CN113051820B - Cross-drainage-basin pneumatic parameter simulation method based on convolutional neural network - Google Patents

Cross-drainage-basin pneumatic parameter simulation method based on convolutional neural network Download PDF

Info

Publication number
CN113051820B
CN113051820B CN202110314455.0A CN202110314455A CN113051820B CN 113051820 B CN113051820 B CN 113051820B CN 202110314455 A CN202110314455 A CN 202110314455A CN 113051820 B CN113051820 B CN 113051820B
Authority
CN
China
Prior art keywords
data
aerodynamic
pneumatic
neural network
flow
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.)
Active
Application number
CN202110314455.0A
Other languages
Chinese (zh)
Other versions
CN113051820A (en
Inventor
李志辉
张子彬
党德鹏
吴俊林
彭傲平
孙学舟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ultra High Speed Aerodynamics Institute China Aerodynamics Research and Development Center
Original Assignee
Ultra High Speed Aerodynamics Institute China Aerodynamics Research and Development Center
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ultra High Speed Aerodynamics Institute China Aerodynamics Research and Development Center filed Critical Ultra High Speed Aerodynamics Institute China Aerodynamics Research and Development Center
Priority to CN202110314455.0A priority Critical patent/CN113051820B/en
Publication of CN113051820A publication Critical patent/CN113051820A/en
Application granted granted Critical
Publication of CN113051820B publication Critical patent/CN113051820B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • 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
    • G06N3/045—Combinations of networks
    • 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

Abstract

The invention discloses a method for manufacturing a semiconductor device, which comprises the following steps: step one, obtaining a part of representative aerodynamic data result based on a unified aerodynamic theory algorithm of a Boltzmann model equation; step two, based on partial representative pneumatic data results, acquiring a fitting model of the outline convolution neural network by means of a convolution neural network algorithm; and thirdly, obtaining aerodynamic state result parameters and a flow field information cloud picture of the required state according to the convolutional neural network fitting model. Compared with different calculation methods used in the past for different watercourses, the cross-watercourse pneumatic parameter simulation method provided by the invention relies on the calculation result of the typical fly height bypass state of a unified algorithm as data, can be used for rapidly calculating the pneumatic characteristics of the whole watercourses, and avoids the situation that the results of the boundary of the methods of all the watercourses are not uniform, especially the result of a thin transition flow area, and has high result reliability.

Description

Cross-drainage-basin pneumatic parameter simulation method based on convolutional neural network
Technical Field
The invention relates to the technical field of aerodynamics of aircrafts, in particular to a cross-river basin pneumatic parameter simulation method based on a convolutional neural network.
Background
Under the conditions that the computer conditions are not advanced enough and the fluid mechanics calculation technology is not fully developed in the past, aerodynamic scientific researchers can only calculate the flow of the atmosphere into four large flow areas such as a continuous flow area, a sliding flow area, a thin transition flow area and a free molecular flow area according to the typical flow state characteristics of different flow areas, for example, the knudsen number is utilized, then the problem of the continuous flow area is solved by utilizing an Euler and N-S equation, the problem of the flow of the sliding flow area is solved by utilizing a Slip-N-S (N-S) equation with sliding boundary conditions, the problem of the pneumatic calculation of the thin transition flow area is solved by utilizing a bypass method and a coupling method, and the problem of the flow from the free molecular flow area to the thin transition flow area is solved by utilizing a DSMC method. But such a solution still has the following drawbacks:
firstly, the flow areas do not have strict boundaries, and the pneumatic state near the boundary of each area lacks a unified and reliable theoretical support basis; secondly, a plurality of flow areas can exist in the same flow field, for example, in the high-altitude jet problem, the same flow field simultaneously has two flow partitions of a continuous flow area and a thin flow area; the thin transition flow area is not complete in the aspects of a simulation method and a test technology, and in theory, system research is lacking, while the bridging method and the coupling method have use values, the bridging method is not smooth in connection, the coupling method is narrow in calculation range, the calculation result depends on experience judgment, and integrated simulation cannot be realized.
Literature such as "unified algorithm research of gas motion theory based on Boltzmann model equation" proposes a cross-basin unified algorithm based on Boltzmann model equation. The algorithm removes continuous dependence of a distribution function on a speed space by using discrete, dimension reduction and other methods, disassembles a calculation task, disperses data storage, enables a process of solving a Boltzmann model equation to be more suitable for a parallel computer operation principle, solves the complex aerodynamic problem of reentry of an aircraft across a river basin, and calculates convergence consistency verification comparison of the flow resistance coefficient of the falling ball of the full-flight river basin by using a unified algorithm (GKUA) as shown in figure 3.
However, due to large calculation amount of a unified algorithm and high resource consumption, a full-river-basin pneumatic data result cannot be obtained by using a small amount of calculation resources in a short time, and a bridge formula method based on the correlation of free molecular flow and continuous flow local bridge function theory is provided in literature on the low-orbit control aerodynamic characteristic integrated modeling and calculation research of the Tiangong aircraft.
Although the method achieves the purpose of acquiring the full-river basin pneumatic data result by using smaller resources in a short time, the related parameters of the bridge formula method need to be manually adjusted, the best fitting result can not be necessarily obtained by visual calibration, and the requirement on the experience level of operators is higher.
Disclosure of Invention
It is an object of the present invention to address at least the above problems and/or disadvantages and to provide at least the advantages described below.
To achieve these objects and other advantages and in accordance with the purpose of the invention, a cross-river basin pneumatic parameter simulation method based on a convolutional neural network is provided, comprising:
step one, obtaining a part of representative aerodynamic data result based on a unified aerodynamic theory algorithm of a Boltzmann model equation;
step two, based on partial representative pneumatic data results, acquiring a fitting model of the outline convolution neural network by means of a convolution neural network algorithm;
and thirdly, obtaining aerodynamic calculation result parameters and flow field information cloud pictures of the required states according to the convolutional neural network fitting model.
Preferably, in step one, the method of calculating the partial representative pneumatic data using a unified algorithm is configured to include:
s10, inputting the Knudsen number, the Plantnumber, the Mach number, the pressure, the temperature, the attack angle, the sideslip angle, the flying height and the aerodynamic profile grid of an incoming flow;
s11, introducing a local equilibrium state distribution function f N And a gas molecular collision relaxation parameter v to obtain a gas molecular velocity distribution function control equation as follows:
Figure GDA0004114126240000021
Figure GDA0004114126240000031
Figure GDA0004114126240000032
Figure GDA0004114126240000033
Figure GDA0004114126240000034
Figure GDA0004114126240000035
wherein f is a dependent position space
Figure GDA0004114126240000036
Molecular speed->
Figure GDA0004114126240000037
And a gas molecular velocity distribution function of time t, f N F is a local equilibrium velocity distribution function M For local Maxwell equilibriumA cloth function; n, P, T are respectively the molecular number density, pressure and temperature of the gas, +.>
Figure GDA0004114126240000038
Pr is Plantt number, and χ is a gas molecular interaction parameter related to a molecular model; lambda (lambda) ∞ The average free path of incoming gas molecules is L, the characteristic length is L, and Kn is the Knudsen number and is used for dividing the flow state control parameters of each river basin; u, V, W the flow velocity in x, y and z directions respectively, +.>
Figure GDA0004114126240000039
Representing the thermal movement speed of gas molecules;
s12, substituting the gas molecular velocity distribution function f obtained through iterative calculation into the following equation to obtain related macroscopic flow physical quantity:
Figure GDA00041141262400000310
Figure GDA00041141262400000311
Figure GDA00041141262400000312
P＝n′T；
Figure GDA00041141262400000313
Figure GDA00041141262400000314
the aerodynamic drag coefficient related aerodynamic result parameters and flow field information cloud pictures of the appearance are obtained under the input condition through macroscopic flow physical quantity;
s13, input data related to the incoming flow Knudsen number, the Plantnumber, the Mach number, the pressure, the temperature, the attack angle, the sideslip angle and the flying height are regulated, and multiple groups of pneumatic data results under one pneumatic profile input condition are obtained through repeating S10-S12.
Preferably, in the second step, the configuration mode of the fitting model includes:
s20, classifying the pneumatic data result obtained in the S13 according to input data and output data, and grouping according to the corresponding relation of input and output, so as to establish a pneumatic result database of the pneumatic appearance unified algorithm;
s21, dividing the database into training data and test data according to the proportion of 7:3, and then delivering the training data and the test data to a convolutional neural network algorithm for operation;
s22, after the convolutional neural network operation is finished, the accuracy of the fitting model is judged by using test data detection, and if the accuracy is lower than a preset value, the adaptation improvement is carried out on the relevant parameters of the convolutional kernel size and the sliding step length so as to obtain the fitting model meeting the requirements.
Preferably, in the third step, the input parameters of the required aerodynamic state are input into the fitting model, so as to obtain aerodynamic calculation result parameters and flow field information cloud images of the required aerodynamic state
The invention at least comprises the following beneficial effects: compared with the prior art that different calculation methods are used for different watercourses, the method uses the operation result of the unified algorithm as a data basis, and can be used for calculating the pneumatic characteristics of the whole watercourses, so that the situation that the results of the method junctions of the watercourses are not uniform, particularly the results of the thin transition flow areas, are avoided, and the result reliability is high.
Compared with the traditional methods such as the bridge formula method based on the correlation of free molecular flow and continuous flow local bridge function theory, the method can generate the flow field information such as the pressure cloud image, the temperature cloud image and the like of the required height besides generating the curve of the aerodynamic parameter changing along with the height, and is beneficial to the development of further work such as aircraft design.
Thirdly, the invention can obtain the full-drainage-basin pneumatic data result with smaller time cost and calculation resources by means of an artificial intelligent Convolutional Neural Network (CNN) technology according to the partial representative pneumatic data result of the unified algorithm.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention.
Drawings
FIG. 1 is a graph showing the comparison of the Results of the unified algorithm (GKUA Results) and the pressure cloud image of the method (CNN Results);
FIG. 2 is a schematic diagram of the result of the unified algorithm pressure cloud;
fig. 3 is a graph of a unified algorithm (GKUA) for calculating convergence consistency of the flow resistance coefficients of the full-flight basin falling ball.
Detailed Description
The present invention is described in further detail below with reference to the drawings to enable those skilled in the art to practice the invention by referring to the description.
It will be understood that terms, such as "having," "including," and "comprising," as used herein, do not preclude the presence or addition of one or more other elements or groups thereof.
The patent has the following three steps:
1. pneumatic theory unified algorithm based on Boltzmann model equation to obtain partial representative pneumatic data result
According to the literature such as Boltzmann model equation-based gas motion theory unified algorithm research, the unified algorithm calculation flow is briefly cited here as follows:
data such as the knudsen number, planttnumber, mach number, pressure, temperature, angle of attack, sideslip angle, altitude, and aerodynamic profile grid of the incoming flow are first entered.
Secondly, introducing a local equilibrium state speed distribution function f N And a gas molecular collision relaxation parameter v, the control equation of the gas molecular velocity distribution function can be obtained as follows:
Figure GDA0004114126240000051
Figure GDA0004114126240000052
Figure GDA0004114126240000053
Figure GDA0004114126240000054
Figure GDA0004114126240000055
Figure GDA0004114126240000056
wherein f is a dependent position space
Figure GDA0004114126240000057
Molecular speed->
Figure GDA0004114126240000058
And a gas molecular velocity distribution function of time t, f N F is a local equilibrium velocity distribution function M Is a local Maxwell equilibrium distribution function; n, P, T are respectively the molecular number density, pressure and temperature of the gas, +.>
Figure GDA0004114126240000059
Pr is Plantt number, and χ is a gas molecular interaction parameter related to a molecular model; lambda (lambda) ∞ Is the mean free path of incoming gas molecules, L is the characteristic length, kn is the Knudsen number, and is used for dividing the flow of each river basinA state control parameter; u, V, W the flow velocity in x, y and z directions respectively, +.>
Figure GDA0004114126240000061
Representing the thermal movement speed of gas molecules;
thirdly, substituting the gas molecular velocity distribution function f obtained by iterative calculation into the following equation to obtain macroscopic flow physical quantities, such as gas density n ', flow velocity n', temperature T, gas pressure P, viscosity stress tau, heat flow vector q and the like:
Figure GDA0004114126240000062
Figure GDA0004114126240000063
Figure GDA0004114126240000064
P＝n′T；
Figure GDA0004114126240000065
/>
Figure GDA0004114126240000066
finally, the aerodynamic result parameters such as aerodynamic drag coefficient (Cd) and the like of the appearance under the input condition can be obtained from the macroscopic flow physical quantity, and flow field information such as pressure cloud picture, temperature cloud picture and the like can be obtained.
By adjusting input data such as incoming gas knudsen number, prandtl number, mach number, pressure, temperature, attack angle, sideslip angle, flying height and the like and repeating the steps, a plurality of groups of aerodynamic data results (including aerodynamic result parameters and flow field information cloud pictures as shown in fig. 2) under different input conditions of a certain aerodynamic shape can be obtained.
2. Based on partial representative aerodynamic data result, acquiring the fitting model of the outline convolution neural network by means of a convolution neural network algorithm
The pneumatic data results are classified according to input data (including knudsen number, prandtl number, mach number, pressure, temperature, attack angle, sideslip angle, flying height and the like) and output data (including pneumatic result parameters and flow field information cloud pictures), and are grouped according to corresponding relations of input and output, so that a pneumatic result database of the pneumatic appearance unified algorithm is established.
The database is divided into training data and test data according to the proportion of 7:3, and then the training data and the test data are transmitted to a convolutional neural network algorithm for operation.
And when the operation is finished, the convolutional neural network algorithm detects the correct rate of the fitting model by using the test data, and if the correct rate is too low, the convolution kernel size, the sliding step length and other super parameters can be properly adjusted until the fitting model with higher correct rate is obtained.
3. According to the convolutional neural network fitting model, obtaining aerodynamic state result parameters of the required state and a flow field information cloud picture
And inputting the input parameters of the required aerodynamic state into the fitting model to obtain aerodynamic calculation result parameters and flow field information cloud pictures of the required aerodynamic state.
The patent provides a cross-river basin pneumatic parameter simulation method based on unified algorithm data and convolutional neural network technology, which replaces the existing method. And convolutional neural network (Convolutional Neural Networks) is a deep learning model or a multi-layer perceptron similar to an artificial neural network, and is one of representative algorithms of artificial intelligence. Compared with the traditional computational fluid dynamics method, the artificial intelligence algorithm does not need clear theoretical logic operation, and can find rules in the input and output data only after enough data are given, so that an output result with higher accuracy is given for unknown input data. And compared with other artificial intelligence algorithms, the convolutional neural network has obvious advantages in the aspect of image recognition.
The specific idea of the invention is as follows: firstly, a part of representative pneumatic data result is obtained by a unified algorithm, the data are analyzed by a convolutional neural network algorithm, so that a convolutional neural network fitting model is established, then the pneumatic data result of the pneumatic appearance full-flow area and a flow field information cloud picture are obtained according to the fitting model, in the cylindrical bypass operation of a certain input state, the unified algorithm (GKUA Results) is compared with the calculation result of the pressure cloud picture of the method (CNN Results) according to the figure, and the Results of the unified algorithm (GKUA Results) and the CNN Results are almost identical from the figure.
Table 1 shows the comparison of the unified algorithm with the aerodynamic result parameters of the method of the present patent in the above cylindrical bypass operation.
Figure GDA0004114126240000071
Table 1 unified algorithm and pneumatic result parameter comparison of the present patent method
The average deviation indicated in the above table is the result of averaging the coefficients of the coordinates of the different positions, and it can be seen from the table that the remaining parts remain identical except for the difference at the end of the coefficient values of the few spatial positions. Overall, the average deviation is less than 10 < -7 >, the average deviation percentage is less than 10 < -6 >, and the use requirement is met.
It should be noted that, the above unified algorithm results take days to obtain by calling thousands of computing cores on the huge computer platform, while the above patent method results in a few minutes to obtain the required data by the personal computer, so the calculation efficiency is high.
In practical projects, the method can be used together with a unified algorithm:
if the actual project requires 30 states of pneumatic data of a certain shape, three strategies can be selected as follows
A. If the unified algorithm is only relied on, a huge computer platform is required to call tens of thousands of calculation cores, and the calculation of 30 states is completed in a period of months, so that 30 groups of pneumatic result parameters and a flow field information cloud image are obtained;
B. if the unified algorithm is used to cooperate with the bridge formula method and other traditional methods, the unified algorithm needs to calculate 10 states, but the bridge formula method cannot obtain the flow field information cloud picture, so that the calculated amount is reduced by 2/3, but only 30 groups of pneumatic result parameters and 10 groups of flow field information cloud pictures obtained by the unified algorithm can be obtained;
C. if the unified algorithm is used to match the method, the unified algorithm only needs to calculate 10 states, and the rest states can be completely obtained by the method, so that the complete 30 groups of pneumatic result parameters and flow field information cloud pictures can be obtained under the condition of reducing the calculated amount by 2/3.
Therefore, the method has higher practical value in practical use.
The above is merely illustrative of a preferred embodiment, but is not limited thereto. In practicing the present invention, appropriate substitutions and/or modifications may be made according to the needs of the user.
The number of equipment and the scale of processing described herein are intended to simplify the description of the present invention. Applications, modifications and variations of the present invention will be readily apparent to those skilled in the art.
Although embodiments of the invention have been disclosed above, they are not limited to the use listed in the specification and embodiments. It can be applied to various fields suitable for the present invention. Additional modifications will readily occur to those skilled in the art. Therefore, the invention is not to be limited to the specific details and illustrations shown and described herein, without departing from the general concepts defined in the claims and their equivalents.

Claims (1)

1. The cross-river basin pneumatic parameter simulation method based on the convolutional neural network is characterized by comprising the following steps of:
step one, obtaining a part of representative aerodynamic data result based on a unified aerodynamic theory algorithm of a Boltzmann model equation;
step two, based on partial representative pneumatic data results, acquiring a convolutional neural network fitting model corresponding to the appearance by means of a convolutional neural network algorithm;
step three, according to the convolutional neural network fitting model, obtaining the aerodynamic calculation result parameters and the flow field information cloud picture of the required state;
in a first step, a method for calculating partial representative pneumatic data using a unified algorithm is configured to include:
s10, inputting the Knudsen number, the Plantnumber, the Mach number, the pressure, the temperature, the attack angle, the sideslip angle, the flying height and the aerodynamic profile grid of an incoming flow;
s11, introducing a local equilibrium state distribution function f N And a gas molecular collision relaxation parameter v to obtain a gas molecular velocity distribution function control equation as follows:
Figure FDA0004114126230000011
Figure FDA0004114126230000012
Figure FDA0004114126230000013
Figure FDA0004114126230000014
Figure FDA0004114126230000015
Figure FDA0004114126230000016
wherein f is a dependent position space
Figure FDA0004114126230000017
Molecular speed->
Figure FDA0004114126230000018
And a gas molecular velocity distribution function of time t, f N F is a local equilibrium velocity distribution function M Is a local Maxwell equilibrium distribution function; n, P, T are respectively the molecular number density, pressure and temperature of the gas, +.>
Figure FDA0004114126230000019
Pr is Plantt number, and χ is a gas molecular interaction parameter related to a molecular model; lambda (lambda) ∞ The average free path of incoming gas molecules is L, the characteristic length is L, and Kn is the Knudsen number and is used for dividing the flow state control parameters of each river basin; u, V, W the flow velocity in x, y and z directions respectively, +.>
Figure FDA0004114126230000021
Representing the thermal movement speed of gas molecules;
s12, substituting the gas molecular velocity distribution function f obtained through iterative calculation into the following equation to obtain related macroscopic flow physical quantity:
Figure FDA0004114126230000022
Figure FDA0004114126230000023
Figure FDA0004114126230000024
P＝n′T；
Figure FDA0004114126230000025
Figure FDA0004114126230000026
the aerodynamic drag coefficient related aerodynamic result parameters and flow field information cloud pictures of the appearance are obtained under the input condition through macroscopic flow physical quantity;
s13, inputting data related to the incoming flow Knudsen number, plantnumber, mach number, pressure, temperature, attack angle, sideslip angle and flying height by adjusting the incoming flow, and repeating S10-S12 to obtain a plurality of groups of pneumatic data results under one pneumatic shape input condition;
in the second step, the configuration mode of the fitting model is configured to include:
s20, classifying the pneumatic data result obtained in the S13 according to input data and output data, and grouping according to the corresponding relation of input and output, so as to establish a pneumatic result database corresponding to the pneumatic appearance through a unified algorithm;
s21, dividing the database into training data and test data according to the proportion of 7:3, and then delivering the training data and the test data to a convolutional neural network algorithm for operation;
s22, after the convolutional neural network operation is finished, judging the correct rate of the fitting model by using test data detection, and if the correct rate is lower than a preset value, carrying out adaptability improvement on the relevant parameters of the convolutional kernel size and the sliding step length so as to obtain the fitting model meeting the requirements;
in the third step, input parameters of the required aerodynamic state are input into the fitting model, so that aerodynamic calculation result parameters and flow field information cloud pictures of the required aerodynamic state are obtained.
CN202110314455.0A 2021-03-24 2021-03-24 Cross-drainage-basin pneumatic parameter simulation method based on convolutional neural network Active CN113051820B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110314455.0A CN113051820B (en) 2021-03-24 2021-03-24 Cross-drainage-basin pneumatic parameter simulation method based on convolutional neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110314455.0A CN113051820B (en) 2021-03-24 2021-03-24 Cross-drainage-basin pneumatic parameter simulation method based on convolutional neural network

Publications (2)

Publication Number Publication Date
CN113051820A CN113051820A (en) 2021-06-29
CN113051820B true CN113051820B (en) 2023-05-16

Family

ID=76514927

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110314455.0A Active CN113051820B (en) 2021-03-24 2021-03-24 Cross-drainage-basin pneumatic parameter simulation method based on convolutional neural network

Country Status (1)

Country Link
CN (1) CN113051820B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114330080B (en) * 2022-03-04 2022-05-13 中国空气动力研究与发展中心计算空气动力研究所 Prediction method for aircraft surface-symmetric cross-basin flow field
CN116296239A (en) * 2023-05-26 2023-06-23 中国空气动力研究与发展中心计算空气动力研究所 Wind tunnel test method, device, equipment and medium based on convolutional neural network

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6751529B1 (en) * 2002-06-03 2004-06-15 Neural Robotics, Inc. System and method for controlling model aircraft
CN106205126A (en) * 2016-08-12 2016-12-07 北京航空航天大学 Large-scale Traffic Network based on convolutional neural networks is blocked up Forecasting Methodology and device
CN106779051A (en) * 2016-11-24 2017-05-31 厦门中控生物识别信息技术有限公司 A kind of convolutional neural networks model parameter processing method and system
CN109410208A (en) * 2018-11-14 2019-03-01 成都极致智造科技有限公司 The machine learning identification of Wear Mechanism of Abrasive Belt and process parameter optimizing method
CN109783828A (en) * 2017-11-13 2019-05-21 北京普华亿能风电技术有限公司 Wind power forecasting method and system in a kind of wind power plant
CN110851957A (en) * 2019-10-15 2020-02-28 南京航空航天大学 Atmospheric data sensing system resolving method based on deep learning
US10766136B1 (en) * 2017-11-03 2020-09-08 Amazon Technologies, Inc. Artificial intelligence system for modeling and evaluating robotic success at task performance
CN112052621A (en) * 2020-07-22 2020-12-08 西安交通大学 CNN-based supercritical carbon dioxide impeller mechanical transcritical region prediction and control method

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111709983A (en) * 2020-06-16 2020-09-25 天津工业大学 Bubble flow field three-dimensional reconstruction method based on convolutional neural network and light field image
CN112329322A (en) * 2020-10-09 2021-02-05 清华大学 Thermal spraying method and device based on convolutional neural network

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6751529B1 (en) * 2002-06-03 2004-06-15 Neural Robotics, Inc. System and method for controlling model aircraft
CN106205126A (en) * 2016-08-12 2016-12-07 北京航空航天大学 Large-scale Traffic Network based on convolutional neural networks is blocked up Forecasting Methodology and device
CN106779051A (en) * 2016-11-24 2017-05-31 厦门中控生物识别信息技术有限公司 A kind of convolutional neural networks model parameter processing method and system
US10766136B1 (en) * 2017-11-03 2020-09-08 Amazon Technologies, Inc. Artificial intelligence system for modeling and evaluating robotic success at task performance
CN109783828A (en) * 2017-11-13 2019-05-21 北京普华亿能风电技术有限公司 Wind power forecasting method and system in a kind of wind power plant
CN109410208A (en) * 2018-11-14 2019-03-01 成都极致智造科技有限公司 The machine learning identification of Wear Mechanism of Abrasive Belt and process parameter optimizing method
CN110851957A (en) * 2019-10-15 2020-02-28 南京航空航天大学 Atmospheric data sensing system resolving method based on deep learning
CN112052621A (en) * 2020-07-22 2020-12-08 西安交通大学 CNN-based supercritical carbon dioxide impeller mechanical transcritical region prediction and control method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Aero-engine sensor fault diagnosis based on convolutional neural network;J. Li 等;《2018 37th Chinese Control Conference》;6049-6054 *
基于深度学习的翼型气动系数预测;陈海 等;《空气动力学学报》;第36卷(第2期);294-299 *
基于深度神经网络的空气动力学数据建模方法研究;时梓芳;《中国优秀硕士学位论文全文数据库基础科学辑》(第（2020）8期);A004-13 *

Also Published As

Publication number Publication date
CN113051820A (en) 2021-06-29

Similar Documents

Publication Publication Date Title
Sekar et al. Fast flow field prediction over airfoils using deep learning approach
Liang et al. Differentiable cloth simulation for inverse problems
CN113051820B (en) Cross-drainage-basin pneumatic parameter simulation method based on convolutional neural network
Alata et al. Optimizing of fuzzy c-means clustering algorithm using GA
Park et al. Machine-learning-based feedback control for drag reduction in a turbulent channel flow
Ntampaka et al. A hybrid deep learning approach to cosmological constraints from galaxy redshift surveys
Forti et al. Convergence of neural networks for programming problems via a nonsmooth Łojasiewicz inequality
Hou et al. Machine-learning-based detection of aerodynamic disturbances using surface pressure measurements
CN108805222A (en) A kind of deep learning digital handwriting body recognition methods based on ARM platforms
CN106023286A (en) Fluid animation accelerated generation method based on data driving
Janny et al. Eagle: Large-scale learning of turbulent fluid dynamics with mesh transformers
CN114444216B (en) Aircraft attitude control method and system under high-altitude condition based on numerical simulation
Yang et al. BPNN and CNN-based AI modeling of spreading and icing pattern of a water droplet impact on a supercooled surface
Hou et al. Machine learning based detection of flow disturbances using surface pressure measurements
CN106407601B (en) A kind of aerodynamic characteristic data processing method based on data mining technology
Wong et al. Graph neural network based surrogate model of physics simulations for geometry design
Rajabi et al. Intelligent prediction of turbulent flow over backward-facing step using direct numerical simulation data
Kretinin et al. Neuronet's method of weighted residuals for computer simulation of hydrodynamics problems
CN108009565B (en) Method for calculating design flood level of river network area under variable environment
Kretinin et al. Method of weighted residuals on the base of neuronet's approximations for computer simulation of hydrodynamics problems
JP4267726B2 (en) Device for determining relationship between operation signal and operation amount in control device, control device, data generation device, input / output characteristic determination device, and correlation evaluation device
Zhang et al. Optimization and uncertainty quantification of spray break-up submodel with regularized multi-task neural nets
Mei et al. Real-time facial expression recognition based on the improved brain emotional learning model
CN108846421A (en) A kind of image classification method based on deep learning
CN112306020B (en) Uniform spherical surface dispersion control method for designated target position by multi-agent cluster

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
GR01 Patent grant
GR01 Patent grant