CN116227316A - Wave field modeling method for aerial fueling head - Google Patents
Wave field modeling method for aerial fueling head Download PDFInfo
- Publication number
- CN116227316A CN116227316A CN202211096167.3A CN202211096167A CN116227316A CN 116227316 A CN116227316 A CN 116227316A CN 202211096167 A CN202211096167 A CN 202211096167A CN 116227316 A CN116227316 A CN 116227316A
- Authority
- CN
- China
- Prior art keywords
- space
- flow field
- constructing
- oil receiving
- head wave
- 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]
- G06F30/20—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
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- 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)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
Abstract
The invention discloses an air refueling head wave field modeling method, which relates to the technical field of data processing and modeling, and has the technical scheme that: s1: constructing the pneumatic shapes of the oiling machine and the oil receiving machine; s2: respectively constructing space grids of the oiling machine and the oil receiving machine; s3: constructing a calculated overlapped grid; s4: constructing a cube region model, respectively meshing the space along three directions, constructing a space grid region of a head wave region, interpolating space grid nodes to be constructed and calculated space flow field parameters through a linear interpolation or proxy model, and outputting the flow field parameters; s5: sample data normalization processing S6: constructing a deep learning neural network learning model; s7: and taking the position coordinates as input parameters to carry out prediction by using the neural network model, thus finishing the modeling of the flow field parameters of different space points of the head wave interference area. The method can rapidly predict aerodynamic parameters of any position of the head wave flow field space on the basis of a certain sample numerical calculation result.
Description
Technical Field
The invention relates to the technical field of data processing and modeling, in particular to an air refueling head wave field modeling method.
Background
With the increasing demand for long voyages and long voyages of aviation aircrafts, if one ground refueling is adopted, the mode of returning the ground after performing the task to refuel is difficult to meet the flying demand of future aircrafts. The air combined refueling mode is the most direct and efficient mode for improving the range and the dead time of the aircraft, so that the air refueling technology has become a 'multiplier' for improving the flight capacity of the aircraft, can increase the mission load of the aircraft and enhance the long-term flight capacity of the remote aircraft.
The existing air refueling technology mainly comprises soft refueling, mainly relates to a refueling platform, an oil receiving machine, a hose-taper sleeve and other systems, and is the most important link in the refueling process, so that the success or failure of the refueling process is directly determined. Because the oil receiver is positioned in the washing field of the oiling machine, the oil receiver is positioned in a complex nonuniform flow field, and various complex dynamics and control problems exist, such as double-machine pneumatic interference, atmospheric wind field disturbance, wake vortex of the oiling machine, butt joint cone sleeve wave interference and the like. In the air refueling process, when the oil receiving machine approaches the taper sleeve, the taper sleeve deviates from the balance position, a phenomenon of rapid backswing occurs, the phenomenon is commonly called as a 'head wave effect', head wave interference can directly influence the accuracy of the docking process, a proper head wave aerodynamic model is established to have an important parameter input function on a docking control system of the oil receiving machine, the head wave aerodynamic model can be obtained by adopting means of theoretical analysis, wind tunnel test, flight test, numerical simulation and the like, wherein the theoretical analysis is generally low in accuracy based on a simplified model, the wind tunnel test and the flight test are high in cost on one hand, long in period and difficult to meet the dynamics similarity, and flow field information of any space position cannot be obtained. The existing Rankine body head wave modeling method is generally suitable for aircraft with specific configuration, is difficult to adapt to more aircraft, and has the advantages of low cost, fast period, complete data and the like based on a computational fluid dynamics method.
Disclosure of Invention
The invention aims to provide an air refueling head wave field modeling method which can be used for rapidly predicting aerodynamic parameters of any position of a head wave flow field space on the basis of a certain sample numerical calculation result.
The technical aim of the invention is realized by the following technical scheme: the modeling method of the aerial filler head wave field specifically comprises the following steps:
s1: constructing pneumatic shapes of the oiling machine and the oil receiving machine, determining the space position of the oil receiving machine at the tail part of a flow field of the oiling machine according to the relative position of a hose taper sleeve, and simultaneously, setting the speed of the oiling machine and the oil receiving machine in the air oiling butt joint process;
s2: respectively constructing space grids of the oiling machine and the oil receiving machine, wherein the oil receiving machine adopts an overlapped grid method for processing;
s3: constructing calculated overlapped grids, placing the oil receiving machine grids at estimated positions for simulating aerial refueling by a moving and rotating method, wherein relative position vectors are (delta X, delta Y and delta Z), and carrying out space flow field solution by adopting a computational fluid dynamics method until a calculation result is converged;
s4: constructing a region to be predicted range of the head wave region according to the flow field change of the head wave interference region, respectively determining the region lengths l, m and n along the three directions of a stagnation point X, Y, Z, constructing a cube region model, respectively meshing the space along the three directions, constructing a space grid region of the head wave region, interpolating the space grid node to be constructed and the calculated space flow field parameters through linear interpolation or proxy model, inputting the space grid node coordinates (xi, yi and zi), and outputting the flow field parameters of the interpolation sample points into a text format;
s5: normalizing the sample data, and dividing the sample data into two types, namely a training set and a testing set;
s6: constructing a deep learning neural network learning model, training the neural network model through a training set, and checking the accuracy of the neural network model through a testing set;
s7: and selecting space position coordinates to be predicted according to the space flow field prediction requirement or the space flow field parameter required by flight control, and taking the position coordinates as input parameters to carry out prediction by using a neural network model, namely completing flow field parameter modeling of different space points of a head wave interference region.
Furthermore, the oil receiving machine in the step S2 adopts an elliptic space flow field, and the size of the flow field is 1.5-2 times of the size of the oil receiving machine in three dimensions.
Further, the neural network learning model in S6 has at least two hidden layers.
In summary, the invention has the following beneficial effects:
1. the modeling method realizes the rapid prediction design of the flow field of the wave interference of the aerial fueling head, so that the accurate prediction of the spatial flow field is realized on the basis of not establishing the parameters of the full flow field, reliable spatial flow field parameter input can be provided for the design of a flight control system, and the accuracy of control law design is improved;
2. the method meets the head wave flow field prediction considering the pneumatic interference of the oil adding/receiving machine in the air refueling process, can realize the fast prediction based on deep learning, and has higher efficiency and engineering application value.
Drawings
FIG. 1 is a flow chart of a head wave interference model deep learning in an embodiment of the invention;
FIG. 2 is a schematic diagram of a neural network training model in an embodiment of the present invention;
FIG. 3 is a graph of neural network residual convergence in an embodiment of the invention;
FIG. 4 is a graph showing a comparison of v-velocity direction predictions for a neural network test sample in an embodiment of the present invention;
FIG. 5 is a graph showing the w-speed direction prediction contrast of a neural network test sample according to an embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to fig. 1-5.
Examples: an aerial filler head wave field modeling method, as shown in fig. 1 to 5, specifically comprises the following steps:
s1: constructing pneumatic shapes of the oiling machine and the oil receiving machine, determining the space position of the oil receiving machine at the tail part of a flow field of the oiling machine according to the relative position of a hose taper sleeve, and simultaneously, setting the speed of the oiling machine and the oil receiving machine in the air oiling butt joint process;
in this embodiment, since the fueling machine and the fueling receiver are relatively stationary during airborne fueling, the flat flight angle of attack of the fueling/fueling receiver can be determined by the speed of flight, the altitude of flight, and the weight of the fueling/fueling receiver itself, with the speed of flight taking a mach number of 0.3.
S2: respectively constructing space grids of the oiling machine and the oil receiving machine, wherein the oil receiving machine adopts an overlapped grid method for processing;
in the embodiment, in order to better simulate the head wave effect, grid encryption processing is carried out on the space flow field of the oiling machine at the predicted oil receiving machine position, so that the boundary grid scale of the oil receiving machine is ensured to be equivalent to the grid scale of the encrypting area of the oiling machine, and the solving precision of the overlapped grids is improved.
S3: constructing calculated overlapped grids, placing the oil receiving machine grids at estimated positions for simulating aerial refueling by a moving and rotating method, wherein relative position vectors are (delta X, delta Y and delta Z), and carrying out space flow field solution by adopting a Computational Fluid Dynamics (CFD) method until a calculation result is converged;
s4: constructing a head wave region to-be-predicted range according to flow field changes of the head wave interference region, respectively determining region lengths l, m and n along three directions of a stagnation point X, Y, Z, constructing a cube region model, respectively meshing a space along the three directions, constructing a space grid region of the head wave region, interpolating space grid nodes to be constructed and calculated space flow field parameters through linear interpolation or a proxy model, and inputting the space gridNode coordinates (x) i ,y i ,z i ) Meanwhile, outputting flow field parameters of the interpolation sample points into a text format; wherein the parameters output are flow field parameters at different nodes, including u i ,v i ,w i ,cp i ,T i Equal parameters, where u i Representing the velocity in the x-direction, v i Representing the velocity in the y-direction, w i Representing the velocity in the z-direction, cp i Expressed as pressure coefficient, T i Temperature is shown.
S5: normalizing sample data, including coordinate position and flow field parameters, wherein the coordinate position is processed by taking a stagnation point as an initial origin, the speed flow field parameters are normalized by taking maximum speeds in three directions as references, and normalization of other parameters is processed in a reasonable mode according to calculation data; after processing, the sample data is required to be divided into two types, namely a training set and a testing set;
s6: constructing a deep learning neural network learning model, training the neural network model through a training set, and checking the accuracy of the neural network model through a testing set;
s7: and selecting spatial position coordinates to be predicted according to flow field prediction requirements or spatial flow field parameters required by flight control, taking the position coordinates as input parameters to carry out prediction by using a neural network model, namely completing flow field parameter modeling of different spatial points of a head wave interference region, and thus establishing a mapping relation between the input parameters and output parameters.
In the embodiment, the oil receiving machine in the S2 preferably adopts an elliptic space flow field, and the size of the flow field is 1.5 times of the size of the oil receiving machine in three dimensions.
And S6, the neural network learning model is provided with at least two hidden layers, the number of units of the hidden layers is determined according to the complexity degree or the test progress, the selection of the training method can be adjusted according to multiple training, the training frequency depends on the convergence degree of residual errors, and the quality of the training result is judged according to the accuracy and the error of the output test model.
The following is a specific modeling method:
as shown in fig. 1, firstly, determining an oiling machine and an oil receiving machine according to an aerial oiling object, determining the approximate movement range of the oil receiving machine according to the length and the size of a soft rope-taper sleeve, completing space grid of the oiling machine by adopting grid division software, and encrypting the space grid according to the approximate range of the oil receiving machine in a washing field, thereby completing grid division of the oiling machine; and determining a flight attack angle in the oiling process according to the flight characteristics of the oiling machine, finishing the rotation of a space grid or an aircraft, ensuring that the incoming flow direction is along the airflow in the numerical simulation process, and not setting an attack angle parameter in the calculation process, wherein the flight speed is Mach number 0.3.
As shown in fig. 1, the aircraft nose of the oil receiver is a main area for generating head wave interference in the air refueling process, because the embodiment adopts an overlapped grid method to perform flow field analysis in numerical simulation, the space grid of the oil receiver is divided independently, in order to ensure the accuracy of head wave interference simulation, the oil receiver adopts an ellipsoidal space grid topological structure, the space size is selected to be about 1.5 times of the three-axis length of the aircraft, and in order to improve the overlapped grid accuracy, the ellipsoidal boundary grid size is basically equivalent to the grid size of the background grid encryption area of the oil charger.
As shown in fig. 1, the spatial position of the taper sleeve relative to the aircraft is determined according to the size and dynamics characteristics of the soft rope-taper sleeve in the real flight process, the position of the taper sleeve, namely the position of the aircraft nose of the oil receiving machine, the grid of the oil receiving machine is moved to the corresponding position in a translation and rotation mode, and meanwhile, the construction of the overlapped grid is completed.
After the grid division and the space arrangement are completed, the aerodynamic disturbance numerical simulation analysis of the space flow field is entered, the incoming flow boundary conditions (including the flying speed, attack angle, sideslip angle, turbulence degree and the like) and the wall boundary conditions of the full-field space grid are given, aerodynamic distribution of the oiling machine, the oil receiving machine component and the like is separately considered in the object plane boundary setting, a turbulence model considering viscosity is selected, the flow field numerical simulation is carried out by adopting Computational Fluid Dynamics (CFD) codes, and the space flow field is output after the convergence condition is reached.
The deep learning module shown in fig. 1 performs post-processing on the calculated space flow field, divides a main area of head wave interference generated by the oil receiver through flow field analysis, constructs a space structure grid block, arranges 30,30 and 30 grid points along the x, y and z directions to construct a sampling space of the head wave area, obtains space coordinates and flow field parameters (including density, three-component speed, pressure coefficient, temperature and the like) of different sample points by adopting a linear interpolation method, sets the flow field parameters of the points to 0 for the space sample points in the oil receiver body, and outputs the flow field parameters of the sample points to a text format.
After the spatial flow field sample points are obtained, because the sample points contain non-physical space solution sets, the sample points which are not in the calculation space need to be removed through data reprocessing, so that sample data containing real flow field information which can be subjected to the next data analysis and processing are obtained.
As shown in fig. 1, after obtaining reliable sample point data, normalization processing is performed on the sample point, where the input of the sample point takes three coordinates of x, y and z, and the output is the velocity components in three directions of u, v and w, firstly, the input coordinates are processed, the initial coordinates are subtracted by taking the engine head as the origin, and for the velocity components in three directions, only the velocity component in the x direction is subtracted by the incoming flow velocity, and then normalization processing may be performed on the velocity components in three directions, and processing may be performed by using maximum velocity values in three directions respectively, or other methods may be adopted, where the purpose of this step is to improve the accuracy.
In the next step, a neural network model is built, the back propagation multi-layer neural network training method is selected, 4 layers are adopted as the hidden layers, 64 are adopted as the node books of each layer, three variables are input values, three space coordinates of a head wave region are respectively adopted as input values, three speed components are output values, the selection of the activation function of each hidden layer is the same, and the activation function of each hidden layer is a sigmoid activation function. The activation functions here may vary in actual training and each layer may choose a different activation function.
As shown in FIG. 3, the neural network training is converged through 20000 iteration steps, the training iteration steps and convergence accuracy of other actual models can be determined according to flow field parameters and flight control requirements, and the training efficiency is improved on the premise of ensuring certain accuracy.
As shown in fig. 4 and fig. 5, the comparison of the predictions of the test samples by using the neural network model obtained by training can show that the speed of the model prediction in the v direction and the w direction is very high in agreement with the numerical calculation data, and the variation trend is basically consistent, so that the feasibility and the accuracy of the method for constructing the head wave model by using the deep learning provided by the patent are verified.
The present embodiment is only for explanation of the present invention and is not to be construed as limiting the present invention, and modifications to the present embodiment, which may not creatively contribute to the present invention as required by those skilled in the art after reading the present specification, are all protected by patent laws within the scope of claims of the present invention.
Claims (3)
1. An aerial filler head wave field modeling method is characterized in that: the method specifically comprises the following steps:
s1: constructing pneumatic shapes of the oiling machine and the oil receiving machine, determining the space position of the oil receiving machine at the tail part of a flow field of the oiling machine according to the relative position of a hose taper sleeve, and simultaneously, setting the speed of the oiling machine and the oil receiving machine in the air oiling butt joint process;
s2: respectively constructing space grids of the oiling machine and the oil receiving machine, wherein the oil receiving machine adopts an overlapped grid method for processing;
s3: constructing calculated overlapped grids, placing the oil receiving machine grids at estimated positions for simulating aerial refueling by a moving and rotating method, wherein relative position vectors are (delta X, delta Y and delta Z), and carrying out space flow field solution by adopting a computational fluid dynamics method until a calculation result is converged;
s4: constructing a head wave region to-be-predicted range according to flow field changes of the head wave interference region, respectively determining region lengths l, m and n along three directions of a stagnation point X, Y, Z, constructing a cube region model, respectively meshing a space along the three directions, constructing a space grid region of the head wave region, interpolating space grid nodes to be constructed and calculated space flow field parameters through linear interpolation or a proxy model, and inputting space grid node coordinates (x i ,y i ,z i )，Meanwhile, outputting flow field parameters of the interpolation sample points into a text format;
s5: normalizing the sample data, and dividing the sample data into two types, namely a training set and a testing set;
s6: constructing a deep learning neural network learning model, training the neural network model through a training set, and checking the accuracy of the neural network model through a testing set;
s7: and selecting space position coordinates to be predicted according to the space flow field prediction requirement or the space flow field parameter required by flight control, and taking the position coordinates as input parameters to carry out prediction by using a neural network model, namely completing flow field parameter modeling of different space points of a head wave interference region.
2. An airborne fueling head wave field modeling method as defined in claim 1, wherein: the oil receiving machine in the step S2 adopts an elliptic space flow field, and the size of the flow field is 1.5-2 times of the size of the oil receiving machine in three dimensions.
3. An airborne fueling head wave field modeling method as defined in claim 1, wherein: the neural network learning model in the S6 is provided with at least two hidden layers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211096167.3A CN116227316A (en) | 2022-09-08 | 2022-09-08 | Wave field modeling method for aerial fueling head |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211096167.3A CN116227316A (en) | 2022-09-08 | 2022-09-08 | Wave field modeling method for aerial fueling head |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116227316A true CN116227316A (en) | 2023-06-06 |
Family
ID=86570276
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211096167.3A Pending CN116227316A (en) | 2022-09-08 | 2022-09-08 | Wave field modeling method for aerial fueling head |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116227316A (en) |
-
2022
- 2022-09-08 CN CN202211096167.3A patent/CN116227316A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110610065B (en) | Aircraft multi-body separation CFD simulation method and system based on hybrid dynamic grid technology | |
| Nielsen et al. | Recent improvements in aerodynamic design optimization on unstructured meshes | |
| Biedron et al. | Recent enhancements to the FUN3D flow solver for moving-mesh applications | |
| Brezillon et al. | Development and application of multi-disciplinary optimization capabilities based on high-fidelity methods | |
| Frink et al. | Computational aerodynamic modeling tools for aircraft loss of control | |
| CN112414668B (en) | Wind tunnel test data static bomb correction method, device, equipment and medium | |
| Ordaz et al. | Conceptual design of low-boom aircraft with flight trim requirement | |
| Vanharen et al. | Near-field anisotropic mesh adaptation for the third AIAA sonic boom workshop | |
| Leifsson et al. | Variable-fidelity aerodynamic shape optimization | |
| Duan et al. | High-order overset flux reconstruction method for dynamic moving grids | |
| US20220229948A1 (en) | Simulation of the behavior of a vehicle in a fluid by Hamiltonian neural network | |
| CN115310325A (en) | Multi-disciplinary coupling analysis framework and method for air refueling of hose taper sleeve | |
| CN113221237B (en) | Large attack angle flutter analysis method based on reduced order modeling | |
| Dalle et al. | Inviscid and viscous CFD analysis of booster separation for the space launch system vehicle | |
| Yaofei et al. | Cognition behavior model for air combat based on reinforcement learning | |
| Li | Feasibility of supersonic aircraft concepts for low-boom and flight trim constraints | |
| Dean et al. | Aircraft stability and control characteristics determined by system identification of CFD simulations | |
| CN116227316A (en) | Wave field modeling method for aerial fueling head | |
| CN111581790A (en) | Combined simulation method for variable-mass lift-off process of stratospheric airship based on ADAMS and MATLAB | |
| Pappalardo et al. | Modeling the Longitudinal Flight Dynamics of a Fixed-Wing Aircraft by using a Multibody System Approach. | |
| Pham et al. | Aerodynamic sensing for hypersonics via scientific machine learning | |
| Rallabhandi | Sonic boom prediction and mitigation using three-dimensional earth effects | |
| Moran et al. | Wind-Tunnel based Free-Flight Testing of a Viscous Optimised Hypersonic Waverider | |
| Duan et al. | A high order overset FR/CPR method for dynamic moving grids | |
| CN111506959A (en) | Universal flight control system and method for flight simulation training |
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 |