CN114004023A - Aircraft pneumatic parameter identification method based on recurrent neural network - Google Patents
Aircraft pneumatic parameter identification method based on recurrent neural network Download PDFInfo
- Publication number
- CN114004023A CN114004023A CN202111249073.0A CN202111249073A CN114004023A CN 114004023 A CN114004023 A CN 114004023A CN 202111249073 A CN202111249073 A CN 202111249073A CN 114004023 A CN114004023 A CN 114004023A
- Authority
- CN
- China
- Prior art keywords
- aerodynamic
- neural network
- flight
- recurrent neural
- aircraft
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Mathematical Analysis (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Mathematical Optimization (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Pure & Applied Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computational Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Automation & Control Theory (AREA)
- Aviation & Aerospace Engineering (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Algebra (AREA)
- Fluid Mechanics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
Abstract
The invention discloses an aircraft pneumatic parameter identification method based on a recurrent neural network. The method comprises the following steps: 1) carrying out a simulated flight test by using a training-level simulator in combination with flight simulation software to obtain flight data; 2) taking an aircraft rigid six-degree-of-freedom dynamic equation set as a state equation of the system, and calculating according to data obtained by tests to obtain corresponding aerodynamic force and aerodynamic moment; 3) taking flight data such as an attack angle, a sideslip angle and the like as input, taking the pneumatic parameters obtained by calculation in the step two as reference data, and training by utilizing a recurrent neural network combined with a real-time recursive learning algorithm to obtain a pneumatic parameter identification model; 4) and (4) selecting a flight data which does not participate in model training and loading the aerodynamic parameter identification model of the recurrent neural network obtained in the third step for parameter identification to obtain corresponding aerodynamic force and aerodynamic moment. The parameter identification model established by the invention has better applicability, can finish accurate modeling aiming at aerodynamic force, and can be popularized and applied.
Description
Technical Field
The invention relates to the field of aircraft pneumatic parameter identification, in particular to an aircraft pneumatic parameter identification method based on a recurrent neural network.
Background
The aircraft is an extremely complex system, and with the development of aircraft research, obtaining accurate aerodynamic characteristics of the aircraft is an important prerequisite and basis for establishing an aircraft model and designing an aircraft control system with excellent performance. And three methods for acquiring the aerodynamic characteristics of the aircraft comprise theoretical calculation, wind tunnel test and aircraft test. In the process of developing the aircraft, a theoretical calculation method is adopted, so that the design period and the research and development cost can be greatly reduced, but the theoretical calculation is limited by imperfect theoretical research and the calculation capability of a computer, and the other two modes cannot be completely replaced. Compared with a flight test, the wind tunnel test is less in cost and higher in flexibility, but the Reynolds number of the wind tunnel test is lower, and tunnel wall interference and support interference exist, so that the wind tunnel test has certain limitation.
Because the limitations of wind tunnel tests of aircrafts and the theoretical calculation are limited by imperfect theoretical research, the application of the pneumatic parameter identification technology to obtain each pneumatic parameter of an aircraft model from flight test data is an integral part of aircraft design and research. The aircraft system identification is to regard an aircraft as a research target, take the relevant theoretical knowledge such as aerodynamics and the like as the main research method by using the control theory such as parameter estimation and the like, and obtain the pneumatic parameters based on flight data so as to complete the construction of a dynamic model. Today, with the development of flight test testing means, the characteristics of high precision and high speed are achieved from sensor signal adjustment, data acquisition, data recording and processing, and how to accurately and rapidly separate real airplane aerodynamic characteristics from various test data results so as to determine an airplane aerodynamic model becomes a problem which needs to be solved urgently by flight test data processing and analyzing personnel. Therefore, the method for determining the aerodynamic parameters of the airplane by researching the airplane parameter identification method is used for finding out a method for accurately and quickly separating the real aerodynamic characteristics of the airplane from the test results, and has great application value for shortening data processing time, reducing flight test period and the like.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an aircraft pneumatic parameter identification method based on a recurrent neural network, which can conveniently and quickly identify the pneumatic power and the pneumatic moment and complete accurate modeling aiming at the pneumatic power.
In order to achieve the purpose, the invention adopts the technical scheme that:
a method for identifying aerodynamic parameters of an aircraft based on a recurrent neural network comprises the following steps:
1) carrying out a simulated flight test by using a training-level simulator in combination with flight simulation software to obtain flight data;
2) taking an aircraft rigid six-degree-of-freedom dynamic equation set as a state equation of the system, and calculating according to the part of flight data acquired in the step 1) to obtain corresponding aerodynamic force and aerodynamic moment;
3) inputting part of flight data acquired in the step 1, wherein the data comprises altitude, airspeed, attack angle, sideslip angle, roll, pitch and yaw angular speeds, elevators, ailerons, rudders and flap rudder deflection angles, taking aerodynamic force and aerodynamic moment obtained by calculation in the step 2) as reference data, and training by utilizing a recurrent neural network and extended Kalman filtering combined with a real-time recursive learning algorithm to obtain a recurrent neural network aerodynamic parameter identification model;
4) and (3) loading the flight data which does not participate in the model training of the step 3) in the step 1) into the pneumatic parameter identification model of the recurrent neural network obtained in the step 3) for parameter identification to obtain corresponding aerodynamic force and aerodynamic moment.
Preferably, the specific steps of using the training-level simulator in combination with flight simulation software to perform a simulated flight test in step 1) to obtain flight data are as follows: adopting a Seiner 172 type flight simulator combined with Prepar 3D software, opening SIMConnect.samples of Prepar 3D on a VS2015 platform, then generating an application program DataHarvester.exe, operating an aircraft to perform a flight test and recording flight data, if the flight test meets the requirement of data acquisition, obtaining corresponding flight test data DataHarvester.csv, and if the flight test does not meet the requirement, deleting csv files under a folder, and re-entering the flight test until the requirement of data acquisition is met.
Preferably, the expression of the aerodynamic force and the aerodynamic moment in the step 2) is as follows:
wherein, CA、CY、CNIs the coefficient of aerodynamic force, Cl、Cm、CnIs the aerodynamic moment coefficient, T is the engine thrust, psiT、For engine mounting angle, nx、ny、nzIs the overload component in three body axis directions, S is the wing area, q∞Is dynamic pressure, Ix、Iy、IzMoment of inertia, I, of the aircraft body axisxzIs the product of inertia of the aircraft body axis.
Preferably, in step 3), firstly, a method of selecting an artificial neural network is selected for the problem of aircraft pneumatic parameter identification, and as the aircraft system dynamically changes with time, the cyclic neural network is further determined to be used, and meanwhile, for online real-time training, the complex nonlinear problem is solved and the operation speed is increased, a real-time recursive learning algorithm is adopted and an extended kalman filter is added, and finally, the cyclic neural network real-time recursive learning algorithm is obtained for parameter identification.
Drawings
FIG. 1 is a flow chart of a simulation calculation method of the present invention;
FIG. 2 is a graph of a portion of raw flight data derived by a simulator;
FIG. 3 is a lift coefficient diagram calculated from rigid six-degree-of-freedom kinetic equations of an aircraft;
FIG. 4 is a resistance coefficient diagram calculated from rigid six-degree-of-freedom kinematic equations of an aircraft;
FIG. 5 is a graph of relative error between the neural network lift coefficient training results and the calculated results;
FIG. 6 is a graph of the relative error of the neural network resistance coefficient training results and the calculated results;
FIG. 7 is a comparison graph of lift coefficient identification results of the aerodynamic parameter identification model.
Detailed Description
The following description of the embodiments of the present invention will be made with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
A method for identifying aerodynamic parameters of an aircraft based on a recurrent neural network specifically comprises the following steps:
1) flight data acquisition by using training-level simulator and flight simulation software to perform simulated flight test
The method comprises the steps of opening SIMConnect.samples of Prepar 3D on a VS2015 platform by adopting a Seina 172 type flight simulator in combination with Prepar 3D software, then generating an application program DataHarvester.exe, setting clear weather in the simulator, setting the turbulence degree to be a normal range and adopting five-sided flight of the field to carry out a test in order to enable flight data to meet the requirement of parameter identification and eliminate external factor interference as much as possible.
And if the flight test meets the data acquisition requirement, acquiring corresponding flight test data DataHarvester. Two sets of data are collected together and used as training parameters for parameter identification, and another set of data is prepared and used for verification. FIG. 2 is a portion of raw flight data derived by the simulator.
2) Taking an aircraft rigid six-degree-of-freedom dynamic equation set as a state equation of a system, and calculating according to data obtained by tests to obtain a corresponding aerodynamic force and aerodynamic moment concrete expression as follows:
wherein, CA、CY、CNIs the coefficient of aerodynamic force, Cl、Cm、CnIs the aerodynamic moment coefficient, T is the engine thrust, psiT、For engine mounting angle, nx、ny、nzIs the overload component in three body axis directions, S is the wing area, q∞Is dynamic pressure, Ix、Iy、IzMoment of inertia, I, of the aircraft body axisxzIs the product of inertia of the aircraft body axis. And (3) lift coefficient and drag coefficient calculated by the two groups of flight data derived in the step one are shown in figures 3 and 4.
3) Inputting part of flight data acquired in the step 1, wherein the data comprises altitude, airspeed, attack angle, sideslip angle, roll, pitch and yaw angular speeds, elevators, ailerons, rudders and flap rudder deflection angles, taking aerodynamic force and aerodynamic moment obtained by calculation in the step 2) as reference data, and training by utilizing a recurrent neural network and extended Kalman filtering combined with a real-time recursive learning algorithm to obtain a recurrent neural network aerodynamic parameter identification model; the method comprises the following specific steps:
firstly, selecting an artificial neural network method aiming at the problem of aircraft pneumatic parameter identification, further determining to use a recurrent neural network because an aircraft system dynamically changes along with time, and simultaneously adopting a real-time recursive learning algorithm and adding an extended Kalman filtering for online real-time training to solve a complex nonlinear problem and accelerate the operation speed so as to finally obtain a recurrent neural network real-time recursive learning algorithm for parameter identification; the specific relative errors of the identification results and the calculation results of the lift coefficient and the resistance coefficient are shown in fig. 5 and fig. 6, the relative errors of the lift coefficient and the calculation results are mainly concentrated between-0.5% and 0.3% and maximally reach 2.4%, the relative errors of the resistance coefficient and the calculation results are more concentrated between-0.1% and 0.1%, and the maximum relative error is 3.1%, so that the recurrent neural network aerodynamic parameter identification model obtained after training has high reliability.
4) And (3) loading flight data which do not participate in model training under the same test condition of the same model into the cyclic neural network aerodynamic parameter identification model obtained in the step 3) for parameter identification, so that corresponding aerodynamic force and aerodynamic moment can be obtained through identification, and the verification result of the lift coefficient is shown in the graph 7, so that the accuracy of the model is reflected.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the technical principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (4)
1. A method for identifying aerodynamic parameters of an aircraft based on a recurrent neural network is characterized by comprising the following steps:
1) carrying out a simulated flight test by using a training-level simulator in combination with flight simulation software to obtain flight data;
2) taking an aircraft rigid six-degree-of-freedom dynamic equation set as a state equation of the system, and calculating according to the part of flight data acquired in the step 1) to obtain corresponding aerodynamic force and aerodynamic moment;
3) inputting part of flight data acquired in the step 1, wherein the data comprises altitude, airspeed, attack angle, sideslip angle, roll, pitch and yaw angular speeds, elevators, ailerons, rudders and flap rudder deflection angles, taking aerodynamic force and aerodynamic moment obtained by calculation in the step 2) as reference data, and training by utilizing a recurrent neural network and extended Kalman filtering combined with a real-time recursive learning algorithm to obtain a recurrent neural network aerodynamic parameter identification model;
4) and (3) loading the flight data which does not participate in the model training of the step 3) in the step 1) into the pneumatic parameter identification model of the recurrent neural network obtained in the step 3) for parameter identification to obtain corresponding aerodynamic force and aerodynamic moment.
2. The method for identifying the aerodynamic parameters of the aircraft based on the recurrent neural network as claimed in claim 1, wherein the specific steps of performing the simulated flight test by using the training-level simulator in combination with the flight simulation software in the step 1) to obtain the flight data are as follows: adopting a Seiner 172 type flight simulator combined with Prepar 3D software, opening SIMConnect.samples of Prepar 3D on a VS2015 platform, then generating an application program DataHarvester.exe, operating an aircraft to perform a flight test and recording flight data, if the flight test meets the requirement of data acquisition, obtaining corresponding flight test data DataHarvester.csv, and if the flight test does not meet the requirement, deleting csv files under a folder, and re-entering the flight test until the requirement of data acquisition is met.
3. The method for identifying the aerodynamic parameters of the aircraft based on the recurrent neural network as claimed in claim 1, wherein the expressions of the aerodynamic force and the aerodynamic moment in step 2) are as follows:
wherein, CA、CY、CNIs the coefficient of aerodynamic force, Cl、Cm、CnIs the aerodynamic moment coefficient, T is the engine thrust, psiT、For engine mounting angle, nx、ny、nzIs the overload component in three body axis directions, S is the wing area, q∞Is dynamic pressure, Ix、Iy、IzMoment of inertia, I, of the aircraft body axisxzIs the product of inertia of the aircraft body axis.
4. The method for identifying the aerodynamic parameters of the aircraft based on the recurrent neural network as claimed in claim 1, wherein in step 3), firstly, an artificial neural network is selected for the problem of identifying the aerodynamic parameters of the aircraft, and as the aircraft system dynamically changes with time, the recurrent neural network is further determined to be used, and meanwhile, in order to train on line in real time, solve the complex nonlinear problem and accelerate the operation speed, a real-time recursive learning algorithm is adopted and an extended kalman filter is added, and finally, a real-time recursive learning algorithm of the recurrent neural network is obtained for identifying the parameters.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111249073.0A CN114004023A (en) | 2021-10-26 | 2021-10-26 | Aircraft pneumatic parameter identification method based on recurrent neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111249073.0A CN114004023A (en) | 2021-10-26 | 2021-10-26 | Aircraft pneumatic parameter identification method based on recurrent neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114004023A true CN114004023A (en) | 2022-02-01 |
Family
ID=79924129
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111249073.0A Pending CN114004023A (en) | 2021-10-26 | 2021-10-26 | Aircraft pneumatic parameter identification method based on recurrent neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114004023A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114444551A (en) * | 2022-04-02 | 2022-05-06 | 西南交通大学 | Aerodynamic load identification method based on wavelet transform and convolution self-encoder |
CN115688288A (en) * | 2023-01-05 | 2023-02-03 | 西北工业大学 | Aircraft pneumatic parameter identification method and device, computer equipment and storage medium |
CN116382071A (en) * | 2023-02-08 | 2023-07-04 | 大连理工大学 | Pneumatic parameter intelligent identification method for deep learning network correction compensation |
CN117521561A (en) * | 2024-01-03 | 2024-02-06 | 中国人民解放军国防科技大学 | Aerodynamic force and thrust online prediction method of cruise aircraft |
CN117589190A (en) * | 2024-01-18 | 2024-02-23 | 西北工业大学 | Atmospheric parameter resolving method based on inertial navigation/flight control |
-
2021
- 2021-10-26 CN CN202111249073.0A patent/CN114004023A/en active Pending
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114444551A (en) * | 2022-04-02 | 2022-05-06 | 西南交通大学 | Aerodynamic load identification method based on wavelet transform and convolution self-encoder |
CN114444551B (en) * | 2022-04-02 | 2022-06-10 | 西南交通大学 | Aerodynamic load identification method based on wavelet transform and convolution self-encoder |
CN115688288A (en) * | 2023-01-05 | 2023-02-03 | 西北工业大学 | Aircraft pneumatic parameter identification method and device, computer equipment and storage medium |
CN116382071A (en) * | 2023-02-08 | 2023-07-04 | 大连理工大学 | Pneumatic parameter intelligent identification method for deep learning network correction compensation |
CN116382071B (en) * | 2023-02-08 | 2023-12-22 | 大连理工大学 | Pneumatic parameter intelligent identification method for deep learning network correction compensation |
CN117521561A (en) * | 2024-01-03 | 2024-02-06 | 中国人民解放军国防科技大学 | Aerodynamic force and thrust online prediction method of cruise aircraft |
CN117521561B (en) * | 2024-01-03 | 2024-03-19 | 中国人民解放军国防科技大学 | Aerodynamic force and thrust online prediction method of cruise aircraft |
CN117589190A (en) * | 2024-01-18 | 2024-02-23 | 西北工业大学 | Atmospheric parameter resolving method based on inertial navigation/flight control |
CN117589190B (en) * | 2024-01-18 | 2024-03-26 | 西北工业大学 | Atmospheric parameter resolving method based on inertial navigation/flight control |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114004023A (en) | Aircraft pneumatic parameter identification method based on recurrent neural network | |
CN108196532B (en) | Fault detection and separation method for longitudinal flight control system of unmanned aerial vehicle based on nonlinear adaptive observer | |
CN109635494B (en) | Flight test and ground simulation aerodynamic force data comprehensive modeling method | |
CN110554606B (en) | Self-adaptive fault-tolerant control method for hypersonic aircraft | |
CN110471313B (en) | Flight simulation subsystem of simulation aircraft | |
CN111695193B (en) | Modeling method and system of globally relevant three-dimensional aerodynamic mathematical model | |
CN102607639A (en) | BP (Back Propagation) neural network-based method for measuring air data in flight state with high angle of attack | |
CN108100301B (en) | Test flight data processing method for objective test of helicopter simulator | |
CN110633790B (en) | Method and system for measuring residual oil quantity of airplane oil tank based on convolutional neural network | |
CN112859898B (en) | Aircraft trajectory prediction method based on two-channel bidirectional neural network | |
CN107367941B (en) | Method for observing attack angle of hypersonic aircraft | |
CN113609600A (en) | Multi-body separation compatibility measurement and characterization method suitable for aircraft | |
CN114329766A (en) | Flight dynamics model reliability evaluation method for deep reinforcement learning | |
CN113221237B (en) | Large attack angle flutter analysis method based on reduced order modeling | |
CN114492176A (en) | Dynamic model parameter identification method and system based on generation countermeasure network | |
CN113049215B (en) | Quantitative assessment and test system for airflow interference resistance of rotor unmanned aerial vehicle position | |
CN112836581B (en) | Sensitive fault feature extraction method and device based on correlation analysis | |
CN211685678U (en) | Simulation analysis system of real-time trail of multi-rotor unmanned aerial vehicle | |
CN108333945A (en) | The distributed fully excitation input signal design method of airplane flutter experiment | |
Battipede et al. | Neural networks for air data estimation: test of neural network simulating real flight instruments | |
Saderla et al. | Parameter estimation of UAV from flight data using neural network | |
CN114722695A (en) | FADS resolving system and method based on dimensionless input and output neural network | |
Martin et al. | Design and Evaluation of a Realtime, Microcontroller Based Gust Sensing System for a Small Unmanned Aerial Vehicle | |
Sun et al. | A gnss/imu-based 5-hole pitot tube calibration algorithm | |
Tai et al. | Test Data Processing of Fly-by-Wire Civil Aircraft in Low-Speed Wind Tunnel Virtual Flight |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |