Digital pilot model generation method, device, equipment and medium
By constructing a digital pilot model and utilizing flight data analysis and operational safety values, the problems of long training cycles and poor model generalization in traditional pilot training have been solved, enabling efficient identification and adaptive improvement of safety risks during flight.
Patent Information
- Application Number
- CN202511706878.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-20
- Publication Date
- 2026-02-17
AI Technical Summary
Traditional pilot training is lengthy and costly, and physiological limitations restrict aircraft performance improvement. Existing digital pilot models have poor generalization capabilities and cannot adapt to individual pilot differences in real time, posing flight safety risks such as pilots' habitual actions, failure to follow instructions, and psychological fluctuations in extreme environments.
A digital pilot model is constructed by analyzing flight data to determine action parameters, building a training dataset and training the target model, including logical expressions and model definitions of behavioral actions and action parameters, and combining operational safety values to generate the digital pilot model.
It enables comprehensive and efficient identification of safety risks during flight, improves flight safety, adapts to individual differences among pilots, and reduces safety hazards caused by operational deviations and psychological fluctuations.
Smart Images
Figure CN121543414A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer technology and related technical fields, specifically to a method, apparatus, device, and medium for generating digital pilot models. Background Technology
[0002] Traditional pilot training is lengthy and costly, and physiological limitations restrict further improvements in aircraft performance. With the accelerated transformation of military intelligence and civil aviation digitalization, the concept of digital pilots is moving from "human control" to "AI autonomy." Therefore, a definition of a digital pilot has been proposed. Digital pilots are applied in the field of aviation safety, with the core objective of simulating, optimizing, and even surpassing the behavioral capabilities of human pilots through artificial intelligence technology, achieving automation, precision, and adaptability. However, building digital pilot models relies on massive amounts of labeled data for training. These models suffer from poor generalization and cannot adapt to individual pilot differences in real time, lacking cognitive decision-making capabilities in complex environments. The main problems are as follows: First, there are pilots' habitual actions, which repeatedly occur unconsciously. Micro-management details are easily overlooked, leading to operational deviations, loss of flight control, or omissions of key procedures, posing a potential threat to flight safety. Second, failure to perform flight maneuvers according to operating procedures or other standards can easily cause accident symptoms and pose safety hazards, requiring strengthened muscle memory training of standard operating procedures. Third, negative psychological fluctuations in pilots under extreme environments (such as high-intensity missions and cross-domain collaborative operations) require pilots to simultaneously process multimodal data information, including visual (situational awareness) and tactile (overload direction determination), in complex air combat.
[0003] Therefore, there is an urgent need to propose a method for generating digital pilot models to solve the problems existing in the current technology. Summary of the Invention
[0004] The embodiments described herein provide a method, apparatus, device, and medium for generating digital pilot models, addressing problems existing in the prior art.
[0005] Firstly, based on the content of this disclosure, a method for generating a digital pilot model is provided, including: Constructing a digital pilot, wherein the digital pilot includes behavioral actions and action parameters; Based on the logical expressions of the model elements of the digital pilots corresponding to different flight sorties, the model definitions of the digital pilots for different flight sorties, and the flight data corresponding to each flight sortie, the action parameters of the digital pilots corresponding to each flight sortie are determined. A training dataset for the digital pilot model is constructed based on the behavior, action parameters, and operational safety values of the digital pilots for each flight. Based on the constructed digital pilot model training dataset, the initial digital pilot model is trained to determine the target digital pilot model.
[0006] In some embodiments of this disclosure, determining the motion parameters of the digital pilot corresponding to each flight sortie based on the logical expressions of the model elements of the digital pilot corresponding to different flight sorties, the model definitions of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie includes: Establish the correspondence between flight data and flight sorties; Construct logical expressions for the elements of the digital pilot model corresponding to different flight sorties; Based on the logical expressions of the model elements of the digital pilots corresponding to different flight sorties, the model definitions of the digital pilots for different flight sorties, and the flight data corresponding to each flight sortie, the action parameters of the digital pilots corresponding to each flight sortie are determined.
[0007] In some embodiments of this disclosure, determining the motion parameters of the digital pilot corresponding to each flight sortie based on the logical expressions of the model elements of the digital pilot corresponding to different flight sorties, the model definitions of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie includes: Based on the logical expression of the model elements of the digital pilot corresponding to the target flight, the target flight data is filtered out from the flight data corresponding to the target flight. In the model definition of digital pilots based on the target number of flight sorties The logical expression and target flight data corresponding to the target flight are used to determine the action parameters in the model definition of the digital pilot for the target flight.
[0008] In some embodiments of this disclosure, the model definition of the digital pilot based on the target flight sorties is as follows: The logical expression and target flight data corresponding to the target flight are used to determine the action parameters in the model definition of the digital pilot for the target flight, including: In the model definition of digital pilots based on the target number of flight sorties The logical expression of the first element and the target flight data corresponding to the target flight number are used to determine the first element in the model definition of the digital pilot; In the model definition of digital pilots based on the target number of flight sorties The logical expression of the second element and the first element determine the action parameters in the model definition of the digital pilot.
[0009] In some embodiments of this disclosure, the step of constructing a digital pilot model training dataset based on the digital pilot's actions and action parameters and operational safety values for each flight includes: Based on the relationship between the action parameters corresponding to each action of the digital pilot of the target flight sortie and the preset action parameters, determine the number of warnings for the digital pilot of the target flight sortie when performing different actions. Based on the relationship between the number of warnings issued by the digital pilots of the target flight sortie during different actions and the preset number of warnings, the operational safety values of the digital pilots of the target flight sortie during different actions are determined. A training dataset for digital pilot models is constructed based on the behaviors, action parameters, and operational safety values of digital pilots for different target flight sorties.
[0010] In some embodiments of this disclosure, determining the warning information for different actions of the digital pilot of the target flight based on the relationship between the action parameters corresponding to each action of the digital pilot of the target flight and preset action parameters includes: Based on the relationship between the action parameters corresponding to each action of the digital pilot for the target flight and the preset action parameters, the difference information between the action parameters of each action and the preset action parameters is determined. Based on the relationship between the difference information and the preset difference information, the number of warnings for the digital pilots of the target flight sortie under different behavioral actions is determined.
[0011] In some embodiments of this disclosure, training an initial digital pilot model based on a constructed digital pilot model training dataset to determine a target digital pilot model includes: The behavior actions and action parameters of each digital pilot in the training dataset of the constructed digital pilot model are input into the initial digital pilot model to obtain the predicted operational safety value of the initial digital pilot model. The initial digital pilot model is trained based on the predicted operational safety value and operational safety value corresponding to each digital pilot until the loss value of the initial digital pilot model meets a preset threshold, thereby obtaining the target digital pilot model.
[0012] Secondly, according to the present disclosure, a digital pilot model generation apparatus is provided, comprising: A digital pilot construction module is used to construct a digital pilot, wherein the digital pilot includes behavioral actions and action parameters; The motion parameter determination module is used to determine the motion parameters of the digital pilot corresponding to each flight sortie based on the logical expressions of the model elements of the digital pilot corresponding to different flight sorties, the model definition of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie. The model training dataset determination module is used to construct a digital pilot model training dataset based on the digital pilot's behavior and action parameters and operational safety values for each flight. The target digital pilot model determination module is used to train the initial digital pilot model based on the constructed digital pilot model training dataset and determine the target digital pilot model.
[0013] Thirdly, according to this disclosure, a computer device is provided, comprising: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors perform the method as described in any of the first aspects.
[0014] Fourthly, according to this disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the methods described in any of the first aspects.
[0015] The digital pilot model generation method, apparatus, device, and medium provided in this disclosure first construct a digital pilot; then, based on the logical expressions of model elements of the digital pilot corresponding to different flight sorties, the model definitions of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie, the action parameters of the digital pilot for each flight sortie are determined; and based on the behavior actions and action parameters of the digital pilot for each flight sortie, as well as operational safety values, a digital pilot model training dataset is constructed; finally, based on the constructed digital pilot model training dataset, the initial digital pilot model is trained to determine the target digital pilot model. First, a digital pilot is proposed from the dimension of pilot behavior actions and action parameters. Then, based on the logical expressions of model elements of the digital pilot and the model definitions of the digital pilot for different flight sorties, the behavior actions and action parameters of the digital pilot are determined. This enables the construction of a digital pilot model training dataset based on the behavior actions and action parameters of the digital pilot for each flight sortie, as well as operational safety values, and the training and generation of the digital pilot model based on the digital pilot model training dataset, providing a foundation for the subsequent comprehensive and efficient identification of safety risks existing during flight.
[0016] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments will be briefly described below. It should be understood that the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure, wherein: Figure 1 This is a flowchart illustrating a digital pilot model generation method provided in an embodiment of this disclosure; Figure 2 This is a schematic diagram of the structure of a digital pilot model generation device provided in an embodiment of this disclosure; Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure.
[0018] In the accompanying diagram, markers with the same last two digits correspond to the same elements. It should be noted that the elements in the diagram are schematic and not drawn to scale. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are also within the scope of protection of this disclosure.
[0020] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this subject matter pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having the meaning consistent with their meaning in the context of the specification and in the relevant art, and shall not be interpreted in an idealized or overly formal form unless otherwise explicitly defined herein. As used herein, the statement of “connecting” or “coupling” two or more parts together shall mean that these parts are directly joined together or joined through one or more intermediate components.
[0021] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of the phrase "embodiment" in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0022] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists, A and B exist simultaneously, or B exists. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.
[0023] Furthermore, in all embodiments of this disclosure, terms such as “first” and “second” are used only to distinguish one component (or part of a component) from another component (or another part of a component).
[0024] In the description of this application, unless otherwise stated, "multiple" means two or more (including two), and similarly, "multiple groups" means two or more (including two groups).
[0025] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.
[0026] In view of the problems existing in the prior art, the present disclosure provides a method for generating a digital pilot model. Figure 1 This is a flowchart illustrating a digital pilot model generation method provided in an embodiment of this disclosure, as shown below. Figure 1 As shown, the digital pilot model generation method includes: S110, Building a Digital Pilot.
[0027] Digital pilots include behavioral actions and action parameters.
[0028] Digital pilots construct pilot operation models based on the actual flight procedures and actions of pilots. ,in It describes the specific operational actions or states at specific moments in the eight phases of pilot behavior, such as: engine start, taxiing out, takeoff, return to base, airspace maneuvers, landing, taxiing back, and engine shutdown. This represents pilot action parameters, such as: release of the parking brake, takeoff speed, takeoff angle of attack, throttle angle, engine speed and temperature, interception altitude, and peak speed. Based on the flight phase and capability category, the pilot action parameters are decomposed to generate a digital pilot profile. By constructing a virtual simulation of the digital pilot's flight process using actions and parameters, the experience and capabilities of human pilots are transformed into an algorithmic model, achieving "human-machine collaboration" or "dual-modal piloting."
[0029] (1) Flight phase division From pre-flight preparation to the shutdown of the aircraft after landing, the pilot's operation involves both pre-set operations according to specific formulas and standards, as well as on-the-spot handling based on specific weather, air situation, aircraft status, and actions. In particular, the randomness of the pilot's operation is significantly increased during the in-flight phase depending on the training subject, which makes the construction of digital pilots quite complex. Based on flight operation standards, equipment usage restrictions, and course flight specifications, the entire flight process is divided into models to achieve refined modeling and organization of pilot operations. Specifically, the start-up phase is the time period from the moment the start button is pressed to the five minutes following; the taxiing phase is the time period from the end of the start-up phase to the moment the throttle first exceeds 62 degrees; the takeoff phase is the time period from the moment the throttle first exceeds 62 degrees to the five minutes following the landing gear retraction; the landing phase is the time period from the moment the aircraft touches down to the moment it reverses 10 minutes; the engine shutdown phase is the time period from the moment the throttle is shut down to the moment it moves forward 300 seconds; the taxiing return phase is the time period from the moment the aircraft lands to the moment the engine shutdown phase begins; the airspace maneuver phase is the time period from the moment the aircraft enters the specific training airspace; and the departure and return phases are the time periods from the end of takeoff to the moment of entering the airspace and from the moment of leaving the airspace to the moment of starting landing.
[0030] (2) Pilot action parameters Considering the complexity of actual pilot operations, this study categorizes pilot actions into eight phases based on their operational characteristics and flight stage: engine start, taxiing, takeoff, return flight, airspace maneuvers, landing, taxiing back, and engine shutdown. Specific action parameters for each phase are analyzed to support the development of a digital pilot system. The resulting breakdown of pilot action parameters according to flight stage and capability category is shown in the table below.
[0031]
[0032] Based on the pilot's actions throughout the flight, a digital pilot is defined as a set of flight operation events, where T = ,in This indicates specific pilot operations, which, combined with the table above, can be represented as: {sequential activation of fuel pump switches, duration of activation of all fuel pump switches, activation of start-up regulator, start-up throttle angle, duration of pressing the start button, handling of starter light failure, handling of starter RPM failure to increase, ..., inertial navigation system before engine shutdown, fuel supply / transmission pump switches before engine shutdown, stop brake disengagement, attitude control switch disengagement} This represents the parameter state value corresponding to each action or state.
[0033] In other words, the actions of a digital pilot are fixed, and the parameters of the digital pilot's actions are determined based on the flight data of the digital pilot during actual operation.
[0034] S120. Based on the logical expressions of the model elements of the digital pilot corresponding to different flight sorties, the model definition of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie, determine the action parameters of the digital pilot corresponding to each flight sortie.
[0035] In a specific implementation, the action parameters of the digital pilot corresponding to each flight sortie are determined based on the logical expressions of the model elements of the digital pilot corresponding to different flight sorties, the model definitions of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie. This includes: constructing the correspondence between flight data and flight sorties; constructing the logical expressions of the model elements of the digital pilot corresponding to different flight sorties; and determining the action parameters of the digital pilot corresponding to each flight sortie based on the logical expressions of the model elements of the digital pilot corresponding to different flight sorties, the model definitions of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie.
[0036] Specifically, based on the logical expressions of the model elements of the digital pilots corresponding to different flight sorties, the model definitions of the digital pilots for different flight sorties, and the flight data corresponding to each flight sortie, the action parameters of the digital pilots corresponding to each flight sortie are determined, including: based on the model definition of the digital pilots for the target flight sortie... The first element in the model definition of a digital pilot is determined by the logical expression of the first element and the target flight data corresponding to the target flight sortie; based on the model definition of the digital pilot for the target flight sortie... The logical expression of the second element and the first element determine the action parameters in the model definition of the digital pilot.
[0037] First, based on the static information and flight data file information contained in the configuration information corresponding to each flight, as well as the flight time period, aircraft number, batch, and pilot information corresponding to the flight activity, the association between flight flights and flight data is established.
[0038] Then, logical expressions for the digital pilot model elements corresponding to different flight sorties are constructed. Specifically, to achieve the specific definition of digital pilot elements, basic elements are defined based on flight sorties. For a specific flight sortie, with aircraft type J, the basic element definition can be expressed as follows: Where X represents the set of all flight data parameters, , Represents the name of each flight data item, such as indicated airspeed, ground speed, landing gear retraction, etc.; E represents the set of symbols for the original parameter constraint expressions. , The symbols represent comparisons such as analog quantity being greater than, analog quantity being less than, analog quantity being between two values, and state quantity being equal to; Y represents the set of key parameters and time values calculated from flight data of flight sorties. , This represents takeoff time, landing time, takeoff speed, landing speed, takeoff angle of attack, and time in the air; F represents the set of function terms representing the parameters. , Represents a time period, moment, or specific numerical calculation function, such as: parameter change rate calculation, related parameter difference calculation, maximum value moment calculation, minimum value moment calculation, start time delay, end time prerequisite, first fulfillment condition, last fulfillment condition, etc.; V represents digital pilot elements. The set of calculation functions corresponding to the parameters. , Representing digital pilots Calculation functions, such as: maximum value function, minimum value function, value function at a specific time, etc.
[0039] The logical expressions for the digital pilot model elements differ for different flight sorties.
[0040] Based on the fundamental definitions of flight data items and calculation functions for the aforementioned flight sorties, the elements of the digital pilot model can be defined as binary tuples. , This indicates the time or time period definition of model elements, such as: "ATBEG[(FLYRG@)&&( ")]", where "FLYRG@" is a specific time period, moment, or key parameter value element defined in set Y within J, representing the airborne flight phase. "middle In the parameter-bound expression of set E in J, the analog quantity is greater than the function, indicating a data item. For time periods greater than 50, “ATBEG[]” is the start time value calculation function in set F of J, representing the start time function of the time period. This string represents the start time of the flight phase. Definitions representing the actual physical meaning of model elements, such as: "MAX[DIFF[ , ]]",in and "DIFF" is a function that calculates the difference between two data items in the set X defined in J, representing two parameter items of the flight data of the aircraft model being processed. and The difference between them, "MAX[]" is the calculation function for the digital pilot elements in the V set defined in J, representing the data item. and The maximum value.
[0041] Based on the above description, the model definition of the digital pilot is completed as follows:
[0042] The digital pilot model element set can be flexibly designed according to actual conditions, and elements can be added, deleted, or modified according to actual conditions, supporting the adaptation and construction of the model in various scenarios.
[0043] Key basic information about flight sorties is crucial data reflecting key stages throughout the flight process. Based on imported flight data, the basic elements and key parameters of different flight sorties are calculated, according to the elements in the digital pilot model T. The definition, using imported flight data as input, completes... middle Analytical calculation, based on The definition of implementation in T The calculation and storage of the solution.
[0044] S130. Based on the behavior and action parameters of the digital pilots for each flight and the operational safety values, construct a training dataset for the digital pilot model.
[0045] In a specific implementation, a digital pilot model training dataset is constructed based on the digital pilot's actions, action parameters, and operational safety values for each flight sortie. This includes: constructing a digital pilot model training dataset based on the digital pilot's actions and action parameters for each flight sortie; determining the number of warnings for different actions of the digital pilot in the target flight sortie based on the relationship between the action parameters corresponding to each action of the digital pilot in the target flight sortie and preset action parameters; determining the operational safety values for different actions of the digital pilot in the target flight sortie based on the relationship between the number of warnings for different actions of the digital pilot in the target flight sortie and preset warning number information; and constructing a digital pilot model training dataset based on the actions, action parameters, and operational safety values included by the digital pilots in different target flight sorties.
[0046] First, digital pilot behavior analysis criteria are constructed. Feature classification methods are used to summarize and analyze the changing trends of pilot control deviations at each stage, helping to identify potential problems and uncover safety risks and alerts. Corresponding criteria are formulated based on the characteristics of each stage. Each criterion examines the pilot's corresponding behavioral operations. Based on the actual situation, appropriate criteria are developed, using the operations and parameters of each stage during a flight. A partial example of the criterion rules is described below.
[0047] i) Start-up: Equipment in use; the regulator is not engaged when starting up. If the electronic regulator switch / generator control switch is engaged when the start button is pressed, it is considered true; otherwise, it is considered false.
[0048] ii) Taxiing out: Equipment in use; taxi lights not turned on before taxiing out. At ground speed of 7-10 km / h, the taxi light switch should be in the "auto" position for true, otherwise false.
[0049] iii) Takeoff: Takeoff and landing techniques; takeoff speed. A true result is when the main landing gear switch is off and the aircraft's indicated speed is 200-400 km / h, or 250 km / h < aircraft speed < 350 km / h; otherwise, it is false.
[0050] iv) Airspace Maneuvers: Tactical Basics; PL-15 Launch Angle of Attack. A true launch signal is given if the aircraft's angle of attack is between -20° and 50°; otherwise, it is false.
[0051] v) Landing: Takeoff and landing techniques; FAF 2km indicated airspeed. The aircraft avionics are in the RWY phase and the landing gear handle is in the "down" position. The distance between the aircraft and the runway threshold is 1.9-2.1km. Time T1 is true if the indicated airspeed at time T1 is 330-350km / h, otherwise it is false.
[0052] vi) Glide back: Follow the rules; deploy the parachute at a high speed. The parachute deployment switch is activated at time T1. If the aircraft's indicated airspeed is less than 280 km / h at T1, it is considered true; otherwise, it is considered false.
[0053] vii) Shutdown: Equipment in use; the inertial navigation system (INS) was not turned off before shutdown. The aircraft receives a shutdown signal at time T1. If the INS is in the off position at T1, it is true; otherwise, it is false.
[0054] Then, based on the criteria for the eight stages of starting, taxiing, takeoff, return, airspace maneuvering, landing, taxiing back, and shutting down the engine in the knowledge base, and combined with the calculated pilot operation data T, the difference information between the action parameters corresponding to each action of the digital pilot of the target flight and the preset action parameters is calculated; then, based on the relationship between the difference information and the preset difference information, the number of warnings for the digital pilot of the target flight at different actions is determined.
[0055] Furthermore, based on the relationship between the number of warnings issued by the digital pilots of the target flight sortie during different actions and the preset number of warnings, the operational safety value of the digital pilots of the target flight sortie during different actions is determined. The operational safety value is a value between 0 and 1, where a larger value indicates higher safety and fewer warnings, and vice versa.
[0056] Finally, based on the solution results of the digital pilot T in step S120 and the operational safety values of the flight sorties, a data matching relationship is constructed and stored. Based on this, the fusion of the digital representation of the pilot and the operational safety values is completed, and the construction of the training dataset for the digital pilot model is realized.
[0057] S140. Based on the constructed digital pilot model training dataset, train the initial digital pilot model to determine the target digital pilot model.
[0058] In a specific implementation, the initial digital pilot model is trained based on the constructed digital pilot model training dataset to determine the target digital pilot model. This includes: inputting the behavior actions and action parameters of each digital pilot in the constructed digital pilot model training dataset into the initial digital pilot model to obtain the predicted operational safety value of the initial digital pilot model; and training the initial digital pilot model according to the predicted operational safety value and operational safety value corresponding to each digital pilot until the loss value of the initial digital pilot model meets a preset threshold to obtain the target digital pilot model.
[0059] Specifically, based on the constructed digital pilot model training dataset, the behavior actions and action parameters of each digital pilot in the digital pilot model training dataset are input into the initial digital pilot model to obtain the predicted operational safety value corresponding to the behavior actions and action parameters of each digital pilot. Then, based on the predicted operational safety value and the operational safety value corresponding to each digital pilot, the loss value is determined. The parameters of the initial digital pilot model are trained and updated based on the loss value until the loss value of the initial digital pilot model meets the preset threshold, thus obtaining the target digital pilot model.
[0060] Given a dataset D, there are N inputs. The corresponding output is If the output dimension is K, then the loss function is defined as follows: The first term in the loss function , The output result when the weight is W. The second term of the loss function represents the difference between the actual output y and the network output f. Regularization. The parameter iteration takes the following form:
[0061]
[0062] In the formula, i represents the number of iterations. It's the learning rate. It is a very small scalar. The stochastic gradient is defined as follows: The formula uses batch training, with a total of M batches of data, each batch containing... Data.
[0063] The posterior distribution of the Bayesian method is ,in, , The calculation involves integral terms, which is computationally intensive. Variational inference is a highly scalable posterior probability method. Approximate calculation method. The ELBO form is:
[0064] Optimize the deep neural network model using the natural gradient variational inference method. When p(w) := hour natural gradient parameters The stochastic gradient optimization iteration under expected regularization loss is as follows: , in the formula This represents the learning rate.
[0065] When the distribution q is Gaussian, the second-order Gaussian variational inference method predicts the mean of the Gaussian posterior. The update and iteration rules for the covariance are as follows:
[0066]
[0067]
[0068] In the formula, , . This represents the learning rate.
[0069] In the reinforcement learning training process, when the learning environment is very complex, the training convergence process will be slow or may not converge. First, a reasonable initial parameter for the model learning environment is confirmed through the training process. Due to the special nature of the parameters (keyword groups of the criterion) of the historical flight data, the keyword groups of the sentence are retained, and the subsequent word groups are randomly distilled out with a certain probability.
[0070] That is, in this embodiment of the disclosure, the input is the behavior actions and action parameters of each digital pilot, and the output is the predicted operational safety value corresponding to the behavior actions and action parameters of each digital pilot. Based on the predicted operational safety value and the actual operational safety value, based on the loss function, the initial digital pilot model is trained.
[0071] The digital pilot model generation method provided in this disclosure first constructs a digital pilot; then, based on the logical expressions of model elements of the digital pilot corresponding to different flight sorties, the model definition of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie, it determines the action parameters of the digital pilot for each flight sortie; and based on the behavior actions and action parameters of the digital pilot for each flight sortie, as well as operational safety values, it constructs a digital pilot model training dataset; finally, based on the constructed digital pilot model training dataset, it trains the initial digital pilot model to determine the target digital pilot model. First, it proposes a digital pilot from the dimension of pilot behavior actions and action parameters. Then, based on the logical expressions of model elements of the digital pilot and the model definition of the digital pilot for different flight sorties, it determines the behavior actions and action parameters of the digital pilot. This enables the construction of a digital pilot model training dataset based on the behavior actions and action parameters of the digital pilot for each flight sortie, as well as operational safety values. Finally, it enables the training and generation of the digital pilot model based on the digital pilot model training dataset, providing a foundation for the subsequent comprehensive and efficient identification of safety risks present during flight.
[0072] Based on the above embodiments, this disclosure also provides a digital pilot model generation device. Figure 2 This is a schematic diagram of the structure of the digital pilot model generation device provided in the embodiments of this disclosure, as shown below. Figure 2 As shown, the digital pilot model generation device includes: Digital pilot construction module 210 is used to construct a digital pilot, wherein the digital pilot includes behavioral actions and action parameters; The motion parameter determination module 220 is used to determine the motion parameters of the digital pilot corresponding to each flight sortie based on the logical expression of the model elements of the digital pilot corresponding to different flight sorties, the model definition of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie. The model training dataset determination module 230 is used to construct a digital pilot model training dataset based on the behavior and action parameters of the digital pilots for each flight and the operational safety values. The target digital pilot model determination module 240 is used to train the initial digital pilot model based on the constructed digital pilot model training dataset to determine the target digital pilot model.
[0073] The digital pilot model generation apparatus provided in this disclosure first constructs a digital pilot; then, based on the logical expressions of model elements of the digital pilot corresponding to different flight sorties, the model definitions of the digital pilot for different flight sorties, and the flight data corresponding to each flight sortie, it determines the action parameters of the digital pilot for each flight sortie; and based on the behavior actions and action parameters of the digital pilot for each flight sortie, as well as operational safety values, it constructs a digital pilot model training dataset; finally, based on the constructed digital pilot model training dataset, it trains the initial digital pilot model to determine the target digital pilot model. First, it proposes a digital pilot from the dimension of pilot behavior actions and action parameters. Then, based on the logical expressions of model elements of the digital pilot and the model definitions of the digital pilot for different flight sorties, it determines the behavior actions and action parameters of the digital pilot. This enables the construction of a digital pilot model training dataset based on the behavior actions and action parameters of the digital pilot for each flight sortie, as well as operational safety values. Finally, it enables the training and generation of the digital pilot model based on the digital pilot model training dataset, providing a foundation for the subsequent comprehensive and efficient identification of safety risks present during flight.
[0074] In a specific implementation, determining the action parameters of the digital pilot corresponding to each flight sortie based on the logical expressions of the model elements of the digital pilot corresponding to different flight sorties, the model definitions of the digital pilots for different flight sorties, and the flight data corresponding to each flight sortie includes: Establish the correspondence between flight data and flight sorties; Construct logical expressions for the elements of the digital pilot model corresponding to different flight sorties; Based on the logical expressions of the model elements of the digital pilots corresponding to different flight sorties, the model definitions of the digital pilots for different flight sorties, and the flight data corresponding to each flight sortie, the action parameters of the digital pilots corresponding to each flight sortie are determined.
[0075] In a specific implementation, determining the action parameters of the digital pilot corresponding to each flight sortie based on the logical expressions of the model elements of the digital pilot corresponding to different flight sorties, the model definitions of the digital pilots for different flight sorties, and the flight data corresponding to each flight sortie includes: Based on the logical expression of the model elements of the digital pilot corresponding to the target flight, the target flight data is filtered out from the flight data corresponding to the target flight. In the model definition of digital pilots based on the target number of flight sorties The logical expression and target flight data corresponding to the target flight are used to determine the action parameters in the model definition of the digital pilot for the target flight.
[0076] In a specific implementation, the model definition of the digital pilot based on the target flight sorties... The logical expression and target flight data corresponding to the target flight are used to determine the action parameters in the model definition of the digital pilot for the target flight, including: In the model definition of digital pilots based on the target number of flight sorties The logical expression of the first element and the target flight data corresponding to the target flight number are used to determine the first element in the model definition of the digital pilot; In the model definition of digital pilots based on the target number of flight sorties The logical expression of the second element and the first element determine the action parameters in the model definition of the digital pilot.
[0077] In a specific implementation, the step of constructing a digital pilot model training dataset based on the digital pilot's actions, action parameters, and operational safety values for each flight includes: Based on the relationship between the action parameters corresponding to each action of the digital pilot of the target flight sortie and the preset action parameters, determine the number of warnings for the digital pilot of the target flight sortie when performing different actions. Based on the relationship between the number of warnings issued by the digital pilots of the target flight sortie during different actions and the preset number of warnings, the operational safety values of the digital pilots of the target flight sortie during different actions are determined. A training dataset for digital pilot models is constructed based on the behaviors, action parameters, and operational safety values of digital pilots for different target flight sorties.
[0078] In a specific implementation, determining the warning information for different actions of the digital pilot of the target flight sortie based on the relationship between the action parameters corresponding to each action of the digital pilot and preset action parameters includes: Based on the relationship between the action parameters corresponding to each action of the digital pilot for the target flight and the preset action parameters, the difference information between the action parameters of each action and the preset action parameters is determined. Based on the relationship between the difference information and the preset difference information, the number of warnings for the digital pilots of the target flight sortie under different behavioral actions is determined.
[0079] In a specific implementation, the step of training an initial digital pilot model based on a constructed digital pilot model training dataset to determine a target digital pilot model includes: The behavior actions and action parameters of each digital pilot in the training dataset of the constructed digital pilot model are input into the initial digital pilot model to obtain the predicted operational safety value of the initial digital pilot model. The initial digital pilot model is trained based on the predicted operational safety value and operational safety value corresponding to each digital pilot until the loss value of the initial digital pilot model meets a preset threshold, thereby obtaining the target digital pilot model.
[0080] This application also provides a computer device, please refer to the following for details. Figure 3 , Figure 3 This is a basic structural block diagram of the computer device in this embodiment.
[0081] The computer device includes a memory 510 and a processor 520 that are interconnected via a system bus. It should be noted that only a computer device with components 510-520 is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components may be implemented alternatively. Those skilled in the art will understand that the computer device described herein is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.
[0082] Computer devices can include desktop computers, laptops, handheld computers, and cloud servers. These devices allow for human-computer interaction with users through keyboards, mice, remote controls, touchpads, or voice-activated devices.
[0083] The memory 510 includes at least one type of readable storage medium, including non-volatile memory or volatile memory, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. RAM may include static RAM or dynamic RAM. In some embodiments, the memory 510 may be an internal storage unit of a computer device, such as the hard disk or memory of the computer device. In other embodiments, the memory 510 may also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, or flash card equipped on the computer device. Of course, the memory 510 may include both internal storage units and external storage devices of the computer device. In this embodiment, the memory 510 is typically used to store the operating system and various application software installed on the computer device, such as the program code of the method described above. In addition, the memory 510 may also be used to temporarily store various types of data that have been output or will be output.
[0084] The processor 520 is typically used to perform the overall operation of a computer device. In this embodiment, the memory 510 is used to store program code or instructions, including computer operation instructions. The processor 520 is used to execute the program code or instructions stored in the memory 510 or to process data, such as program code that runs the methods described above.
[0085] In this article, the bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus system can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.
[0086] Another embodiment of this application also provides a computer-readable medium, which may be a computer-readable signal medium or a computer-readable medium. A processor in a computer reads computer-readable program code stored in the computer-readable medium, enabling the processor to execute the functional actions specified in each step or combination of steps in the above method; and to generate means for implementing the functional actions specified in each block or combination of blocks in the block diagram.
[0087] Computer-readable media include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared memory or semiconductor systems, devices or apparatuses, or any suitable combination thereof, wherein the memory is used to store program code or instructions, the program code including computer operation instructions, and the processor is used to execute the program code or instructions of the above-described methods stored in the memory.
[0088] The definitions of memory and processor can be found in the description of the foregoing computer device embodiments, and will not be repeated here.
[0089] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0090] In the various embodiments of this application, the functional units or modules can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0091] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0092] Unless otherwise expressly indicated by the context, the singular form of words used herein and in the appended claims includes the plural form, and vice versa. Thus, when referring to the singular, the plural form of the corresponding term is generally included. Similarly, the terms “comprising” and “including” shall be interpreted as including rather than exclusively. Likewise, the terms “including” and “or” shall be interpreted as including unless such interpretation is expressly prohibited herein. Where the term “example” is used herein, particularly when it follows a set of terms, the “example” is merely exemplary and illustrative and should not be considered exclusive or extensive.
[0093] Further aspects and scope of adaptation become apparent from the description provided herein. It should be understood that various aspects of this application may be implemented individually or in combination with one or more other aspects. It should also be understood that the descriptions and specific embodiments herein are for illustrative purposes only and are not intended to limit the scope of this application.
[0094] Several embodiments of this disclosure have been described in detail above. However, it is obvious that those skilled in the art can make various modifications and variations to the embodiments of this disclosure without departing from the spirit and scope of this disclosure. The scope of protection of this disclosure is defined by the appended claims.
Claims
1. A digital pilot model generation method characterized by, The method comprises the following steps: constructing a digital pilot, wherein the digital pilot comprises behavior actions and action parameters; determining the action parameters of the digital pilot corresponding to each flight mission according to the model element logical expression of the digital pilot corresponding to different flight missions, the model definition of the digital pilot of different flight missions, and the flight data corresponding to each flight mission; constructing a digital pilot model training data set according to the behavior actions and action parameters of the digital pilot of each flight mission and the operation safety value; training the initial digital pilot model based on the constructed digital pilot model training data set to determine a target digital pilot model.
2. The method of claim 1, wherein, The method of determining the action parameters of the digital pilot corresponding to each flight mission according to the model element logical expression of the digital pilot corresponding to different flight missions, the model definition of the digital pilot of different flight missions, and the flight data corresponding to each flight mission comprises the following steps: constructing a corresponding relationship between the flight data and the flight mission; constructing the model element logical expression of the digital pilot corresponding to different flight missions; determining the action parameters of the digital pilot corresponding to each flight mission according to the model element logical expression of the digital pilot corresponding to different flight missions, the model definition of the digital pilot of different flight missions, and the flight data corresponding to each flight mission.
3. The method of claim 2, wherein, The method of determining the action parameters of the digital pilot corresponding to each flight mission according to the model element logical expression of the digital pilot corresponding to different flight missions, the model definition of the digital pilot of different flight missions, and the flight data corresponding to each flight mission comprises the following steps: screening the target flight data from the flight data corresponding to the target flight mission according to the model element logical expression of the digital pilot corresponding to the target flight mission; In the model definition of digital pilots based on the target number of flight sorties The logical expression and target flight data corresponding to the target flight are used to determine the action parameters in the model definition of the digital pilot for the target flight.
4. The method of claim 3, wherein, The model definition of the digital pilot of the target flight mission is determined according to the logic expression in the model definition of the digital pilot of the target flight mission and the target flight data corresponding to the target flight mission. The action parameter in the model definition of the digital pilot of the target flight mission is determined by the logic expression in the model definition of the digital pilot of the target flight mission and the target flight data corresponding to the target flight mission. In the model definition of digital pilots based on the target number of flight sorties The logical expression of the first element and the target flight data corresponding to the target flight number are used to determine the first element in the model definition of the digital pilot; In the model definition of digital pilots based on the target number of flight sorties The logical expression of the second element and the first element determine the action parameters in the model definition of the digital pilot.
5. The method of claim 1, wherein, The method of constructing a digital pilot model training data set according to the behavior actions and action parameters of the digital pilot of each flight mission and the operation safety value comprises the following steps: determining the pre-warning quantity information of the digital pilot of the target flight mission in different behavior actions according to the relationship between the action parameters corresponding to each behavior action of the digital pilot of the target flight mission and the preset action parameters; determining the operation safety value of the digital pilot of the target flight mission in different behavior actions according to the relationship between the pre-warning quantity information of the digital pilot of the target flight mission in different behavior actions and the preset pre-warning quantity information; constructing a digital pilot model training data set according to the behavior actions and action parameters and operation safety value of the digital pilot of different target flight missions.
6. The method of claim 5, wherein, The method of determining the pre-warning information of the digital pilot of the target flight mission in different behavior actions according to the relationship between the action parameters corresponding to each behavior action of the digital pilot of the target flight mission and the preset action parameters comprises the following steps: determining the difference value information of the action parameters of each behavior action and the preset action parameters according to the relationship between the action parameters corresponding to each behavior action of the digital pilot of the target flight mission and the preset action parameters; determining the pre-warning quantity information of the digital pilot of the target flight mission in different behavior actions according to the relationship between the difference value information and the preset difference value information.
7. The method of claim 1, wherein, The method comprises the following steps: The behavior action and the action parameter of each digital pilot included in the constructed digital pilot model training data set are input to the initial digital pilot model, and a predicted operation safety value of the initial digital pilot model is obtained; The initial digital pilot model is trained according to the predicted operation safety value and the operation safety value corresponding to each digital pilot until the loss value of the initial digital pilot model meets a preset threshold value, and a target digital pilot model is obtained.
8. A digital pilot model generation apparatus characterized by comprising: The method comprises the following steps: A digital pilot construction module is configured to construct a digital pilot, wherein the digital pilot comprises a behavior action and an action parameter; An action parameter determination module is configured to determine the action parameter of the digital pilot corresponding to each flight mission according to a model element logical expression of the digital pilot corresponding to each flight mission, a model definition of the digital pilot of each flight mission and flight data corresponding to each flight mission; A model training data set determination module is configured to construct a digital pilot model training data set according to the behavior action and the action parameter of the digital pilot of each flight mission and an operation safety value; A target digital pilot model determination module is configured to train an initial digital pilot model based on the constructed digital pilot model training data set, and determine a target digital pilot model.
9. A computer device, comprising: The method comprises the following steps: One or more processors; A storage device configured to store one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1-7.
10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the method of any one of claims 1-7.