node of j, represents the delivery time of the truck from i to j, represents the delivery time of the drone from i to j, represents the arc generated by the truck between any two delivery nodes i and j, It can be 0 or 1. represents the arc generated by the drone between any two delivery nodes i and j, It can be 0 or 1; A represents the arc set of any two delivery nodes on the corresponding candidate delivery path, and N represents the arc set of any two delivery nodes on the corresponding candidate delivery path; represents the earliest time to provide delivery service for delivery node l; It represents the latest time to provide delivery service for delivery node l, and the other parameters are the parameters obtained by dual transformation of the parameters in the second model to be solved.

[0236] Formula (19) is used as a subproblem, that is, the second linear programming model is used as a subproblem. In Benders decomposition, at each iteration, when the subproblem is bounded, the subproblem will generate an optimality cut for the main problem; when the subproblem is unbounded, the subproblem will generate a feasibility cut for the main problem. represents the set of optimal cuts, represents the set of feasible cuts, then the main problem is expressed as follows:

[0237]

[0238] The process of solving the second model by Benders decomposition is similar to the previous one. You can refer to the previous description to understand it. In the process of solving, we first calculate the lower bound of the main problem. , the upper bound of the subproblem , feasible set and other parameters are initialized, and then the iteration begins. The first step is to solve the subproblem (19). When the optimal objective value of the subproblem is unbounded, add a feasibility cut; when the optimal objective value of the subproblem is bounded, add an optimality cut. The second step is to observe the convergence level. When the upper and lower bounds of the problem reach the convergence conditions, stop the algorithm and set the current third delivery path Output as the optimal target delivery path of RTDRP-TW. If it does not converge, go to step 3. Step 3: Solve the main problem (20) and obtain the optimal solution of the main problem , and Assign a lower bound to the main problem ,Right now , then return to the first step and iterate again.

[0239] In the benders decomposition of RTDRP-TW, due to The two-level planning cannot be solved directly, so the path planning and service delay risk problems are divided into two stages for calculation. Through the main problem (20), the solution of the path planning can be solved to obtain the third delivery path; the solution is passed to the sub-problem (19), and the fourth delivery path corresponding to the worst-case service delay risk can be obtained; by judging the nature of the sub-problem, it is decided to add specific feasibility cuts or optimality cuts to the main problem. By splitting the decision into two stages (corresponding to the main problem and the sub-problem), the dilemma that the two-line min max problem cannot be solved directly is effectively solved, because given the solution of the main problem path planning, the sub-problem can be converted into a mixed integer linear programming problem for solution, which can increase the speed of solution.

[0240] The embodiment of the present application obtains multiple node location information corresponding to multiple distribution nodes that need to be coordinated and generates multiple candidate distribution paths based on the multiple node location information; wherein each distribution node in each candidate distribution path is distributed by at least one of a truck and a drone; obtains the distribution time window of each distribution node, and for each candidate distribution path, determines the target distribution time of each distribution node in a recursive manner from the first distribution time when the truck arrives at the distribution node, the second distribution time when the drone arrives at the distribution node, and the distribution time window; determines the service delay risk index of the corresponding candidate distribution path based on the target distribution time of the multiple distribution nodes and the corresponding multiple distribution time windows; inputs the service delay risk index into the target model to obtain the model to be solved; solves the minimized distribution delay risk of each candidate distribution path for the model to be solved, and obtains the target distribution path of the coordinated distribution. In this way, the distribution delay risk of each candidate distribution path can be measured by summarizing the service delay risk index of all distribution nodes, and the distribution delay risk of the candidate distribution path can be minimized. On this basis, the coordinated distribution path is optimized, thereby improving the timeliness of logistics distribution and user satisfaction.

[0241] Please refer to Figure 3 , in some embodiments, Figure 3 It is a target delivery path generated according to the method of the present application. Exemplarily, the truck and the drone start at 0, the truck serves customers 1 and 2 alone, the drone serves customer 3 alone, and after the drone and the truck meet up at customer 2, they travel together to customer 4. After serving customer 4, the truck waits on the spot until the drone finishes serving customer 5 and then returns; the truck and the drone travel together to customer 6 and then separate, the truck continues to serve customers 6 and 8, and the drone serves customer 7 and then flies to customer 8 to meet up with the truck; then the truck returns to the starting point, and the drone continues to serve customer 9 and then returns to the starting point. The target delivery path includes three delivery methods: the truck driving alone path ( Figure 3 The arc of the truck is shown in the middle), and the drone's individual flight path ( Figure 3The drone arc in the middle), the truck / drone co-path ( Figure 3 This means that in the target delivery path, the dispatching of trucks and drones is highly coordinated.

[0242] See also Figure 4 The embodiment of the present application further provides a collaborative delivery path planning device based on trucks and drones, which can implement the collaborative delivery path planning method based on trucks and drones. The collaborative delivery path planning device based on trucks and drones includes:

[0243] A generation module 41 is used to obtain multiple node location information corresponding to multiple delivery nodes that need to be coordinated for delivery, and generate multiple candidate delivery paths based on the multiple node location information; wherein each delivery node in each candidate delivery path is delivered by at least one of a truck and a drone;

[0244] An acquisition module 42 is used to acquire the delivery time window of each delivery node, and for each candidate delivery route, determine the target delivery time of each delivery node from the first delivery time when the truck arrives at the delivery node, the second delivery time when the drone arrives at the delivery node, and the delivery time window in a recursive manner;

[0245] A determination module 43, configured to determine a service delay risk index of a corresponding candidate delivery path based on target delivery times of a plurality of delivery nodes and corresponding plurality of delivery time windows;

[0246] An input module 44 is used to input the service delay risk index into the target model to obtain a model to be solved;

[0247] The solution module 45 is used to solve the delivery delay risk minimization of each candidate delivery path for the model to be solved, and obtain the target delivery path for collaborative delivery.

[0248] The specific implementation of the collaborative delivery path planning device based on trucks and drones is basically the same as the specific implementation of the collaborative delivery path planning method based on trucks and drones described above, and will not be repeated here. On the premise of meeting the requirements of the embodiments of this application, the collaborative delivery path planning device based on trucks and drones can also be provided with other functional modules to implement the collaborative delivery path planning method based on trucks and drones in the above embodiments.

[0249] The embodiment of the present application also provides a computer device, the computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned truck-based and drone-based collaborative delivery path planning method when executing the computer program. The computer device can be any intelligent terminal including a tablet computer, a vehicle-mounted computer, etc.

[0250] See also Figure 5 , Figure 5 The hardware structure of a computer device according to another embodiment is shown, and the computer device includes:

[0251] The processor 51 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0252] The memory 52 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 52 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 52, and the processor 51 calls and executes the collaborative delivery path planning method based on trucks and drones in the embodiment of this application;

[0253] Input / output interface 53, used to implement information input and output;

[0254] Communication interface 54, used to realize communication interaction between the device and other devices, which can be realized by wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);

[0255] A bus 55 that transmits information between the various components of the device (e.g., the processor 51, the memory 52, the input / output interface 53, and the communication interface 54);

[0256] The processor 51 , the memory 52 , the input / output interface 53 and the communication interface 54 are connected to each other in communication within the device via a bus 55 .

[0257] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned collaborative delivery path planning method based on trucks and drones.

[0258] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0259] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0260] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0261] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0262] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0263] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0264] It should be understood that in the present application, "at least one (item)" and "several" refer to one or more, and "plurality" refers to two or more. "And / or" is used to describe the association relationship of associated objects. The parameters can exist in three relationships. For example, "A and / or B" can be parameters: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after the parameter are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can be parameters: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0265] In the several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0266] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0267] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0268] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0269] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. A collaborative delivery path planning method based on trucks and drones, characterized in that: The method comprises: Acquire multiple node location information corresponding to multiple delivery nodes that need to be coordinated for delivery, and generate multiple candidate delivery paths based on the multiple node location information; wherein each of the delivery nodes in each of the candidate delivery paths is delivered by at least one of a truck and a drone; Obtaining the delivery time window of each delivery node, and for each candidate delivery path, recursively determining the target delivery time of each delivery node from the first delivery time when the truck arrives at the delivery node, the second delivery time when the drone arrives at the delivery node, and the delivery time window; Determine the service delay risk index of the corresponding candidate delivery path based on the target delivery time of the plurality of delivery nodes and the corresponding plurality of delivery time windows; Inputting the service delay risk index into the target model to obtain a model to be solved; The model to be solved is used to solve the minimized delivery delay risk of each candidate delivery path to obtain a target delivery path for collaborative delivery.

2. The collaborative delivery path planning method based on trucks and drones according to claim 1 is characterized in that: The step of inputting the service delay risk index into the target model to obtain a model to be solved includes: When the number of historical samples of path planning exceeds a first threshold, the target model is a first planning model constructed based on empirical distribution; then the service delay risk index is input into the target model to obtain a model to be solved, including: inputting the service delay risk index into the first planning model to obtain a first model to be solved; When the number of historical samples of the path planning does not exceed the first threshold, the target model is a second planning model constructed based on an uncertain set; then the service delay risk index is input into the target model to obtain a model to be solved, including: inputting the service delay risk index into the second planning model to obtain a second model to be solved.

3. The collaborative delivery path planning method based on trucks and drones according to claim 2 is characterized in that: The first planning model is constructed in the following way: Obtaining a plurality of historical samples, and determining a plurality of historical delivery times of the plurality of historical samples; Building a first initial model based on the average of multiple historical delivery times and the multiple candidate delivery routes; Constructing a first constraint model based on the historical delivery device corresponding to any historical delivery node of each of the historical samples, the delivery sequence between the historical delivery nodes, and the preset value range of the historical delivery time of each of the historical delivery nodes; A first planning model is constructed based on the first constraint model and the first initial model.

4. The collaborative delivery path planning method based on trucks and drones according to claim 3 is characterized in that: The model to be solved is a first model to be solved; solving the minimized delivery delay risk of each candidate delivery path for the model to be solved to obtain a target delivery path for collaborative delivery includes: Determining an initial first delivery path from the plurality of candidate delivery paths; Decomposing a first linear programming model corresponding to a service delay risk index from the first model to be solved based on the initial first delivery path; Obtaining a first dual model of the first linear programming model, and solving the first dual model to obtain a second delivery path and at least one first constraint for path planning; Based on the at least one first constraint, a new first delivery path is repeatedly determined from the multiple candidate delivery paths, based on the first delivery path, a first linear programming model corresponding to the service delay risk index is decomposed from the first model to be solved, a first dual model of the first linear programming model is obtained, and the first dual model is solved to obtain a new second delivery path and at least one new first constraint for path planning, and the new first delivery path is compared with the new second delivery path until the comparison results of the new first delivery path and the new second delivery path are consistent, then the determination of the new first delivery path is stopped, and the current new first delivery path is determined as the target delivery path.

5. The collaborative delivery path planning method based on trucks and drones according to claim 2 is characterized in that: The second planning model is constructed in the following way: Acquire the time fluctuation range of each of the delivery nodes, and construct a time fluctuation function based on the time fluctuation ranges of the plurality of delivery nodes; Building a second initial model based on the time fluctuation function and multiple candidate delivery routes; Constructing a second constraint model based on the delivery order between the delivery nodes in each of the candidate delivery paths and the preset value range of the delivery time of each of the delivery nodes; A second planning model is constructed according to the second initial model and the second constraint model.

6. The collaborative delivery path planning method based on trucks and drones according to claim 5 is characterized in that: The model to be solved is a second model to be solved; solving the minimized delivery delay risk of each candidate delivery path for the model to be solved to obtain a target delivery path for collaborative delivery includes: Performing a dual transformation on the second model to be solved to obtain a second dual model; Determining an initial third delivery path from the plurality of candidate delivery paths; Decomposing a second linear programming model corresponding to a service delay risk index from the second dual model based on the initial third delivery path, and solving the second linear programming model to obtain a fourth delivery path and at least one second constraint for path planning; Repeatingly determining a new third delivery path from the multiple candidate delivery paths based on the at least one second constraint, decomposing a second linear programming model corresponding to a service delay risk index from the second model to be solved based on the third delivery path, acquiring a second linear programming model of the second linear programming model, solving the second linear programming model to obtain a new fourth delivery path and at least one new second constraint for path planning, and comparing the new third delivery path with the new fourth delivery path until the comparison results of the new third delivery path and the new fourth delivery path are consistent, stopping determining the new third delivery path, and determining the current new third delivery path as the target delivery path.

7. The method for collaborative delivery path planning based on trucks and drones according to claim 5 is characterized in that: The obtaining of the time fluctuation range of each of the delivery nodes comprises: Obtaining the nominal value and the maximum deviation value of the truck and the drone at each delivery node respectively; The time fluctuation range of each of the delivery nodes is determined based on the nominal value and the maximized deviation value.

8. A collaborative delivery path planning device based on trucks and drones, characterized in that: The device comprises: A generation module, used to obtain multiple node location information corresponding to multiple delivery nodes that need to be coordinated for delivery, and generate multiple candidate delivery paths based on the multiple node location information; wherein each of the delivery nodes in each of the candidate delivery paths is delivered by at least one of a truck and a drone; an acquisition module, configured to acquire a delivery time window of each delivery node, and for each candidate delivery route, determine a target delivery time of each delivery node from a first delivery time when a truck arrives at the delivery node, a second delivery time when a drone arrives at the delivery node, and the delivery time window in a recursive manner; A determination module, configured to determine a service delay risk index of a corresponding candidate delivery path based on target delivery times of a plurality of delivery nodes and a corresponding plurality of delivery time windows; An input module, used for inputting the service delay risk index into a target model to obtain a model to be solved; The solution module is used to solve the minimized delivery delay risk of each candidate delivery path in the model to be solved, and obtain the target delivery path of collaborative delivery.

9. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the collaborative delivery path planning method based on trucks and drones as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the collaborative delivery path planning method based on trucks and drones described in any one of claims 1 to 7 is implemented.