Aircraft adaptive interval management and control method and system

Abstract

The invention discloses an aircraft adaptive interval management and control method and system, and relates to the technical field of air traffic management. An aircraft adaptive interval management and control system comprises a flight conflict basic unit module, a flight conflict scene generation module, a flight conflict resolution module and a flight conflict learning module. According to the method, the aircraft flight conflict scene training set can be constructed by constructing the basic flight unit of the aircraft and generating the flight trajectory vector set and combining the flight conflict generation model; the training set is utilized to train an aircraft autonomous interval management and control model through an improved Markov decision model, so that effective strategy output is realized when a flight conflict occurs; the model is updated through analysis of the test result, the adaptability and accuracy of the autonomous interval management and control model of the aircraft are improved, the flight safety of the aircraft in a complex environment is improved in the process, and the prediction and processing capacity of the model for flight conflicts is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of air traffic management, and in particular to an aircraft adaptive interval control method and system. Background Art

[0002] The method of adaptive aircraft interval control is of great significance in modern air traffic management. It not only improves flight safety, but also improves the efficiency of airspace use. It can adjust the distance between aircraft in real time to ensure that no collision occurs at any stage of flight. The main difficulties faced by the method of adaptive aircraft interval control include reaction time and decision-making requirements for route changes. First, the reaction time requirement is extremely high. Aircraft must respond to emergencies within milliseconds to ensure safe intervals. Second, the decision on route changes needs to comprehensively consider many factors, such as airspace restrictions, the dynamics of other aircraft, flight efficiency, etc. The decision-making process is complex and needs to be completed quickly. In addition, aircraft must also have a high degree of reliability and fault tolerance to cope with various uncertainties and abnormal situations to ensure flight safety and efficient use of airspace. Summary of the invention

[0003] The present invention aims to provide an aircraft adaptive interval control method and system.

[0004] An aircraft adaptive separation control method comprises the following steps: Step S1: Construct the basic flight unit F of the aircraft n , F n =(A n , B n , C n , D n ); where n represents the aircraft certification number, belonging to aircraft H n ; A n Indicates the aircraft flight direction angle, B n Indicates the aircraft flight speed, C n Indicates the minimum unit time of aircraft flight, D n Indicates the flight time of an aircraft unit; using the basic flight unit F of the aircraft n For aircraft H n Randomly generate aircraft flight trajectory vectors to obtain aircraft flight trajectory vector set X n ; Step S2: Generate a model based on flight conflicts and a vector set X of all aircraft flight trajectories nGenerate aircraft flight conflict scenarios; combine the generated aircraft flight conflict scenarios to obtain an aircraft flight conflict scenario training set; the flight conflict generation model includes a path generation layer, a path selection layer, a path optimization layer and a result output layer, which is constructed according to the CNN model and is used to construct and screen aircraft flight conflict scenarios; Step S3: using the aircraft flight conflict scenario training set to train the aircraft autonomous separation control model; the aircraft autonomous separation control model includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer and a control strategy output layer, which is constructed based on the improved Markov decision model and is used to output strategies when an aircraft flight conflict occurs; Step S4: For aircraft H n A simulation test is performed with the aircraft autonomous separation control model to obtain an aircraft flight conflict simulation test result; the aircraft autonomous separation control model is updated according to the aircraft flight conflict simulation test result to obtain a target aircraft autonomous separation control model.

[0005] As a preferred technical solution of the present invention, the flight conflict generation model in step S2 includes a path generation layer, a path selection layer, a path optimization layer and a result output layer; The path generation layer is used to generate the path from the aircraft H n In the random algorithm, I aircraft H is selected n Perform path selection to obtain an aircraft simulation generation group; The path selection layer is used to select the aircraft Hn from the aircraft flight trajectory vector set X n A preset number of aircraft flight trajectory vectors are randomly selected and spliced ​​to obtain the aircraft simulated flight trajectory G n ; The path optimization layer is used to generate conflict scenarios based on the aircraft simulation generation group to obtain aircraft flight conflict scenarios; The result output layer is used to traverse several aircraft simulation generation groups to obtain the aircraft flight conflict scenario training set.

[0006] As a preferred technical solution of the present invention, the specific steps of generating a conflict scenario in the path optimization layer include: Step E1: Generate the corresponding aircraft simulation flight trajectory G according to the aircraft simulation generation group n , get the aircraft simulation flight trajectory set; Step E2: generating a conflict scenario for the aircraft simulated flight trajectory set to obtain the aircraft flight conflict scenario to be optimized; Step E3: feature screening is performed on the aircraft flight conflict scenario to be optimized to obtain the aircraft flight conflict scenario; The specific steps for training the path optimization layer include: Collect several groups of standard aircraft flight conflict scenario training samples; combine several groups of standard aircraft flight conflict scenario training samples to obtain an aircraft flight conflict scenario training set; use the CNN model to construct an initial path optimization layer; use the aircraft flight conflict scenario training set to train the initial path optimization layer to obtain a path optimization layer to be evaluated; perform model evaluation on the path optimization layer to be evaluated to obtain a model evaluation result of the path optimization layer to be evaluated; if the model evaluation result of the path optimization layer to be evaluated is passed, the path optimization layer to be evaluated is used as the path optimization layer; otherwise, continue model training using the aircraft flight conflict scenario training set.

[0007] As a preferred technical solution of the present invention, the aircraft autonomous separation control model in step S3 includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer and a control strategy output layer; The data extraction layer is used to obtain the flight data of the aircraft to which the current aircraft belongs, and obtain the flight data of the aircraft to be analyzed; The aircraft interval sensing layer is used to obtain the aircraft flight data of the disturbing aircraft within the preset interval sensing range, and obtain the flight data of the disturbing aircraft to be analyzed; The initial interval control strategy calculation layer is used to perform strategy analysis based on the flight data of the aircraft to be analyzed and the flight data of the disturbing aircraft to be analyzed, and obtain the aircraft control strategy set to be evaluated; the initial interval control strategy calculation layer is trained and constructed by the improved Markov decision model and the aircraft flight conflict scenario training set; The feedback monitoring layer is used to make predictions based on the aircraft control strategy set to be evaluated and obtain the aircraft separation control feedback results; the aircraft control strategy set to be evaluated is judged based on the aircraft separation control feedback results to obtain the optimal aircraft adaptive control strategy in the aircraft control strategy set to be evaluated; The control strategy output layer is used to output the aircraft adaptive control strategy.

[0008] As a preferred technical solution of the present invention, the specific steps of training the initial interval control strategy calculation layer include: Construct training aircraft and aircraft conflict digital twin models; deploy the aircraft flight conflict scenario training set in the aircraft conflict digital twin model; input the training aircraft into the aircraft conflict digital twin model to simulate aircraft flight conflicts; The Markov decision model is used to generate K aircraft separation control strategy individuals L for simulated aircraft flight conflicts. k , k=1,2,…,K; K aircraft interval control strategy individuals L kCombine to obtain the iterative population of aircraft interval control strategy; when performing population iteration, set the maximum number of iterations; When performing population iteration, calculate the interval control time T of the aircraft interval control strategy individual Lk in the aircraft interval control strategy iteration population k , interval control time T k It is obtained by weighted addition of avoidance time and return time; the interval control time T k The inverse of the aircraft spacing control strategy L k The fitness Y k ; At each population iteration, all fitness values ​​Y are retained k The aircraft interval control strategy individual L with a fitness greater than the preset optimal fitness k , and the remaining aircraft interval control strategy individuals L in the aircraft interval control strategy iteration population k The Markov decision model is used to update and obtain the new aircraft interval control strategy individual L k ; All aircraft interval control strategy individuals L with fitness greater than the preset optimal fitness k and all new aircraft separation control strategy individuals L k Combine to obtain an updated aircraft interval control strategy iteration population, thereby updating the aircraft interval control strategy iteration population; When the maximum number of iterations is reached, all aircraft interval control strategy individuals L whose fitness is greater than the preset optimal fitness are output. k , as a set of candidate aircraft control strategies for training aircraft; Collect candidate aircraft control strategy sets for several groups of training aircraft; combine candidate aircraft control strategy sets for several groups of training aircraft to obtain an aircraft control strategy training set; use the aircraft control strategy training set to perform model training on the initial interval control strategy calculation layer to obtain the initial interval control strategy calculation layer to be evaluated; perform model evaluation on the initial interval control strategy calculation layer to be evaluated to obtain a model evaluation result of the initial interval control strategy calculation layer to be evaluated; if the model evaluation result of the initial interval control strategy calculation layer to be evaluated is passed, use the trained initial interval control strategy calculation layer to be evaluated as the initial interval control strategy calculation layer; otherwise, continue model training using the aircraft control strategy training set.

[0009] As a preferred technical solution of the present invention, the specific steps of training the feedback monitoring layer include: Collecting several groups of aircraft separation control strategy prediction samples with target value annotations, where the target value is a quantitative evaluation result of the aircraft separation control strategy; combining several groups of aircraft separation control strategy prediction samples with target value annotations to obtain an aircraft control strategy feedback training set; The feedback monitoring layer is trained using the aircraft control strategy feedback training set to obtain the initial feedback monitoring layer; a model evaluation is performed on the initial feedback monitoring layer to obtain an initial feedback monitoring layer model evaluation result; if the initial feedback monitoring layer model evaluation result is passed, the initial feedback monitoring layer is used as the feedback monitoring layer; otherwise, the model training is continued using the aircraft control strategy feedback training set.

[0010] As a preferred technical solution of the present invention, the specific steps of step S4 include: Step S41: Collect aircraft H n Using the aircraft autonomous separation control model to simulate the control simulation data of aircraft adaptive separation control P n And record the corresponding aircraft flight conflict simulation test results; Step S42: Using the control simulation data P n Construct a target aircraft flight conflict training set with the corresponding aircraft flight conflict simulation test results; Step S43: Use the target aircraft flight conflict training set to update the parameters of the aircraft autonomous separation control model to obtain the target aircraft autonomous separation control model.

[0011] An aircraft adaptive separation control system, comprising: The flight conflict basic unit module includes a construction unit for constructing the basic flight unit F of an aircraft. n , F n =(A n , B n , C n , D n ); where n represents the aircraft certification number, belonging to aircraft H n ; A n Indicates the aircraft flight direction angle, B n Indicates the aircraft flight speed, C n Indicates the minimum unit time of aircraft flight, D n Indicates the flight time of an aircraft unit; using the basic flight unit F of the aircraft n For aircraft H n Randomly generate aircraft flight trajectory vectors to obtain aircraft flight trajectory vector set X n ; The flight conflict scenario generation module includes a generation unit for generating a flight conflict model and a flight trajectory vector set X of all aircraft based on the flight conflict generation model. nGenerate aircraft flight conflict scenarios; combine the generated aircraft flight conflict scenarios to obtain an aircraft flight conflict scenario training set; the flight conflict generation model includes a path generation layer, a path selection layer, a path optimization layer and a result output layer, which is constructed according to the CNN model and is used to construct and screen aircraft flight conflict scenarios; The flight conflict resolution module includes a training unit for training an autonomous aircraft separation control model using an aircraft flight conflict scenario training set; the autonomous aircraft separation control model includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer, and a control strategy output layer, which is constructed based on an improved Markov decision model and is used for outputting strategies when an aircraft flight conflict occurs; The flight conflict learning module includes a learning unit for aircraft H n A simulation test is performed with the aircraft autonomous separation control model to obtain an aircraft flight conflict simulation test result; the aircraft autonomous separation control model is updated according to the aircraft flight conflict simulation test result to obtain a target aircraft autonomous separation control model.

[0012] The present invention has the following advantages: 1. The present invention constructs an aircraft basic flight unit and generates a flight trajectory vector set, and combines it with a flight conflict generation model to construct an aircraft flight conflict scenario training set; using the training set, the aircraft autonomous interval control model is trained through an improved Markov decision model, and effective strategy output is achieved when a flight conflict occurs; the simulation test further verifies the effectiveness of the model, and updates the model by analyzing the test results, thereby improving the adaptability and accuracy of the aircraft autonomous interval control model. This process not only improves the flight safety of aircraft in complex environments, but also enhances the model's ability to predict and handle flight conflicts, thereby providing a more scientific and efficient solution for aircraft interval management.

[0013] 2. The present invention realizes effective management of aircraft flight conflicts by constructing an aircraft autonomous interval control model, including a data extraction layer, an interval perception layer, an initial interval control strategy calculation layer, a feedback monitoring layer and a control strategy output layer; by using an improved Markov decision model and a flight conflict scenario training set, it is possible to generate and iteratively optimize interval control strategies; by simulating conflicts and evaluating fitness, it ultimately outputs the optimal aircraft adaptive control strategy. This process not only improves the aircraft's autonomous interval control capability in a complex flight environment, but also enhances the model's adaptability and accuracy, thereby improving flight safety; in addition, through the digital twin model and iterative population method, the model can continuously learn and optimize, providing a more scientific and efficient solution for aircraft interval control. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A schematic diagram of the structure of an aircraft adaptive separation control system adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0016] Embodiment 1, an aircraft adaptive separation control method, comprising the following steps: Step S1: Construct the basic flight unit F of the aircraft n , F n =(A n , B n , C n , D n ); where n represents the aircraft certification number, belonging to aircraft H n ; A n Indicates the aircraft flight direction angle, B n Indicates the aircraft flight speed, C n Indicates the minimum unit time of aircraft flight, D n Indicates the flight time of an aircraft unit; using the basic flight unit F of the aircraft n For aircraft H n Randomly generate aircraft flight trajectory vectors to obtain aircraft flight trajectory vector set X n ; By defining clear flight units such as direction angle, speed, minimum unit time and unit flight duration, the flight path of the aircraft can be calculated and predicted more accurately, which is of great significance for optimizing flight plans, reducing flight time and fuel consumption; the flight trajectory vector generated based on detailed flight unit information can help the air traffic management system to more effectively monitor and manage the flow of aircraft in the airspace, avoid flight conflicts and ensure flight safety; it helps to provide a basis for the automatic generation of aircraft flight trajectories and realize the intelligent flight control, such as automatic obstacle avoidance and intelligent scheduling, so as to improve the operating efficiency and service quality of the entire aviation system; at the same time, various simulated flight trajectories can be flexibly generated to provide theoretical support for subsequent model training and other steps; Step S2: Generate a model based on flight conflicts and a vector set X of all aircraft flight trajectories n Generate aircraft flight conflict scenarios; combine the generated aircraft flight conflict scenarios to obtain an aircraft flight conflict scenario training set; the flight conflict generation model includes a path generation layer, a path selection layer, a path optimization layer and a result output layer, which is constructed according to the CNN model and is used to construct and screen aircraft flight conflict scenarios; The flight conflict generation model in step S2 includes a path generation layer, a path selection layer, a path optimization layer and a result output layer; The path generation layer is used to generate the path from the aircraft H n In the random algorithm, I aircraft H is selected n Perform path selection to obtain an aircraft simulation generation group; The path selection layer is used to target aircraft H n Vector set of aircraft flight tracks X n A preset number of aircraft flight trajectory vectors are randomly selected and spliced ​​to obtain the aircraft simulated flight trajectory G n ; The path optimization layer is used to generate conflict scenarios based on the aircraft simulation generation group to obtain aircraft flight conflict scenarios; The result output layer is used to traverse several aircraft simulation generation groups to obtain the aircraft flight conflict scenario training set; The specific steps for generating conflict scenarios in the path optimization layer include: Step E1: Generate the corresponding aircraft simulation flight trajectory G according to the aircraft simulation generation group n , get the aircraft simulation flight trajectory set; Step E2: generating a conflict scenario for the aircraft simulated flight trajectory set to obtain the aircraft flight conflict scenario to be optimized; Step E3: feature screening is performed on the aircraft flight conflict scenario to be optimized to obtain the aircraft flight conflict scenario; The specific steps for training the path optimization layer include: Collect several groups of standard aircraft flight conflict scenario training samples; combine several groups of standard aircraft flight conflict scenario training samples to obtain an aircraft flight conflict scenario training set; use the CNN model to construct an initial path optimization layer; use the aircraft flight conflict scenario training set to train the initial path optimization layer to obtain a path optimization layer to be evaluated; perform model evaluation on the path optimization layer to be evaluated to obtain a model evaluation result of the path optimization layer to be evaluated; if the model evaluation result of the path optimization layer to be evaluated is passed, the path optimization layer to be evaluated is used as the path optimization layer; otherwise, continue model training using the aircraft flight conflict scenario training set; By generating and optimizing the simulated flight trajectory of aircraft, potential flight conflict points can be discovered in advance, so that measures can be taken to avoid the risk of collision in actual flight, greatly improving flight safety; the path selection layer and the path optimization layer work together. In the process of path optimization, multiple goals can be comprehensively considered for optimization, and conflict paths that meet the conflict criteria can be screened out, reducing meaningless simulated flight path conflicts; using the CNN model to train the path optimization layer can improve the accuracy and robustness of the model through continuous learning and iteration, so as to better cope with complex flight environments; Step S3: using the aircraft flight conflict scenario training set to train the aircraft autonomous separation control model; the aircraft autonomous separation control model includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer and a control strategy output layer, which is constructed based on the improved Markov decision model and is used to output strategies when an aircraft flight conflict occurs; In step S3, the aircraft autonomous separation control model includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer, and a control strategy output layer; The data extraction layer is used to obtain the flight data of the aircraft to which the current aircraft belongs, and obtain the flight data of the aircraft to be analyzed; The aircraft interval sensing layer is used to obtain the aircraft flight data of the disturbing aircraft within the preset interval sensing range, and obtain the flight data of the disturbing aircraft to be analyzed; The initial interval control strategy calculation layer is used to perform strategy analysis based on the flight data of the aircraft to be analyzed and the flight data of the disturbing aircraft to be analyzed, and obtain the set of aircraft control strategies to be evaluated; the initial interval control strategy calculation layer is trained and constructed by the improved Markov decision model and the aircraft flight conflict scenario training set; the real-time digital twin model is constructed using the flight data of the aircraft to be analyzed and the flight data of the disturbing aircraft to be analyzed for strategy generation; The feedback monitoring layer is used to make predictions based on the aircraft control strategy set to be evaluated and obtain the aircraft separation control feedback results; the aircraft control strategy set to be evaluated is judged based on the aircraft separation control feedback results to obtain the optimal aircraft adaptive control strategy in the aircraft control strategy set to be evaluated; The control strategy output layer is used to output the aircraft adaptive control strategy; The aircraft interval perception layer can obtain the flight data of surrounding disturbing aircraft in real time, detect potential flight conflicts in time, and improve flight safety; through the feedback monitoring layer and the control strategy output layer, the model can adaptively adjust the interval control strategy according to the current flight conditions to avoid flight conflicts and ensure flight safety; the initial interval control strategy calculation layer is based on the improved Markov decision model, which can comprehensively consider multiple goals and generate the optimal interval control strategy; the feedback monitoring layer can dynamically adjust the control strategy according to the actual flight situation to ensure that the flight path and interval are always in the optimal state; the data extraction layer and the aircraft interval perception layer can collect a large amount of flight data. By analyzing these data, flight patterns and laws can be discovered, providing a scientific basis for flight planning and air traffic management; The specific steps for training the initial interval control strategy calculation layer include: Construct training aircraft and aircraft conflict digital twin models; deploy the aircraft flight conflict scenario training set in the aircraft conflict digital twin model; input the training aircraft into the aircraft conflict digital twin model to simulate aircraft flight conflicts; use the aircraft flight conflict scenario training set to randomly generate multiple flight conflict scenarios and deploy them in the aircraft conflict digital twin model; The Markov decision model is used to generate K aircraft separation control strategy individuals L for simulated aircraft flight conflicts. k , k=1,2,…,K; K aircraft interval control strategy individuals L k Combine to obtain the iterative population of aircraft interval control strategy; when performing population iteration, set the maximum number of iterations; the maximum number of iterations is set by professional technicians according to actual conditions; When performing population iteration, calculate the interval control time T of the aircraft interval control strategy individual Lk in the aircraft interval control strategy iteration population k , interval control time T k It is obtained by weighted addition of avoidance time and return time; the interval control time T k The inverse of the aircraft spacing control strategy L k The fitness Y k ; At each population iteration, all fitness values ​​Y are retained k The aircraft interval control strategy individual L with a fitness greater than the preset optimal fitness k , and the remaining aircraft interval control strategy individuals L in the aircraft interval control strategy iteration population k The Markov decision model is used to update and obtain the new aircraft interval control strategy individual L k ; All aircraft interval control strategy individuals L with fitness greater than the preset optimal fitness k and all new aircraft separation control strategy individuals Lk Combine to obtain an updated iterative population of aircraft interval control strategy, and realize the update of the iterative population of aircraft interval control strategy; the preset optimal fitness is set by professional and technical personnel according to actual conditions; When the maximum number of iterations is reached, all aircraft interval control strategy individuals L whose fitness is greater than the preset optimal fitness are output. k , as a set of candidate aircraft control strategies for training aircraft; Collect candidate aircraft control strategy sets for several groups of training aircraft; combine candidate aircraft control strategy sets for several groups of training aircraft to obtain aircraft control strategy training sets; use the aircraft control strategy training sets to perform model training on the initial interval control strategy calculation layer to obtain the initial interval control strategy calculation layer to be evaluated; perform model evaluation on the initial interval control strategy calculation layer to be evaluated to obtain a model evaluation result of the initial interval control strategy calculation layer to be evaluated; if the model evaluation result of the initial interval control strategy calculation layer to be evaluated is passed, use the trained initial interval control strategy calculation layer to be evaluated as the initial interval control strategy calculation layer; otherwise, continue model training using the aircraft control strategy training sets; By building a digital twin model of aircraft and aircraft conflicts and simulating flight conflicts in the model, potential flight conflicts can be detected and predicted more accurately, so that effective avoidance measures can be taken; the model can adjust the interval control strategy in real time during flight, respond to emergencies in a timely manner, and ensure flight safety; it can automatically generate and adjust the flight interval strategy, reducing the need for human intervention and improving the efficiency of air traffic management. It can dynamically adjust the flight interval strategy based on real-time flight data and environmental changes, enhancing flexibility and adaptability; the improved optimization algorithm is used to screen aircraft control strategies, while retaining the better solutions in the population, improving the convergence speed; The specific steps of training the feedback monitoring layer include: Collecting several groups of aircraft separation control strategy prediction samples with target value annotations, where the target value is a quantitative evaluation result of the aircraft separation control strategy; combining several groups of aircraft separation control strategy prediction samples with target value annotations to obtain an aircraft control strategy feedback training set; The feedback monitoring layer is trained using the aircraft control strategy feedback training set to obtain an initial feedback monitoring layer; a model evaluation is performed on the initial feedback monitoring layer to obtain an initial feedback monitoring layer model evaluation result; if the initial feedback monitoring layer model evaluation result is passed, the initial feedback monitoring layer is used as the feedback monitoring layer; otherwise, the model training is continued using the aircraft control strategy feedback training set; Through training feedback monitoring layer, flight conflicts can be better predicted and evaluated, so that effective control strategies can be adopted to reduce the risk of mid-air collisions and ensure flight safety. Through simulation testing and model evaluation, the autonomous separation control model can be continuously updated and optimized to improve the model's ability to predict and handle flight conflicts, thereby enhancing the ability of autonomous separation control of aircraft. Step S4: For aircraft H n Conduct simulation tests with the aircraft autonomous separation control model to obtain aircraft flight conflict simulation test results; update the aircraft autonomous separation control model according to the aircraft flight conflict simulation test results to obtain a target aircraft autonomous separation control model; The specific steps of step S4 include: Step S41: Collect aircraft H n Using the aircraft autonomous separation control model to simulate the control simulation data of aircraft adaptive separation control P n And record the corresponding aircraft flight conflict simulation test results; control simulation data P n Based on the flight data and control strategy generation, the aircraft flight conflict simulation test results are evaluated and determined by professional technicians according to the actual situation; Step S42: Using the control simulation data P n Construct a target aircraft flight conflict training set with the corresponding aircraft flight conflict simulation test results; Step S43: using the target aircraft flight conflict training set to update the parameters of the aircraft autonomous separation control model to obtain the target aircraft autonomous separation control model; By collecting and analyzing control simulation data P n And the corresponding aircraft flight conflict simulation test results, the model can be optimized based on actual data to build a more targeted aircraft autonomous separation control model. A more adaptable aircraft autonomous separation control model can be built for different aircraft to improve safety when encountering flight conflicts; ensuring that its performance in actual applications is always in the best condition.

[0017] Example 2, an aircraft adaptive separation control system, see Figure 1 As shown, including: The flight conflict basic unit module includes a construction unit for constructing the basic flight unit F of an aircraft. n , F n =(A n , B n , C n , D n ); where n represents the aircraft certification number, belonging to aircraft H n ; A n Indicates the aircraft flight direction angle, Bn Indicates the aircraft flight speed, C n Indicates the minimum unit time of aircraft flight, D n Indicates the flight time of an aircraft unit; using the basic flight unit F of the aircraft n For aircraft H n Randomly generate aircraft flight trajectory vectors to obtain aircraft flight trajectory vector set X n ; The flight conflict scenario generation module includes a generation unit for generating a flight conflict model and a flight trajectory vector set X of all aircraft based on the flight conflict generation model. n Generate aircraft flight conflict scenarios; combine the generated aircraft flight conflict scenarios to obtain an aircraft flight conflict scenario training set; the flight conflict generation model includes a path generation layer, a path selection layer, a path optimization layer and a result output layer, which is constructed according to the CNN model and is used to construct and screen aircraft flight conflict scenarios; The flight conflict resolution module includes a training unit for training an autonomous aircraft separation control model using an aircraft flight conflict scenario training set; the autonomous aircraft separation control model includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer, and a control strategy output layer, which is constructed based on an improved Markov decision model and is used for outputting strategies when an aircraft flight conflict occurs; The flight conflict learning module includes a learning unit for aircraft H n A simulation test is performed with the aircraft autonomous separation control model to obtain an aircraft flight conflict simulation test result; the aircraft autonomous separation control model is updated according to the aircraft flight conflict simulation test result to obtain a target aircraft autonomous separation control model.

[0018] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. An aircraft adaptive separation control method, characterized in that: The following steps are involved: Step S1: Construct the basic flight unit F of the aircraft n , F n =(A n , B n , C n , D n ); where n represents the aircraft certification number, belonging to aircraft H n ; A n Indicates the aircraft flight direction angle, B n Indicates the aircraft flight speed, C n Indicates the minimum unit time of aircraft flight, D n Indicates the flight time of an aircraft unit; using the basic flight unit F of the aircraft n For aircraft H n Randomly generate aircraft flight trajectory vectors to obtain aircraft flight trajectory vector set X n ; Step S2: Generate a model based on flight conflicts and a vector set X of all aircraft flight trajectories n Generate aircraft flight conflict scenarios; combine the generated aircraft flight conflict scenarios to obtain an aircraft flight conflict scenario training set; the flight conflict generation model includes a path generation layer, a path selection layer, a path optimization layer and a result output layer, which is constructed according to the CNN model and is used to construct and screen aircraft flight conflict scenarios; Step S3: using the aircraft flight conflict scenario training set to train the aircraft autonomous separation control model; the aircraft autonomous separation control model includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer and a control strategy output layer, which is constructed based on the improved Markov decision model and is used to output strategies when an aircraft flight conflict occurs; Step S4: For aircraft H n A simulation test is performed with the aircraft autonomous separation control model to obtain an aircraft flight conflict simulation test result; the aircraft autonomous separation control model is updated according to the aircraft flight conflict simulation test result to obtain a target aircraft autonomous separation control model.

2. The method for adaptive aircraft separation control according to claim 1, characterized in that: The flight conflict generation model in step S2 includes a path generation layer, a path selection layer, a path optimization layer and a result output layer; The path generation layer is used to generate the path from the aircraft H n In the random algorithm, I aircraft H is selected n Perform path selection to obtain an aircraft simulation generation group; The path selection layer is used to target aircraft H n Vector set of aircraft flight tracks X n A preset number of aircraft flight trajectory vectors are randomly selected and spliced ​​to obtain the aircraft simulated flight trajectory G n ; The path optimization layer is used to generate conflict scenarios based on the aircraft simulation generation group to obtain aircraft flight conflict scenarios; The result output layer is used to traverse several aircraft simulation generation groups to obtain the aircraft flight conflict scenario training set.

3. The method for adaptive aircraft separation control according to claim 2, characterized in that: The specific steps for generating conflict scenarios in the path optimization layer include: Step E1: Generate the corresponding aircraft simulation flight trajectory G according to the aircraft simulation generation group n , get the aircraft simulation flight trajectory set; Step E2: generating a conflict scenario for the aircraft simulated flight trajectory set to obtain the aircraft flight conflict scenario to be optimized; Step E3: feature screening is performed on the aircraft flight conflict scenario to be optimized to obtain the aircraft flight conflict scenario; The specific steps for training the path optimization layer include: Collect several groups of standard aircraft flight conflict scenario training samples; combine several groups of standard aircraft flight conflict scenario training samples to obtain an aircraft flight conflict scenario training set; use the CNN model to construct an initial path optimization layer; use the aircraft flight conflict scenario training set to train the initial path optimization layer to obtain a path optimization layer to be evaluated; perform model evaluation on the path optimization layer to be evaluated to obtain a model evaluation result of the path optimization layer to be evaluated; if the model evaluation result of the path optimization layer to be evaluated is passed, the path optimization layer to be evaluated is used as the path optimization layer; otherwise, continue model training using the aircraft flight conflict scenario training set.

4. The method for adaptive aircraft separation control according to claim 3, characterized in that: In step S3, the aircraft autonomous separation control model includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer and a control strategy output layer; The data extraction layer is used to obtain the aircraft flight data to which the current aircraft belongs, and obtain the aircraft flight data to be analyzed; The aircraft interval sensing layer is used to obtain the aircraft flight data of the disturbing aircraft within the preset interval sensing range, and obtain the flight data of the disturbing aircraft to be analyzed; The initial interval control strategy calculation layer is used to perform strategy analysis based on the flight data of the aircraft to be analyzed and the flight data of the disturbing aircraft to be analyzed, and obtain the aircraft control strategy set to be evaluated; the initial interval control strategy calculation layer is trained and constructed by the improved Markov decision model and the aircraft flight conflict scenario training set; The feedback monitoring layer is used to make predictions based on the aircraft control strategy set to be evaluated and obtain the aircraft separation control feedback results; the aircraft control strategy set to be evaluated is judged based on the aircraft separation control feedback results to obtain the optimal aircraft adaptive control strategy in the aircraft control strategy set to be evaluated; The control strategy output layer is used to output the aircraft adaptive control strategy.

5. The method for adaptive aircraft separation control according to claim 4, characterized in that: The specific steps for training the initial interval control strategy calculation layer include: Construct training aircraft and aircraft conflict digital twin models; deploy the aircraft flight conflict scenario training set in the aircraft conflict digital twin model; input the training aircraft into the aircraft conflict digital twin model to simulate aircraft flight conflicts; The Markov decision model is used to generate K aircraft separation control strategy individuals L for simulated aircraft flight conflicts. k , k=1,2,…,K; K aircraft interval control strategy individuals L k Combine to obtain the iterative population of aircraft interval control strategy; when performing population iteration, set the maximum number of iterations; When performing population iteration, calculate the interval control time T of the aircraft interval control strategy individual Lk in the aircraft interval control strategy iteration population k , interval control time T k It is obtained by weighted addition of avoidance time and return time; the interval control time T k The inverse of the aircraft spacing control strategy L k The fitness Y k ; At each population iteration, all fitness values ​​Y are retained k The aircraft interval control strategy individual L with a fitness greater than the preset optimal fitness k , and the remaining aircraft interval control strategy individuals L in the aircraft interval control strategy iteration population k The Markov decision model is used to update and obtain the new aircraft interval control strategy individual L k ; All aircraft interval control strategy individuals L with fitness greater than the preset optimal fitness k and all new aircraft separation control strategy individuals L k Combine to obtain an updated aircraft interval control strategy iteration population, thereby updating the aircraft interval control strategy iteration population; When the maximum number of iterations is reached, all aircraft interval control strategy individuals L whose fitness is greater than the preset optimal fitness are output. k , as a set of candidate aircraft control strategies for training aircraft; Collect candidate aircraft control strategy sets for several groups of training aircraft; combine candidate aircraft control strategy sets for several groups of training aircraft to obtain an aircraft control strategy training set; use the aircraft control strategy training set to perform model training on the initial interval control strategy calculation layer to obtain the initial interval control strategy calculation layer to be evaluated; perform model evaluation on the initial interval control strategy calculation layer to be evaluated to obtain a model evaluation result of the initial interval control strategy calculation layer to be evaluated; if the model evaluation result of the initial interval control strategy calculation layer to be evaluated is passed, use the trained initial interval control strategy calculation layer to be evaluated as the initial interval control strategy calculation layer; otherwise, continue model training using the aircraft control strategy training set.

6. The method for adaptive aircraft separation control according to claim 5, characterized in that: The specific steps of training the feedback monitoring layer include: Collecting several groups of aircraft separation control strategy prediction samples with target value annotations, where the target value is a quantitative evaluation result of the aircraft separation control strategy; combining several groups of aircraft separation control strategy prediction samples with target value annotations to obtain an aircraft control strategy feedback training set; The feedback monitoring layer is trained using the aircraft control strategy feedback training set to obtain the initial feedback monitoring layer; a model evaluation is performed on the initial feedback monitoring layer to obtain an initial feedback monitoring layer model evaluation result; if the initial feedback monitoring layer model evaluation result is passed, the initial feedback monitoring layer is used as the feedback monitoring layer; otherwise, the model training is continued using the aircraft control strategy feedback training set.

7. The method for adaptive aircraft separation control according to claim 6, characterized in that: The specific steps of step S4 include: Step S41: Collect aircraft H n Using the aircraft autonomous separation control model to simulate the control simulation data of aircraft adaptive separation control P n And record the corresponding aircraft flight conflict simulation test results; Step S42: Using the control simulation data P n Construct a target aircraft flight conflict training set with the corresponding aircraft flight conflict simulation test results; Step S43: Use the target aircraft flight conflict training set to update the parameters of the aircraft autonomous separation control model to obtain the target aircraft autonomous separation control model.

8. An aircraft adaptive separation control system, characterized in that: The system applies an aircraft adaptive separation control method according to any one of claims 1 to 7, including: The flight conflict basic unit module includes a construction unit for constructing the basic flight unit F of an aircraft. n , F n =(A n , B n , C n , D n ); where n represents the aircraft certification number, belonging to aircraft H n ; A n Indicates the aircraft flight direction angle, B n Indicates the aircraft flight speed, C n Indicates the minimum unit time of aircraft flight, D n Indicates the flight time of an aircraft unit; using the basic flight unit F of the aircraft n For aircraft H n Randomly generate aircraft flight trajectory vectors to obtain aircraft flight trajectory vector set X n ; The flight conflict scenario generation module includes a generation unit for generating a flight conflict model and a flight trajectory vector set X of all aircraft based on the flight conflict generation model. n Generate aircraft flight conflict scenarios; combine the generated aircraft flight conflict scenarios to obtain an aircraft flight conflict scenario training set; the flight conflict generation model includes a path generation layer, a path selection layer, a path optimization layer and a result output layer, which is constructed according to the CNN model and is used to construct and screen aircraft flight conflict scenarios; The flight conflict resolution module includes a training unit for training an autonomous aircraft separation control model using an aircraft flight conflict scenario training set; the autonomous aircraft separation control model includes a data extraction layer, an aircraft separation perception layer, an initial separation control strategy calculation layer, a feedback monitoring layer, and a control strategy output layer, which is constructed based on an improved Markov decision model and is used for outputting strategies when an aircraft flight conflict occurs; The flight conflict learning module includes a learning unit for aircraft H n A simulation test is performed with the aircraft autonomous separation control model to obtain an aircraft flight conflict simulation test result; the aircraft autonomous separation control model is updated according to the aircraft flight conflict simulation test result to obtain a target aircraft autonomous separation control model.

Images