Aircraft escape trajectory prediction method and device based on intelligent confrontation and storage medium

Abstract

The invention discloses an aircraft escape trajectory prediction method and device based on intelligent confrontation and a storage medium, and relates to the field of aircraft intelligent simulation. The method comprises the following steps: simulating escape trajectory data of an aircraft for avoiding a striking object in a simulated flight environment model; based on the escape trajectory data, training an aircraft escape trajectory prediction model through a loss function comprising a trajectory prediction error term and a physical constraint penalty term; and outputting an escape trajectory prediction result of the aircraft through the trained aircraft escape trajectory prediction model. According to the flight vehicle escape trajectory prediction model, the flight vehicle escape trajectory which is small in error and meets the flight vehicle dynamics constraint can be output according to the state of the striking object and the flight vehicle, and then the prediction precision of the flight vehicle escape trajectory is remarkably improved.

Description

Technical Field

[0001] This application relates to the field of intelligent aircraft simulation, specifically to a method, device, and storage medium for predicting aircraft escape trajectories based on intelligent countermeasures. Background Technology

[0002] When evading an attack, aircraft often employ complex flight maneuvers to increase their success rate in avoiding it.

[0003] Existing methods for predicting aircraft escape trajectories mainly fall into two categories:

[0004] Physical model-based prediction methods, such as Kalman filtering and interactive multi-model algorithms, perform well in predicting uniform or uniformly variable speed targets, but they are difficult to accurately predict the motion of aircraft under threat and have large prediction errors in highly dynamic combat environments.

[0005] Data-driven prediction methods: With the development of deep learning technology, methods such as recurrent neural networks and long short-term memory networks have been introduced into trajectory prediction tasks. However, these methods typically rely on historical observation data for training, facing the problem of scarce high-quality adversarial trajectory data in real-world adversarial scenarios.

[0006] At the same time, the above methods often treat trajectory prediction as a pure data fitting problem, without fully considering the constraints of aircraft dynamics, which makes the predicted trajectory physically infeasible, for example, the predicted trajectory is impossible for the aircraft to reach.

[0007] Therefore, the accuracy of existing methods for predicting aircraft escape trajectories is relatively low. Summary of the Invention

[0008] In view of the deficiencies in the existing technology, the technical problem to be solved by this application is: how to improve the prediction accuracy of the escape trajectory of an aircraft.

[0009] To achieve the above objectives, in a first aspect, embodiments of this application provide a method for predicting the escape trajectory of an aircraft based on intelligent countermeasures, the method comprising the following steps: Construct a simulated flight environment model, which includes a mathematical model of the impact object and a mathematical model of the aircraft. In a simulated flight environment model, escape trajectory data generated by the interaction between the mathematical model of the aircraft and the mathematical model of the target object is obtained. Based on escape trajectory data, the escape trajectory prediction model of the aircraft is trained using a loss function that includes trajectory prediction error term and physical constraint penalty term; The escape trajectory prediction model for the aircraft, once trained, outputs the predicted escape trajectory of the aircraft.

[0010] In conjunction with the first aspect, in one implementation, the process of acquiring escape trajectory data generated by the interaction between the aircraft mathematical model and the target mathematical model includes: in a simulated flight environment model, adversarially training the maneuver decision model of the aircraft mathematical model based on the aircraft mathematical model and the target mathematical model. The aircraft mathematical model, including the trained maneuver decision model, is subjected to adversarial simulation with the target mathematical model. The escape trajectory data generated by the interaction between the aircraft mathematical model and the target mathematical model during the adversarial simulation is used.

[0011] In conjunction with the first aspect, in one implementation, the process of adversarially training the maneuver decision model of the aircraft mathematical model includes: Set a multi-objective reward function based on the escape reward parameters of the aircraft; Escape reward parameters include survival time, ability to evade impacts, and physical constraint penalties; Survival time is the core reward of the multi-objective reward function; Survival time and the ability to evade attacks are both directly proportional to the reward score of the multi-objective reward function; physical constraint penalties are inversely proportional to the reward score of the multi-objective reward function.

[0012] In conjunction with the first aspect, in one implementation, the state space required for training the maneuver decision model includes the position and velocity of the aircraft, the position and velocity of the target, and the distance and line-of-sight angle between the target and the aircraft. The action space required for training the maneuver decision model includes the aircraft's angle of attack, throttle, and roll angle.

[0013] In conjunction with the first aspect, in one implementation, the training process of the aircraft escape trajectory prediction model includes: After the escape trajectory data is input into the aircraft escape trajectory prediction model, the aircraft escape trajectory prediction model outputs the corresponding predicted escape trajectory data. The difference between the predicted escape trajectory data and the corresponding escape trajectory data obtained from the simulated flight environment model is calculated using the loss function and used as the loss value. The escape trajectory prediction model for the aircraft is updated based on the loss value.

[0014] In conjunction with the first aspect, in one implementation, the loss function is:

[0015] For trajectory prediction error term, , These are the weighting coefficients; To predict the step size; The actual position at time t; Predict the position at time t; This is a physical constraint penalty term;

[0016] The maximum rate of change of heading angle, This represents the maximum available overload.

[0017] In conjunction with the first aspect, in one embodiment, the mathematical model of the striking object includes a six-degree-of-freedom dynamics and guidance model of the striking object; the mathematical model of the aircraft also includes a dynamics and flight control model.

[0018] In conjunction with the first aspect, in one implementation, the aircraft escape trajectory prediction model employs a hybrid neural network architecture, including: The spatiotemporal feature extraction layer is used to extract the motion features of the aircraft by sliding a one-dimensional convolution along the time dimension and outputting a dimensionality-reduced feature sequence. The temporal dependency modeling layer is used to: employ a BiGRU network, where the forward GRU learns the evolutionary logic from the past to the present maneuver, and the backward GRU learns the trajectory information from the future to the present during the training phase; The attention mechanism layer is used to: employ an additive attention model to calculate the weights of features at each time step; and perform a weighted summation of the BiGRU output. The output layer is used to: feed the weighted feature representation into the fully connected layer, map the preset output dimension, and generate the future trajectory information of the spacecraft.

[0019] Secondly, embodiments of this application provide an intelligent adversarial aircraft escape trajectory prediction device, the intelligent adversarial aircraft escape trajectory prediction device including a processor, a memory, and an intelligent adversarial aircraft escape trajectory prediction program stored in the memory and executable by the processor, wherein when the intelligent adversarial aircraft escape trajectory prediction program is executed by the processor, the method provided in the first aspect is implemented.

[0020] Thirdly, embodiments of this application provide a computer-readable storage medium storing a vehicle escape trajectory prediction program based on intelligent countermeasures, wherein the computer program, when executed, implements the method provided in the first aspect.

[0021] Compared with the prior art, the advantages of this application are: (1) This application first simulates the escape trajectory data of the aircraft to avoid the attack object through a simulated flight environment model, thereby solving the problem of the scarcity of high-quality combat trajectory data.

[0022] Furthermore, this application trains the aircraft escape trajectory prediction model using a loss function that includes a trajectory prediction error term and a physical constraint penalty term. This forces the model to learn a trajectory that conforms to real physical laws, ensuring the physical feasibility of the output results and overcoming the limitation of unreasonable prediction results in traditional data-driven models.

[0023] Based on this, the aircraft escape trajectory prediction model can output an escape trajectory prediction result with small error and in accordance with the dynamic constraints of the aircraft, according to the state of the impact object and the aircraft, thereby significantly improving the prediction accuracy of the aircraft escape trajectory.

[0024] (2) In this application, an escape reward function based on survival time is set in the escape trajectory data of the simulated aircraft evading the attacking object. The escape reward function also includes the ability to evade the attacking object and physical constraint penalties, thereby improving the physical reliability of the training data. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart illustrating the aircraft escape trajectory prediction method based on intelligent countermeasures in the embodiments of this application; Figure 2 This is a schematic diagram illustrating the training of aircraft escape trajectory data in the embodiments of this application; Figure 3 This is a schematic diagram of the aircraft training reward curve in the embodiments of this application; Figure 4 This is a schematic diagram of the predicted loss variation curve of the aircraft escape trajectory in the embodiments of this application; Figure 5 This is a schematic diagram of the hardware structure of the aircraft escape trajectory prediction device based on intelligent countermeasures involved in the embodiments of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0029] First, a brief description of the research and development principles of this application: Existing methods for predicting escape trajectories of aircraft have significant limitations. These limitations include: a lack of effective modeling of intelligent evasion behavior of aircraft under threat of attack objects; insufficient and limited training data that fails to cover complex tactical maneuvers; a lack of deep integration of prediction models with physical constraints, resulting in output trajectories that often exceed the aircraft's maneuverability; and a disconnect between the prediction system and subsequent decision-making processes, making it unable to directly support real-time tactical decision-making.

[0030] Based on this, in order to make the objectives, technical solutions and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0031] Firstly, embodiments of this application provide a method for predicting aircraft escape trajectories based on intelligent countermeasures, see [link to relevant documentation]. Figure 1 As shown, the steps of this method include: Step A: Construct a simulated flight environment model, which includes a mathematical model of the target (specifically, a six-degree-of-freedom dynamics and guidance model of the target) and a mathematical model of the aircraft (specifically, a dynamics and flight control model of the aircraft, as well as a maneuver decision model).

[0032] It should be noted that those skilled in the art can obtain corresponding mathematical models of aircraft and targets based on real information about aircraft and targets, according to different needs.

[0033] Step B: Obtain escape trajectory data generated by the interaction between the aircraft mathematical model and the impact object mathematical model; the specific process includes: In the simulated flight environment model, the maneuver decision model of the aircraft mathematical model is subjected to adversarial training based on the aircraft mathematical model and the target mathematical model (i.e., the relative positional relationship between the aircraft and the target, etc.) until its strategy converges. In the simulated flight environment model, the aircraft mathematical model, which includes a trained maneuver decision model, will be used to simulate an adversarial interaction with the target mathematical model. Escape trajectory data generated by the interaction between the aircraft mathematical model and the target mathematical model will be collected. The escape trajectory data will be used to characterize the attack capability boundary of the target and provide support for the construction of targeted cooperative kill zones.

[0034] Steps A and B generate escape trajectory data for aircraft by training a maneuvering decision-making model (agent) and having it engage in combat against an attacking object in a simulated environment, thus solving the problem of scarce high-quality combat trajectory data.

[0035] Step C: Based on the escape trajectory dataset obtained in Step B, train the aircraft escape trajectory prediction model offline. During training, optimize the model parameters by minimizing a composite loss function. This loss function is composed of a weighted sum of the trajectory prediction error term and the physical constraint penalty term. This forces the model to learn a trajectory that conforms to the real physical laws, ensuring the physical feasibility of the output results and breaking through the limitation of unreasonable prediction results of traditional data-driven models.

[0036] Step D: Deploy the aircraft escape trajectory prediction model after training. When it is necessary to evade an attack object, the aircraft escape trajectory prediction model outputs the predicted escape trajectory of the aircraft based on the state of the attack object and the aircraft. The predicted escape trajectory specifically consists of multiple discrete trajectories of the aircraft over a future period and their corresponding probabilities. These discrete trajectories are used to generate a continuous probability escape region for the aircraft through a Gaussian mixture model. This region quantitatively represents the future spatial uncertainty of the aircraft, providing direct input for subsequent tactical decisions.

[0037] In one embodiment, the state space required for training the maneuver decision model in step B includes the position and velocity of the aircraft, the position and velocity of the target, and the distance and line-of-sight angle between the target and the aircraft. The action space required for training the maneuver decision model in step B includes the aircraft's angle of attack, throttle, and roll angle.

[0038] In one embodiment, the process of adversarial training of the maneuver decision model of the aircraft mathematical model in step B includes: A multi-objective reward function is set based on the escape reward parameters of the aircraft. These parameters include multiple dimensions such as survival time, object avoidance capability, and physical constraint penalty. Survival time is the core reward of the multi-objective reward function. Both survival time and object avoidance capability are directly proportional to the reward score of the multi-objective reward function, while the physical constraint penalty is inversely proportional to the reward score.

[0039] During adversarial training, when the cumulative reward score obtained by the maneuver decision model in a certain round exceeds a specified threshold, the decision model parameters for that round are saved.

[0040] Specifically, step B is implemented using deep reinforcement learning training. In this embodiment, the architecture employs offline experience replay and a dual-Q value network (specifically, existing technology) to improve learning stability and sample efficiency.

[0041] See Figure 2 As shown, the training process of the dual-Q value network architecture used in this application is as follows: Interaction and Storage: In a combat environment where objects lock onto aircraft, the states (position, velocity, etc.) of both sides are input into the policy network, and the network outputs the aircraft's maneuvers (such as overload commands). After the environment executes the action, a new state and an immediate reward are generated, and this experience data (state, action, reward, new state) is stored in the experience replay pool.

[0042] Sampling and updating: When the amount of data in the experience replay pool reaches a specified scale, a batch of data is randomly sampled, and the target Q value is calculated using two value networks (Q1 and Q2). The smaller value is then used to update the network parameters to reduce overestimation of value.

[0043] Iterative loop: The above interaction-update process is repeated tens of thousands of times. Driven by rewards, the agent fully explores various maneuvers within its flight envelope, thereby learning complex and efficient escape strategies. The training results of the aircraft are as follows: Figure 3 As shown, the cumulative reward (escape trajectory score) increases continuously with the number of training rounds, indicating that the strategy is continuously optimized.

[0044] Therefore, it can be seen that the above method can realize the intelligent generation of batch training data (escape trajectory data) covering multiple tactical scenarios based on deep reinforcement learning; moreover, the design of a reward function oriented towards survival time guides the maneuver decision model to make evasion decisions that conform to the laws of dynamics, fundamentally improving the physical rationality and adversarial realism of the data itself.

[0045] In one embodiment, to simultaneously capture the spatiotemporal characteristics of aircraft motion and the long-term evolution of tactical intent, the aircraft escape trajectory prediction model in step C adopts a hybrid neural network architecture, specifically including: The spatiotemporal feature extraction layer is used to extract the motion features of the aircraft by sliding one-dimensional convolution (Conv1D) along the time dimension and output the dimensionality-reduced feature sequence. The temporal dependency modeling layer is used to: employ a bidirectional gated recurrent unit (BiGRU) network, where the forward GRU learns the evolution logic of maneuvers from history to the present, and the backward GRU learns trajectory information from the future to the present during the training phase, jointly capturing the evolution of tactical intentions; the gating mechanism can effectively alleviate the gradient vanishing problem.

[0046] The attention mechanism layer is used to: employ an additive attention model to calculate the weights of features at each time step; and to perform a weighted summation of the BiGRU output to focus the model on key tactical moments (such as turning and acceleration phases).

[0047] The output layer is used to: feed the weighted feature representation into the fully connected layer, map the preset output dimension, and thus generate the future trajectory information of the aircraft.

[0048] In one embodiment, the process of training the aircraft escape trajectory prediction model offline based on the escape trajectory dataset obtained in step B in step C includes: After the escape trajectory dataset is input into the aircraft escape trajectory prediction model, the aircraft escape trajectory prediction model outputs the corresponding escape trajectory prediction value; The difference between the predicted escape trajectory value and the corresponding escape trajectory data (i.e., the real trajectory in the escape trajectory data obtained from the simulated flight environment model) is calculated using the above loss function and used as the loss value. The model parameters are updated based on the loss value using the backpropagation algorithm to minimize the prediction error.

[0049] See Figure 4 As shown, the loss value continues to decrease with the increase of training rounds, indicating that the prediction accuracy of the model is constantly improving.

[0050] Furthermore, the loss function in step C is:

[0051] in For trajectory prediction error term, , These are the weighting coefficients; To predict the step size; The actual position at time t; Predict the position at time t; This is a physical constraint penalty term;

[0052] The maximum rate of change of heading angle, This represents the maximum available overload.

[0053] In summary, the effects of this application are as follows: (1) Effectively solves the problem of scarce training data in aircraft escape trajectory prediction: Through deep reinforcement learning to simulate intelligent avoidance of aircraft, a large amount of diverse and physically feasible adversarial trajectory data is automatically generated.

[0054] (2) Improved prediction accuracy: The model integrates time-dependent modeling and attention mechanism, which can effectively capture the evolution of aircraft tactical intentions and improve prediction accuracy.

[0055] (3) High physical credibility: Dynamic constraints are explicitly embedded in the loss function to ensure that the predicted trajectory conforms to the actual maneuverability of the aircraft and avoids outputting prediction results that are physically infeasible.

[0056] (4) Strong system adaptability: The generated adversarial data can cover a variety of combat scenarios, enabling the model to generalize to new evasion strategies.

[0057] Secondly, embodiments of this application provide an aircraft escape trajectory prediction device based on intelligent countermeasures. The aircraft escape trajectory prediction device based on intelligent countermeasures can be a personal computer (PC), laptop computer, server, or other device with data processing capabilities.

[0058] Reference Figure 5 , Figure 5 This is a schematic diagram of the hardware structure of the aircraft escape trajectory prediction device based on intelligent countermeasures involved in the embodiments of this application. In the embodiments of this application, the aircraft escape trajectory prediction device based on intelligent countermeasures may include a processor, a memory, a communication interface, and a communication bus.

[0059] The communication bus can be of any type and is used to interconnect the processor, memory, and communication interface.

[0060] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces used for interconnecting internal components of the intelligent warfare-based aircraft escape trajectory prediction device, as well as interfaces used for interconnecting the intelligent warfare-based aircraft escape trajectory prediction device with other devices (such as other computing devices or user equipment). Physical interfaces can be Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc.; user equipment can be displays, keyboards, etc.

[0061] Memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0062] The processor can be a general-purpose processor, which can call a spacecraft escape trajectory prediction program based on intelligent adversarial tactics stored in memory and execute the spacecraft escape trajectory prediction method based on intelligent adversarial tactics provided in the embodiments of this application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the spacecraft escape trajectory prediction program based on intelligent adversarial tactics is called can be referred to in the various embodiments of the spacecraft escape trajectory prediction method based on intelligent adversarial tactics of this application, and will not be repeated here.

[0063] Those skilled in the art will understand that Figure 5 The hardware structure shown does not constitute a limitation of this application and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0064] Thirdly, embodiments of this application also provide a computer-readable storage medium.

[0065] The computer-readable storage medium of this application stores a vehicle escape trajectory prediction program based on intelligent countermeasures, wherein when the vehicle escape trajectory prediction program based on intelligent countermeasures is executed by a processor, it implements the steps of the vehicle escape trajectory prediction method based on intelligent countermeasures as described above.

[0066] The method implemented when the intelligent adversarial aircraft escape trajectory prediction program is executed can be referred to in the various embodiments of the intelligent adversarial aircraft escape trajectory prediction method of this application, and will not be repeated here.

[0067] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0068] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0069] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0070] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0071] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0072] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0073] The above are merely specific embodiments of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the scope of the claims.

Claims

1. A method for predicting the escape trajectory of an aircraft based on intelligent adversarial tactics, characterized in that, The method includes the following steps: Construct a simulated flight environment model, which includes a mathematical model of the impact object and a mathematical model of the aircraft. In a simulated flight environment model, escape trajectory data generated by the interaction between the mathematical model of the aircraft and the mathematical model of the target object is obtained. Based on escape trajectory data, the escape trajectory prediction model of the aircraft is trained using a loss function that includes trajectory prediction error term and physical constraint penalty term; The escape trajectory prediction model for the aircraft, once trained, outputs the predicted escape trajectory of the aircraft.

2. The aircraft escape trajectory prediction method based on intelligent countermeasures as described in claim 1, characterized in that, The process of acquiring escape trajectory data generated by the interaction between the aircraft mathematical model and the target mathematical model includes: in a simulated flight environment model, adversarially training the maneuver decision model of the aircraft mathematical model based on the aircraft mathematical model and the target mathematical model; The aircraft mathematical model, including the trained maneuver decision model, is subjected to adversarial simulation with the target mathematical model. The escape trajectory data generated by the interaction between the aircraft mathematical model and the target mathematical model during the adversarial simulation is used.

3. The aircraft escape trajectory prediction method based on intelligent countermeasures as described in claim 2, characterized in that, The process of adversarially training the maneuver decision model of the aircraft mathematical model includes: Set a multi-objective reward function based on the escape reward parameters of the aircraft; Escape reward parameters include survival time, ability to evade impacts, and physical constraint penalties; Survival time is the core reward of the multi-objective reward function; Survival time and the ability to evade attacks are both directly proportional to the reward score of the multi-objective reward function; physical constraint penalties are inversely proportional to the reward score of the multi-objective reward function.

4. The aircraft escape trajectory prediction method based on intelligent countermeasures as described in claim 2, characterized in that, The state space required for training the maneuver decision model includes the position and velocity of the aircraft, the position and velocity of the target, and the distance and line-of-sight angle between the target and the aircraft. The action space required for training the maneuver decision model includes the aircraft's angle of attack, throttle, and roll angle.

5. The aircraft escape trajectory prediction method based on intelligent countermeasures as described in claim 1, characterized in that, The training process for the aircraft escape trajectory prediction model includes: After the escape trajectory data is input into the aircraft escape trajectory prediction model, the aircraft escape trajectory prediction model outputs the corresponding predicted escape trajectory data. The difference between the predicted escape trajectory data and the corresponding escape trajectory data obtained from the simulated flight environment model is calculated using the loss function and used as the loss value. The escape trajectory prediction model for the aircraft is updated based on the loss value.

6. The aircraft escape trajectory prediction method based on intelligent countermeasures as described in claim 5, characterized in that, The loss function is: For trajectory prediction error term, , These are the weighting coefficients; To predict the step size; The actual position at time t; Predict the position at time t; This is a physical constraint penalty term; The maximum rate of change of heading angle, This represents the maximum available overload.

7. The aircraft escape trajectory prediction method based on intelligent countermeasures as described in any one of claims 1 to 4, characterized in that, The mathematical model of the impact object includes a six-degree-of-freedom dynamics and guidance model of the impact object; the mathematical model of the aircraft also includes a dynamics and flight control model.

8. The aircraft escape trajectory prediction method based on intelligent countermeasures as described in any one of claims 1 to 4, characterized in that, The aircraft escape trajectory prediction model adopts a hybrid neural network architecture, including: The spatiotemporal feature extraction layer is used to extract the motion features of the aircraft by sliding a one-dimensional convolution along the time dimension and outputting a dimensionality-reduced feature sequence. The temporal dependency modeling layer is used to: employ a BiGRU network, where the forward GRU learns the evolutionary logic from the past to the present maneuver, and the backward GRU learns the trajectory information from the future to the present during the training phase; The attention mechanism layer is used to: employ an additive attention model to calculate the weights of features at each time step; and perform a weighted summation of the BiGRU output. The output layer is used to: feed the weighted feature representation into the fully connected layer, map the preset output dimension, and generate the future trajectory information of the spacecraft.

9. A device for predicting the escape trajectory of an aircraft based on intelligent countermeasures, characterized in that, The intelligent adversarial-based aircraft escape trajectory prediction device includes a processor, a memory, and an intelligent adversarial-based aircraft escape trajectory prediction program stored in the memory and executable by the processor, wherein when the intelligent adversarial-based aircraft escape trajectory prediction program is executed by the processor, it implements the steps of the intelligent adversarial-based aircraft escape trajectory prediction method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a vehicle escape trajectory prediction program based on intelligent countermeasures, wherein when the vehicle escape trajectory prediction program based on intelligent countermeasures is executed, it implements the steps of the vehicle escape trajectory prediction method based on intelligent countermeasures as described in any one of claims 1 to 8.

Images