Aircraft flight path planning method and device based on large language model and medium

Abstract

The invention discloses an aircraft flight path planning method and device based on a large language model and a medium, and belongs to the technical field of aviation path planning. According to the method, the initial flight path of the target aircraft capable of avoiding the dangerous weather area is obtained, then the initial flight path is optimized, and the optimal path planning strategy for preventing the target aircraft from colliding with other aircrafts is determined, so that the flight path of the aircraft can be reasonably planned according to the weather and the airspace condition around the aircraft; flight safety is improved, and the problem that at present, an aircraft can only be guided to change the path through ground command is solved.

Description

Technical Field

[0001] The present invention relates to a method, device and medium for aircraft flight path planning based on a large language model, belonging to the technical field of air route planning. Background Art

[0002] When an aircraft encounters sudden situations such as weather, the original flight route cannot be maintained, and it is necessary to adjust the flight track in real time to perform evasive actions to get out of the dangerous area. This dynamic bypass behavior not only breaks the execution chain of the existing control and command plans, but also significantly increases the decision-making difficulty and workload of the air traffic control department, and is extremely likely to induce risk events of "omission, forgetting and missing" caused by command mistakes, thereby causing the expansion of economic losses and even the derivation of secondary safety accidents.

[0003] In the air traffic control operation system, ground-air voice communication is the core means of interaction between air traffic controllers and flight crews, and relies on the very high frequency radio system to achieve real-time instruction transmission. This interaction process includes a triple confirmation mechanism: the air traffic controller needs to issue instruction information through voice, the pilot must repeat the instruction content completely, and then the air traffic controller makes a final confirmation of the repeated content, so as to construct a closed-loop verification system for instruction transmission. In view of the dual requirements of civil aviation operation for safety margin and operation efficiency, the existing voice recognition system still cannot be directly applied to the real-time command scenario, and its function positioning is still limited to the post-analysis level. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art, and provide a method, device and medium for aircraft flight path planning based on a large language model, which can reasonably plan the flight path of the aircraft according to the weather and the airspace conditions around the aircraft, improve flight safety, and solve the problem that the aircraft can only change its path through ground command and guidance at present.

[0005] To solve the above technical problems, the present invention is implemented by adopting the following technical solutions:

[0006] The first aspect of the present invention provides a method for aircraft flight path planning based on a large language model, including:

[0007] Obtain the real-time weather information of the cities along the flight mission of the target aircraft;

[0008] Input the real-time weather information into a pre-constructed large language area prediction model to obtain an initial flight path for the target aircraft to avoid dangerous weather areas;

[0009] Optimize the initial flight path to determine an optimal path planning strategy that prevents the target aircraft from colliding with other aircraft in the surrounding airspace;

[0010] Control the target aircraft to adjust its flight path according to the optimal path planning strategy.

[0011] Further, the step of inputting the real-time weather information into a pre-constructed large language area prediction model to obtain an initial flight path for the target aircraft to avoid dangerous weather areas includes:

[0012] According to the real-time weather information, use the Graham scanning method to obtain an initial set of boundary points for the dangerous weather area;

[0013] Based on the initial set of boundary points, use a grey prediction model to predict the future boundary of the dangerous weather area to obtain a flight restricted area;

[0014] According to the flight restricted area, obtain the initial flight path.

[0015] Even further, the step of optimizing the initial flight path to determine an optimal path planning strategy for the target aircraft not to collide with other aircraft in the surrounding airspace includes:

[0016] Use the SARSA algorithm to map the target aircraft and its initial flight path to an observable virtual environment space;

[0017] Construct a reward function corresponding to the actions performed by the target aircraft in the virtual environment space;

[0018] Make the target aircraft act in the virtual environment space to determine an optimal path planning strategy in which the target aircraft does not collide with other aircraft, can reach the destination, and obtains the highest reward value.

[0019] Even further, the step of using the SARSA algorithm to map the target aircraft and its initial flight path to an observable virtual environment space includes:

[0020] Represent the observable virtual environment space through a tuple:

[0021] ;

[0022] Where:

[0023] represents the state space, which consists of the states corresponding to the target aircraft at each moment;

[0024] represents the action space, which consists of the actions performed by the target aircraft at each moment;

[0025] represents the reward function, which consists of the rewards given after the target aircraft performs actions at each moment.

[0026] Furthermore, the reward function includes:

[0027] A distance reward function, which is expressed by the following formula:

[0028] ;

[0029] ;

[0030] ;

[0031] represents the intruder reward function at time represents the distance threshold between the target aircraft and other aircraft in the vicinity of the airspace, represents the closest distance between the target aircraft and other aircraft in the vicinity of the airspace;

[0032] represents the remaining distance reward function at time represents the function for calculating the distance between the target aircraft and the destination;

[0033] represents the arrival at destination reward function at time

[0034] A weather reward function, which is expressed by the following formula:

[0035] ;

[0036] ;

[0037] ;

[0038] represents the rainfall reward function at time represents the rainfall in the environment where the target aircraft is located;

[0039] represents the visibility reward function at time represents the visibility in the environment where the target aircraft is located;

[0040] represents the cloud cover reward function at time represents the cloud cover in the environment where the target aircraft is located;

[0041] A flight restricted area reward function, which is expressed by the following formula:

[0042] ;

[0043] represents the flight restricted area reward function, represents the maximum reward, represents the distance between the aircraft and the center of the minimum enclosing circle corresponding to the flight restricted area, represents the radius of the minimum enclosing circle corresponding to the flight restricted area.

[0044] Furthermore, the optimal path planning strategy for making the target aircraft act in the virtual environment space, determining that the target aircraft does not collide with other aircraft, can reach the destination, and obtains the highest reward value, includes:

[0045] Updating the current state-action pair of the target aircraft using the state-action value function, enabling the target aircraft to select the optimal path planning strategy for action, and updating the current state-action pair of the target aircraft is expressed as:

[0046] ;

[0047] Where:

[0048] represents the estimated value of the current state-action pair ;

[0049] represents the update of the estimated value of the current state-action pair ;

[0050] represents the learning rate, used to control the update step size;

[0051] represents the time series error, ;

[0052] represents the state corresponding to the target aircraft at the current moment, represents the action executed by the target aircraft at the current moment;

[0053] represents the reward at the next moment after the target aircraft executes the action ; represents the state corresponding to the target aircraft at the next moment, represents the action executed by the target aircraft at the next moment;

[0054] represents the estimated value of the state-action pair at the next moment;

[0055] denotes the discount factor;

[0056] The reward value corresponding to the action of the target aircraft in the virtual environment space is calculated by the following formula:

[0057] ;

[0058] denotes 、 and the minimum value among

[0059] Furthermore, based on the initial boundary point set, a grey prediction model is used to predict the future boundary of the hazardous weather area to obtain a flight restricted area, including:

[0060] Observing the initial boundary point set at a fixed time interval to obtain a set of original observation sequences representing the change of the initial boundary over time;

[0061] Calculating the grey parameters of the original observation sequence corresponding to each initial boundary point;

[0062] Inputting the grey parameters into the grey prediction model to obtain the boundary point coordinates at future moments;

[0063] Obtaining the flight restricted area according to the boundary point coordinates at each future moment.

[0064] Furthermore, the calculation of the grey parameters of the original observation sequence corresponding to each initial boundary point includes:

[0065] Calculating the grey parameters of the original observation sequence corresponding to each initial boundary point by the following formula:

[0066] ;

[0067] ;

[0068] Where:

[0069] denotes the grey parameter of the original observation sequence of the abscissa corresponding to the initial boundary point;

[0070] denotes the cumulative data of the abscissa of the initial boundary point, denotes the cumulative matrix of the abscissa, denotes the constant vector of the abscissa;

[0071] denotes the grey parameter of the original observation sequence of the ordinate corresponding to the initial boundary point;

[0072] denotes the cumulative data of the ordinate of the initial boundary point, The cumulative matrix representing the ordinate; The ordinate constant vector.

[0073] The second aspect of the present invention provides an aircraft flight path planning device based on a large language model, including:

[0074] An acquisition module for acquiring real-time weather information of cities along the flight mission of the target aircraft;

[0075] A path planning module for inputting the real-time weather information into a pre-constructed large language area prediction model to obtain an initial flight path for the target aircraft to avoid dangerous weather areas;

[0076] A path optimization module for optimizing the initial flight path to determine an optimal path planning strategy for preventing the target aircraft from colliding with other aircraft;

[0077] A processing module for controlling the target aircraft to adjust its flight path according to the optimal path planning strategy.

[0078] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the aircraft flight path planning method based on a large language model as described above.

[0079] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0080] 1. The present invention utilizes a pre-constructed large language area prediction model to obtain an initial flight path for the target aircraft to avoid dangerous weather areas according to the real-time weather information of cities along the flight mission of the target aircraft, and then optimizes the initial flight path to determine an optimal path planning strategy for preventing the target aircraft from colliding with other aircraft. The present invention can reasonably plan the flight path of the aircraft according to the weather and the airspace situation around the aircraft, improving flight safety and solving the problem that currently only ground command can be used to guide the aircraft to change its path.

[0081] 2. The present invention uses the SARSA algorithm to map the target aircraft and its initial flight path to an observable virtual environment space, enabling the target aircraft to act in the virtual environment space and evaluating the state of the target aircraft after performing corresponding actions in combination with a reward function, which can quickly and effectively optimize the initial flight path. Description of the Drawings

[0082] Figure 1 It is a flowchart of an aircraft flight path planning method based on a large language model provided by an embodiment of the present invention;

[0083] Figure 2It is a schematic diagram of weather reward values at each moment in four cities under hazy weather provided by an embodiment of the present invention;

[0084] Figure 3 It is a schematic diagram of weather reward values at each moment in four cities under sunny weather provided by an embodiment of the present invention;

[0085] Figure 4 It is a schematic diagram of weather reward values at each moment in four cities under light snow weather provided by an embodiment of the present invention;

[0086] Figure 5 It is a schematic diagram of the average reward value of aircraft under three typical weather scenarios provided by an embodiment of the present invention;

[0087] Figure 6 It is a schematic diagram of the trend of the number of modifications of the aircraft heading angle provided by an embodiment of the present invention;

[0088] Figure 7 It is a schematic diagram of the trend of the number of modifications of the aircraft speed provided by an embodiment of the present invention;

[0089] Figure 8 It is a schematic diagram of the trend of the number of modifications of the aircraft altitude provided by an embodiment of the present invention. Detailed implementation manners

[0090] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention.

[0091] Embodiment 1

[0092] As Figure 1 shown, a method for aircraft flight path planning based on a large language model includes:

[0093] Obtaining real-time weather information of cities along the flight mission of the target aircraft;

[0094] Inputting the real-time weather information into a pre-constructed large language area prediction model to obtain an initial flight path for the target aircraft to avoid dangerous weather areas. Specifically:

[0095] In this embodiment, the real-time weather information is a satellite cloud image;

[0096] Performing denoising and merging processing on the areas with dangerous weather in the satellite cloud image, ignoring areas with a size less than 7 miles, merging areas with a minimum distance less than 20 km, selecting a reference point near the area as the coordinate origin O, constructing a plane rectangular coordinate system, with the due north direction as the positive direction of the Y-axis and the direction 90° east of due north as the positive direction of the X-axis;

[0097] Use the Graham scan method to obtain the set of initial boundary points of the convex polygon formed by the hazardous weather area at time , , where represents the coordinates of the -th initial boundary point of the hazardous weather area at time . The -th initial boundary point has coordinates , , where represents the abscissa of the -th initial boundary point at time , and represents the ordinate of the -th initial boundary point at time

[0098] , represents a total of time instants, , represents a total of initial boundary points;

[0099] Based on the set of initial boundary points, use the grey prediction model to predict the future boundary of the hazardous weather area to obtain the flight restricted area, including:

[0100] Observe the set of initial boundary points at fixed time intervals to obtain the set of original observation sequences representing the change of the initial boundary over time;

[0101] Calculate the grey parameters of the original observation sequence corresponding to each initial boundary point, including: Calculate the grey parameters of the original observation sequence corresponding to each initial boundary point through the following formula:

[0102] ;

[0103] ;

[0104] represents the grey parameter of the original observation sequence of the abscissa corresponding to the initial boundary point. The original observation sequence of the abscissa of the -th initial boundary point is expressed as ; ;

[0105] represents the cumulative data of the abscissa of the initial boundary point, ;

[0106] ; represents the abscissa of the -th initial boundary point at the -th time instant among a total of time instants;

[0107] The cumulative matrix representing the abscissa;

[0108] ;

[0109] The constant vector representing the abscissa;

[0110] ;

[0111] Represents the grey parameter of the original observation sequence of the ordinate corresponding to the initial boundary point. The original observation sequence of the ordinate of the th initial boundary point is expressed as ; ;

[0112] Represents the cumulative data of the ordinate of the initial boundary point, ;

[0113] ; Represents the th moment among a total of th moment and the th initial boundary point of the ordinate;

[0114] The cumulative matrix representing the ordinate;

[0115] ;

[0116] The constant vector representing the ordinate;

[0117] ;

[0118] Input the grey parameter into the grey prediction model to obtain the set of boundary point coordinates at future moments ;

[0119]

[0120] Represents the coordinates of the th predicted boundary point of the dangerous weather area at future th moment, ; Represents the abscissa of the th predicted boundary point at future th moment, Represents the ordinate of the th predicted boundary point at future th moment;

[0121] Calculate the midpoint position of the convex polygon at the future moment by the following formula :

[0122] ;

[0123] Calculate the slope of the line passing through the midpoint of the convex polygon and the predicted boundary point by the following formula:

[0124] ;

[0125] In this embodiment, when the target aircraft enters the spare buffer zone which is widened by 25 km outside the convex polygon, it is considered to enter the dangerous weather area. Calculate the boundary point coordinates at the future moment after widening by the following formula :

[0126] ;

[0127] ;

[0128] Obtain the initial flight path according to the convex polygon flight restricted area surrounded by the boundary point coordinates at the future moment after widening.

[0129] Optimize the initial flight path to determine the optimal path planning strategy that enables the target aircraft not to collide with other aircraft in the surrounding airspace. Specifically:

[0130] Use the SARSA algorithm to map the target aircraft and its initial flight path to the observable virtual environment space, including:

[0131] Represent the observable virtual environment space by a tuple:

[0132] ;

[0133] represents the state space, which is composed of the states corresponding to each moment ; , represents the position coordinates of the target aircraft, represents the speed of the target aircraft, represents the heading angle of the target aircraft;

[0134] represents the action space, which is composed of the actions executed by the target aircraft at each moment ; , Indicates the change in the heading angle of the target aircraft, Indicates the change in the speed of the target aircraft, Indicates the change in the altitude of the target aircraft; The aircraft mainly changes the heading angle, speed, and altitude through the actions in Table 1;

[0135] Table 1 Specific action description

[0136]

[0137] Indicates the reward function, Composed of the rewards given after the target aircraft executes actions at each moment ;

[0138] Construct the reward function corresponding to the actions executed by the target aircraft in the virtual environment space, where the reward function includes:

[0139] Distance reward function, the distance reward function is represented by the following formula:

[0140] ;

[0141] ;

[0142] ;

[0143] Indicates The reward function of the intruder at time Indicates the distance threshold between the target aircraft and other aircraft around the airspace, Indicates the closest distance between the target aircraft and other aircraft around the airspace;

[0144] Indicates The remaining distance reward function at time Indicates the function for calculating the distance between the target aircraft and the destination;

[0145] Indicates The reward function for arriving at the destination at time

[0146] Weather reward function, the weather reward function is represented by the following formula:

[0147] ;

[0148] ;

[0149] ;

[0150] Indicates Instantaneous rainfall reward function, representing the rainfall in the environment where the target aircraft is located;

[0151] indicating Instantaneous visibility reward function, representing the visibility in the environment where the target aircraft is located;

[0152] indicating Instantaneous cloud cover reward function, representing the cloud cover in the environment where the target aircraft is located;

[0153] The reward function also includes a flight restricted area reward function. Specifically:

[0154] Based on the convex polygon flight restricted area, calculate the radius of the smallest circumscribed circle that can enclose the entire flight restricted area, and construct the flight restricted area reward function as follows:

[0155] ;

[0156] representing the flight restricted area reward function, representing the maximum reward, ;

[0157] representing the distance between the aircraft and the center of the smallest circumscribed circle corresponding to the flight restricted area, representing the radius of the smallest circumscribed circle corresponding to the flight restricted area.

[0158] Make the target aircraft act in the virtual environment space, and determine the optimal path planning strategy in which the target aircraft does not collide with other aircraft, can reach the destination, and obtains the highest reward value, including:

[0159] Use the state-action value function to update the current state-action pair of the target aircraft, so that the target aircraft can select the optimal path planning strategy to act. The update of the current state-action pair of the target aircraft is expressed as:

[0160] ;

[0161] representing the estimated value of the current state-action pair ; representing the update of the estimated value of the current state-action pair ; representing the learning rate, used to control the update step size; representing the time series error, ; Indicates the state corresponding to the target aircraft at the current moment, Indicates the action performed by the target aircraft at the current moment; Indicates the reward at the next moment after performing the current action, Indicates the state corresponding to the target aircraft at the next moment, Indicates the action performed by the target aircraft at the next moment;

[0162] Indicates the estimated value of the state-action pair at the next moment ;

[0163] Indicates the discount factor;

[0164] The reward value corresponding to the action of the target aircraft in the virtual environment space is calculated using the following formula:

[0165] ;

[0166] Indicates , and the minimum value among;

[0167] The target aircraft is controlled to fly using the action corresponding to the optimal path planning strategy.

[0168] To verify the method proposed in this example, the following data is used for simulation experiments:

[0169] Software: Python, select the Chinese map for spatial mapping;

[0170] As Figure 2 , Figure 3 and Figure 4 shown, select April 28, 2024, July 31, 2024, and December 10, 2024, as three typical weather scenarios of haze, sunny, and snowy days respectively, and conduct an example analysis on the two-way passenger route of ZGGG (Guangzhou Baiyun International Airport) - ZWWW (Urumqi Tianshan International Airport). In the actual weather system experiment, the route mainly passes through two nodal cities, Lanzhou and Changsha, and real-time flight restricted areas are constructed using grey prediction in the two nodal cities;

[0171] As Figure 5As shown, in the initial stage of training, as the number of rounds increases, the average reward value of the aircraft increases rapidly and reaches a peak. Subsequently, due to the influence of weather penalties, the average reward value slowly decreases. After approximately 5,000 times of training, it fluctuates slightly near 1510, 1264, and 1179 respectively, showing a stable trend, indicating that the aircraft has adapted to the negative impacts brought by the environment after a large number of iterative trainings;

[0172] As Figure 6 , Figure 7 and Figure 8 shown, the number of times the aircraft's heading angle is modified is significantly more than the number of times the speed and altitude are modified. This indicates that when the aircraft faces potential conflicts, it preferentially adjusts the heading angle to avoid collisions, which is highly consistent with the command operations of traditional air traffic controllers. The method proposed in this embodiment is in line with human experience in terms of decision-making logic.

[0173] Embodiment 2

[0174] An aircraft flight path planning device based on a large language model, comprising:

[0175] An acquisition module for acquiring real-time weather information of cities along the flight mission of the target aircraft;

[0176] A path planning module for inputting the real-time weather information into a pre-constructed large language area prediction model to obtain an initial flight path for the target aircraft to avoid dangerous weather areas;

[0177] A path optimization module for optimizing the initial flight path to determine an optimal path planning strategy for the target aircraft not to collide with other aircraft;

[0178] A processing module for controlling the target aircraft to adjust its flight path according to the optimal path planning strategy.

[0179] Embodiment 3

[0180] A computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the following method are implemented:

[0181] Acquire real-time weather information of cities along the flight mission of the target aircraft;

[0182] Input the real-time weather information into a pre-constructed large language area prediction model to obtain an initial flight path for the target aircraft to avoid dangerous weather areas;

[0183] Perform path optimization on the initial flight path to determine an optimal path planning strategy for the target aircraft not to collide with other aircraft in the surrounding airspace;

[0184] According to the optimal path planning strategy, control the target aircraft to adjust its flight path.

[0185] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, apparatus, or computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0186] The present application is described with reference to the flowcharts of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each process in the flowchart can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate an apparatus for implementing the functions specified in Figure 1 one process or multiple processes or boxes Figure 1 one box or multiple boxes.

[0187] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction apparatus that implements the functions specified in Figure 1 one process or multiple processes.

[0188] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one process or multiple processes.

[0189] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit of the present invention and the scope protected by the claims. These all fall within the protection scope of the present invention.

Claims

1. An aircraft flight path planning method based on a large language model, characterized in that Including: Obtain the real-time weather information of the cities along the flight mission route of the target aircraft; Input the real-time weather information into a pre-constructed large language regional prediction model to obtain the initial flight path for the target aircraft to avoid dangerous weather areas; Optimize the initial flight path to determine the optimal path planning strategy that enables the target aircraft not to collide with other aircraft in the surrounding airspace; According to the optimal path planning strategy, control the target aircraft to adjust its flight path.

2. The aircraft flight path planning method based on a large language model according to claim 1, characterized in that, The step of inputting the real-time weather information into a pre-constructed large language regional prediction model to obtain the initial flight path for the target aircraft to avoid dangerous weather areas includes: According to the real-time weather information, use the Graham scan method to obtain the initial boundary point set of the dangerous weather area; Based on the initial boundary point set, use the grey prediction model to predict the future boundary of the dangerous weather area to obtain the flight restricted area; According to the flight restricted area, obtain the initial flight path.

3. The method for aircraft flight path planning based on a large language model according to claim 1, wherein, The step of optimizing the initial flight path to determine the optimal path planning strategy that enables the target aircraft not to collide with other aircraft in the surrounding airspace includes: Use the SARSA algorithm to map the target aircraft and its initial flight path to an observable virtual environment space; Construct a reward function corresponding to the actions of the target aircraft in the virtual environment space; Make the target aircraft act in the virtual environment space to determine the optimal path planning strategy in which the target aircraft does not collide with other aircraft, can reach the destination, and obtains the highest reward value.

4. The method for aircraft flight path planning based on a large language model according to claim 3, wherein, The step of using the SARSA algorithm to map the target aircraft and its initial flight path to an observable virtual environment space includes: Represent the observable virtual environment space by a tuple: ； Where: Denote the state space, which is composed of the states corresponding to the target aircraft at each moment; Represents the action space, which is composed of the actions performed by the target aircraft at each moment; represents the reward function, which consists of the rewards given after the target aircraft executes actions at each moment.

5. The method for aircraft flight path planning based on a large language model according to claim 3, characterized in that, The reward function includes: A distance reward function, and the distance reward function is represented by the following formula: ； ； ； Indicates the moment intruder reward function, indicates the distance threshold between the target aircraft and other aircraft around the airspace, indicates the closest distance between the target aircraft and other aircraft around the airspace; Indicates Time remaining distance reward function Indicates the function for calculating the distance between the target aircraft and the destination Indicate Reward function for arriving at the destination at a certain time; A weather reward function, and the weather reward function is represented by the following formula: ； ； ； Indicates instantaneous rainfall reward function indicating the rainfall in the environment where the target aircraft is located; Indicates Visibility reward function at a moment Indicates the visibility of the environment where the target aircraft is located; Indicate Instant cloud cover reward function indicates the cloud cover of the environment where the target aircraft is located; A flight restricted area reward function, and the flight restricted area reward function is represented by the following formula: ； Represents the flight restricted area reward function, Represents the maximum reward, Represents the distance between the aircraft and the center of the minimum circumscribed circle corresponding to the flight restricted area, Represents the radius of the minimum circumscribed circle corresponding to the flight restricted area.

6. The method for aircraft flight path planning based on a large language model according to claim 5, wherein The step of making the target aircraft act in the virtual environment space to determine the optimal path planning strategy in which the target aircraft does not collide with other aircraft, can reach the destination, and obtains the highest reward value includes: Use the state-action value function to update the current state-action pair of the target aircraft, enabling the target aircraft to select the optimal path planning strategy for action. The update of the current state-action pair of the target aircraft is represented as: ； Where: Represents the estimated value of the current state-action pair ; Indicates an update to the estimated value of the current state-action pair ; Denotes the learning rate, which is used to control the update step size; represents the time series error, ; Indicates the state corresponding to the target aircraft at the current moment, Indicates the action performed by the target aircraft at the current moment; Indicates the action performed by the target aircraft The reward at the next moment after the action is performed, Indicates the state corresponding to the target aircraft at the next moment, Indicates the action performed by the target aircraft at the next moment; Represents the estimated value of the next moment state-action pair ; represents a discount factor; Calculate the reward value corresponding to the action of the target aircraft in the virtual environment space using the following formula: ； represent and as well as the minimum value among 7. The method for aircraft flight path planning based on a large language model according to claim 2, wherein The step of using the grey prediction model to predict the future boundary of the dangerous weather area based on the initial boundary point set to obtain the flight restricted area includes: Observe the initial boundary point set at fixed time intervals to obtain a set of original observation sequences representing the change of the initial boundary over time; Calculate the grey parameters of the original observation sequence corresponding to each initial boundary point; Input the grey parameters into the grey prediction model to obtain the boundary point coordinates at future times; According to the boundary point coordinates at each future time, obtain the flight restricted area.

8. The method for aircraft flight path planning based on a large language model according to claim 7, wherein The step of calculating the grey parameters of the original observation sequence corresponding to each initial boundary point includes: Calculate the grey parameters of the original observation sequence corresponding to each initial boundary point using the following formula: ； ； Where: Indicates the grey parameter of the original observation sequence corresponding to the abscissa of the initial boundary point; The cumulative data representing the abscissa of the initial boundary point, The cumulative matrix representing the abscissa, The constant vector representing the abscissa; Denote the grey parameter of the original observation sequence corresponding to the ordinate of the initial boundary point; Represents the cumulative data of the ordinate of the initial boundary point, Represents the cumulative matrix of the ordinate; Represents the ordinate constant vector.

9. An aircraft flight path planning device based on a large language model, characterized in that, Including: An acquisition module, configured to acquire real-time weather information of cities along the flight route of a target aircraft; A path planning module, configured to input the real-time weather information into a pre-constructed large language regional prediction model to obtain an initial flight path for the target aircraft to avoid dangerous weather regions; A path optimization module, configured to optimize the initial flight path to determine an optimal path planning strategy for preventing the target aircraft from colliding with other aircraft; A processing module, configured to control the target aircraft to adjust its flight path according to the optimal path planning strategy.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the aircraft flight path planning method based on a large language model as described in any one of claims 1 to 8.

Images