A training task planning system and method

Abstract

The present invention relates to the field of flight management technology, and in particular to a training task planning system and method, the method comprising: obtaining multiple flight tasks, each flight task is a combination of multiple tasks; using the KM algorithm to perform multiple non-repetitive matching of multiple flight tasks with multiple aircraft to obtain multiple matching results; obtaining the routes of the aircraft corresponding to each flight task in the matching results, and obtaining the matching index of all routes, the matching index characterizing the pros and cons of all routes; calculating the product of the sum of the edge weights of each matching result and the matching index of the corresponding matching result, taking the matching result corresponding to the maximum product as the best matching result, and taking all routes corresponding to the best matching result as the best routes. That is, the scheme of the present invention can reasonably allocate flight tasks and improve the efficiency of flight tasks.

Description

Technical Field

[0001] The present invention relates to the field of flight management technology, and more specifically, to a training mission planning system and method. Background Art

[0002] With the rapid development of general aircraft in the fields of forest fire prevention, aerial surveying and mapping, emergency rescue, etc., the types of operational missions have also become diversified, requiring aircraft to be equipped with different mission payloads for different operational tasks during flight, including optoelectronic pods, infrared, radar and aerial cameras, etc., to achieve different mission objectives.

[0003] However, flight command and mission management have always been weak links, especially when flight missions become increasingly busy. At this time, how to plan flight training missions also faces the problem of difficulty in timely and reasonable deployment, resulting in low efficiency in flight mission execution and management.

[0004] In order to solve the above problems, in the related technologies, such as the patent application document with publication number CN118226881A, entitled A flight mission scheduling method, system and medium based on a drone platform, it is disclosed that a number of corresponding flight missions are established based on flight request information, and a flight plan table is established according to the priority of the flight mission; a number of flight missions are sorted according to the flight plan table, and the drone flight is controlled in sequence according to the currently executed flight mission based on the sorting results. The above scheme controls the drone to execute flight missions in sequence by analyzing the priorities of different flight missions, and during the flight, the flight missions can be flexibly switched according to the judgment results of the task completion status, thereby improving the efficiency of drone flight scheduling.

[0005] However, the tasks are assigned only by the priority of the flight tasks without considering the situation of the aircraft itself, which makes it difficult to make reasonable allocation of the flight tasks when executing the flight tasks, resulting in low efficiency of the flight tasks.

[0006] Therefore, it is particularly important to reasonably match the flight mission with the aircraft to improve flight efficiency. Summary of the invention

[0007] The purpose of the present invention is to propose a training mission planning system and method to solve the problem in the prior art that it is difficult to make reasonable allocation of flight missions when performing flight missions, resulting in low efficiency of flight missions; to this end, the present invention provides solutions in the following two aspects.

[0008] In a first aspect, the present invention provides a training task planning method, comprising:

[0009] Acquire multiple flight missions, each flight mission is a combination of multiple tasks;

[0010] Using the KM algorithm, multiple flight missions and multiple aircraft are matched multiple times without duplication to obtain multiple matching results;

[0011] Among them, when matching, the edge weight between each aircraft and each flight mission is calculated for: ; is a natural constant, represents normalization, represents the historical mutual information entropy between the aircraft and the jth task, represents the parameter value of the jth mission provided by the ith aircraft, Indicates The value of the jth task in the flight tasks; is the set of historical values ​​of the j-th task, is a collection of aircraft types;

[0012] Obtaining the routes of the aircraft corresponding to each flight mission in the matching results, and obtaining matching indicators of all routes, wherein the matching indicators represent the pros and cons of all routes;

[0013] The product of the sum of the edge weights of each matching result and the matching index of the corresponding matching result is calculated, and the matching result corresponding to the maximum product is taken as the best matching result, and all routes corresponding to the best matching result are taken as the best routes.

[0014] The above scheme matches multiple flight missions with multiple aircraft, introduces the importance of the aircraft in performing historical flight missions during the matching, and uses the importance as the weight for the current matching to adjust the edge weight, so as to achieve more accurate matching; further, after the matching, the matching index is obtained by evaluating the corresponding routes after the matching, and the matching index is combined with the edge weight sum of the matching results, so as to more accurately determine the best matching result and the corresponding best route.

[0015] Optionally, the mutual information entropy is: ;

[0016] in, represents the mutual information entropy between the jth task and the aircraft, represents the information entropy of the aircraft, represents the conditional entropy between the j-th task and the aircraft.

[0017] The above scheme calculates the importance between aircraft and tasks by utilizing the relationship between multiple aircraft and each task in historical flight missions, and can provide a reference for historical flights for matching subsequent flight missions with aircraft.

[0018] Optionally, the parameter values ​​include but are not limited to the maximum range and maximum load weight data of the aircraft; the various tasks include but are not limited to range and load.

[0019] Optionally, the specific process of obtaining the route of the aircraft corresponding to each flight mission in the matching result and obtaining the matching index of all routes is:

[0020] Determine multiple basic flight paths corresponding to the aircraft according to the origin and destination of the flight mission, then use a genetic algorithm to select the optimal flight path from the multiple flight paths, use the optimal flight path as the route of the aircraft, obtain the economic index of the route corresponding to each aircraft, and use the sum of all the economic indexes as the matching index of the matching result;

[0021] The economic index is the product of the planned flight path mileage and the fuel consumption per unit mileage.

[0022] The above scheme provides a method for quickly obtaining the optimal route and indicators for route evaluation.

[0023] Optionally, the specific process of obtaining the route of the aircraft corresponding to each flight mission in the matching result and obtaining the matching index of all routes is:

[0024] Based on the path information and environmental information corresponding to each flight mission, and combined with the reinforcement learning algorithm, the routes of the corresponding aircraft in each matching result are determined, and the reward value of the corresponding route is obtained; the sum of the reward values ​​of all routes is used as the matching index;

[0025] The setting of the reward value is related to the distance to the destination and the navigation time.

[0026] The above scheme provides a method for quickly and accurately obtaining the optimal route and indicators for route evaluation.

[0027] Optionally, the reinforcement learning algorithm adopts a Q-Learning algorithm.

[0028] Optionally, the method further includes a step of normalizing the parameter values.

[0029] The above scheme can eliminate the dimension of the parameter value.

[0030] In a second aspect, the present invention provides a training task planning system, comprising:

[0031] processor;

[0032] The memory stores computer instructions for training task planning. When the computer instructions are executed by the processor, the system executes the above-mentioned training task planning method.

[0033] The beneficial effects of the present invention are:

[0034] The scheme of the present invention is divided into two steps. The first is to obtain the historical mutual information entropy of historical flights with the help of historical flight data of multiple aircraft, and use the historical mutual information entropy as the weight of the edge weight when matching multiple aircraft with current multiple flight missions, so as to correct the edge weight in the matching result; the second is to obtain the matching index of the route by evaluating the route with the help of the aircraft in the matching result and the route of the flight mission, and then use the product of the matching index and the sum of the corrected edge weights as the indicator for judging the matching result, so as to select the best matching result and its corresponding route, thereby realizing accurate matching and planning of multiple flight missions and multiple aircraft. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0036] Figure 1 A flowchart of the steps of a training task planning method in this embodiment is schematically shown;

[0037] Figure 2 The structural block diagram of a training task planning system in this embodiment is schematically shown. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0039] like Figure 1 As shown, a training task planning method in this embodiment includes the following steps:

[0040] Step S1, obtaining multiple flight missions, each flight mission is a combination of multiple tasks. The tasks in the flight mission in this embodiment can be load requirements and range requirements.

[0041] Specifically, load requirement refers to the weight of the transported goods, and range requirement refers to the transport distance, transport time, etc.

[0042] Step S2, using the KM algorithm to perform multiple non-repetitive matching of multiple flight missions with multiple aircraft to obtain multiple matching results.

[0043] The KM (Kuhn-Munkres) algorithm is an algorithm for solving the optimal matching problem in weighted bipartite graphs. It is suitable for finding a perfect matching with weights from one set of nodes to another set of nodes in a bipartite graph. Since the KM algorithm is an existing technology, it will not be described in detail here.

[0044] When matching multiple flight missions with multiple aircraft, multiple non-repetitive matches can be performed, and each match can avoid being exactly the same as other matching results, so multiple matching results can be obtained. It should be noted that each matching result includes multiple matching pairs, and a matching pair consists of a flight mission and an aircraft.

[0045] For example, when the flight missions are A, B, C, and D, and the aircraft are numbered 1, 2, 3, and 4, assuming that the matching pairs of the first matching results are (A, 1), (B, 2), (C, 3), and (D, 4), then when performing the second matching, there are non-repeating matching pairs between the second matching results and the first matching results, namely (A, 2), (B, 1), (C, 3), and (D, 4), that is, there are at least two different matching pairs between the second matching results and the first matching results, which is a non-repeating match.

[0046] Considering that a single matching result may result in unsafe or unfeasible route planning, multiple non-repetitive matchings are performed.

[0047] At each match, the group with the largest sum of the edge weights is selected as the matching result. Since the specific matching process of the KM (Kuhn-Munkres) algorithm is an existing technology, it will not be described in detail here.

[0048] In the matching process, since there are multiple tasks in one flight mission, the edge weights are calculated according to the different tasks in each flight mission in this embodiment.

[0049] Specifically, calculate the edge weight between each aircraft and each flight mission for:

[0050] ; is a natural constant, represents normalization, represents the historical mutual information entropy between the aircraft and the jth task, represents the parameter value of the jth mission provided by the ith aircraft, Indicates The value of the jth task in the flight tasks; is the set of historical values ​​of the j-th task, A collection of aircraft types.

[0051] in, The larger the value, the greater the difference between the i-th aircraft and the The smaller the difference between the two flight missions, the higher the matching degree between the two. As the i-th aircraft and the The edge weights when matching all tasks in a flight mission can estimate the importance of the i-th aircraft to the j-th task.

[0052] The parameter values ​​mentioned above are standard flight parameters of the aircraft, including the maximum range and maximum load weight data of the aircraft, which correspond one to one with the tasks in the flight mission. For example, when one of the flight missions includes the flight distance and load, the standard flight parameters of the selected aircraft include the maximum range and maximum load weight data of the aircraft.

[0053] Furthermore, the standard flight parameters of the aircraft also include the best cruising speed and fuel consumption index. The historical mutual information entropy mentioned above is obtained by analyzing the historical flight missions and multiple aircraft.

[0054] Specifically, the conditional entropy between the j-th task and the aircraft is calculated, which reflects the uncertainty in aircraft selection after the introduction of the j-th task. The greater the difference between this uncertainty and the aircraft uncertainty before the introduction of the j-th task, the greater the increase in the certainty of the j-th task on the aircraft, and the more important it is.

[0055] The calculation process of conditional entropy is as follows: ;

[0056] in, represents the conditional entropy between the j-th task and the aircraft, represents the joint probability of the aircraft and the jth mission, Indicates the selection of aircraft when the jth mission is determined The probability of is the historical value of the j-th task, is the aircraft type, is the set of aircraft types, is the set of historical values ​​of the j-th task.

[0057] Among them, the calculation formula of historical mutual information entropy is as follows: ;

[0058] in, represents the mutual information entropy between the jth task and the aircraft, represents the information entropy of the aircraft, represents the conditional entropy between the jth task and the aircraft. The greater the mutual information entropy between the jth task and the aircraft, the greater the reduction in uncertainty in selecting the aircraft under the premise of determining the jth task, and the more important the jth task is to the selection of the aircraft. Since historical mutual information entropy, conditional entropy and information entropy are existing technologies, they will not be described in detail here.

[0059] It should be noted that the above historical mutual information entropy is obtained by analyzing the relationship between each task and multiple aircraft in multiple historical flight missions, and counting the importance of aircraft to different tasks in historical flight missions. The importance is used as the weight of the same task in the current flight mission, and the edge weight is adjusted in the current matching process, which can achieve more accurate and reasonable matching.

[0060] Step S3, obtaining the routes of the aircraft corresponding to each flight mission in the matching results, and obtaining matching indicators of all routes.

[0061] In one embodiment, the method for obtaining the route of the aircraft in each matching result may be: determining multiple basic flight paths corresponding to the aircraft according to the origin and destination of the flight mission, then selecting the optimal flight path from the multiple flight paths, and using the optimal flight path as the route of the aircraft.

[0062] The optimal flight path can be obtained by using genetic algorithms, particle swarm optimization, etc. Of course, corresponding restriction rules can also be determined in combination with flight conditions and surrounding environment to finally obtain the optimal flight path. Since genetic algorithms and particle swarm optimization are both existing technologies, the specific steps will not be described in detail here.

[0063] When obtaining the optimal flight path, the environment includes airspace information (the flight path does not pass through the no-fly zone), geographic information (the flight altitude corresponding to the flight path is greater than the highest altitude or the highest building in the intended flight airspace), population density information (the population density of the intended flight area corresponding to the flight path is lower than the preset population density), noise information, etc. The flight conditions include whether the meteorological information meets the flight requirements, whether the maximum range of the aircraft meets the flight path, etc.

[0064] When obtaining the optimal flight path, the economic index of the optimal flight path corresponding to each aircraft may be obtained, and the sum of all economic indexes may be used as the matching index of the matching result.

[0065] The economic indicator is the product of the aircraft's planned flight path mileage and the fuel consumption per unit mileage, wherein the fuel consumption between the start and end points of the flight phase is obtained, and the ratio of the planned flight path mileage to the fuel consumption is calculated as the fuel consumption per unit mileage.

[0066] In another embodiment, a reinforcement learning algorithm may be used to obtain the path information of the best route of each aircraft in the matching results, and obtain the reward value corresponding to the best route; and the sum of the reward values ​​of all routes is used as the matching index of the matching results.

[0067] The above reinforcement learning algorithm can be Q-learning, DDPG (deep deterministic policy gradient) or PPO (Proximal Policy Optimization).

[0068] In this embodiment, taking Q-learning as an example, a simulation analysis is performed on the routes of the aircraft of each matching result corresponding to the same flight mission, as follows:

[0069] (1) Environmental information: Obtain the flight parameters, path information, and environmental information of historical flight missions and corresponding aircraft. Path information includes the coordinates of take-off and landing locations and three-dimensional geographic information; aircraft flight parameters include maximum range, optimal cruising speed, and fuel consumption indicators; environmental information includes meteorological conditions and airspace information (no-fly zones, busy airspace).

[0070] (2) State space :The status of the aircraft, including: optimal cruising speed, aircraft fuel consumption, weather information of the passing area, and airspace information.

[0071] Among them, in the state Execute action when , get the status , will be in state Execute action when The mapping of states and actions is called a state-action pair. >.

[0072] (3) Action Space :The action is the operating instruction of the aircraft, such as the future navigation direction, navigation distance, and navigation speed.

[0073] (4) Set reward value :The setting of the reward value is related to the distance to the destination and the flight time. If the flight time is within the transportation time, the reward value is large, and if it exceeds the transportation time, a negative reward value is obtained. The closer the distance to the destination, the greater the reward value, and the reward value is negative if there is a collision with other aircraft or objects.

[0074] The reward value in this embodiment is the sum of the reward Rd related to the distance to the destination and the reward Rt related to the navigation time.

[0075] In one embodiment, the negative reward (penalty) Rt related to the flight time increases with the increase of the flight time, so that the flight mission can be completed in a shorter time. Exemplarily, the reward value is 100 if the flight time is within the transportation time, and the reward value is -100 if the flight time exceeds the transportation time.

[0076] The reward Rd related to the distance to the destination is: reaching the destination will get a larger reward, the closer the distance d to the destination, the larger the reward, and encountering an obstacle will get a negative reward, that is:

[0077] ;in, is the distance to the destination.

[0078] The termination condition for reaching the final state in this embodiment is reaching the destination, flying to the map boundary, or having completed the set number of explorations. Finally, after continuous iterative training, an accurate Q-learning algorithm can be obtained.

[0079] In this embodiment, the Q-learning algorithm is pre-trained using the historical flight missions and the flight parameters of the corresponding aircraft to obtain the trained Q-learning. Then, according to the current flight mission and the flight parameters of the aircraft, the trained Q-learning is used to output the route of the path coordinate sequence of the aircraft in each matching result.

[0080] It should be noted that since the method of using reinforcement learning algorithm to obtain the optimal solution is an existing technology, it will not be introduced in detail here.

[0081] Step S4, obtaining the best matching result and its corresponding best route.

[0082] Taking the matching index as the sum of the reward values ​​of all routes as an example, the score of each matching result is obtained. The specific calculation formula is as follows: ;

[0083] in, represents the score of the tth matching result, represents the sum of the reward values ​​of the route corresponding to the tth matching result, Represents the sum of all edge weights of the tth matching result.

[0084] The larger the score of the matching result, the higher the matching degree between the aircraft and the flight mission under the matching result, and the safer the route of each aircraft under the matching result. The route corresponding to the matching result with the largest score is used as the optimal route to realize the transportation of subsequent aircraft flight missions.

[0085] It should be noted that, in this embodiment, the reward value of the route through each matching result under the pre-trained Q-Learning algorithm reflects the feasibility of the corresponding matching result; the larger the reward value, the more reliable the route planning is.

[0086] Therefore, on the basis of the obtained matching results, combined with the routes of the aircraft in the matching results, the feasibility of the matching results is verified to select the best matching results, which can accurately realize the reasonable allocation of flight missions and aircraft.

[0087] The solution of the present invention can perform non-repetitive matching of multiple flight missions and aircraft in advance to obtain each matching result, and then use the sum of all edge weights of the matching results and the pros and cons of the flight missions and the routes of the corresponding aircraft in the matching results to select the best matching result from the multiple matching results, and then obtain the best route of the best matching result.

[0088] The present invention also provides a training task planning system. Figure 2 As shown, the system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a training task planning method according to the present invention is implemented.

[0089] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface, whose configuration and functions are known in the art and thus will not be described in detail here.

[0090] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0091] In the description of this specification, “plurality” means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0092] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A training task planning method, characterized in that: include: Acquire multiple flight missions, each flight mission is a combination of multiple tasks; Using the KM algorithm, multiple flight missions and multiple aircraft are matched multiple times without duplication to obtain multiple matching results; Among them, when matching, the edge weight between each aircraft and each flight mission is calculated for: ; is a natural constant, represents normalization, represents the historical mutual information entropy between the aircraft and the jth task, represents the parameter value of the jth mission provided by the ith aircraft, Indicates The value of the jth task in the flight tasks; is the set of historical values ​​of the j-th task, It is a collection of aircraft types; the parameter value is the standard flight parameters of the aircraft, including the maximum range and maximum load weight data of the aircraft; various missions include range and load weight; Obtaining the routes of the aircraft corresponding to each flight mission in the matching results, and obtaining matching indicators of all routes, wherein the matching indicators represent the pros and cons of all routes; The product of the sum of the edge weights of each matching result and the matching index of the corresponding matching result is calculated, and the matching result corresponding to the maximum product is taken as the best matching result, and all routes corresponding to the best matching result are taken as the best routes.

2. A training task planning method according to claim 1, characterized in that: The historical mutual information entropy is: ; in, represents the historical mutual information entropy between the j-th task and the aircraft, represents the information entropy of the aircraft, represents the conditional entropy between the j-th task and the aircraft.

3. A training task planning method according to claim 1, characterized in that: The specific process of obtaining the routes of the aircraft corresponding to each flight mission in the matching results and obtaining the matching indexes of all routes is as follows: Determine multiple basic flight paths corresponding to the aircraft according to the origin and destination of the flight mission, then use a genetic algorithm to select the optimal flight path from the multiple flight paths, use the optimal flight path as the route of the aircraft, obtain the economic index of the route corresponding to each aircraft, and use the sum of all the economic indexes as the matching index of the matching result; The economic index is the product of the planned flight path mileage and the fuel consumption per unit mileage.

4. A training task planning method according to claim 1, characterized in that: The specific process of obtaining the routes of the aircraft corresponding to each flight mission in the matching results and obtaining the matching indexes of all routes is as follows: Based on the path information and environmental information corresponding to each flight mission, and combined with the reinforcement learning algorithm, the routes of the corresponding aircraft in each matching result are determined, and the reward value of the corresponding route is obtained; the sum of the reward values ​​of all routes is used as the matching index; The setting of the reward value is related to the distance to the destination and the navigation time.

5. A training task planning method according to claim 4, characterized in that: The reinforcement learning algorithm adopts the Q-Learning algorithm.

6. A training task planning method according to claim 1, characterized in that: Before calculating the edge weight, the method further includes a step of normalizing the parameter value.

7. A training task planning system, characterized in that: include: processor; A memory storing computer instructions for training task planning. When the computer instructions are executed by the processor, the system executes a training task planning method according to any one of claims 1 to 6.

Images