Flight training adaptive optimization method, system, device and storage medium based on operation big data

Abstract

The application relates to the field of operation big data technology, and provides a flight training self-adaptive optimization method, system, device and storage medium based on operation big data, which solves the problems of low control precision and poor stability of high-maneuvering flight training. The method comprises the following steps: collecting attitude time sequence data, structure deformation monitoring data and historical operation data of a flight target in a current period under a maneuvering state; performing environment interference suppression on the attitude time sequence data, and extracting motion characteristic parameters; converting the structure deformation monitoring data into deformation compensation data, and fusing the motion characteristic parameters and the deformation compensation data by using an attitude fusion algorithm; based on the historical operation data, performing multi-period feature matching on the fusion result of the current period, and generating six-dimensional attitude information of the flight target; and generating an optimized control instruction by using an adaptive control strategy, wherein the optimized control instruction is used for optimizing the response performance of the flight target. The application improves the control precision and stability of high-maneuvering flight training.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of operating big data, and in particular to a flight training adaptive optimization method and system based on operating big data, a device, and a storage medium. BACKGROUND

[0002] In the process of civil aviation passenger plane flight training, especially in the complex maneuvering stages such as take-off climb, encountering airflow, approach and landing, dynamic environmental disturbances (such as clear air turbulence, wind shear) and structural deformation (such as wing elastic deformation) can significantly affect the flight attitude stability and control accuracy, and may also cause the flight trajectory to deviate or the passenger comfort to decrease.

[0003] The prior art usually adopts a static optimization method based on a preset compensation coefficient, corrects the sensor data by a fixed weight, and generates a control instruction in combination with a preset rule. This method establishes a benchmark model using historical operation data, and adjusts the control parameters by comparing the deviation between real-time data and the model.

[0004] The static compensation coefficient is difficult to adapt to the dynamic changes in the maneuvering process, and the data correction and control instruction generation links are independent of each other, resulting in response delay in complex maneuvering states and limited suppression effect on sudden disturbances. SUMMARY

[0005] The present application provides a flight training adaptive optimization method and system based on operating big data, a device, and a storage medium, to solve the problems of low control precision and poor stability in high maneuvering flight training in the prior art.

[0006] In a first aspect, the present application provides a flight training adaptive optimization method based on operating big data, comprising:

[0007] Collecting attitude time series data, structural deformation monitoring data, and historical operation data of a flight target in a current period of maneuvering state;

[0008] Performing environmental disturbance suppression processing on the attitude time series data to extract motion feature parameters;

[0009] Converting the structural deformation monitoring data into deformation compensation data, and fusing the motion feature parameters and the deformation compensation data using an attitude fusion algorithm;

[0010] Based on the historical operation data, performing multi-period feature matching on the fusion result of the current period to generate six-dimensional attitude information of the flight target;

[0011] Based on the six-dimensional attitude information, generating an optimized control instruction through an adaptive control strategy, the optimized control instruction being used to optimize the response performance of the flight target.

[0012] Optionally, the converting the structure deformation monitoring data into deformation compensation data, fusing the motion feature parameters and the deformation compensation data by a pose fusion algorithm, comprises:

[0013] Performing a modal decoupling operation on the structure deformation monitoring data to obtain a plurality of decoupled modal components;

[0014] Extracting corresponding key deformation feature vectors from each of the modal components;

[0015] Mapping all the key deformation feature vectors to a body coordinate system to generate deformation compensation data, the body coordinate system being established based on the geometric configuration of the flight target;

[0016] Establishing a spatial trimming matrix between the motion feature parameters and the deformation compensation data;

[0017] Performing a heterogeneous data coupling operation on the spatial trimming matrix by a pose fusion algorithm to generate a target fusion pose feature set, the target fusion pose feature set being a fusion result.

[0018] Optionally, the performing a heterogeneous data coupling operation on the spatial trimming matrix by a pose fusion algorithm to generate a target fusion pose feature set, comprises:

[0019] Assigning the motion feature parameters to a first coupling domain of the spatial trimming matrix;

[0020] Assigning the deformation compensation data to a second coupling domain of the spatial trimming matrix;

[0021] Activating a tensor fusion core of the pose fusion algorithm to perform a cross-projection operation on the first coupling domain and the second coupling domain based on the tensor fusion core on the spatial trimming matrix;

[0022] Generating an initial fusion pose feature set based on the projection operation result;

[0023] Performing a degree-of-freedom dimension reduction operation on the initial fusion pose feature set to generate a target fusion pose feature set.

[0024] Optionally, the performing a cross-projection operation on the first coupling domain and the second coupling domain based on the tensor fusion core on the spatial trimming matrix, comprises:

[0025] Generating a maneuver feature vector based on the first coupling domain and extracting a structure constraint vector from the second coupling domain;

[0026] Calling a projection direction calculation method of the tensor fusion core to determine a projection direction in combination with a current flight state of the flight target;

[0027] Orthogonal projection of the maneuvering eigenvector and the structure constraint vector is superimposed along the projection direction.

[0028] Optionally, the environment interference suppression processing is performed on the attitude time series data to extract the motion feature parameter, and the method comprises the following steps:

[0029] The frequency domain transformation is performed on the attitude time series data to obtain a corresponding frequency energy distribution.

[0030] The vibration residual amount is identified from the frequency energy distribution through a noise separation mechanism, the vibration residual amount is filtered out, and denoising time series data is formed.

[0031] The angular motion feature component and the linear motion feature component are separated from the denoising time series data, and the angular motion feature component and the linear motion feature component are combined as the motion feature parameter.

[0032] Optionally, the six-dimensional attitude information of the flight target is generated by performing multi-period feature matching on the fusion result of the current period based on the historical running data, and the method comprises the following steps:

[0033] According to the historical running data, a cross-period feature library is constructed, and the cross-period feature library contains a historical fusion feature sequence and a corresponding compensation coefficient.

[0034] The deviation feature parameter between each parameter in the fusion result of the current period and the corresponding parameter in the historical fusion feature sequence is calculated.

[0035] For each parameter in the fusion result, the compensation coefficient of the corresponding parameter in the historical fusion feature sequence with the smallest value of the deviation feature parameter is taken as the optimal compensation coefficient to construct an optimal compensation coefficient combination.

[0036] Based on the optimal compensation coefficient combination, the six-dimensional attitude information is generated.

[0037] Optionally, the optimal control instruction is generated through an adaptive control strategy based on the six-dimensional attitude information, and the method comprises the following steps:

[0038] The six-dimensional attitude information is input into a strategy iterator, and the strategy iterator stores a dynamic response rule set.

[0039] According to the dynamic response rule set, an initial control instruction is generated.

[0040] The initial control instruction is subjected to a disturbance simulation test.

[0041] Based on the test result, the gain distribution of the initial control instruction is optimized to output an optimized control instruction.

[0042] In a second aspect, the application provides a flight training adaptive optimization system based on operation big data, comprising:

[0043] A collection module is configured to collect attitude time series data, structure deformation monitoring data and historical operation data of a flight target in a current period under a maneuvering state;

[0044] An extraction module is configured to perform environment interference suppression processing on the attitude time series data to extract motion feature parameters;

[0045] A conversion module is configured to convert the structure deformation monitoring data into deformation compensation data, and fuse the motion feature parameters and the deformation compensation data by using a posture fusion algorithm;

[0046] A matching module is configured to perform multi-period feature matching on a fusion result of the current period based on the historical operation data to generate six-dimensional attitude information of the flight target;

[0047] A generation module is configured to generate an optimized control instruction by using an adaptive control strategy based on the six-dimensional attitude information, and the optimized control instruction is used to optimize the response performance of the flight target.

[0048] In a third aspect, the application provides a computing device comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the flight training adaptive optimization method based on operation big data according to any one of the first aspect.

[0049] In a fourth aspect, the application provides a computer storage medium storing computer program instructions, wherein the computer program instructions are executed by a processor to implement the flight training adaptive optimization method based on operation big data according to any one of the first aspect.

[0050] In the application, a flight training adaptive optimization method based on operation big data is provided, which comprises the following steps: collecting attitude time series data, structure deformation monitoring data and historical operation data of a flight target in a current period under a maneuvering state; performing environment interference suppression processing on the attitude time series data to extract motion feature parameters; converting the structure deformation monitoring data into deformation compensation data, and fusing the motion feature parameters and the deformation compensation data by using a posture fusion algorithm; performing multi-period feature matching on a fusion result of the current period based on the historical operation data to generate six-dimensional attitude information of the flight target; and generating an optimized control instruction by using an adaptive control strategy based on the six-dimensional attitude information, and the optimized control instruction is used to optimize the response performance of the flight target.

[0051] The technical scheme provided by the application has the following beneficial effects:

[0052] The application obtains real-time state data and historical operation records in a comprehensive manner during dynamic motion, and provides complete data basis for subsequent processing. The application effectively eliminates the interference of environmental factors such as air flow disturbance on motion data, and improves the accuracy of feature extraction. The application converts structural deformation monitoring data into compensation parameters that can be directly used for control calculation. The application realizes the collaborative optimization of motion features and structural deformation, and improves the comprehensiveness and accuracy of state perception. The application uses historical optimal data to intelligently correct the current state, and enhances the adaptive ability of the system. The application generates optimal control instructions based on accurate attitude information, and significantly improves the dynamic response performance.

[0053] Further, the application also extracts a key feature vector from the structural deformation monitoring data through a modal decoupling operation, maps it to a body coordinate system to generate deformation compensation data, establishes a spatial trimming matrix with the motion feature parameters, and finally realizes accurate coupling of heterogeneous data by using an attitude fusion algorithm, and outputs an optimized fused attitude feature set.

[0054] Moreover, the technology realizes the deep collaboration of motion features and structural deformation data, effectively solves the nonlinear coupling problem of mechanical characteristics and structural state in the process of maneuvering, and significantly improves the accuracy of attitude perception and the reliability of control instructions in complex motion states.

[0055] These aspects or other aspects of the application will be more apparent in the following description of the embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can also be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0057] Figure 1 A flowchart of a flight training adaptive optimization method based on operation big data provided by an embodiment of the application;

[0058] Figure 2 A structural schematic diagram of a flight training adaptive optimization system based on operation big data provided by an embodiment of the application;

[0059] Figure 3 A structural schematic diagram of a computing device provided by an embodiment of the application. DETAILED DESCRIPTION

[0060] In order to enable personnel in the technical field to better understand the application scheme, the technical solutions in the embodiments of the application will be described clearly and completely in the following with reference to the drawings in the embodiments of the application.

[0061] In some of the flowcharts described in the specification and claims of the present application and in the above description, a plurality of operations are included which occur in a particular order, but it should be clearly understood that the operations can be performed in the order in which they appear herein or in parallel, and the serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, the flowcharts can include more or fewer operations, and the operations can be performed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. herein are used to distinguish different messages, devices, modules, etc., and do not represent the order of precedence, nor do "first" and "second" represent different types.

[0062] In the process of civil aviation passenger aircraft flight training, especially in the complex maneuvering stages such as take-off climb, encountering airflow, approach and landing, dynamic environmental disturbances (such as clear air turbulence, wind shear) and structural deformation (such as wing elastic deformation) will significantly affect the flight attitude stability and control accuracy. The existing technology usually adopts a static optimization method based on preset compensation coefficients, which corrects the sensor data by fixing the weight and generates control instructions combined with preset rules. This method uses historical operation data to establish a baseline model, and adjusts the control parameters by comparing the deviation between real-time data and the model.

[0063] To solve the above problems, the present application proposes a flight training adaptive optimization method based on operation big data, which realizes accurate state perception and control optimization by real-time fusion of motion characteristics and structural deformation data. Specifically, first, collect motion state data and eliminate environmental disturbances, and at the same time convert structural deformation information into compensation parameters; then, through intelligent algorithm, the motion characteristics and deformation compensation data are deeply fused, and the current state is dynamically corrected combined with the historical optimal data; finally, the control instructions adapted to the real-time motion state are generated. This method breaks through the limitations of traditional static processing, effectively solves the problems of response delay and insufficient anti-interference ability through dynamic coordination of motion characteristics and structural state, and significantly improves the control accuracy and stability in maneuvering state.

[0064] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0065] Figure 1 A flowchart of a flight training adaptive optimization method based on operation big data provided by the embodiments of the present application is shown in Figure 1 The method comprises:

[0066] Step 101: Collecting attitude time series data, structure deformation monitoring data and historical operation data of the flight target in the current period of maneuvering state.

[0067] In step 101, the maneuvering state refers to the dynamic flight state of the flight target when performing specific training actions, including high angle of attack turning, rapid climbing, high speed diving and other flight actions that require complex control. These states will cause rapid changes in the attitude, speed and force of the aircraft, and require multi-dimensional data collection and real-time processing to ensure flight safety and training effectiveness. The flight target can refer to a civil passenger aircraft. The attitude time series data represents the attitude information of the flight target changing over time during the maneuvering process, including pitch angle, roll angle, yaw angle, etc., which are used to describe the flight state. The structure deformation monitoring data represents the deformation of the structure (such as wings, fuselage) of the flight target under the maneuvering, reflecting the force state of the structure. The historical operation data represents the recorded attitude, control command and environmental data in the past flight training, which is used to optimize the current control strategy.

[0068] In the embodiments of the present application, the attitude time series data in the current maneuvering state is collected in real time by sensors installed on the flight target, including angle, angular velocity and other dynamic parameters; at the same time, strain sensors or optical fiber sensors are used to monitor the structure deformation data of key parts such as wings and fuselage; in addition, the historical flight data stored in the database is retrieved, including flight records under similar maneuvering state. These data together constitute the basis for subsequent optimization.

[0069] For example, in the maneuvering training, the flight target A performs high angle of attack turning, and the sensors collect attitude data such as roll angular velocity and pitch angular velocity in real time, while monitoring the deformation of the wing leading edge. The system retrieves flight records under similar maneuvering for the past 10 times from the database, including attitude adjustment parameters and structure deformation compensation values, to provide a reference for subsequent optimization.

[0070] Step 102: Environmental interference suppression processing is performed on the attitude time series data to extract motion feature parameters.

[0071] In step 102, environmental interference suppression processing refers to eliminating the influence of airflow disturbance, mechanical vibration and other noise on attitude data. Motion feature parameters refer to the core motion information extracted after denoising, including angular motion components (such as roll, pitch) and linear motion components (such as acceleration).

[0072] In the embodiments of the present application, frequency domain analysis is performed on the collected attitude time series data to identify and filter out high frequency noise components (such as engine vibration); smooth denoising time series data is generated through time domain reconstruction; finally, angular motion components (such as roll angular velocity) and linear motion components (such as vertical acceleration) are separated and combined as motion feature parameters.

[0073] For example, the original attitude data of the flight target A contains roll angle fluctuations caused by air flow disturbance. After frequency domain filtering, the abnormal high frequency components are removed to obtain a smooth roll angle velocity curve. Then, stable roll and pitch motion features are extracted from the curve for subsequent fusion calculation.

[0074] Step 103: converting the structure deformation monitoring data into deformation compensation data, and fusing the motion feature parameters with the deformation compensation data by using an attitude fusion algorithm.

[0075] In step 103, the deformation compensation data represents key deformation features extracted from the structure deformation monitoring data after modal decomposition, which are used to correct the motion parameters.

[0076] In the embodiments of the present application, the structure deformation monitoring data is subjected to modal decomposition to extract feature vectors of dominant deformation modes. The feature vectors are mapped to the body coordinate system to generate deformation compensation data. A spatial trimming matrix is established to couple the motion feature parameters and the deformation compensation data according to weights, and a fused attitude feature set is output.

[0077] For example, the wing deformation data of the flight target A is decomposed to obtain bending and torsion modes. Key deformation vectors are extracted and converted into compensation coefficients. In the spatial trimming matrix, the roll angle velocity and the wing deformation compensation coefficients are fused according to weights to output corrected roll motion parameters.

[0078] Step 104: based on the historical operation data, performing multi-period feature matching on the fusion result of the current period to generate six-dimensional attitude information of the flight target.

[0079] In step 104, the six-dimensional attitude information refers to a complete motion state description of the flight target in a three-dimensional space, which includes three angular motion components (roll angle, pitch angle, and yaw angle) and three linear motion components (forward and backward acceleration, left and right acceleration, and vertical acceleration). The six-dimensional parameters together constitute a quantitative representation of the spatial pose and motion trend of the flight vehicle, which is used for accurate control of the flight attitude.

[0080] In the embodiments of the present application, a feature library is constructed from the historical data, and the deviation of the current fusion result from the historical sequence is calculated. The historical compensation coefficient with the smallest deviation is selected to generate the six-dimensional attitude information.

[0081] For example, the current roll angle velocity of the flight target A is closest to the third group of data in the historical library. The corresponding compensation coefficient 0.9 is used to adjust the current parameter to output the optimized roll angle velocity.

[0082] Step 105: based on the six-dimensional attitude information, generating an optimized control instruction through an adaptive control strategy, wherein the optimized control instruction is used to optimize the response performance of the flight target.

[0083] In step 105, the optimization control instruction includes angle control (such as rudder deflection angle), speed control (such as thrust adjustment), and dynamic response gain adjustment (such as damping coefficient, sensitivity coefficient, etc.), which optimizes the attitude stability, maneuvering accuracy, and anti-interference ability of the flight target by comprehensively adjusting these parameters.

[0084] In the embodiments of the present application, the six-dimensional attitude information is input into the strategy iterator to generate an initial instruction; and the gain is adjusted through disturbance simulation test to output an optimized control instruction.

[0085] For example, the flight target A generates a rudder deflection instruction according to the optimized roll angular velocity, which is adjusted to 3 degrees of deflection and the damping gain is increased after simulation test to ensure the turning stability.

[0086] The method effectively improves the flight accuracy and anti-interference ability under maneuvering by collecting and fusing motion and deformation data in real time, combining historical optimal strategies, and dynamically generating control instructions, solving the response lag problem of traditional static optimization methods.

[0087] In order to solve the problem of structural deformation interference on flight attitude during maneuvering, in some embodiments, step 103: the structure deformation monitoring data is converted into deformation compensation data, and the motion characteristic parameters and the deformation compensation data are fused by using an attitude fusion algorithm, including:

[0088] Step 201: performing modal decoupling operation on the structure deformation monitoring data to obtain a plurality of decoupled modal components.

[0089] In step 201, the modal decoupling operation refers to decomposing the structure deformation monitoring data into a plurality of independent vibration mode components, each component representing the deformation characteristics of the structure under different force conditions.

[0090] In the embodiments of the present application, the collected structure deformation data is decomposed into basic deformation modes such as bending and torsion by mathematical methods, and these mode components can more clearly reflect the force characteristics of the structure during maneuvering.

[0091] Step 202: extracting corresponding key deformation feature vectors from each of the modal components.

[0092] In step 202, the key deformation feature vector refers to the deformation feature parameter with the greatest influence on the flight attitude extracted from each modal component.

[0093] In the embodiments of the present application, each modal component is analyzed, and the features with large amplitude and long duration are selected as key deformation feature vectors, which can accurately represent the main influence of structure deformation on flight attitude.

[0094] Step 203: Map all the key deformation feature vectors to the body coordinate system, which is established based on the geometric configuration of the flight target, to generate deformation compensation data.

[0095] In step 203, the body coordinate system is a reference coordinate system established with the geometric center of the flight target as the origin, used to uniformly describe the spatial relationship between structural deformation and motion features. The geometric configuration refers to the three-dimensional physical structural features of the flight target, including key size parameters such as fuselage length, wing span, control surface position, etc. These data are obtained from the design drawings or three-dimensional digital models of the aircraft, through measurement or modeling, to accurately define the reference direction and origin position of each axis of the body coordinate system.

[0096] In the embodiments of the present application, the extracted key deformation feature vectors are converted to the coordinate system to generate deformation compensation data that can be directly used for motion compensation, ensuring that the deformation influence is consistent with the correction direction of the motion parameters.

[0097] Step 204: Establish a spatial balancing matrix between the motion feature parameters and the deformation compensation data.

[0098] In step 204, the spatial balancing matrix is a calculation framework used to coordinate motion feature parameters and deformation compensation data, with rows corresponding to motion parameters and columns corresponding to deformation compensation coefficients.

[0099] In the embodiments of the present application, when establishing the matrix, the roll angular velocity, pitch angular velocity, and other motion parameters are arranged according to physical correlation with deformation compensation coefficients of parts such as wings and fuselages, forming the basis structure for data fusion.

[0100] Step 205: Perform heterogeneous data coupling operations on the spatial balancing matrix through an attitude fusion algorithm to generate a target fused attitude feature set, which is the fusion result.

[0101] In step 205, the heterogeneous data coupling operation refers to the process of collaboratively calculating different types of data (motion parameters and deformation compensation) in a unified framework.

[0102] In the embodiments of the present application, the motion parameters and deformation compensation data are weighted and fused through a specific algorithm on the spatial balancing matrix, finally outputting a fused attitude feature set that considers both motion state and structural deformation.

[0103] The following is a specific example:

[0104] In the maneuvering training, when the flight target A makes a high angle of attack turn, the deformation monitoring data of the wing leading edge is collected by the distributed sensor, showing that the maximum deformation of the left wing is 0.5 mm and that of the right wing is 0.3 mm; first, modal decoupling operation is performed on the deformation data, and decomposition is performed to obtain a bending modal proportion of 65% and a torsional modal proportion of 35%, wherein the deformation characteristic vector corresponding to the bending modal is [0.65, 0], and the torsional modal is [0, 0.35]; the characteristic vectors are mapped to the body coordinate system established with the center of gravity of the flight target A as the origin to generate deformation compensation data, wherein the left wing compensation amount is 0.5*0.65=0.325 mm, and the right wing compensation amount is 0.3*0.35=0.105 mm; at the same time, the current roll angular velocity 150 degrees per second and the pitch angular velocity 80 degrees per second are obtained from the motion characteristic parameters, a spatial trimming matrix is established, and the rows of the matrix represent the motion parameters and the columns represent the deformation compensation amount; coupling calculation is performed through an attitude fusion algorithm, the roll angular velocity 150 degrees per second and the left wing compensation amount 0.325 mm are fused according to the weight coefficient 0.8 to obtain a corrected value 150-0.325*0.8*100=124 degrees per second, and the pitch angular velocity 80 degrees per second and the right wing compensation amount 0.105 mm are fused according to the weight coefficient 0.6 to obtain a corrected value 80-0.105*0.6*100=73.7 degrees per second, and the finally generated target fusion attitude feature set contains the corrected roll angular velocity 124 degrees per second and the pitch angular velocity 73.7 degrees per second, which takes into account both the motion state and the structural deformation influence, wherein the weight coefficients 0.8 and 0.6 are determined according to the optimal compensation effect of the same kind of maneuvering in the historical data, and 0.325*0.8*100 in the calculation formula represents that the deformation amount 0.325 mm is converted into angular velocity correction amount according to the weight 0.8, and 100 is a unit conversion coefficient.

[0105] In the embodiments of the present application, the deep cooperation of motion characteristics and structural deformation is realized through the above steps, the control error problem caused by structural deformation during maneuvering is effectively solved, and the stability and control accuracy of the flight attitude are significantly improved.

[0106] In order to solve the cooperative optimization problem of motion characteristics and structural deformation data in maneuvering, in some embodiments, step 205: the heterogeneous data coupling operation is performed on the spatial trimming matrix through the attitude fusion algorithm to generate a target fusion attitude feature set, comprising:

[0107] Step 301: assigning the motion characteristic parameters to the first coupling domain of the spatial trimming matrix.

[0108] In step 301, the spatial alignment matrix contains two coupling domains, and the matrix serves as a data bearing framework, the first coupling domain and the second coupling domain are independent data storage areas specially divided therefrom, and are respectively used for storing motion characteristic parameters and deformation compensation data, which are kept spatially isolated in the matrix but can interact through operation. The first coupling domain refers to a sub-area in the spatial alignment matrix specially used for storing motion characteristic parameters, including roll angular velocity, pitch angular velocity and other dynamic motion parameters.

[0109] In the embodiment of the present application, the extracted motion characteristic parameters are sequentially arranged in the designated row area of the matrix according to their physical dimensions, forming a data set that can be directly called by an algorithm.

[0110] Step 302: allocating the deformation compensation data to the second coupling domain of the spatial alignment matrix.

[0111] In step 302, the first coupling domain is specially used for storing motion characteristic parameters (such as angular velocity, acceleration and other dynamic behavior data), and the second coupling domain is specially used for storing deformation compensation data (such as structural deformation amount and other static state data), and the difference between the two lies in the data type and the action dimension, the first coupling domain reflects the flight mechanics characteristics, and the second coupling domain reflects the structural mechanics characteristics.

[0112] In the embodiment of the present application, the converted deformation compensation data is stored in the column area of the matrix according to the corresponding structure part, and a spatial corresponding relationship is formed with the motion parameters.

[0113] Step 303: activating the tensor fusion core of the attitude fusion algorithm, and performing cross projection operation on the first coupling domain and the second coupling domain based on the tensor fusion core on the spatial alignment matrix.

[0114] In step 303, the tensor fusion core refers to the core processing unit for realizing multi-source data fusion calculation. The cross projection operation refers to the process of coupling calculation of two types of data in a specific direction.

[0115] In the embodiment of the present application, the optimal projection direction is determined according to the current flight state, so that the motion parameters and the deformation compensation data are calculated in a weighted interaction along the direction, and coupled features are generated.

[0116] Step 304: generating an initial fusion attitude feature set based on the projection operation result.

[0117] In step 304, the initial fusion attitude feature set refers to the intermediate result containing redundant information generated after the cross projection operation.

[0118] In the embodiment of the present application, the multi-dimensional coupled features obtained by the projection operation are combined according to the physical dimensions, forming a data set containing complete motion-deformation correlation features.

[0119] Step 305: Perform a degree of freedom dimension reduction operation on the initial fusion attitude feature set to generate a target fusion attitude feature set.

[0120] In step 305, the degree of freedom dimension reduction operation refers to a process of extracting core features from the initial feature set and simplifying the data dimension. This operation removes redundant information and adapts to the real-time processing requirements of the flight control system. If not dimensionally reduced, it will result in excessively high computational complexity and non-focused control instructions.

[0121] In the embodiments of the present application, through feature importance analysis, the feature components that have the greatest impact on flight control are retained, redundant information is removed, and a simplified feature set that can be directly used for control decision-making is generated.

[0122] The following is a specific example:

[0123] In a maneuvering training, when the flight target A makes a high angle of attack turn, first, the roll angular velocity 150 degrees per second and the pitch angular velocity 80 degrees per second in the motion feature parameters are respectively assigned to the first row and the second row of the first coupling domain of the spatial trim matrix, and the left wing shape deformation compensation amount 0.325 millimeters and the right wing shape deformation compensation amount 0.105 millimeters are assigned to the first column and the second column of the second coupling domain; after activating the tensor fusion core of the attitude fusion algorithm, the projection direction vector is calculated as [0.7, 0.3] according to the current flight state parameters Mach number 4.5 and angle of attack 30 degrees, indicating the weight proportion of aerodynamic effect and structural effect; along the projection direction, the roll angular velocity 150 degrees per second of the first coupling domain and the left wing compensation amount 0.325 millimeters of the second coupling domain are executed orthogonal projection superposition operation, the calculation formula is 150*0.7+0.325*0.3=105.0975, and the pitch angular velocity 80 degrees per second and the right wing compensation amount 0.105 millimeters are executed the same operation to get 80*0.7+0.105*0.3=56.0315, generating an initial fusion attitude feature set containing the two calculation results; finally, the degree of freedom dimension reduction operation is performed on the feature set, and the roll axis result 105.0975 is combined with the pitch axis result 56.0315 by weighted average calculation according to the weight 0.6 and the weight 0.4, to get the final target fusion attitude feature set value 105.0975*0.6+56.0315*0.4=85.0655, wherein the weight coefficients 0.6 and 0.4 are the importance proportions of the roll axis and the pitch axis determined according to the historical optimal control strategy. The feature value will be directly used as the basis for the flight control system to generate rudder deflection instructions, ensuring that the aircraft can maintain the predetermined attitude and effectively suppress the interference caused by wing shape deformation during high angle of attack turning.

[0124] In the embodiments of the present application, the depth collaboration of motion parameters and deformation data is realized through the above steps, which not only retains the key motion characteristics but also effectively compensates for the influence of structural deformation, significantly improving the control accuracy and stability in the mobile state.

[0125] To solve the problem of accurate fusion of motion characteristics and structural deformation data in the mobile state, in some embodiments, step 303: performing cross-projection operation on the first coupling domain and the second coupling domain based on the tensor fusion kernel on the spatial trimming matrix, comprises:

[0126] Step 401: generating a mobile feature vector based on the first coupling domain, and extracting a structural constraint vector from the second coupling domain.

[0127] In step 401, the mobile feature vector refers to the core motion feature combination extracted from the motion characteristic parameters of the first coupling domain. The structural constraint vector refers to the key deformation feature combination extracted from the deformation compensation data of the second coupling domain.

[0128] In the embodiments of the present application, the roll angular velocity and pitch angular velocity and other parameters in the first coupling domain are combined into a mobile feature vector according to the physical dimension. The specific process is based on the motion characteristic parameters (such as roll angular velocity 120 degrees / second, pitch angular velocity 80 degrees / second) in the first coupling domain, and the mobile feature vector is generated through normalization and principal component extraction operation. The specific implementation is as follows: first, divide each motion parameter by its maximum range value (for example, divide the roll angular velocity 120 degrees / second by the range 200 degrees / second to get 0.6, and divide the pitch angular velocity 80 degrees / second by the range 150 degrees / second to get 0.53), then select the principal components with a contribution rate of more than 85% through eigenvalue decomposition, and finally generate the mobile feature vector [0.58, 0.42], where 0.58=0.6×0.8+0.53×0.2 (0.8 and 0.2 are principal component weight coefficients), which condenses the core motion feature. At the same time, the deformation compensation amount of the key positions such as the leading edge of the wing and the middle part of the fuselage is extracted from the second coupling domain to form a structural constraint vector, forming a feature group that can be calculated interactively.

[0129] Step 402: calling the projection direction calculation method of the tensor fusion kernel, combining the current flight state of the flight target, and determining the projection direction.

[0130] In step 402, the current flight state is a real-time quantitative representation of the maneuvering state, including dynamic parameters (such as Mach number, angle of attack) and structural state (such as deformation gradient), which provides a transient environmental benchmark for cross-projection operation; while the maneuvering state is a macroscopic action type defined by the training task (such as high angle of attack turn), and the two constitute a mapping relationship between real-time data and preset scene - by dynamically adjusting the projection direction vector according to the current flight state parameters, so that data fusion adapts to the physical constraints of specific maneuvering state. The projection direction refers to the optimal fusion direction for coordinating motion features and deformation features determined according to the current flight state.

[0131] In the embodiments of the present application, by analyzing the current flight speed, attitude angle and other state parameters, a projection direction vector that can balance the aerodynamic effect and structural effect is calculated, which determines the fusion weight ratio of motion features and deformation features.

[0132] Step 403: Orthogonal projection and superposition of the maneuvering feature vector and the structural constraint vector along the projection direction.

[0133] In step 403, orthogonal projection and superposition refers to the process of weighted fusion calculation of the maneuvering feature vector and the structural constraint vector along the projection direction.

[0134] In the embodiments of the present application, according to the determined projection direction vector, the components of the maneuvering feature vector and the corresponding components of the structural constraint vector are proportionally superimposed to generate a fusion feature value that reflects both the motion state and the structural deformation.

[0135] The following is a specific example:

[0136] In the maneuvering training, when the flight target A performs a high angle of attack turn, first generate a maneuvering feature vector [150, 80] based on the roll angular velocity 150 degrees per second and the pitch angular velocity 80 degrees per second in the first coupling domain, and at the same time extract the left wing deformation compensation amount 0.325 millimeters and the right wing deformation compensation amount 0.105 millimeters from the second coupling domain to form a structural constraint vector [0.325, 0.105]; call the projection direction calculation method of the tensor fusion kernel, combine the current flight state parameters Mach number 4.5 and angle of attack 30 degrees, and determine the projection direction vector [0.7, 0.3] by calculating the balance relationship between aerodynamic force and structural force, where 0.7 represents the motion feature weight and 0.3 represents the deformation feature weight; along the projection direction, the maneuvering feature vector and the structural constraint vector are orthogonally projected and superimposed, wherein the fusion calculation process of the roll angular velocity 150 degrees per second and the left wing compensation amount 0.325 millimeters is 150x0.7+0.325x0.3=105.0975, and the fusion calculation process of the pitch angular velocity 80 degrees per second and the right wing compensation amount 0.105 millimeters is 80x0.7+0.105x0.3=56.0315.

[0137] In the embodiments of the present application, the intelligent fusion of the motion features and the deformation features is realized through the above steps, so that the generated control instructions can not only accurately reflect the changes in the flight state, but also effectively compensate for the interference caused by the structural deformation, thereby significantly improving the accuracy and reliability of the maneuvering flight control.

[0138] To solve the problem of the influence of environmental interference on the attitude data during maneuvering, in some embodiments, step 102: performing environmental interference suppression processing on the attitude time series data to extract motion feature parameters, includes:

[0139] Step 501: performing frequency domain transformation on the attitude time series data to obtain a corresponding frequency energy distribution.

[0140] In step 501, the frequency energy distribution refers to converting the attitude data in the time domain into the energy intensity distribution in the frequency domain, which is used to identify the energy features of different frequency components.

[0141] In the embodiments of the present application, the data of the attitude angle changing with time collected is converted into the corresponding relationship between frequency and energy intensity through a mathematical transformation method, forming an energy distribution map that can analyze the contribution degree of each frequency component.

[0142] Step 502: identifying the vibration residual amount from the frequency energy distribution through a noise separation mechanism, filtering out the vibration residual amount to form denoised time series data.

[0143] In step 502, the vibration residual amount refers to the abnormal high-frequency component in the frequency energy distribution caused by mechanical interference such as engine vibration and airflow impact, and its energy amplitude is significantly higher than the basic frequency domain features of the normal flight state, which is manifested as a burst energy peak in the 5-50Hz frequency band, and needs to be identified and filtered out to ensure data authenticity.

[0144] In the embodiments of the present application, by analyzing the features of the frequency energy distribution, the high-frequency component exceeding the energy range of the normal flight state is identified, and digital filtering technology is used to remove these interference components from the original data.

[0145] Step 503: separating the angular motion feature component and the linear motion feature component from the denoised time series data, and combining the angular motion feature component and the linear motion feature component into a motion feature parameter.

[0146] In step 503, the angular motion characteristic component refers to the motion parameters of the aircraft rotating around each axis. The linear motion characteristic component refers to the motion parameters of the aircraft translating along each axis. The denoised time series data is the original time series signal after interference suppression (such as the smoothed pitch angle curve), and the motion characteristic parameter is a quantitative index (such as the pitch angle velocity 20 degrees per second) extracted from the signal. The former is the processed "data carrier", and the latter is the extracted "feature information". The two are in a data hierarchy relationship.

[0147] In the embodiments of the present application, the angular velocity parameters describing the rotational motion and the acceleration parameters describing the translational motion are extracted from the processed time series data to form a complete motion characteristic description.

[0148] The following is a specific example:

[0149] In the maneuvering training, when the flight target A makes a high angle of attack turn, the sensor collects the original roll angle data containing airflow disturbance, and the data shows that the angle fluctuates from 20 degrees to 25 degrees and then falls to 22 degrees within 0.5 seconds. First, the time series data is subjected to fast Fourier transform to obtain the frequency energy distribution, and it is found that there is an abnormal peak with an amplitude of 3 times the normal value in the 8-12Hz frequency band, which is determined to be an interference component caused by engine vibration. After filtering out the signal components in this frequency band by designing a digital band-stop filter, the time domain is reconstructed to obtain a smoothed roll angle change curve, which shows that the angle increases smoothly from 20 degrees to 22 degrees. The roll angle velocity is calculated to be 22-20 / 0.5=4 degrees per second, and the lateral acceleration 1.5g is extracted from the accelerometer data, where 1.5g is obtained by measuring the lateral force 3000N divided by the mass of the aircraft 2000kg. Finally, the roll angle velocity 4 degrees per second and the lateral acceleration 1.5g are combined into the motion characteristic parameter [4, 1.5], which not only eliminates the influence of vibration interference, but also accurately reflects the actual motion state of the flight target.

[0150] In the embodiments of the present application, the influence of environmental interference on attitude data is effectively eliminated through the above steps, and the motion characteristics reflecting the actual flight state are extracted, providing a reliable data basis for subsequent precise control and optimization.

[0151] In order to solve the problem of optimizing the application of historical experience data in current flight control, in some embodiments, step 104: based on the historical operation data, multi-period characteristic matching is performed on the fusion result of the current period to generate six-dimensional attitude information of the flight target, including:

[0152] Step 601: constructing a cross-period feature library according to the historical operation data, the cross-period feature library containing a historical fusion feature sequence and a corresponding compensation coefficient.

[0153] In step 601, the cross-cycle feature library refers to a set of effective control parameters verified in the historical flight process, including a historical fusion feature sequence and a corresponding compensation coefficient. The attitude timing data and the structural deformation monitoring data in the historical operation data are subjected to the same process as the environmental interference suppression process, that is, the denoising timing data are obtained through frequency domain transformation and dynamic noise separation, the angular motion feature component and the linear motion feature component are separated to form a motion feature parameter, the deformation monitoring data are converted into deformation compensation data, and finally the motion feature parameter and the deformation compensation data are coupled on a spatial balancing matrix through an attitude fusion algorithm to generate a fusion result of each historical cycle, which is arranged in time sequence to form a historical fusion feature sequence.

[0154] In the embodiment of the present application, the fusion result of the verified motion feature and the deformation compensation data is extracted from the historical flight database, stored according to the type of maneuver, and each historical record contains complete feature parameters and the optimal compensation coefficient used at that time.

[0155] Step 602: Calculate the deviation feature parameter between each parameter in the fusion result of the current cycle and the corresponding parameter in the historical fusion feature sequence.

[0156] In step 602, the deviation feature parameter refers to the difference between the current flight state parameter and the historical optimal state parameter.

[0157] In the embodiment of the present application, the fusion feature parameters obtained in the current flight are compared with the corresponding parameters of the historical records in the feature library one by one, the difference in each dimension parameter is calculated, and a deviation feature set that can be quantitatively compared is formed.

[0158] Step 603: For each parameter in the fusion result, the compensation coefficient of the corresponding parameter in the historical fusion feature sequence with the smallest value of the deviation feature parameter is taken as the optimal compensation coefficient to construct an optimal compensation coefficient combination.

[0159] In step 603, the optimal compensation coefficient combination refers to the most suitable compensation parameter set selected from the historical data for the current flight state.

[0160] In the embodiment of the present application, by analyzing the deviation features of the current parameters and each historical record, the compensation coefficient used by the historical record with the smallest deviation is selected to form an optimal compensation scheme suitable for the current flight state.

[0161] Step 604: Generate six-dimensional attitude information based on the optimal compensation coefficient combination.

[0162] In the embodiment of the present application, the optimal compensation coefficient set is applied to the current fusion result, and the parameters of each motion dimension are adjusted accordingly to output accurate attitude information that meets the current state requirements and historical optimal experience. The specific implementation process is as follows: the optimal compensation coefficient is weighted and synthesized with the current cycle fusion result, wherein the optimal compensation coefficient is used as an attitude correction weight to proportionally adjust the angular motion component and linear motion component in the fusion result, and the adjusted components are combined into the six-dimensional attitude information through kinematics inverse solution. For example, in the high-speed aircraft large-angle turning training, the current cycle fusion result shows that there is a roll axis oscillation feature, and the optimal compensation coefficient (such as the roll angular velocity damping coefficient 0.85) for such oscillation is obtained by matching the historical feature library. After the coefficient is applied to the current fusion result, the roll axis overshoot component is suppressed by 20%, and stable six-dimensional attitude information (including the corrected roll angular velocity 120° / s, pitch angle 45°, etc.) is generated for subsequent control command generation.

[0163] The following is a specific example:

[0164] In the maneuver training, when the flight target A makes a large-angle turn, the system first retrieves the historical records of the past 10 similar maneuver trainings from the database to construct a cross-cycle feature library, which contains the fusion results of the roll angular velocity, pitch angular velocity, etc. motion parameters and wing deformation compensation values and the corresponding compensation coefficients; calculate the deviation of the current roll angular velocity 124 degrees per second from the historical records in the feature library, find that the deviation from the third group of historical data 125 degrees per second is the smallest, which is 1 degree per second, and the compensation coefficient corresponding to this group of data is 0.9; At the same time, calculate the deviation of the current pitch angular velocity 73.7 degrees per second, and the deviation of the fifth group of historical data 74 degrees per second is the smallest, which is 0.3 degrees per second, and the compensation coefficient is 0.85; Apply the two optimal compensation coefficients 0.9 and 0.85 to the current parameters respectively, wherein the roll angular velocity is adjusted to 124x0.9=111.6 degrees per second, and the pitch angular velocity is adjusted to 73.7x0.85=62.645 degrees per second. The calculation formula of the two adjustment values is that 124x0.9 represents the historical optimal compensation proportion based on the current roll angular velocity, and 73.7x0.85 is the same; combined with the motion parameters of the other four dimensions, the final six-dimensional attitude information is generated, including the adjusted roll angular velocity 111.6 degrees per second, the pitch angular velocity 62.645 degrees per second, and the unadjusted yaw angular velocity, three-axis acceleration parameters, which reflect the current actual flight state and also integrate historical optimal control experience, providing a reliable basis for subsequent generation of accurate control commands.

[0165] In the embodiments of the present application, the intelligent application of historical flight experience in current control is realized through the above steps, so that the generated attitude information not only conforms to the current flight state, but also inherits the advantages of the historical optimal control strategy, significantly improving the accuracy and reliability of flight control.

[0166] To solve the problem of accurate generation of control instructions in the maneuvering state, in some embodiments, step 105: generating optimized control instructions through an adaptive control strategy based on the six-dimensional attitude information, includes:

[0167] Step 701: input the six-dimensional attitude information into a strategy iterator, and the strategy iterator stores a dynamic response rule set.

[0168] In step 701, the strategy iterator is the carrier of the adaptive control strategy, the stored dynamic response rule set is the static knowledge base of the strategy, and the gain adjustment through disturbance simulation test is the dynamic optimization process of the strategy. The combination of the two realizes the self-iterative update of the strategy. The adaptive control strategy is embodied through the two steps of "the strategy iterator stores a dynamic response rule set" and "adjusting the gain distribution of the initial instruction sequence based on test feedback", the former defines the strategy rule base, and the latter executes the strategy optimization, which together constitute a closed-loop adaptive control. The dynamic response rule set contains verified effective control parameter combinations in different maneuvering states.

[0169] In the embodiments of the present application, the attitude information containing six dimensions of roll angular velocity, pitch angular velocity, etc. is input into the module, and the system automatically matches the control rule category corresponding to the current state.

[0170] Step 702: generating an initial control instruction according to the dynamic response rule set.

[0171] In step 702, the initial control instruction refers to the preliminary execution command generated according to the basic control rule.

[0172] In the embodiments of the present application, based on the matched control rule category, the corresponding reference rudder deflection angle and thrust parameter are extracted to form a preliminary control scheme.

[0173] Step 703: performing disturbance simulation test on the initial control instruction.

[0174] In step 703, the disturbance simulation test refers to the verification of the execution effect of the initial instruction in a virtual disturbance environment.

[0175] In the embodiments of the present application, by constructing disturbance scenarios such as crosswind and airflow mutation, the response performance of the initial control instruction in actual flight is simulated and calculated.

[0176] Step 704: based on the test results, optimizing the gain distribution of the initial control instruction to output an optimized control instruction.

[0177] In step 704, the gain distribution refers to the control intensity proportion relationship of each execution component (such as a rudder surface, an engine) in the control instruction, which is derived from the verified optimal control parameter combination in different maneuvering states in historical flight data, and the reference gain value is determined by analyzing the contribution of each execution component in a similar flight scenario. Gain distribution optimization refers to adjusting the action intensity proportion of the control parameters according to the test results.

[0178] In the embodiments of the present application, the response effect of each control parameter in the simulation test is analyzed, and the weight of the parameters that are insufficient or excessive is adjusted to form the final optimized instruction. The specific implementation process is as follows: first, perturb the initial control instruction, and record the response deviation of each execution component; then calculate the proportional relationship between the deviation and the allowable threshold, and increase the gain of the components with insufficient response in proportion, and decrease the gain of the components with excessive response. For example, if the initial rudder instruction of 5 degrees produces a 15-degree overshoot (allowable 10 degrees) in the crosswind test, the gain adjustment is 1-(15-10) / 10=0.5 times, and the final optimized instruction is 5*0.5=2.5 degrees.

[0179] The following is a specific example:

[0180] In the maneuvering training, when the flight target A generates control instructions according to the optimized six-dimensional attitude information (including roll angular velocity 111.6 degrees per second, pitch angular velocity 62.645 degrees per second, etc.), first, the attitude information is input into the strategy iterator, matched to the dynamic response rule of the high angle of attack turning category, and the initial control instruction including rudder deflection 4 degrees and elevator deflection 7 degrees is generated; in the simulation test link, a 10 meters per second crosswind disturbance condition is set, and the test finds that the rudder deflection of 4 degrees causes the roll angular velocity deviation to reach 18 degrees per second (allowable threshold is 12 degrees per second), and the pitch angular velocity deviation is 5 degrees per second (allowable threshold is 8 degrees per second); based on the test results, the gain distribution is optimized, wherein the rudder gain is adjusted to 1-(18-12) / 12=0.5 times, and the final instruction is 4*0.5=2 degrees, and the pitch control is kept at 7 degrees because the deviation is within the allowable range, but the response coefficient is increased to 1.1, which is obtained by rounding 1+5 / 8=1.625; the final output optimized control instruction includes rudder deflection 2 degrees with damping gain 0.8 and elevator deflection 7 degrees with response gain 1.1.

[0181] In the embodiments of the present application, the dynamic optimization of the control instruction is realized through the above steps, so that the flight system can quickly respond to attitude changes and effectively resist external disturbances, and the control quality and safety of complex maneuvering flight are significantly improved.

[0182] Figure 2A structural schematic diagram of a flight training adaptive optimization system based on operation big data provided by an embodiment of the present application is shown in Figure 2 The system comprises:

[0183] A collection module 21 is configured to collect attitude time series data of a flight target in a current period, structure deformation monitoring data, and historical operation data.

[0184] An extraction module 22 is configured to perform environment interference suppression processing on the attitude time series data to extract motion feature parameters.

[0185] A conversion module 23 is configured to convert the structure deformation monitoring data into deformation compensation data, and fuse the motion feature parameters and the deformation compensation data by using a posture fusion algorithm.

[0186] A matching module 24 is configured to perform multi-period feature matching on a fusion result of the current period based on the historical operation data, and generate six-dimensional attitude information of the flight target.

[0187] A generation module 25 is configured to generate an optimized control instruction by using an adaptive control strategy based on the six-dimensional attitude information, and the optimized control instruction is used to optimize response performance of the flight target.

[0188] Figure 2 The flight training adaptive optimization system based on operation big data can perform Figure 1 The implementation principle and technical effects of the flight training adaptive optimization method based on operation big data are not described again. The specific operation modes of each module and unit of the flight training adaptive optimization system based on operation big data in the above embodiment have been described in detail in the embodiment related to the method, and will not be described in detail here.

[0189] In a possible design, Figure 2 The flight training adaptive optimization system based on operation big data can be implemented as a computing device, as shown in Figure 3 The computing device can include a storage component 31 and a processing component 32.

[0190] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0191] The processing component 32 is configured to perform the flight training adaptive optimization method based on operation big data of the above embodiment. Figure 1 The embodiment of the flight training adaptive optimization method based on operation big data.

[0192] The processing component 32 can include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more Application-Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Process Device (DSPD), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components, for executing the above method.

[0193] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage devices or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0194] Of course, the computing device can also include other components, such as an input / output interface, a display component, a communication component, etc.

[0195] The input / output interface provides an interface between the processing component and peripheral interface modules, which can be output devices, input devices, etc.

[0196] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0197] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform, and the computing device can refer to a cloud server, and the processing component, the storage component, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0198] The embodiment of the application further provides a computer storage medium, which stores a computer program, and the computer program can realize the above method when being executed by a computer. Figure 1 The embodiment of the application further provides a flight training adaptive optimization method based on operation big data.

[0199] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, and will not be described here.

[0200] The device embodiments described above are only schematic, and the units described as separate components can or can not be physically separated, and the components displayed as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment scheme. Those skilled in the art can understand and implement without creative labor.

[0201] Through the description of the foregoing embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and a necessary general hardware platform, and of course, can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0202] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the application, and not to limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application.

Claims

1. A flight training adaptive optimization method based on operation big data, characterized in that, The method comprises the following steps: Collecting attitude time series data, structure deformation monitoring data and historical operation data of the flight target in the current period of maneuvering state; Performing environment interference suppression processing on the attitude time series data to extract motion feature parameters; Converting the structure deformation monitoring data into deformation compensation data, and fusing the motion feature parameters and the deformation compensation data by using a posture fusion algorithm; Based on the historical operation data, multi-period feature matching is performed on the fusion result of the current period to generate six-dimensional attitude information of the flight target; Based on the six-dimensional attitude information, an adaptive control strategy is used to generate an optimized control instruction, which is used to optimize the response performance of the flight target; The method of converting the structure deformation monitoring data into deformation compensation data and fusing the motion feature parameters and the deformation compensation data by using a posture fusion algorithm comprises the following steps: Performing modal decoupling operation on the structure deformation monitoring data to obtain a plurality of decoupled modal components; Extracting corresponding key deformation feature vectors from each of the modal components; Mapping all the key deformation feature vectors to a body coordinate system to generate deformation compensation data, wherein the body coordinate system is established based on the geometric configuration of the flight target; Establishing a spatial trimming matrix between the motion feature parameters and the deformation compensation data; Performing heterogeneous data coupling operation on the spatial trimming matrix by using a posture fusion algorithm to generate a target fusion attitude feature set, which is a fusion result.

2. The operational big data based flight training adaptive optimization method according to claim 1, wherein, The method of performing heterogeneous data coupling operation on the spatial trimming matrix by using a posture fusion algorithm to generate a target fusion attitude feature set comprises the following steps: Assigning the motion feature parameters to a first coupling domain of the spatial trimming matrix; Assigning the deformation compensation data to a second coupling domain of the spatial trimming matrix; Activating a tensor fusion core of the posture fusion algorithm to perform cross-projection operation on the first coupling domain and the second coupling domain based on the tensor fusion core on the spatial trimming matrix; Based on the projection operation result, an initial fusion attitude feature set is generated; Performing degree of freedom dimension reduction operation on the initial fusion attitude feature set to generate a target fusion attitude feature set.

3. The operational big data based flight training adaptive optimization method according to claim 2, wherein, The method of performing cross-projection operation on the first coupling domain and the second coupling domain based on the tensor fusion core on the spatial trimming matrix comprises the following steps: Based on the first coupling domain, a maneuvering feature vector is generated, and a structure constraint vector is extracted from the second coupling domain; A projection direction calculation method of the tensor fusion core is called to determine a projection direction in combination with the current flight state of the flight target; The maneuvering feature vector and the structure constraint vector are orthogonally projected and superimposed along the projection direction.

4. The operational big data based flight training adaptive optimization method of claim 1, wherein, The method of performing environment interference suppression processing on the attitude time series data to extract motion feature parameters comprises the following steps: Performing frequency domain transformation on the attitude time series data to obtain corresponding frequency energy distribution; Identifying vibration residual amount from the frequency energy distribution by using a noise separation mechanism, filtering out the vibration residual amount to form denoising time series data; Separate angular motion feature components and linear motion feature components from the de-noised time series data, and combine the angular motion feature components and the linear motion feature components into motion feature parameters.

5. The operational big data based flight training adaptive optimization method according to claim 1, wherein, The historical running data is used to perform multi-period feature matching on a fusion result of the current period, so as to generate six-dimensional attitude information of the flight target, including: According to the historical running data, a cross-period feature library is constructed, and the cross-period feature library contains historical fusion feature sequences and corresponding compensation coefficients; A deviation feature parameter between each parameter in the fusion result of the current period and a corresponding parameter in the historical fusion feature sequence is calculated. For each parameter in the fusion result, a compensation coefficient of a corresponding parameter in a historical fusion feature sequence with the smallest value of the deviation feature parameter is taken as an optimal compensation coefficient, so as to construct an optimal compensation coefficient combination. Based on the optimal compensation coefficient combination, six-dimensional attitude information is generated.

6. The operational big data based flight training adaptive optimization method according to claim 1, wherein, Based on the six-dimensional attitude information, an adaptive control strategy is used to generate an optimized control instruction, including: The six-dimensional attitude information is input into a strategy iterator, and the strategy iterator stores a dynamic response rule set; According to the dynamic response rule set, an initial control instruction is generated; A disturbance simulation test is performed on the initial control instruction; Based on a test result, a gain distribution of the initial control instruction is optimized, so as to output an optimized control instruction.

7. A flight training adaptive optimization system based on operational big data, characterized in that, It includes: A collection module is configured to collect attitude time series data, structure deformation monitoring data and historical running data of a flight target in a current period under a maneuvering state; An extraction module is configured to perform environmental interference suppression processing on the attitude time series data, so as to extract motion feature parameters; A conversion module is configured to convert the structure deformation monitoring data into deformation compensation data, and fuse the motion feature parameters and the deformation compensation data by using a attitude fusion algorithm; A matching module is configured to perform multi-period feature matching on a fusion result of the current period based on the historical running data, so as to generate six-dimensional attitude information of the flight target; A generation module is configured to generate an optimized control instruction by using an adaptive control strategy based on the six-dimensional attitude information, and the optimized control instruction is used to optimize a response performance of the flight target. The structure deformation monitoring data is converted into deformation compensation data, and the motion feature parameters and the deformation compensation data are fused by using a attitude fusion algorithm, including: A modal decoupling operation is performed on the structure deformation monitoring data, so as to obtain a plurality of decoupled modal components; Corresponding key deformation feature vectors are extracted from the modal components; All the key deformation feature vectors are mapped to a body coordinate system, so as to generate deformation compensation data, and the body coordinate system is established based on a geometric configuration of the flight target; A spatial trimming matrix between the motion feature parameters and the deformation compensation data is established; By using a attitude fusion algorithm, a heterogeneous data coupling operation is performed on the spatial trimming matrix, so as to generate a target fusion attitude feature set, and the target fusion attitude feature set is a fusion result.

8. A computing device, comprising: The method comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to realize the method for adaptive optimization of flight training based on operation big data according to any one of claims 1-6.

9. A computer storage medium, characterized in that The computer program is stored and is executed by a computer to realize the method for adaptive optimization of flight training based on operation big data according to any one of claims 1-6.

Images