CN115470726A - Hypersonic inlet channel flow field rapid prediction method based on deep learning - Google Patents
Hypersonic inlet channel flow field rapid prediction method based on deep learning Download PDFInfo
- Publication number
- CN115470726A CN115470726A CN202211119244.2A CN202211119244A CN115470726A CN 115470726 A CN115470726 A CN 115470726A CN 202211119244 A CN202211119244 A CN 202211119244A CN 115470726 A CN115470726 A CN 115470726A
- Authority
- CN
- China
- Prior art keywords
- flow field
- inlet channel
- air inlet
- hypersonic
- deep learning
- 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
- G06F30/28—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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- 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
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- 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)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Algebra (AREA)
- Computing Systems (AREA)
- Fluid Mechanics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
Abstract
The invention provides a hypersonic inlet channel flow field rapid prediction method based on deep learning, which is used for establishing a hypersonic inlet channel flow field rapid prediction model based on the deep learning, directly utilizing wall surface pressure sensor data to predict a flow field and being applied to rapid/real-time prediction of a hypersonic inlet channel flow field and accurate judgment of an inlet channel state. Compared with the traditional method for acquiring the flow field based on the CFD, the method can be used for quickly acquiring the flow field of the hypersonic inlet channel with high precision and high accuracy. In addition, the method starts from the actual flight state of the hypersonic aircraft, and takes the real-time wall surface pressure data obtained by the pressure sensor on the wall surface of the air inlet passage as input, so that the real-time judgment of the flow state of the hypersonic air inlet passage and the real-time monitoring of the flow field of the air inlet passage can be realized, and the safe and efficient operation of the aircraft is further ensured.
Description
Technical Field
The invention relates to the field of computational fluid mechanics and the field of artificial intelligence, in particular to a hypersonic inlet channel flow field rapid prediction method based on deep learning.
Background
The scramjet engine is a main power device of an air-breathing hypersonic aircraft and provides power support for stable flight of the aircraft under high Mach number, wherein a hypersonic air inlet channel is an important pneumatic component of the scramjet engine, and whether the hypersonic air inlet channel can be smoothly started determines whether the scramjet engine can efficiently, safely and stably work. However, the working environment of the air inlet duct is very complex, and the change of working conditions may cause the air inlet duct not to start, so that the air inlet duct cannot work normally during combustion, and even the structure of the engine and the air inlet duct is damaged. Therefore, it is important to quickly predict the flow field of the intake duct and accurately judge the state of the intake duct.
In the field of fluid mechanics, computational Fluid Dynamics (CFD) dominates the conventional means of acquiring flow fields. However, the flow field of the hypersonic inlet is complex, a large number of wave system structures exist, the flow field needs to be rapidly and accurately acquired in the aspect of starting/non-starting early warning of the hypersonic inlet, but the traditional CFD method is limited by the algorithm and the grid quantity, and therefore the calculation efficiency and the flow field solving precision cannot be considered at the same time. In the process of repeatedly performing CFD calculation, the characteristics of the flow field are ignored, a large amount of repeated calculation is performed, and the calculation efficiency is reduced to a certain extent.
In recent years, the appearance and development of deep learning provide a new idea and method for predicting the flow field of the air inlet channel. The deep learning has strong learning ability on high-order complex functions, has unique advantages in the aspect of feature extraction, and can perform rapid and accurate prediction work. At present, the research of flow field prediction based on a deep learning method is mainly based on the speed field or the combination of the speed field and geometric parameters of an air inlet channel to establish input parameters for flow field prediction, and is difficult to apply to the actual configuration
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a hypersonic inlet flow field rapid prediction method based on deep learning, which is a new concept that the method is feasible and has wide application prospect, wherein a hypersonic inlet flow field rapid prediction model is established based on the deep learning, the flow field prediction is carried out by directly utilizing the data of a wall surface pressure sensor, and the method is applied to the rapid/real-time prediction of the hypersonic inlet flow field and the accurate judgment of the inlet state. Compared with the traditional method for acquiring the flow field based on the CFD, the method can be used for quickly acquiring the flow field of the hypersonic inlet channel with high precision and high accuracy. In addition, the method starts from the actual flight state of the hypersonic aircraft, and takes the real-time wall surface pressure data obtained by the pressure sensor on the wall surface of the air inlet passage as input, so that the real-time judgment of the flow state of the hypersonic air inlet passage and the real-time monitoring of the flow field of the air inlet passage can be realized, and the safe and efficient operation of the aircraft is further ensured.
The technical scheme of the invention is as follows:
a method for quickly predicting a flow field of a hypersonic inlet based on deep learning comprises the following steps:
step 1: generating a sample flow field data set of an air inlet channel;
and 2, step: constructing a deep neural network model for quickly predicting the flow field of the hypersonic inlet, and training the neural network model by using sample flow field data;
and 3, step 3: and the trained model is used for quickly predicting the flow field of the air inlet channel.
Further, the specific steps of generating the air inlet sample flow field data set in step 1 are as follows:
step 1.1: sampling in the working range of the air inlet channel to obtain the calculation condition of each sample point in a flow field sample set;
step 1.2: generating a CFD computational grid of the air inlet channel model;
step 1.3: carrying out CFD numerical calculation on the sample to obtain a CFD flow field calculation result corresponding to the sample;
step 1.4: and intercepting a space grid from the front of the inner flow and inner contraction section to the first-stage compression surface, and mapping the intercepted grid and flow field variables from physical space coordinates to uniform calculation space coordinates through coordinate transformation to obtain a basic flow field data set.
The research finds that the performance of the air inlet and the flow structure of the concerned air inlet are concentrated in the area of the position of the inner flow and the inner contraction section forward to the first-stage compression surface, and in order to reduce the training cost of the model and improve the efficiency, the space grid of the area is intercepted, and the flow field of the area is predicted in practical application.
Further, in step 1.1, a Latin Hypercube Sampling (LHS) method is adopted to perform sampling in the working range of the air inlet channel.
Further, in step 1.2, aiming at the air inlet channel model, grid generation is carried out by adopting a plurality of structured grids, encryption is carried out on an object plane and an inner runner area, and the height of a first layer of a boundary layer in the generated calculation grid is set to be 4.3 multiplied by 10 -7 m, mesh Reynolds number Re under cruising condition Δ ＝5。
Further, in step 1.3, solving the RANS equation by a finite volume method to perform CFD numerical simulation, wherein a turbulence model adopted in the numerical simulation process is an SST turbulence model, a flux format adopts a roe format, and a flux limiter adopts a minmod flux limiter.
Further, in step 2, the process of building and training the deep neural network model is as follows:
step 2.1: the deep neural network model is built by adopting a multilayer perceptron (MLP) network of a full connection layer, the pressure of the wall surface of a compression surface of an air inlet channel in a sample is selected as input, and the output of the network is a flow parameter of each grid coordinate;
step 2.2: and (3) training the deep neural network by using the sample set obtained in the step (1), constructing a loss function by using the root mean square error of the flow field parameters predicted in the flow field in the training process by referring to the definition of the root Mean Square Error (MSE), and completing the training by matching an optimization algorithm until the loss function of the training set is not reduced any more.
Advantageous effects
Compared with the prior art, the invention has the following advantages:
1. according to the method, from the actual flight state of the hypersonic aircraft, in practical application, compared with the situation that the input parameters are established by combining a speed field and geometric parameters of an air inlet channel in the prediction of a passing hypersonic flow field, the wall surface pressure data can be obtained in real time through a pressure sensor on the wall surface of the air inlet channel. In addition, various shock wave reflection and shock wave interference phenomena exist in the internal flow field of the hypersonic air inlet channel, and the flowing phenomena determine the wall surface pressure distribution form of the air inlet channel. Therefore, the wall surface pressure data is highly related to the flow structure of the air inlet, the wall surface pressure data is used as the input parameter of the flow field rapid prediction model to establish the hypersonic air inlet flow field rapid prediction model, and the high-precision and high-accuracy reconstructed flow field can be obtained. And the model is further applied to the reconstruction of the flow field of the air inlet channel of the aircraft, so that the real-time judgment of the flow state of the air inlet channel and the real-time monitoring of the flow field of the air inlet channel can be realized.
3. The method uses the LHS sampling method to establish the sample set, can ensure that the sampling points cover the whole sample space within the limit of limited sample quantity, ensures the uniformity of sample point distribution, can ensure that the model obtained by training is more suitable for the flow field prediction problem within a specific range, can further improve the prediction precision, and ensures the generalization capability of the model to a certain extent.
4. The invention intercepts the flow field of the area in front of the inner flow and inner contraction section of the air inlet channel for prediction, and can ensure the sufficient flow field prediction and feature extraction precision while reducing the data volume and improving the training efficiency. In the flow field prediction process based on limited wall surface pressure data, the output quantity is less, and higher prediction precision is obtained more favorably. In addition, if the prediction of the whole flow field is performed, the error of the flow field of the part is considered by the trained network in consideration of the prediction accuracy of the non-concerned area, and the flow field prediction accuracy of the concerned area (the inner flow part) is further influenced.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
FIG. 1 is a sample set calculation operating condition distribution range.
Fig. 2 is a schematic diagram of meshing of a hypersonic inlet, in which a selected region for flow field prediction is inside a red wire frame.
Fig. 3 is a coordinate transformation illustration. The left side is the mesh before transformation and the right side is the mesh after transformation.
FIG. 4 is a diagram of a deep learning model.
FIG. 5 shows the prediction result of the inlet inactive flow field. (a) a pressure calculated for the CFD, (b) a pressure predicted for the deep learning model, (c) a mach number calculated for the CFD, (d) a mach number predicted for the deep learning model, (e) an X-direction velocity component calculated for the CFD, and (f) an X-direction velocity component predicted for the deep learning model.
FIG. 6 shows the port startup flow field prediction. (a) a pressure calculated for the CFD, (b) a pressure predicted for the deep learning model, (c) a mach number calculated for the CFD, (d) a mach number predicted for the deep learning model, (e) an X-direction velocity component calculated for the CFD, and (f) an X-direction velocity component predicted for the deep learning model.
Fig. 7 shows the prediction result of the non-starting flow field of the air inlet, wherein the abscissa is the flow parameter and the ordinate is the normal station (meter). The prediction result of the section pressure at the outlet, (b) the prediction result of the Mach number of the section at the outlet, (c) the prediction result of the section pressure at the throat, and (d) the prediction result of the section Mach number at the throat.
Fig. 8 shows the prediction result of the starting flow field of the air inlet, wherein the abscissa is the flow parameter and the ordinate is the normal station (meter). The prediction result of the section pressure at the outlet, (b) the prediction result of the Mach number of the section at the outlet, (c) the prediction result of the section pressure at the throat, and (d) the prediction result of the section Mach number at the throat.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings and specific embodiments.
The method for quickly predicting the flow field of the hypersonic inlet based on the deep neural network comprises the steps of generating a flow field data set required by the training of the hypersonic neural network, building a deep neural network model, training the model by using the data set, and finally realizing the quick prediction of the flow field of the hypersonic inlet by using the trained deep neural network model. The method comprises the following specific steps:
step 1: generating a flow field sample set required by deep neural network training:
1) In this embodiment, a binary hypersonic inlet channel is taken as an example, and a Latin Hypercube Sampling (LHS) method is adopted to perform sampling in a flight working range (a flight height interval is [19,22] ㎞, and a flight mach number interval is [2.5,5 ]) of the binary hypersonic inlet channel, so as to obtain a calculation condition of each sample point in a flow field sample set, as shown in fig. 1. It should be noted that in this sample space, the intake duct may exhibit two states of starting and non-starting under different conditions, which also better corresponds to the actual operating state of the intake duct. And dividing the sample points according to the proportion of the training set to the verification set of 5:1 for subsequent deep neural network training.
2) Aiming at the air inlet channel model, a plurality of structured grids are adopted for grid generation, and in order to more accurately capture the details of a flow field, encryption is carried out in areas with severe flow parameter changes, such as an object plane, an inner flow channel and the like. The height of the first layer of the boundary layer in the generated computational grid is set to be 4.3 multiplied by 10 -7 m, mesh Reynolds number Re under cruising condition Δ And (5). The generated CFD computational mesh is shown in fig. 2.
3) And solving the RANS equation by a finite volume method to carry out CFD numerical simulation to obtain an initial flow field data set. The turbulence model adopted in the numerical simulation process is an SST turbulence model, the flux format adopts roe format, and the flux limiter adopts minmod fluxA quantity limiter. The calculation conditions in this embodiment were Ma =4.03, re =11.7 × 10 6 ，T ∞ ＝216.6K，P ∞ =8728.8Pa, α =0 °, where Ma is the free incoming flow mach number, re free incoming flow reynolds number, T ∞ Is the free incoming flow temperature, P ∞ Alpha inlet angle of attack for free incoming flow pressure.
In order to improve the training efficiency and the flow field prediction efficiency of the deep neural network model, a space grid from the front of the inner flow and inner contraction section to the first-stage compression surface is intercepted and used for training and testing the deep neural network model, the size of the finally selected grid point is 41 multiplied by 153 (normal direction multiplied by flow direction), and the area is circled by a red frame in fig. 2. The area comprises an area in an incoming flow capturing area of the air inlet channel and all internal flow fields of the air inlet channel, and the area is enough for obtaining relevant performance parameters of the air inlet channel and judging the working state of the air inlet channel.
4) And mapping the selected grid coordinates from the physical space coordinates to uniform calculation space coordinates through coordinate transformation so as to facilitate subsequent deep neural network training, wherein the coordinate transformation method is shown in fig. 3.
Step 2: building and training a deep neural network model for flow field prediction:
1) Utilizing a fully-connected multilayer perceptron (MLP) to build a deep neural network, and inputting pressure data (P) at the pressure measuring point of the flow field sample concentration wall surface obtained in the step 1 1 ，P 2 ,...,P 22 ) In which P is n For the pressure data at the nth pressure measurement point, the output is the flow parameters (u, v, p, ρ, ma) at the grid point, where u and v are the velocity components in the x and y directions, respectively, p is the pressure, ρ is the density, and Ma mach number. The constructed deep neural network model is shown in fig. 4.
2) And (3) training the deep neural network by using the flow field sample set obtained in the step (1). In the training process, a loss function is constructed by using a concept of reference root mean square error, and training is carried out by matching with an ADAM random gradient optimization algorithm by taking the minimum loss function as an optimization target until the training is finished when the loss function of the training set is not reduced any more. The constructed loss function is as follows:
in the above formula, N is the number of samples, and in each flow parameter, the variable labeled c is the CFD calculation result, and the variable labeled p is the prediction result.
And step 3: fast prediction of a hypersonic inlet channel flow field:
and aiming at the trained deep neural network model, testing by using a training set and a testing set and quickly predicting the flow field of the hypersonic inlet. The air inlet channels in the training set sample 122 and the training set sample 235 are respectively in two representative states of non-starting and starting, the flow field prediction results are respectively shown in fig. 5 and 6, and the normal distribution of parameters of a part of cross sections in the flow field is shown in fig. 7 and 8. And randomly selecting samples from the results and calculating the performance parameters of the air inlet, wherein the total pressure recovery coefficient result is shown in table 1, and the average Mach number result of the outlet is shown in table 2. In addition, the computational efficiency of the deep neural network model and the CFD method were compared, and the comparison results are shown in table 3.
TABLE 1
TABLE 2
TABLE 3
Although embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are exemplary and not to be construed as limiting the present invention, and that those skilled in the art may make variations, modifications, substitutions and alterations within the scope of the present invention without departing from the spirit and scope of the present invention.
Claims (6)
1. A hypersonic inlet channel flow field rapid prediction method based on deep learning is characterized by comprising the following steps: the method comprises the following steps:
step 1: generating a gas inlet channel sample flow field data set;
step 2: constructing a deep neural network model for quickly predicting the flow field of the hypersonic inlet, and training the neural network model by using sample flow field data; the input of the deep neural network model is the wall pressure of a compression surface of an air inlet channel, and the output of the network is the flow parameter of each grid coordinate;
and step 3: and the trained model is used for quickly predicting the flow field of the air inlet channel.
2. The method for quickly predicting the flow field of the hypersonic inlet based on the deep learning as claimed in claim 1, is characterized in that: the specific steps of generating the air inlet channel sample flow field data set in the step 1 are as follows:
step 1.1: sampling in the working range of the air inlet channel to obtain the calculation condition of each sample point in the flow field sample set;
step 1.2: generating a CFD computational grid of the air inlet channel model;
step 1.3: performing CFD numerical calculation on the sample to obtain a CFD flow field calculation result corresponding to the sample;
step 1.4: and intercepting the space grid from the front of the inner flow and inner contraction section to the first-stage compression surface, and mapping the intercepted grid and flow field variables from physical space coordinates to uniform calculation space coordinates through coordinate transformation to obtain a basic flow field data set.
3. The method for quickly predicting the flow field of the hypersonic inlet based on the deep learning as claimed in claim 2, wherein: in step 1.1, sampling is performed within the working range of the air inlet by adopting a Latin Hypercube Sampling (LHS) method.
4. The method for quickly predicting the flow field of the hypersonic inlet based on the deep learning as claimed in claim 2, wherein: in step 1.2, aiming at an air inlet channel model, grid generation is carried out by adopting a plurality of structured grids, the object plane and the inner runner area are encrypted, and the height of a first layer of a boundary layer in the generated computational grid is set to be 4.3 multiplied by 10 -7 m, mesh Reynolds number Re under cruising condition Δ ＝5。
5. The method for quickly predicting the flow field of the hypersonic inlet based on the deep learning as claimed in claim 2, wherein: in step 1.3, solving the RANS equation by a finite volume method to carry out CFD numerical simulation, wherein a turbulence model adopted in the numerical simulation process is an SST turbulence model, a flux format adopts a roe format, and a flux limiter uses a minmod flux limiter.
6. The method for quickly predicting the flow field of the hypersonic inlet based on the deep learning of claim 1 is characterized by comprising the following steps: in step 2, the process of building and training the deep neural network model is as follows:
step 2.1: the deep neural network model is built by adopting a multi-layer sensor network of a full connection layer, the wall pressure of a compression surface of an air inlet channel in a sample is selected as input, and the output of the network is a flow parameter of each grid coordinate;
step 2.2: and (3) training the deep neural network by using the sample set obtained in the step (1), constructing a loss function by using the root mean square error of the flow field parameters predicted in the flow field in the training process, and completing the training by matching an optimization algorithm until the loss function of the training set is not reduced any more.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211119244.2A CN115470726A (en) | 2022-09-13 | 2022-09-13 | Hypersonic inlet channel flow field rapid prediction method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211119244.2A CN115470726A (en) | 2022-09-13 | 2022-09-13 | Hypersonic inlet channel flow field rapid prediction method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115470726A true CN115470726A (en) | 2022-12-13 |
Family
ID=84332745
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211119244.2A Pending CN115470726A (en) | 2022-09-13 | 2022-09-13 | Hypersonic inlet channel flow field rapid prediction method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115470726A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117034815A (en) * | 2023-10-08 | 2023-11-10 | 中国空气动力研究与发展中心计算空气动力研究所 | Slice-based supersonic non-viscous flow intelligent initial field setting method |
-
2022
- 2022-09-13 CN CN202211119244.2A patent/CN115470726A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117034815A (en) * | 2023-10-08 | 2023-11-10 | 中国空气动力研究与发展中心计算空气动力研究所 | Slice-based supersonic non-viscous flow intelligent initial field setting method |
| CN117034815B (en) * | 2023-10-08 | 2024-01-23 | 中国空气动力研究与发展中心计算空气动力研究所 | Slice-based supersonic non-viscous flow intelligent initial field setting method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kong et al. | A deep learning approach for the velocity field prediction in a scramjet isolator | |
| CN108304684B (en) | Rocket engine tail jet flow simulation method and system | |
| Afshari et al. | On numerical methods; optimization of CFD solution to evaluate fluid flow around a sample object at low Re numbers | |
| Zhou et al. | Fault diagnosis of gas turbines with thermodynamic analysis restraining the interference of boundary conditions based on STN | |
| CN111859746B (en) | Method for predicting variable working condition performance of turbomachinery based on flow field reconstruction | |
| Semlitsch et al. | Flow effects due to valve and piston motion in an internal combustion engine exhaust port | |
| CN114254460B (en) | Turbomachine pneumatic robustness optimization method based on graph convolution neural network | |
| CN112818573A (en) | Method for acquiring boundary layer non-local variable information for unstructured grid | |
| CN115470726A (en) | Hypersonic inlet channel flow field rapid prediction method based on deep learning | |
| Mittal et al. | LES of gas exchange in IC engines | |
| Shuang et al. | An adaptive compressor characteristic map method based on the Bezier curve | |
| Cummings et al. | Supersonic, turbulent flow computation and drag optimization for axisymmetric afterbodies | |
| CN114462336B (en) | Method for calculating average temperature of coolant of main pipeline of nuclear reactor | |
| KR102484587B1 (en) | Method to Build a Reduced Order Model based on Eigenmodes from Simulation and Eigenvalues based on Machine Learning by Measurement Data | |
| Sureshbabu et al. | Deep-Learning Methods for Non-Linear Transonic Flow-Field Prediction | |
| CN115983137A (en) | Turbine flow field prediction method based on similarity principle and deep learning and related device | |
| CN113536640B (en) | Optimization design method for internal flow channel structure of air distributor based on orthogonal test | |
| Yang et al. | Surrogate Model-Based Acoustic Optimisation of Jet Nozzle Exit Geometry | |
| Tong et al. | Optimization of aero-engine combustion chambers with the assistance of Hierarchical-Kriging surrogate model based on POD downscaling method | |
| Wang et al. | High-fidelity surface flow data-driven aerodynamic solution strategy for non-smooth configurations: Study of compressor cascade with micro riblet surface | |
| Li et al. | Research on time sequence prediction of supersonic cascade flow field based on compressed sensing artificial neural network | |
| Vedam et al. | Evaluation of Gradient and Curvature-Based Adaptive Mesh Refinement for Viscous Transonic Flows | |
| CN116644673B (en) | State data time sequence change prediction method of aircraft thermal management system | |
| CN116502484B (en) | Deformation calculation method and device for asymmetric section force transmission structure | |
| Baturin et al. | Identification of the Mathematical Model of a Gas Turbine Engine Taking Into Account the Uncertainty of the Initial Data |
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 |