Truck-unmanned aerial vehicle combined transportation scheduling method and system based on disaster relief scene

Abstract

The invention relates to a truck-unmanned aerial vehicle combined transportation scheduling method and system based on a disaster relief scene in the technical field of transportation scheduling. According to the truck-unmanned aerial vehicle combined transportation scheduling method, H trucks and H unmanned aerial vehicles are used for cooperatively transporting materials, the h truck and the h unmanned aerial vehicle form a cooperative group, and the scheduling method of the h cooperative group comprises the steps that the number of rescue points and the positions of the rescue points are determined; according to the number of rescue points, the positions of the rescue points and constraint conditions, generating an initial truck path based on a greedy principle of the shortest distribution distance; generating an initial unmanned aerial vehicle path based on the initial truck path; and optimizing the initial truck path and the initial unmanned aerial vehicle path to obtain an optimal material transportation path. A truck-unmanned aerial vehicle collaborative transportation mechanism is innovatively constructed, the traditional efficiency bottleneck is broken through through a collaborative three-dimensional transportation system of transportation tools, and the emergency material delivery efficiency, reliability and timeliness in a complex catastrophe environment are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of transportation scheduling, in particular to a truck-unmanned aerial vehicle (UAV) combined transportation scheduling method and system based on a disaster relief scenario. BACKGROUND

[0002] Disasters are often accompanied by traffic network paralysis, rescue channel interruption and other secondary problems, which seriously hinder the rescue and disaster relief process. The demand for emergency resources such as medical supplies and life support in the disaster area shows a strong time efficiency. With the rapid development of unmanned aerial vehicle technology, unmanned aerial vehicle distribution has become a research and application hotspot. However, the uncertainty and suddenness of disasters, the harsh environment of disaster areas, and the urgency of disaster relief make the distribution path problem a difficult problem to be solved. In order to complete the rescue goal within an effective time, three core challenges are mainly faced: insufficient coordination of transportation tools, defects of path planning algorithm and lag of disaster response.

[0003] Firstly, the inefficiency of single transportation tool is prominent. The traditional rescue mode relies on single tool of truck or unmanned aerial vehicle, which leads to low efficiency of resource utilization. The truck is blocked by the damaged road in the disaster area, and the transportation time is doubled. Although the unmanned aerial vehicle has the advantage of flexibility, it is limited by the endurance and payload, and the single machine distribution scale cannot meet the large-scale material demand. More importantly, there is a lack of coordination mechanism between transportation tools, and the repeated path and resource waste further aggravate the efficiency loss.

[0004] Secondly, the lag of disaster response has a profound impact on rescue work. The traditional system cannot adjust the emergency plan in time when the disaster occurs, especially when the disaster mutates, the response time delay of path re-planning is significant, which leads to the loss of a large number of best disposal windows of rescue demand. Due to the long response time of path re-planning, many key rescue tasks cannot be executed within the golden rescue window period, missing the most critical opportunity. Moreover, the shortcomings of the existing system in path planning lead to large differences in the demand satisfaction of each node, and there is significant lag. These problems expose the short board of the technical system in actual disaster rescue, which cannot effectively meet the decision-making time efficiency requirement of the 72-hour golden rescue period, and seriously affects the rescue efficiency and the recovery process after the disaster. SUMMARY

[0005] In order to solve the technical problems of low efficiency and lag of material transportation caused by using single transportation tool in the existing disaster area transportation, the present application provides a truck-UAV combined transportation scheduling method and system based on a disaster relief scenario.

[0006] In a first aspect, the present application provides a truck-UAV joint transportation scheduling method based on a disaster relief scene, which uses H trucks and H UAVs to cooperatively transport materials; wherein the hth truck and the hth UAV form a cooperative group; h∈[1,H]; the scheduling method of the hth cooperative group comprises:

[0007] Confirming the number and location of rescue points and defining the constraints of the trucks and UAVs;

[0008] Defining an objective function: the time for the last truck or UAV to return to the final destination; generating a basic truck path based on the principle of shortest distance of distribution according to the number and location of rescue points and the constraints; separating a UAV path from the basic truck path to make the value of the objective function optimal as an initial UAV path; and transforming the basic truck path into an initial truck path after separating the initial UAV path;

[0009] Optimizing the initial truck path and the initial UAV path through an adaptive large neighborhood search algorithm to obtain an optimal material transportation path of the hth cooperative group;

[0010] Scheduling the trucks and UAVs to travel along the optimal material transportation path to minimize the total transportation time.

[0011] In a second aspect, the present application further provides a truck-UAV joint transportation scheduling system based on a disaster relief scene, which uses the truck-UAV joint transportation scheduling method based on a disaster relief scene in the first aspect. The system comprises an information collection module, an initial path planning module, and a path optimization module.

[0012] The information collection module is configured to confirm the number and location of rescue points and define the constraints of the trucks and UAVs; the initial path planning module is configured to define an objective function: the time for the last truck or UAV to return to the final destination; generate a basic truck path based on the principle of shortest distance of distribution according to the number and location of rescue points and the constraints; separate a UAV path from the basic truck path to make the value of the objective function optimal as an initial UAV path; and transform the basic truck path into an initial truck path after separating the initial UAV path; the path optimization module is configured to optimize the initial truck path and the initial UAV path through an adaptive large neighborhood search algorithm to obtain an optimal material transportation path; and schedule the trucks and UAVs to travel along the optimal material transportation path to minimize the total transportation time.

[0013] The present application has the following beneficial effects:

[0014] 1. The truck-UAV cooperative transportation mechanism is innovatively constructed, and the traditional efficiency bottleneck is broken through the cooperative three-dimensional transportation system of the transportation tools. In the truck-UAV cooperative transportation mechanism, the truck plays the advantage of load capacity and transports the materials in batches to the front of the disaster area. When the road between the disaster points is blocked or in order to speed up the transportation efficiency, the UAV takes off from the truck to complete the independent transportation operation. Relying on the flexibility and speed advantage of the UAV, it can directly fly from the current disaster point to the next series of disaster points, and carry out accurate delivery, directly penetrate the road damage area and reach the demand point, which significantly improves the emergency material delivery efficiency, reliability and timeliness in complex disaster environment.

[0015] 2. A series of different targeted destruction and repair operators are innovatively constructed for the path planning problem in the dynamic disaster environment. These operators comprehensively consider the multiple restrictions of path, mileage and capacity, realize the synchronous planning of truck and UAV path, and aim to minimize the transportation time, thereby providing a powerful solution framework for rapid response to disaster relief. This global optimization strategy makes the resource scheduling more reasonable and efficient, while meeting the disaster relief demand, and maximizes the distribution efficiency.

[0016] 3. The adaptive large domain algorithm is improved. Through the core mechanism of phased cooperative optimization, the solving ability is significantly improved, and the characteristics and constraint conditions of the problem are further considered, and reasonable algorithm parameters and stopping criteria are set, so as to ensure the stability and convergence of the algorithm. In addition, the improved adaptive large domain algorithm adopts an adaptive probability regulation mechanism, and uses a dynamic weight updating mechanism to evaluate the performance of the operator in real time. This not only avoids the low efficiency problem caused by the fixed number of destruction tasks in the traditional large neighborhood search, but also optimizes the truck path reconstruction and other multi-dimensional constraints through multi-operator cooperative parallel optimization. Finally, the balance optimization of minimizing the total distribution time and prioritizing high emergency demand is realized in the complex logistics scene. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 It is a flow chart of the truck-UAV joint transportation scheduling method based on the disaster relief scene;

[0018] Figure 2 It is a schematic diagram of the truck-UAV joint transportation scheduling based on the disaster relief scene;

[0019] Figure 3 It is a schematic diagram of the initial truck path and the initial UAV path;

[0020] Figure 4 It is a schematic diagram of the random destruction of the UAV take-off node by the random destruction operator;

[0021] Figure 5Fig. 1 is a schematic diagram of a random destruction of a UAV landing node by a random destruction operator;

[0022] Figure 6 Fig. 2 is a schematic diagram of a random destruction of a UAV takeoff node and a UAV landing node by a random destruction operator;

[0023] Figure 7 Fig. 3 is a schematic diagram of a random destruction of a UAV landing node by a sequential point destruction operator;

[0024] Figure 8 Fig. 4 is a schematic diagram of a destruction of a node by a distance dependent point destruction operator;

[0025] Figure 9 Fig. 5 is a schematic diagram of a destruction of a UAV individual service node by a UAV itinerary operator;

[0026] Figure 10 Fig. 6 is a schematic diagram of a repair of a node by a random repair operator;

[0027] Figure 11 Fig. 7 is a schematic diagram of a repair of a node by a greedy repair operator;

[0028] Figure 12 Fig. 8 is a schematic diagram of a repair of a node by an exchange repair operator;

[0029] Figure 13 Fig. 9 is a schematic diagram of a repair of a node by an optimal repair operator;

[0030] Figure 14 Fig. 10 is a schematic diagram of a takeoff and landing search strategy;

[0031] Figure 15 Fig. 11 is a schematic diagram of a path change search strategy. DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the present application will be described clearly and completely below. Obviously, the described embodiments are only some of the embodiments of the present application, but not all of them. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0033] It should be noted that when a component is referred to as being "mounted on" another component, it can be directly on the other component or there can be a middle component. When a component is referred to as being "disposed on" another component, it can be directly disposed on the other component or there can be a middle component. When a component is referred to as being "fixed on" another component, it can be directly fixed on the other component or there can be a middle component.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of the application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0035] The present application is directed to natural disaster emergency material distribution scenarios, and is particularly suitable for disaster environments with severely damaged roads and complex terrain, such as typical severe natural disasters such as earthquakes, landslides, mudslides, floods, etc. In view of the urgent needs of the 72-hour golden rescue period in the early stage of the disaster, the problem of multi-mode collaborative transportation is focused on. The main idea is to start from the front warehouse with a truck carrying disaster relief materials and a drone, and go to a series of disaster relief points. While the truck is distributing materials, the drone is allowed to carry materials to other disaster areas to complete the rescue mission, effectively overcoming the "last mile" obstacle. Unlike traditional emergency distribution of materials in disaster areas, the present application uses a truck-drones collaborative distribution of emergency materials to ensure the timeliness and accuracy of emergency material transportation. For this purpose, a truck-drones joint transportation scheduling method based on disaster relief scenarios is proposed, as shown in Figure 1 , which includes the following steps:

[0036] S1, first confirm the number of rescue points and the location of the rescue points. It is assumed that there are multiple disaster relief points in a certain disaster area, the material demand of the disaster relief points is known, the location of the disaster relief points and the warehouse is known, and whether the truck can pass between certain disaster relief points is known. Because of the road grade, road damage or deviation from the truck rescue path, only using a truck as a transportation tool cannot guarantee fast rescue for some rescue points, so the truck-drones collaborative transportation mode is used. As shown in Figure 2 , in order to ensure that the disaster relief points can be rescued quickly, H trucks (one truck can carry only one drone) are dispatched according to the number of rescue points and the severity of the disaster to carry out disaster relief. The hth truck and the hth drone form a collaborative group, h∈[1,H]. The hth collaborative group is taken as an example for description below. The truck departs from the warehouse starting point (with sufficient materials to meet the demand of all disaster relief points) carrying materials and a drone. The truck needs to carry a certain amount of materials to meet the material demand on the rescue path. The drone can access multiple points after taking off from the truck, but it needs to meet the weight and power constraints. In order to improve the efficiency and accuracy of material distribution, in addition to basic factors such as transportation cost and material demand, the following core constraint conditions are strictly considered and defined:

[0037] 1. Path constraint: the path design of the truck and the drone needs to meet the inflow-outflow balance and the start-end point constraint to ensure that each disaster relief point is accessed only once by the truck or the drone.

[0038] 2. Capacity constraint: considering the weight limit of the UAV, ensure that the weight of the supplies for each take-off is within the safe range. In addition, allow the UAV to take off and land multiple times to fully utilize its speed advantage and improve transportation efficiency.

[0039] 3. Mileage constraint: considering the limited battery of the UAV, ensure that the UAV still has enough power to reach the disaster point after each flight. Assuming that the UAV can replace the battery on the truck before take-off, ensure the continuity and safety of the flight.

[0040] In addition, key decision elements (such as truck path sequence, UAV task allocation, meeting point location and time of truck and UAV, etc.) are defined, and a clear core goal is set to minimize the total time from the warehouse to the completion of service to all disaster points (i.e. the time when the truck or UAV finally arrives at the end point), ensuring that the most critical needs are met first. By precisely designing the above constraints and objective functions (in this invention, the objective function is used to calculate the time spent on the path, i.e. the smaller the objective function value, the better the solution. In the case of path determination, the objective function value can be calculated based on the driving speed of the truck and the UAV), the technical solution is ensured to be realistic, feasible and targeted to the core goals of emergency disaster relief (fast response and priority protection of critical needs).

[0041] S2. In the hth coordination group, based on the number of rescue points, the location of the rescue points, and the constraints, the basic truck path is generated based on the greedy principle of shortest distance distribution. Specifically, each truck randomly selects the first rescue point from the starting point, finds the nearest next node as the next rescue point of the truck path based on the current node reached by the truck, and completes the construction of the basic truck path until all rescue points are accessed by the truck. The same principle is used to generate the initial truck path for other coordination groups, which needs to avoid overlapping with the rescue points in the hth coordination group, so as to generate the initial truck path for the H coordination groups facing the entire disaster area.

[0042] Due to the limitations of road grade, road damage or high detour cost (such as remote disaster points) on some points on the rescue path of each truck, the UAV needs to compensate for the rescue. After the basic truck path of each truck is determined, the UAV path is separated from the basic truck path, and the main separation methods include:

[0043] The zth target function value optimal candidate path is constructed based on the zth node on the base truck path. The target function values of the Z candidate paths are compared, and the candidate path corresponding to the optimal target function value is taken as the initial UAV path. z∈[1,Z]. Wherein, the construction method of the candidate path includes: the UAV journey includes 1 take-off node and 1 landing node. In a rescue mission, the UAV may have several journeys. All UAV journeys in a rescue mission are combined to define the UAV path. The take-off node of the e-th UAV journey is the landing node of the e-1-th UAV journey. The nodes following the take-off node of the e-th UAV journey are taken in turn as the landing nodes of the e-th UAV journey to find the e-th UAV journey with the optimal target function value. The e UAV journeys are taken as a candidate path. e∈[1,E]. When e=1, the take-off node is the zth node on the base truck path. D represents the maximum number of UAV journeys when the zth node on the base truck path is taken as the take-off node.

[0044] In a popular way, the separation step of separating the UAV path from the base truck path includes:

[0045] S21, take each node in the truck path as the take-off node of the UAV in turn. For each take-off node, perform the following process until all nodes are traversed:

[0046] S211, under the premise of meeting the UAV endurance range limit and the material load constraint, start from the current take-off node and traverse all nodes after the take-off node.

[0047] S212, select a node that optimizes the target function value of the UAV path as the landing node of the UAV to form the first UAV journey.

[0048] S213, take the landing node as the take-off node of the next UAV journey. Repeat S211-S213 until all nodes in the truck path are traversed to form a complete UAV path composed of one or more UAV journeys.

[0049] S22, after traversing all nodes in the truck path as the take-off node of the first UAV journey, obtain several complete UAV paths. Select the optimal solution of the target function value from all UAV paths as the initial UAV path.

[0050] The base truck path is transformed into an initial truck path after the initial UAV path is separated. After the initial truck path and the initial UAV path are constructed, the service mode of each node on the path can be known. As shown in Table 1:

[0051] Table 1: representation of initial truck path and initial UAV path.

[0052]

[0053] In Table 1, r i s denotes the node, denotes the service mode of the node. The initial truck path is: 0-4-3-2-8. The initial drone path has two paths, which are: 4-1-7-3, 3-5-6-2 respectively. When denotes that the corresponding node service mode is truck carrying drone service. When denotes that the corresponding node service mode is drone alone service. When denotes that the corresponding node service mode is truck alone service. The path represented by Table 1 is converted into a path graph as shown in Figure 3 .

[0054] S3, the initial truck path, the initial drone path is optimized by adaptive large neighborhood search algorithm to obtain the optimal material transportation path. In this embodiment, the adaptive large neighborhood search algorithm is improved, which includes the following steps:

[0055] A set X containing a plurality of destruction operators, a set Y containing a plurality of repair operators, and two local search strategies for local search of nodes are constructed. The search probability of the local search strategy is set, and the selection probability of each destruction operator and repair operator is updated once in each calculation period by an adaptive probability control method. B iterations constitute a calculation period. In the tthiteration, the new solution S n，t-1 is taken as the initial solution. According to the selection probability, the destruction operator in X is selected to obtain the solution S d after the initial solution is destroyed. According to the selection probability, the repair operator in Y is selected to obtain the solution S d after the repair of S r , and it is taken as the new solution S n，t . According to the search probability, it is determined whether to perform local search. If local search is performed, the local search strategy is first used to obtain the solution S r after local search S l ; then S l and S n，t-1 are compared, and the solution with better objective function value is selected as the new solution S n，t . Whether S n，t is accepted or not is determined based on the objective function value using the simulated annealing method. If accepted, S n，t is taken as the initial solution of the t+1thiteration for iteration. If not accepted, S n，t-1 is taken as the initial solution of the t+1thiteration for iteration. Until the maximum number of iterations is reached or no improvement is made for K consecutive iterations, the optimal material transportation path is output.

[0056] Among them, in terms of destruction, repair operator, for the path planning problem under dynamic disaster environment, the application also innovatively constructs a series of destruction operators with different pertinence, repair operators. These operators comprehensively consider the multiple restrictions of path, mileage and capacity, realize the synchronous planning of truck and unmanned aerial vehicle path, aim to minimize the transportation time, and provide a powerful solution framework for disaster relief rapid response.

[0057] The operators used and improved in the embodiment are introduced one by one as follows.

[0058] I. Destruction operator

[0059] The destruction operator refers to the operation of removing nodes (rescue points) from the path in different ways, wherein the nodes are divided into five types according to whether they are transported by a truck or an unmanned aerial vehicle: truck carrying unmanned aerial vehicle service node, unmanned aerial vehicle take-off node, truck alone service node, unmanned aerial vehicle alone service node, and unmanned aerial vehicle landing node.

[0060] (1) Random destruction operator

[0061] The random destruction operator randomly selects q nodes from the path for destruction operation. For the three types of nodes, truck carrying unmanned aerial vehicle service node, truck alone service node, and unmanned aerial vehicle alone service node, direct destruction is performed. For unmanned aerial vehicle take-off node and unmanned aerial vehicle landing node, the unmanned aerial vehicle take-off node and unmanned aerial vehicle landing node need to be changed, which can be divided into the following three cases:

[0062] a) When the unmanned aerial vehicle take-off node is destroyed, iterate through the node set A to find an optimal node as the new unmanned aerial vehicle take-off node to make the target function value optimal after destruction. Among them, set A is composed of nodes in the truck path before reaching the unmanned aerial vehicle landing node in the current unmanned aerial vehicle path. If there are other unmanned aerial vehicle paths before the current unmanned aerial vehicle path, set A is composed of nodes between the unmanned aerial vehicle landing node in the current unmanned aerial vehicle path and the unmanned aerial vehicle landing node in the previous unmanned aerial vehicle path. The destruction process is as shown in Figure 4 .

[0063] b) When the unmanned aerial vehicle landing node is destroyed, iterate through the node set B to find an optimal node as the new unmanned aerial vehicle landing node to make the target function value optimal after destruction. Among them, set B is composed of nodes in the truck path after reaching the unmanned aerial vehicle take-off node in the current unmanned aerial vehicle path. If there are other unmanned aerial vehicle paths after the current unmanned aerial vehicle path, set B is composed of nodes between the unmanned aerial vehicle take-off node in the current unmanned aerial vehicle path and the unmanned aerial vehicle take-off node in the next unmanned aerial vehicle path. The destruction process is as shown in Figure 5 .

[0064] c) When the destroyed node is both the UAV landing node in the previous UAV path and the UAV take-off node in the next UAV path, then the new UAV take-off node and the new UAV landing node are selected from the set C to be the nodes that make the objective function value optimal. The set C is the nodes between the UAV take-off node in the previous UAV path and the UAV landing node in the next UAV path. It should be noted that the arrival time of the new UAV landing node needs to be between the new UAV take-off nodes. In addition, if the new UAV take-off node or the UAV landing node found makes the UAV path not satisfy the mileage constraint, then the second best node is selected. If no node serves as the new UAV take-off or landing node, then the destruction of this node is skipped. The destruction process is as shown in Figure 6 The random destruction operator has the advantages of strong randomness and fast running speed, although the probability of appearing a poor solution is high, but it helps the diversity of search and jumps out of the local optimal solution.

[0065] (2) Continuous point destruction operator

[0066] The continuous point destruction operator introduces the disturbance characteristic, which mainly consists of two steps: the first step is to randomly select the nth node in the path for destruction and as the destruction starting point. The second step is to select m nodes on both ends of the path to form a set D, and then randomly select q (q (≤ m) nodes in the set D for destruction operation. The destruction process of the five types of nodes is the same as that of the random destruction operator. For example, refer to Figure 7 , the first node is randomly selected as node 7, which is a truck path node, and the number of adjacent nodes m = 3. The set D = {4, 7, 3}. The number of destroyed nodes q = 2. The destroyed node 3 is randomly selected again, then the truck path destruction node set is {7, 3}, and the UAV path has no destroyed node.

[0067] (3) Distance-related point destruction operator

[0068] The operation steps of the distance-related point destruction operator are similar to those of the continuous point destruction operator. The continuous point destruction operator first randomly selects the nth node in the path for destruction and as the destruction starting point, and then destroys the q nodes closest to the destruction starting point. The destruction process of the five types of nodes is the same as that of the random destruction operator. As shown in Figure 8 , the first destroyed node is node 4, and the number of destroyed nodes q = 2. The node closest to node 4 in geographical position is node 5, so nodes 4 and 5 are destroyed, and the truck path destruction node set is {4}, and the UAV path destruction node set is {5}.

[0069] (4) UAV journey operator

[0070] The purpose of the UAV tour operator is to reduce the time of a single UAV trip, randomly select a UAV path for the current solution existing UAV path, and destroy all UAV single service nodes in it. As shown in Figure 9 , the current solution exists twice UAV tour is respectively tour 1: 0-4-1-3 and tour 2: 6-3-8. Randomly select to destroy tour 1, then destroy UAV single service nodes 4, 1 in tour 1. The UAV path destruction node set is {4, 1}, and the truck path has no destruction node set.

[0071] II. Repair operator

[0072] According to the truck path destruction node set and the UAV path destruction node set obtained by the destruction operator, the existing path is repaired. The repair operator is divided into two repair operators according to whether the original node service mode (i.e. change) is considered in the repair. One of them is the original node service mode immutable repair, which uses the truck path destruction node set for truck path repair and the UAV path destruction node set for UAV path repair, and then updates the node service mode after the repair is completed. This repair type is defined as the path immutable type. The other is that the original node service mode can change (i.e. change), in which the truck path destruction node set can be used for UAV path repair, and the UAV path destruction node set can also be used for truck path repair, and the service mode of the node does not change before and after the destruction repair. This repair type is defined as the path variable type. In this embodiment, the path immutable type includes: random repair operator, greedy repair operator. The path variable type includes: exchange repair operator, optimal repair operator.

[0073] (1) Random repair operator

[0074] The random repair operator refers to determining the repair position of the node based on the objective function value of the repaired solution. First, put the truck path destruction node set and the UAV path destruction node set into a set of nodes to be repaired. The nth destruction node in the set of nodes to be repaired is selected by random traversal. Determine whether the nth destruction node is a truck path destruction node or a UAV path destruction node, and traverse the path it belongs to to find a position for insertion that optimizes the objective function value. Then remove the successfully repaired destruction node from the set of nodes to be repaired. Repeat the above steps until the set of nodes to be repaired is empty. It is worth noting that the repaired UAV path satisfies the mileage constraint of the UAV. As shown in Figure 10As shown, the truck path broken node set is {4}, the drone path broken node set is {5}, and the to-be-repaired node set is {4, 5}. The first to-be-repaired node is randomly selected as 4, which belongs to the truck path broken node set, and 4 is inserted into all possible positions of the truck path to find the optimal insertion position between nodes 7 and 3, and node 4 is removed from the to-be-repaired node set. At this time, the to-be-repaired node set is {5}, and the to-be-repaired node 5 is selected, which belongs to the drone path broken node set, and 5 is inserted into all possible insertion positions (mileage constraint) of the drone to find the optimal insertion position between nodes 1 and 3.

[0075] (2) Greedy repair operator

[0076] The greedy repair operator is similar to the random repair operator. First, the truck path broken node set and the drone path broken node set are put into a to-be-repaired node set, and then the to-be-repaired node set is traversed, and the objective function value of inserting all broken nodes in the to-be-repaired node set into the optimal position is calculated. By comparing the objective function values of all broken nodes, the corresponding broken nodes are inserted into the position with the optimal objective function value from good to bad based on the objective function value, and this process is repeated until all broken nodes are inserted. If the position of the drone repair does not meet the mileage constraint, the second best repair position is found to repair the drone node. As shown, Figure 11 As shown, there is no broken node set on the truck path, and the drone path broken node set is {1, 9}, and the to-be-repaired node set is {1, 9}. Based on the repair node 1 of the current solution, the optimal position is between nodes 7 and 5, and the objective function value v1 = 10. Based on the repair node 9 of the current solution, the optimal position is between nodes 5 and 3, and the objective function value v2 = 8. A decision is made, and node 9 is repaired first because the objective function value v2 is smaller (i.e., the total transportation time is smaller, which is better). After repairing the drone journey: 7-5-9-3. At this time, the to-be-repaired node set is {1}, and based on the repaired solution after the first repair, the optimal position is between nodes 5 and 9. The to-be-repaired node set is empty, and the repair is completed. At this time, the drone journey is: 7-5-1-9-3.

[0077] (3) Exchange repair operator

[0078] The exchange repair operator realizes the expansion of the search range by simultaneously selecting the UAV destruction node and the truck destruction node for exchange. First, the original positions and paths of all destruction nodes are recorded (the destruction positions can be found by comparing the post-destruction path with the pre-destruction path). Then, the destroyed truck nodes and UAV nodes are combined to form a set of nodes to be repaired. The set of nodes to be repaired is then randomly inserted into the original positions after changing the paths, and the service mode at the original positions is kept unchanged. The insertion order of all destruction nodes is traversed to find the optimal arrangement order of the path objective function for repair. The exchange repair process is as shown in Figure 12 .

[0079] (4) Optimal repair operator

[0080] The optimal repair operator is similar to the greedy repair operator, but the difference lies in that the node service mode can be changed. The UAV destruction node can be repaired to the truck path, and the truck destruction node can also be repaired to the UAV path. The repair process is as shown in Figure 13 .

[0081] The search probability for determining whether to perform local search and the probability for selecting which local search strategy can be customized. The purpose of setting local search is to find a better solution by using the take-off and landing search strategy and the path change search strategy based on the current UAV path and truck path solution. The destruction repair operator used in the take-off and landing search strategy is different from the above destruction repair operator in that the order of the UAV-only service nodes does not change, and the UAV path take-off and landing node with the optimal objective function value is tried to be found for the UAV path in the current solution based on the current truck path node. In addition, since the above destruction repair operator lacks an operator that increases the number of UAV paths, the path change search strategy is set to try to find a solution with a better objective function value by reducing or increasing the number of UAV paths.

[0082] (1) Take-off and landing search strategy

[0083] The purpose of the take-off and landing search strategy is to find the optimal take-off and landing node for the existing UAV path on the existing truck path. It also consists of two stages of destruction and repair. The destruction stage is to destroy all UAV-only service nodes in the order of the nodes in the UAV path, but the number and order of the visited disaster points of each UAV path do not change (i.e., the entire UAV path is destroyed from the truck path). The repair stage is to traverse the take-off nodes and landing nodes on the UAV path and insert them into the nodes on the truck path, and the solution with the optimal objective function value after insertion is taken as the optimal take-off and landing node. The destruction and repair process is as shown in Figure 14 .

[0084] (2) Path change search strategy

[0085] The purpose of the path change search strategy is to change the service mode of the node, so as to change the unmanned aerial vehicle service node into the truck service node and change the truck service node into the unmanned aerial vehicle service node, thereby expanding the search space and finding a better solution. The specific steps are as follows: first, randomly destroy the nth node, and then randomly insert the nth node into any path. All nodes are traversed to find the optimal position of the target function value. The destruction and repair process is as shown in Figure 15 .

[0086] In the adaptive large neighborhood search algorithm used in the embodiment, an adaptive probability regulation method is designed to dynamically adjust the selection probability of each operator in each calculation period, so as to realize self-learning and adjustment of the adaptive large neighborhood search algorithm. Specifically, the adaptive probability regulation method includes two stages: a score acquisition stage and an update weight stage.

[0087] (1) Score acquisition stage

[0088] In the Tth calculation period, the destruction operator and the repair operator used are scored: in the tth iteration, if S n，t is better than S n，t-1 , the corresponding destruction operator and repair operator score +3. If S n，t is not only better than S n，t-1 , but also better than the known new solution, the score +5. If S n，t is worse than S n，t-1 , but is still accepted (for example, in simulated annealing or other heuristic mechanisms), the score +1. If S n，t is worse than S n，t-1 , but is not accepted, the score +0. Through the above scoring criteria, the system can effectively evaluate the contribution of each operator to the search quality, and balance exploration and utilization.

[0089] (2) Update weight stage

[0090] In the update weight stage, the system updates the selection probability of each operator according to the score accumulated by each operator in the past calculation period. The new operator weight is calculated by using the moving average method Then, the of all operators is converted into the selection probability p i by using the Softmax function, so as to realize dynamic adjustment of the operator priority, and the calculation formula is as follows:

[0091]

[0092] wherein, represents the score of the ith destruction operator or repair operator in the Tth calculation period. α ∈ (0, 1) represents the smoothing coefficient of weight update. p iPi,j represents the selection probability of the ith destruction operator or repair operator. N represents the number of destruction operators or repair operators. j represents the traversal index of the destruction operator or repair operator. Pj represents the weight of the jth destruction operator or repair operator in the Tth calculation period. The adaptive probability regulation method has good adaptability and dynamic adjustment capability, and is suitable for scenarios of multi-operator collaborative optimization, such as combinatorial optimization, evolutionary computation or large-scale search problems.

[0093] S4, dispatch the truck and the UAV to travel along the optimal material transportation path to minimize the total transportation time.

[0094] In another embodiment, a truck-UAV joint transportation scheduling system based on a disaster relief scenario is also proposed, which uses the truck-UAV joint transportation scheduling method based on a disaster relief scenario in the above embodiment. The truck-UAV joint transportation scheduling system based on a disaster relief scenario includes an information collection module, an initial path planning module, and a path optimization module.

[0095] The information collection module is used to confirm the number of rescue points and the positions of the rescue points, and define the constraint conditions of the truck and the UAV. The initial path planning module is used to define the objective function: the time when the last one of the truck and the UAV returns to the final destination. According to the number of rescue points, the positions of the rescue points, and the constraint conditions, a basic truck path is generated based on the principle of shortest distance for distribution. A UAV path that makes the value of the objective function optimal is separated from the basic truck path as an initial UAV path. The basic truck path is transformed into an initial truck path after the initial UAV path is separated. The path optimization module is used to optimize the initial truck path and the initial UAV path through an adaptive large neighborhood search algorithm to obtain an optimal material transportation path; the truck and the UAV are dispatched to travel along the optimal material transportation path to minimize the total transportation time.

[0096] The technical features of the above-described embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described, but as long as the combinations of the technical features do not contradict, they should be considered within the scope of the present disclosure.

[0097] The above-described embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the protection scope of the present patent should be subject to the appended claims.

Claims

1. A truck-drone joint transportation scheduling method based on disaster relief scenarios, characterized in that, It uses H trucks and H drones to transport goods collaboratively; among them, the h-th truck and the h-th drone form a collaborative group; h∈[1,H]; the scheduling method of the h-th collaborative group includes: Confirm the number and location of rescue points, and define the constraints for trucks and drones; Define the objective function as the time when the last truck or drone returns to its final destination; generate a basic truck path based on the principle of shortest delivery distance, considering the number and location of rescue points and constraints; extract a drone path from the basic truck path that optimizes the objective function value as the initial drone path; transform the basic truck path into the initial truck path after separating the initial drone path. The initial truck route and initial drone route are optimized using an adaptive large neighborhood search algorithm to obtain the optimal material transportation route for the h-th collaborative group. Dispatch trucks and drones along the optimal material transport route to minimize total transport time.

2. The truck-drone joint transportation scheduling method based on disaster relief scenarios according to claim 1, characterized in that, The constraints include: path constraints, capacity constraints, and mileage constraints; among which, the path constraint is used to ensure that each disaster relief point is visited only once by a truck or drone; the capacity constraint is used to constrain the payload limit of the drone; and the mileage constraint is used to constrain the flight range of the drone. And / or, methods for separating the initial drone path from the base truck path include: Construct the candidate path with the optimal objective function value based on the z-th node on the basic truck path; compare the objective function values ​​of the Z candidate paths, and use the candidate path corresponding to the optimal objective function value as the initial UAV path; z∈[1,Z]; The methods for constructing candidate paths include: The drone journey includes one takeoff node and one landing node; the takeoff node of the e-th drone journey is the landing node of the (e-1)-th drone journey; the nodes following the takeoff node of the e-th drone journey are successively used as the landing nodes of the e-th drone journey, in order to find the e-th drone journey with the optimal objective function value; the E-th drone journey is used as a candidate path; e∈[1,E]; when e=1, the takeoff node is the z-th node on the basic truck path; D represents the maximum number of drone journeys when the z-th node on the basic truck path is the takeoff node.

3. The truck-drone joint transportation scheduling method based on disaster relief scenarios according to claim 1, characterized in that, Adaptive large neighborhood search algorithms include: Construct a set X containing multiple destruction operators, a set Y containing multiple repair operators, and two local search strategies for local search of nodes; Set the search probability for the local search strategy, and update the selection probability of each destruction operator and repair operator once in each calculation cycle through an adaptive probability adjustment method; B iterations constitute one calculation cycle; In the t-th iteration, the new solution S from the (t-1)-th iteration will be... n，t-1 As the initial solution; select the destruction operator in X according to the selection probability to obtain the solution S after the initial solution is destroyed. d ; Select the repair operator in Y based on the selection probability to obtain S. d The repaired solution r And as a new solution S n，t ; Whether to perform a local search is determined based on the search probability; if a local search is performed, a local search strategy is first used to obtain S. r Solution S after local search l Next, compare S. l and S n，t-1 And select the solution with the better objective function value as the new solution S. n，t ; Based on the objective function value, simulated annealing method is used to determine S. n，t Whether it is accepted; if accepted, then S n，t Use it as the initial solution for iteration t+1; if not accepted, then S n，t-1 The optimal material transportation route is output after iterating from the initial solution of the (t+1)th iteration until the maximum number of iterations is reached or after K consecutive iterations without improvement.

4. The truck-drone joint transportation scheduling method based on disaster relief scenarios according to claim 3, characterized in that, There are four destruction operators: random destruction operator, continuous point destruction operator, distance-dependent point destruction operator, and UAV journey operator. Among them, the random destruction operator randomly selects q nodes from the path to perform destruction operations; The continuous point destruction operator first randomly selects the nth node in the path to destroy and uses it as the destruction starting point. Then, it randomly selects q nodes from the m nodes adjacent to the destruction starting point and the destruction node to perform destruction operations. The distance-related point destruction operator first randomly selects the nth node in the path to destroy as the destruction starting point, and then performs destruction operations on the q nodes closest to the destruction starting point; The drone journey operator is a destructive operation performed on all individual service nodes of a drone along a random drone path.

5. The truck-drone joint transportation scheduling method based on disaster relief scenarios according to claim 4, characterized in that, Based on the service method, nodes are divided into: truck-borne drone service nodes, truck-only service nodes, drone-only service nodes, drone take-off nodes, and drone landing nodes. When the destruction operator destroys the drone take-off node or the drone landing node, the drone take-off node and the drone landing node are changed on the path.

6. The truck-drone joint transportation scheduling method based on disaster relief scenarios according to claim 5, characterized in that, Repair operators are divided into two categories based on whether they change the path of the original node: immutable path type and mutable path type. Path-immutable types include: random repair operator and greedy repair operator; path-mutable types include: swap repair operator and optimal repair operator. The random repair operator is to traverse and insert the nth damaged node into all insertable nodes on the same path, so that the nth damaged node finds the position where the objective function value is optimal after insertion; The greedy repair operator first traverses all damaged nodes and inserts them into all insertable nodes on the same path, and calculates the objective function value corresponding to each damaged node when it is inserted into the optimal position. Then, it compares the objective function values ​​of all damaged nodes and inserts the corresponding damaged nodes into the positions with the optimal objective function values ​​in descending order of objective function value, until all damaged nodes have been inserted. The swap repair operator first records the original position and path of all damaged nodes, then changes the path of all damaged nodes and randomly inserts them into their original positions, and then iterates through the insertion order of all damaged nodes to find the optimal arrangement order of the path objective function for repair. The optimal repair operator is to traverse and insert the nth damaged node into all insertable nodes on any path, so that the nth damaged node finds the position where the objective function value is optimal after insertion.

7. The truck-drone joint transportation scheduling method based on disaster relief scenarios according to claim 3, characterized in that, Local search strategies include: takeoff and landing search strategies, and path change search strategies; Among them, the take-off and landing search strategy is used to find the optimal take-off and landing nodes for the UAV on the truck path; The path change search strategy is used to expand the search space and find better solutions by changing the service methods of nodes.

8. The truck-drone joint transportation scheduling method based on disaster relief scenarios according to claim 7, characterized in that, The takeoff and landing search strategy includes: first, destroying the entire UAV path from the truck path; then, traversing and inserting the takeoff and landing nodes on the UAV path into the nodes on the truck path; and taking the solution with the optimal objective function value after insertion as the optimal takeoff and landing node. The path change search strategy includes: first, randomly destroying the nth node, then randomly inserting the nth node into any path; traversing all nodes to find the position with the optimal objective function value.

9. The truck-drone joint transportation scheduling method based on disaster relief scenarios according to claim 3, characterized in that, The adaptive probability control method includes the following steps: In the Tth calculation cycle, the destruction operator and the repair operator used are scored: In the tth iteration, if S n，t Superior to S n，t-1 Then the corresponding destruction operator and repair operator score +3; if S n，t Not only superior to S n，t-1 If the solution is better than the known new solution, the score is +5; if S n，t Inferior to S n，t-1 However, if it is still accepted, the score is increased by 1; if S n，t Inferior to S n，t-1 However, if it is not accepted, the score is +0; Calculate the weights corresponding to the destruction operator and the repair operator based on the scores. Then use the Softmax function to... Convert to selection probability p i The calculation formula is as follows: in, Let p represent the score of the i-th destruction or repair operator in the T-th calculation period, α represent the smoothing coefficient for weight updates, and p represent the score of the i-th destruction or repair operator. i Let represent the probability of choosing the i-th destruction or repair operator, N represent the number of destruction or repair operators, and j represent the traversal index of the destruction or repair operator. This represents the weight of the j-th destruction or repair operator in the T-th calculation cycle.

10. A truck-drone joint transportation dispatch system based on disaster relief scenarios, characterized in that, It uses the truck-drone joint transportation scheduling method based on disaster relief scenarios as described in any one of claims 1 to 9; It includes: The information collection module is used to confirm the number and location of rescue points, and to define the constraints for trucks and drones. The initial path planning module defines the objective function: the time when the last truck or drone returns to the final destination; it generates a basic truck path based on the principle of shortest delivery distance, according to the number and location of rescue points and constraints; it separates a drone path from the basic truck path that optimizes the objective function value as the initial drone path; and it transforms the basic truck path into the initial truck path after separating the initial drone path. The route optimization module is used to optimize the initial truck and drone routes using an adaptive large neighborhood search algorithm to obtain the optimal material transportation route; and to schedule trucks and drones to travel along the optimal material transportation route to minimize the total transportation time.

Images