A multi-aircraft multi-site delivery scheduling planning method, a storage medium and an equipment
By employing a multi-aircraft, multi-site delivery scheduling planning method and a neural network iterative sorting algorithm, the optimal scheduling problem for large-scale aircraft accessing multiple sites was solved, the access order of aircraft was optimized, and the goal of minimizing the total delivery time of materials was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HARBIN INST OF TECH
- Filing Date
- 2025-02-25
- Publication Date
- 2026-06-26
Smart Images
Figure CN120146473B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of aircraft delivery scheduling and planning technology, specifically relating to a multi-aircraft, multi-site delivery scheduling and planning method, storage medium, and equipment. Background Technology
[0002] Unmanned aerial vehicles (UAVs) are small to medium-sized unmanned aircraft with high lift-to-drag ratio gliding capabilities and large internal cargo capacity. They can replace manned aircraft to complete cargo transportation tasks. With the rapid development of the low-altitude economy and the Internet of Things, higher requirements are being placed on the timeliness and economy of air freight.
[0003] However, existing collaborative planning methods for long-range aircraft mainly consider time or space coordination, and have not yet considered the optimal scheduling problem of a large number of aircraft sequentially visiting multiple targets. There is a lack of research on related optimization models and optimization problems. Moreover, conventional sorting algorithms sort all data at once, which is insufficient in terms of computational speed and accuracy. They do not consider the scheduling and sorting problem of large-scale aircraft visiting multiple sites, and therefore cannot obtain the optimal scheduling scheme. Summary of the Invention
[0004] The purpose of this invention is to solve the problem that existing methods cannot obtain the optimal scheduling scheme for large-scale aircraft to access multiple sites, and to propose a multi-aircraft multi-site delivery scheduling planning method, storage medium and device.
[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is: a multi-aircraft, multi-site delivery scheduling and planning method, the method specifically including the following steps:
[0006] Step 1: Denote the number of aircraft as N and the number of ground stations as M, and ensure that each aircraft spends the same amount of time waiting at each ground station.
[0007] Step 2: Define constraints (1), (2), and (3):
[0008] Constraint (1) An aircraft may arrive at a ground station at most once;
[0009] Constraint (2): At any given time, a ground station may only accept one aircraft.
[0010] Constraint (3) An aircraft arrives at each ground station in the order in which it is deployed to the ground stations;
[0011] Establish a scheduling model based on the defined constraints (1), (2), and (3);
[0012] Step 3: Establish the objective function for scheduling planning based on the scheduling model;
[0013] Step 4: Solve the objective function established in Step 3 to obtain the scheduling plan result.
[0014] Furthermore, the waiting time incurred by the aircraft at ground stations includes loading and unloading time, refueling time, and maintenance time.
[0015] Furthermore, the scheduling model is as follows:
[0016]
[0017] C(R1,j)=C(R1,j-1)+P 1,j j = 2, 3, ..., M
[0018]
[0019] Where C(R1,1) represents the time when the first aircraft arrives at the first ground station and departs from the first ground station, and the time when the first aircraft arrives at the first ground station is taken as time 0; This represents the waiting time spent at the first ground station by the first aircraft arriving at the first ground station; C(R) i-1 ,1) represents the time when the (i-1)th aircraft that arrived at the first ground station departed from the first ground station; C(R) represents the waiting time spent by the i-th aircraft at the first ground station upon arrival; i C(R1,j) represents the time when the i-th aircraft arrives at the first ground station and departs from the first ground station; C(R1,j) represents the time when the first aircraft departs from the j-th ground station it arrived at; C(R1,j-1) represents the time when the first aircraft departs from the (j-1)-th ground station it arrived at; P 1,j C(R) represents the waiting time spent by the first aircraft at the j-th ground station it arrives at; i C(R,j) represents the time when the i-th aircraft arrives at the j-th ground station and departs from the j-th ground station; i-1 C(R,j) represents the time when the (i-1)th aircraft arrives at the j-th ground station and departs from the j-th ground station; i ,j-1) represents the time when the i-th aircraft, which has arrived at the j-th ground station, departs from the previous ground station it arrived at; This represents the waiting time spent by the i-th aircraft at the j-th ground station before it arrives.
[0020] Furthermore, the objective function of the scheduling plan is:
[0021]
[0022] Among them, R space R represents the set of all scheduling schemes; * Represents the optimal scheduling scheme; C(R) * C(R,M) represents the time when the last aircraft leaves the last ground station after being delivered, under the optimal scheduling scheme; C(R,M) represents the time when the last aircraft leaves the last ground station after being delivered, under scheduling scheme R.
[0023] Furthermore, in step four, the objective function established in step three is solved using a neural network sorting algorithm.
[0024] Furthermore, the specific process of step four is as follows:
[0025] Step 4.1: Initialize the number of iterations p = 1;
[0026] Step 42: Input the delivery sequence of each aircraft to various ground stations into the neural network, and denote the delivery sequence of the nth aircraft to various ground stations as R. n n = 1, 2, ..., N;
[0027] For any given ground station, a neural network is used to output the access order of each aircraft at that ground station;
[0028] Step 43: Determine the ordered array and conflicting array in the neural network output, and add the ordered array to the ordered array set;
[0029] Step 4: Determine whether the conflict array dimension is less than a set threshold or the number of iterations is greater than a set threshold.
[0030] If the conditions are met, the ordered array set and the conflict array are merged. The merged result is adjusted manually to obtain the final access order of each aircraft to each ground station, which is the final delivery scheduling plan result.
[0031] If the conditions are not met, proceed to steps four and five.
[0032] Step 45: Let the iteration count p = p + 1, remove the delivery scheduling tasks corresponding to the ordered array set from the total delivery scheduling tasks, use the remaining delivery scheduling tasks as the input of the neural network, and return to execute step 43.
[0033] Furthermore, determining the ordered array and the conflicting array in the neural network output specifically involves:
[0034] Step 431: Initialize the time step count k = 1;
[0035] Step 432: In the output of the neural network, determine whether there is a spacecraft access conflict in the k-th time step, that is, determine whether there is a situation where different spacecraft access the same ground station in the k-th time step.
[0036] If it does not exist, the access order corresponding to the kth time step in the neural network output is an ordered array, and step 433 continues to be executed;
[0037] If it exists, the access order corresponding to the kth time step in the neural network output is a conflict array, and it is determined that the access order in subsequent time steps in the neural network output is a conflict array, and the process ends.
[0038] Step 433: Determine if the time step count k has reached its maximum;
[0039] If the number of time steps k reaches its maximum, the process ends.
[0040] If the number of time steps k has not reached its maximum, let k = k + 1 and return to step four three two.
[0041] A computer storage medium storing at least one instruction, which is loaded and executed by a processor to implement the multi-aircraft multi-site delivery scheduling planning method.
[0042] A multi-aircraft, multi-site delivery scheduling and planning device includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the multi-aircraft, multi-site delivery scheduling and planning method.
[0043] The beneficial effects of this invention are:
[0044] This invention addresses the scheduling problem of multiple aircraft arriving at multiple stations in succession, taking into account processes such as unloading, loading, refueling, and maintenance. It optimizes the order in which aircraft access each station, establishes a multi-aircraft, multi-station delivery scheduling model, and proposes a discrete data iterative sorting algorithm based on neural networks to solve the problem of complex sorting of a large number of aircraft. By decomposing and iteratively solving the sorting problem of high-dimensional inputs layer by layer, it achieves the goal of minimizing the total delivery time of materials and obtains a better delivery scheduling scheme. Attached Figure Description
[0045] Figure 1 This is a diagram of a multi-aircraft, multi-site delivery scheduling model;
[0046] Figure 2 This is a flowchart of a neural network discrete data iterative sorting algorithm. Detailed Implementation
[0047] Specific implementation method one: Combining Figure 1This embodiment describes a multi-aircraft, multi-site delivery scheduling and planning method, which specifically includes the following steps:
[0048] Step 1: Denote the number of aircraft as N and the number of ground stations as M. The waiting time for each aircraft at each ground station is the same (the waiting time for each aircraft at each ground station is independent of the arrival order and the waiting time for each aircraft at each ground station will not stop midway. For the same ground station, the waiting time for each aircraft at that ground station is the same. For the same aircraft, the waiting time for that aircraft at each ground station is the same).
[0049] Step 2: Define constraints (1), (2), and (3):
[0050] Constraint (1) An aircraft may arrive at a ground station at most once;
[0051] Constraint (2): At any given time, a ground station may only accept one aircraft.
[0052] Constraint (3) An aircraft arrives at each ground station in the order in which it is deployed to the ground stations;
[0053] Establish a scheduling model based on the defined constraints (1), (2), and (3);
[0054] Step 3: Establish the objective function for scheduling planning based on the scheduling model;
[0055] Step 4: Solve the objective function established in Step 3 to obtain the scheduling plan result.
[0056]
[0057] The main features of this invention are as follows:
[0058] (1) Multi-aircraft multi-site delivery scheduling model
[0059] Considering scenarios where large numbers of aircraft transport supplies to multiple ground targets, and taking into account time factors such as loading and unloading time, refueling time, and maintenance time, a mathematical model is established to optimize the order in which each aircraft arrives at each target, thus forming an optimization model that minimizes the total time spent transporting goods.
[0060] (2) Neural Network Discrete Data Iterative Sorting Algorithm
[0061] The scheduling scheme involves dozens of aircraft and multiple targets, resulting in a very large problem space and numerous possible scheduling solutions, making direct solutions difficult. Although neural networks have strong data processing capabilities, they can still lead to incorrect sorting when faced with such a large-scale problem. Therefore, this invention proposes an iterative neural network sorting method that completes the sorting task through multiple iterations.
[0062] Specific Implementation Method Two: This implementation method differs from Specific Implementation Method One in that the waiting time spent by the aircraft at the ground station includes loading and unloading time, refueling time, and maintenance time.
[0063] The other steps and parameters are the same as in Specific Implementation Method 1.
[0064] Specific Implementation Method Three: This implementation method differs from Specific Implementation Method One or Two in that the scheduling model is as follows:
[0065]
[0066] C(R1,j)=C(R1,j-1)+P 1,j j = 2, 3, ..., M
[0067]
[0068] Where C(R1,1) represents the time when the first aircraft arrives at the first ground station and departs from the first ground station, and the time when the first aircraft arrives at the first ground station is taken as time 0; This represents the waiting time spent at the first ground station by the first aircraft arriving at the first ground station; C(R) i-1 ,1) represents the time when the (i-1)th aircraft that arrived at the first ground station departed from the first ground station; C(R) represents the waiting time spent by the i-th aircraft at the first ground station upon arrival; i C(R1,j) represents the time when the i-th aircraft arrives at the first ground station and departs from the first ground station; C(R1,j) represents the time when the first aircraft departs from the j-th ground station it arrived at; C(R1,j-1) represents the time when the first aircraft departs from the (j-1)-th ground station it arrived at; P 1,j C(R) represents the waiting time spent by the first aircraft at the j-th ground station it arrives at; i C(R,j) represents the time when the i-th aircraft arrives at the j-th ground station and departs from the j-th ground station; i-1 C(R,j) represents the time when the (i-1)th aircraft arrives at the j-th ground station and departs from the j-th ground station; i,j-1) represents the time when the i-th aircraft, which has arrived at the j-th ground station, departs from the previous ground station it arrived at; This represents the waiting time spent by the i-th aircraft at the j-th ground station before it arrives.
[0069] Other steps and parameters are the same as in specific implementation method one or two.
[0070] Specific Implementation Method Four: This implementation method differs from Specific Implementation Methods One to Three in that the objective function of the scheduling plan is:
[0071]
[0072] Among them, R space R represents the set of all scheduling schemes; * Represents the optimal scheduling scheme; C(R) * C(R,M) represents the time when the last aircraft leaves the last ground station after being delivered, under the optimal scheduling scheme; C(R,M) represents the time when the last aircraft leaves the last ground station after being delivered, under scheduling scheme R.
[0073] The other steps and parameters are the same as those in one of the specific implementation methods one to three.
[0074] Specific Implementation Method Five: This implementation method differs from Specific Implementation Methods One to Four in that, in step four, a neural network sorting algorithm is used to solve the objective function established in step three.
[0075] The other steps and parameters are the same as those in one of the specific implementation methods one to four.
[0076] Specific Implementation Method Six: Combination Figure 2 This embodiment is described below. The difference between this embodiment and any one of embodiments one through five is that the specific process of step four is as follows:
[0077] Step 4.1: Initialize the number of iterations p = 1;
[0078] Step 42: Input the delivery sequence of each aircraft to various ground stations into a neural network (the neural network used in this invention can be a DNN network, but is not limited to a DNN network), and denote the delivery sequence of the nth aircraft to various ground stations as R. n n = 1, 2, ..., N;
[0079] For each aircraft, the order in which it visits ground stations is fixed and unchanging. For any given ground station, a neural network is used to output the order in which each aircraft visits that ground station.
[0080] Step 43: Determine the ordered array and conflicting array in the neural network output, and add the ordered array to the ordered array set;
[0081] Step 4: Determine whether the conflict array dimension is less than a set threshold or the number of iterations is greater than a set threshold.
[0082] If the conditions are met, the ordered array set and the conflict array are merged. The merged result is adjusted manually to obtain the final access order of each aircraft to each ground station, which is the final delivery scheduling plan result.
[0083] If the conditions are not met, proceed to steps four and five.
[0084] Step 45: Let the iteration number p = p + 1. Remove the delivery scheduling tasks corresponding to the ordered array set from the total delivery scheduling tasks. Take the remaining delivery scheduling tasks (based on the currently determined ordered array, we can initially obtain the access order of some spacecraft to some ground stations, and then obtain which spacecraft still need to access which ground stations, i.e. the remaining tasks) as the input of the neural network, and return to execute step 43.
[0085] The other steps and parameters are the same as those in one of the specific implementation methods one to five.
[0086] Sort R = [R1, R2, ..., R N The dimension of the optimization variable is the number of aircraft or the number of aircraft in the system. Clearly, if the scheduling scheme involves dozens of aircraft and multiple targets, there will be (N!) permutations, resulting in a very large problem space and numerous possible scheduling schemes, making direct solution difficult. For example, swarm intelligence algorithms such as particle swarm optimization, genetic algorithms, and fish swarm optimization are often used to solve multi-parameter optimization problems, but their efficiency becomes very low when faced with high-dimensional optimization variables.
[0087] The sorting algorithm using a neural network is used to obtain the sorted sequence R = [R1, R2, ..., R]. N From a mathematical perspective, sorting is the process of mapping data elements from an unordered state to an ordered state. Ideally, a neural network should sort a large amount of input data at once, which requires huge training resources and may result in two different input data being mapped to the same output, thus affecting the correctness of the neural network's sorting. To address this, this invention proposes an iterative sorting method based on a neural network. Instead of directly training a complex model to sort all input data at once, this method uses a simpler model to complete the sorting task through multiple iterations.
[0088] Specific Implementation Method Seven: This implementation method differs from Specific Implementation Methods One through Six in that the determination of the ordered array and the conflict array in the neural network output result specifically involves:
[0089] Step 431: Initialize the number of time steps k = 1 (the length of each time step is equal to the waiting time of an aircraft at a ground station);
[0090] Step 432: In the output of the neural network, determine whether there is a spacecraft access conflict in the k-th time step, that is, determine whether there is a situation where different spacecraft access the same ground station in the k-th time step.
[0091] If it does not exist, the access order corresponding to the kth time step in the neural network output is an ordered array, and step 433 continues to be executed;
[0092] If it exists, the access order corresponding to the kth time step in the neural network output is a conflict array, and it is determined that the access order in subsequent time steps in the neural network output is a conflict array, and the process ends.
[0093] Step 433: Determine if the time step count k has reached its maximum;
[0094] If the number of time steps k reaches its maximum, the process ends.
[0095] If the number of time steps k has not reached its maximum, let k = k + 1 and return to step four three two.
[0096] The other steps and parameters are the same as those in one of the specific implementation methods one to six.
[0097] This invention optimizes the order in which all aircraft access the same ground station, based on the fixed access order of each aircraft to each ground station, to obtain the final delivery scheduling plan result. It should be noted that the dimension of the conflict array in this invention refers to the total number of time steps corresponding to the conflict array in the output of the neural network.
[0098] Specific Implementation Method Eight: A computer storage medium of this embodiment stores at least one instruction, which is loaded and executed by a processor to implement the multi-aircraft multi-site delivery scheduling planning method.
[0099] It should be understood that the instructions include any computer program product, software, or computerized method corresponding to any method described in this invention; the instructions can be used to program a computer system or other electronic device. Computer storage media may include readable media on which instructions are stored, and may include, but are not limited to, magnetic storage media, optical storage media; magneto-optical storage media include read-only memory, random access memory, erasable programmable memory (e.g., and flash memory layers), or other types of media suitable for storing electronic instructions.
[0100] Specific Implementation Method Nine: This implementation method is a multi-aircraft multi-site delivery scheduling and planning device. The device includes a processor and a memory. It should be understood that it includes any device including a processor and a memory described in this invention. The device may also include other units and modules that perform display, interaction, processing, control and other functions through signals or instructions.
[0101] The memory stores at least one instruction, which is loaded and executed by the processor to implement the multi-aircraft multi-site delivery scheduling planning method.
[0102] The above examples of the present invention are merely illustrative of the computational model and process of the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is impossible to exhaustively list all possible implementations here. Any obvious variations or modifications derived from the technical solutions of the present invention are still within the scope of protection of the present invention.
Claims
1. A multi-aircraft, multi-site delivery scheduling and planning method, characterized in that, The method specifically includes the following steps: Step 1: Denote the number of aircraft as N and the number of ground stations as M, and ensure that each aircraft spends the same amount of time waiting at each ground station. Step 2: Define constraint 1, constraint 2, and constraint 3: Constraint 1: An aircraft may visit a ground station at most once; Constraint 2: At any given time, a ground station may only accept one aircraft. Constraint 3: An aircraft arrives at each ground station in the order in which it is deployed to the ground stations. Establish a scheduling model based on the defined constraints 1, 2, and 3; The scheduling model is as follows: in, This indicates the time when the first aircraft arrives at the first ground station and departs from the first ground station. The time when the first aircraft arrives at the first ground station is taken as time 0. This indicates the waiting time spent by the first aircraft at the first ground station upon arrival. This indicates the number of people arriving at the first ground station. The moment the aircraft departs from the first ground station; This indicates the number of people arriving at the first ground station. The waiting time that an aircraft spends at the first ground station; This indicates the number of people arriving at the first ground station. The moment the aircraft departs from the first ground station; This indicates the first aircraft to leave itself and reach the [missing information - likely a specific location or destination]. The time of each ground station; This indicates the first aircraft to leave itself and reach the [missing information - likely a specific location or destination]. The time of each ground station; This represents the waiting time spent by the first aircraft at the j-th ground station it arrives at; Indicates reaching the th The first ground station The aircraft departed from the The time of each ground station; Indicates reaching the th The first ground station The aircraft departed from the The time of each ground station; Indicates reaching the th The first ground station The moment an aircraft departs from its previous ground station; Indicates reaching the th The first ground station The waiting time of an aircraft at the j-th ground station; Step 3: Establish the objective function for scheduling planning based on the scheduling model; Step 4: Solve the objective function established in Step 3 to obtain the scheduling plan results; The specific process of step four is as follows: Step 4.1: Initialize the number of iterations p = 1; Step 42: Input the sequence of deliveries from each aircraft to various ground stations into the neural network, and then input the sequence of deliveries from each aircraft to various ground stations into the neural network. The sequence in which aircraft deliver supplies to various ground stations is denoted as... , ; For any given ground station, a neural network is used to output the access order of each aircraft at that ground station; Step 43: Determine the ordered array and conflicting array in the neural network output, and add the ordered array to the ordered array set; Step 4: Determine whether the conflict array dimension is less than a set threshold or the number of iterations is greater than a set threshold. If the conditions are met, the ordered array set and the conflict array are merged. The merged result is adjusted manually to obtain the final access order of each aircraft to each ground station, which is the final delivery scheduling plan result. If the conditions are not met, proceed to steps four and five. Step 45: Let the iteration count p = p + 1, remove the delivery scheduling tasks corresponding to the ordered array set from the total delivery scheduling tasks, use the remaining delivery scheduling tasks as the input of the neural network, and return to execute step 43.
2. The multi-aircraft, multi-site delivery scheduling and planning method according to claim 1, characterized in that, The waiting time for the aircraft at the ground station includes loading and unloading time, refueling time, and maintenance time.
3. The multi-aircraft, multi-site delivery scheduling and planning method according to claim 2, characterized in that, The objective function of the scheduling plan is: in, Represents the set of all scheduling schemes; Indicates the optimal scheduling scheme; This represents the moment when the last aircraft leaves the last ground station after being delivered, under the optimal scheduling scheme. Representing the scheduling scheme Below, the moment when the last aircraft departs from the last ground station after being deployed.
4. The multi-aircraft, multi-site delivery scheduling planning method according to claim 3, characterized in that, In step four, the objective function established in step three is solved using a neural network sorting algorithm.
5. The multi-aircraft, multi-site delivery scheduling planning method according to claim 4, characterized in that, The determination of the ordered array and the conflict array in the output of the neural network specifically involves: Step 431: Initialize the time step count k=1; Step 432: In the output of the neural network, determine whether there is a spacecraft access conflict in the k-th time step, that is, determine whether there is a situation where different spacecraft access the same ground station in the k-th time step. If it does not exist, the access order corresponding to the kth time step in the neural network output is an ordered array, and step 433 continues to be executed; If it exists, the access order corresponding to the kth time step in the neural network output is a conflict array, and it is determined that the access order in subsequent time steps in the neural network output is a conflict array, and the process ends. Step 433: Determine if the time step count k has reached its maximum; If the number of time steps k reaches its maximum, the process ends. If the number of time steps k has not reached its maximum, let k = k + 1 and return to step four, three, two.
6. A computer storage medium, characterized in that, The storage medium stores at least one instruction, which is loaded and executed by a processor to implement the multi-aircraft multi-site delivery scheduling planning method according to any one of claims 1 to 5.
7. A multi-aircraft, multi-site delivery scheduling and planning device, characterized in that, The device includes a processor and a memory, the memory storing at least one instruction, which is loaded and executed by the processor to implement the multi-aircraft multi-site delivery scheduling planning method according to any one of claims 1 to 5.