CN110083895A - A kind of surface heat flux three-dismensional effect modification method neural network based - Google Patents
A kind of surface heat flux three-dismensional effect modification method neural network based Download PDFInfo
- Publication number
- CN110083895A CN110083895A CN201910294868.XA CN201910294868A CN110083895A CN 110083895 A CN110083895 A CN 110083895A CN 201910294868 A CN201910294868 A CN 201910294868A CN 110083895 A CN110083895 A CN 110083895A
- Authority
- CN
- China
- Prior art keywords
- fluid
- hot
- neural network
- temperature
- heat flux
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- 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
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Automation & Control Theory (AREA)
- Aviation & Aerospace Engineering (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The invention discloses a kind of surface heat flux three-dismensional effect modification methods neural network based.If the modification method is that the inner wall in stationary point hot-fluid peripheral region installs dry temperature sensor, first with the temperature data of internal each temperature point, the hot-fluid on corresponding generating surface point is obtained by one-dimensional heat flux method, then artificial neural network algorithm is introduced, measuring point each in previous step is recognized into the list entries after hot-fluid normalized as neural network accordingly, heat flux value of the renormalization result exported in neural network by training as interest region.Modification method proposed by the present invention avoids the time complexity of three-dimensional identification, in combination with the strong nonlinearity of sequential function method good noise immunity and neural network, conventional model can be greatly simplified, the identification precision of stationary point hot-fluid is improved, ensure that the real-time of on-line identification.
Description
Technical field
The present invention relates to a kind of calculation methods of surface heat flux and neural network modification method more particularly to a kind of base
In the surface heat flux three-dismensional effect modification method of neural network.
Background technique
Hypersonic flight faces serious Aerodynamic Heating problem.Since air is by strong friction and compression,
A large amount of kinetic energy are converted into thermal energy, and aircraft ambient air temperature is caused sharply to increase, high temperature influence aircraft structural strength and
Rigidity, or even cause the ablative degradation of outer surface.Thermal protection system design is the important of hypersonic flight technology fast development
Support, researching and designing need the test data of a large amount of flight test.Temperature, hot-fluid during being on active service to aircraft etc. close
The measurement of bond parameter be evaluate thermally protective materials service performance, the pneumatic thermal model of verifying and algorithm, instruct heat protection design must
Want means.But the region big for heat flow densities such as stationary points, generally can not directly dispose temperature sensor or heat flow transducer into
On the one hand row measurement is that structural strength is caused to decline and gap heating due to structure aperture, the structures such as cause ablation asynchronous
Matching problem；On the other hand, some sensor body materials cannot bear excessively high thermal force, and the insertion of sensor brings wall
The discontinuous and circumferential interference of temperature, leading to heat-flow measurement result is not true Aerodynamic Heating.Therefore, by measuring aircraft
Inner structural wall temperature carrys out the Aerodynamic Heating discrimination method of inverting outer surface hot-fluid and temperature, becomes and obtains the important of pneumatic thermal environment
Solution.
Aerodynamic Heating identification belongs to a kind of heat conduction inverse problem, and basic principle is surveyed by measuring the temperature of Heat Conduction Material inner wall
The temperature history of point, is finally inversed by the hot-fluid time history in outer wall-heated face.Currently, having been carried out greatly to heat conduction inverse problem both at home and abroad
Quantifier elimination, it is common practice that choose suitable objective function, convert optimization problem solving for identification problem.Qian Weiqi difference
One-dimensional surface heat flux method is had studied with sequential function method and conjugate gradient method, and is extended to two and three dimensions irregular figure
Surface heat flux.Heat transfer physical process has damping and amortization and retardance, and damping and amortization shows as the variation opposite side of boundary hot-fluid
Temperature near boundary generates big influence, and with the increase from frontier distance, the influence of changes of heat flux will reduce；Retardance is then
Show as internal temperature has retardance to the reaction of boundary hot-fluid in time.According to these features, sequential function method is to heat
The identification of stream is gradually to promote progress in chronological order, i.e. introducing time step factor r, is relied on the identification of certain moment hot-fluid
It is the measured temperature of r time step after the moment.Conjugate gradient method is Iteration Regularized method, and optimization problem is decomposed
These three well-posed problems are solved for heat transfer direct problem, sensitivity solution and adjoint variable to be solved.
In addition, inner heat source intensity leads warm system in Xue Qiwen application Tikhonov technique study One-dimensional heat transfer indirect problem
Several and boundary condition more arguments identification, it is anti-to be applied to Nonlinear Heat Conduction using the distance weighted function of Bregman as regular terms
The solution of problem is based on a kind of precise algorithm in time domain and spatial spreading technology, considers the non-linear of heat source item, establishes thermal transient biography
Correcting/indirect problem mathematical model has carried out combination identification to one-dimensional multiple thermal parameters.Cui Miao uses nondimensionalization
Target equation carries out parameter identification to thermo-fluid model, but is confined to known hot-fluid functional form.Qian Weiqi is in view of high ultrasound
The material ablation of speed retreats, and using simplified pyrolysis face ablating model, is studied one-dimensional ablated surface heat flux,
And it is used for test result analysis of the bluff nose Carbon-phenolic material Narmco4028 test specimen in ceramic heat wind-tunnel, it was demonstrated that
The reasonability of ablating model and the validity of method.Zhang Cong has been carried out hypersonic using a simplified peacekeeping Two-Dimensional Heat model
The identification of combustion chamber wall surface hot-fluid achieves preferable effect under axisymmetric model.
Neural network algorithm is widely used due to its powerful nonlinear fitting ability in every field.
S.DENG and intelligence understand strong et al. the artificial neural network and genetic algorithms that have studied and solve heat conduction inverse problem, are considering simple hot-fluid
Under load condition, the letter of internal temperature field and unknown surface heat flow or material thermal physical property parameter is approached using artificial neural network
Number relationship, while the extremal optimization problem under appropriate objective function is converted by indirect problem, utilize the global optimization of genetic algorithm
Method finds indirect problem optimal solution.
The present invention studies three-dismensional effect modification method on the basis of one-dimensional heat flux, by the one-dimensional heat of sequential function method
Stream identification algorithm combines with neural network algorithm, a set of amendment for obtaining stationary point hot-fluid accurate identification result in real time of proposition
Method.
Summary of the invention
The purpose of the present invention is to provide a kind of surface heat flux three-dismensional effect modification methods neural network based.
Method main technical schemes proposed by the present invention are as follows:
A kind of surface heat flux three-dismensional effect modification method neural network based is in stationary point hot-fluid peripheral region
If inner wall installs dry temperature sensor, first with the temperature data of internal each temperature point, pass through one-dimensional heat flux side
Method obtains the hot-fluid on corresponding generating surface point, then introduces artificial neural network algorithm, and measuring point each in previous step is corresponding
Identification hot-fluid normalized after list entries as neural network, in neural network by it is trained exported it is anti-
Normalize heat flux value of the result as interest region.The specific method is as follows:
First with sequential function method, the temperature data of inner structural wall temperature sensor is passed through into one-dimensional heat flux estimation method
It is finally inversed by the heat flow density of corresponding generating surface point.
For one-dimensional and unsteady state heat conduction problem, the governing equation of temperature T (x, t), primary condition and boundary condition can be with
It indicates are as follows:
T (x, 0)=T0(x) (2)
In formula, t indicates the time, and material initial temperature is distributed as T0(x), length L, one end load hot-fluid q, and the other end is exhausted
Heat, ρ are the density of material, CpFor specific heat capacity, k (T) is the thermal coefficient varied with temperature.
The corresponding indirect problem of direct problem is exactly according to the anti-hot-fluid for pushing away current time of material inner wall measuring point temperature, that is, root
Identical surface heat flow is found according to the temperature-responsive put in model.Geometry symmetry mode is diffused, therefore can be according to heat
The physical process of conduction, estimates hot-fluid sequentially in time.So-called sequential function method is exactly to pass through t according to this thoughtM,
tM+1,…,tM+r-1Total r future time walks measured temperature to estimate tMThe heat flow value at moment.
Measuring point xmThe temperature measuring data of place's temperature sensor can indicate are as follows:
In formula: xmIt is measuring point position, v (t) is measurement noise.
Parametric optimization problem can be converted into the processing of parameter identification problem in engineering, it is minimum former according to output error
Then, suitable parameter is found out.According to this thinking, heat flux problem can be converted into the measured value according to temperature point, look for
To the solution most identical with true hot-fluid, optimization aim is to solve for the measuring point temperature that direct problem obtains and the accumulation of temperature measured value misses
It is poor minimum, therefore following objective function can be found:
Wherein, T (xm, t, q) and it is measuring point temperature calculations,It is then corresponding measured value, | | | | represent certain model
Number, there are commonly 1- norm and 2- norm, it is upper more convenient in derivation optimization that 2- norm compares 1- norm.2- norm is applied to
Formula (6) is available:
Consider the retardance of heat transfer problem, it is assumed that tMMoment hot-fluid is only capable of influencing tMTo tM+r-1Between measuring point temperature, and
And consider numerical solution in the time it is discrete, then when identification tMHot-fluid q (the t at momentM) when (be abbreviated as q hereinafterM), excellent
The function of change may be expressed as:
For objective function as above, it can use optimization method and it optimized, the purpose of optimization is formula
(8) minimalization.It was verified that Newton-Raphson algorithm is to solve for this type of optimization problem most efficient method.
The necessary condition that J (q) is minimized is:
Omit that second order is small to measure Δ qkMeet following formula:
M is known as information matrix in formula.Remember q (ti)=qi, due to:
WhereinIt is first derivative of the quantity of state to location parameter, referred to as sensitivity.Further derivation can obtain:
It is found that for convergent solution, the first item in formula (12) can level off to zero quickly for analysis, can directly ignore, because
This can be obtained:
As previously mentioned, solving temperature field TM+r-1Depend not only on qM, also by qM,…,qM+r-1Influence, therefore recognize qMWhen
It is also required to know tMTo tM+r-1Between hot-fluid, that is, need to establish qM+iWith qMRelationship.Assuming that hot-fluid is linearly to become during this period
Change, when the sampling interval is fixed, have:
The iterated revision formula of hot-fluid in this way can indicate are as follows:
γ is the iterated revision step-length of Newton-Raphson method, is also relaxation factor, and k is the number of iterations.It is being not introduced into pine
The unreasonable situation that relaxation can shake in some cases by the period of the day from 11 p.m. to 1 a.m, identification result, this is because going out during iterated revision
It is excessive caused to have showed correction amount, such as positive excessive correction amount has occurred in certain iterative process, in order to eliminate this
The influence of a correction amount can generate a bigger negative correction amount in next iteration step, in order to eliminate this negative amendment
The influence of amount can generate a bigger positive correction amount in next iteration step, and and so on there have been be result shake
The case where swinging.And relaxation factor γ can control correction amount in a lesser range, and then improve result shake well
The case where swinging.
Temperature is denoted as the sensitivity of hot-fluidIts solution can be sought hot-fluid by heat transfer differential equation
It leads to obtain:
In conclusion the calculating process of sequential function method identification surface heat flow may be summarized to be: based on initial hot-fluid q0, first
The hot-fluid at moment after estimating first with formula (14) solves the temperature value at subsequent r moment in conjunction with initial temperature, then by asking
Sensitivity of the temperature to the moment hot-fluid is solved, then hot-fluid is iterated using Newton-Raphson method according to formula (15)
Amendment, until meeting required precision, the heat flow value being calculated at this time is exactly the identifier at current time, forward along time orientation
It promotes, the heat flow value of following instant can be released using same method.
It is finally inversed by the heat flow value of corresponding generating surface point in the one-dimensional identification algorithm using sequential function method, as BP
The input of neural network, by exporting stationary point changes of heat flux after training.The number of nodes of BP network input layer is set as n, hidden layer section
Points are set as l, and output layer number of nodes is m, and list entries is expressed as vector X, and output sequence is then expressed as Y.It is fixed according to input X
Weight between each node of adopted input layer and hidden node is ωi,j, hidden layer threshold value is denoted as a, by summation unit and activation
After function effect, hidden layer output H can be indicated are as follows:
Wherein f is the activation primitive of hidden layer.Note output layer threshold value is b, using hidden layer result as the input of output layer, hidden layer
Weight between each node and output node layer is ωj,k, the reality output O of output layer can equally be calculated:
The then error of this wheel training are as follows:
ek=Yk-OkK=1,2 ..., m (19)
Update is iterated to weight using gradient descent method, for input layer and hidden layer, right value update formula difference
It can indicate are as follows:
Wherein η is the iteration step length in optimization process, referred to as learning rate.
Similarly, in gradient descent method network threshold more new formula are as follows:
BP neural network is built according to above thinking and carries out learning training, and the mature neural network of training has powerful
Nonlinear Mapping and generalization ability.Studies have shown that the Multi-layered Feedforward Networks comprising single hidden layer, as long as hidden neuron is enough
It is more, then the continuous function of arbitrarily complicated degree can be approached with arbitrary accuracy.Therefore it can use possessed by neural network
Advantage is used for the amendment of surface heat flux three-dismensional effect, first with the temperature data of internal each measuring point, passes through one
Dimension heat flux method obtains the hot-fluid on corresponding generating surface point, then introduces artificial neural network algorithm, will be in previous step
Each measuring point is used as list entries after recognizing the normalized that heat flow density passes through data accordingly, passes through in neural network
Heat flux value of the renormalization result that training is exported as interest region.
The beneficial effects of the present invention are:
Modification method proposed by the present invention is examined on existing Research foundation of the sequential function method to one-dimensional surface heat flux
The difficulty for considering three-dimensional identification real-time, combines in conjunction with neural network and sequential function method, it is quasi- in real time to finally obtain surface heat flow
True identification result.Modification method proposed by the present invention avoids the time complexity of three-dimensional identification, in combination with sequence letter
The strong nonlinearity of number method good noise immunity and neural network, can greatly simplify conventional model, improve distinguishing for stationary point hot-fluid
Know precision, ensure that the real-time of on-line identification.
Detailed description of the invention
Fig. 1 is that the present invention is based on the flow charts of the surface heat flux three-dismensional effect modification method of neural network.
Fig. 2 is Three-dimensional Heat-transfer model geometric schematic diagram.
Fig. 3 is internal temperature measuring point position.
Fig. 4 is that the one-dimensional identification result under three-dismensional effect compares.
Fig. 5 is stationary point hot-fluid neural network prediction result.
Fig. 6 is stationary point hot-fluid Relative Error.
Fig. 7 is different moments stationary point hot-fluid prediction result.
Specific embodiment
The present invention is described in detail specific embodiment with reference to the accompanying drawing, the objects and effects of the present invention will become brighter
It is aobvious.
A kind of surface heat flux three-dismensional effect modification method neural network based of the invention, core content be
If the inner wall of stationary point hot-fluid peripheral region installs dry temperature sensor, with one-dimensional sequential function method by each temperature sensor
Measurement data be converted into corresponding heat flux sequence, the input after the normalized of data as BP network, net
The renormalization result of network output is exactly the identification result of stationary point hot-fluid.Neural network amendment process of the invention is detailed in Fig. 1.
This example verifies example by one group of three-dismensional effect correction model to show the superiority of this method.Example model is adopted
With outer diameter 0.2m, the hollow hemisphere shell of thickness 0.03m, as shown in Fig. 2, material specific heat Cp=500J/ (kgK), density p=
8000kg/m3, thermal coefficient k=80W/ (mK), initial temperature 300K, time step 0.2s, total heating time 200s, the time
Step r takes 30, and temperature measuring point quantity is 9, and radially distributed in inner wall, temperature measuring point is laid out in bottom surface perspective view such as Fig. 3
It is shown.Measuring point 1 is located at spherical shell end, and inside 2-5 measuring point is surveyed apart from measuring point 1 spherical distance about 0.03m, outside 6-9 measuring point distance
1 about 0.06m of point.Spherical shell outer surface apply at any time with the hot-fluid of spatial variations, hot-fluid functional form are as follows:In order to assess the influence of three-dismensional effect, using only temperature point 1
Temperature data recognizes stationary point hot-fluid, and Fig. 4 is the result of one-dimensional identification compared with the hot-fluid of stationary point, comparing result, which is shown, does not consider three
There is great deviation in the heat flux result of dimension effect, or even cannot obtain the normal solution for meeting physical significance, for space-time
The small conductive structure of the hot-fluid and radius of curvature of complex distribution, three-dismensional effect be can not ignore, it is necessary to be corrected.
Using method of the invention, the temperature data of 9 measuring points is picked out accordingly using one-dimensional sequential function method respectively
Hot-fluid sequence, as the input of training sample, sample output be known stationary point hot-fluid sequence.Emulation obtains 1000 groups
Sample data randomly selects and is wherein used for network training for 900 groups, and in addition 100 groups of samples are used for network test.Such neural network
There are 9 input neurons, 1 output neuron.Fig. 5 is the neural network stationary point hot-fluid prediction result and reality after training is mature
Stationary point hot-fluid result compares, and Fig. 6 is the prediction error of stationary point hot-fluid.Since test sample is the different moments knot randomly selected
Test sample is sequentially arranged by fruit for convenience of observing, and the prediction result comparison of different moments is as shown in Figure 7.It can be seen that three
The revised stationary point hot-fluid prediction error of effect is tieed up less than 1%, three-dismensional effect correction model neural network based is effective.
It can be seen that identification of the method proposed by the present invention for peak value hot-fluid by the numerical testing of numerical simulation
Result precision is high, and model training lower online can carry out, avoid three-dimensional identification it is time-consuming it is big, precision is low, cannot achieve and exists in real time
The shortcomings that line recognizes, while there is good noise immunity and stability, in the online real-time identification of spacecraft surface peak hot-fluid
Have broad application prospects.
Claims (3)
1. a kind of surface heat flux three-dismensional effect modification method neural network based, which is characterized in that
If this method is that the inner wall in stationary point hot-fluid peripheral region installs dry temperature sensor, surveyed first with internal each temperature
The temperature data of point, obtains the hot-fluid on corresponding generating surface point by one-dimensional heat flux method, then introduces artificial neuron
Measuring point each in previous step is recognized the list entries after hot-fluid normalized as neural network by network algorithm accordingly,
Heat flux value of the renormalization result exported in neural network by training as interest region.
2. surface heat flux three-dismensional effect modification method neural network based according to claim 1, feature exist
In,
One-dimensional heat flux is carried out using sequential function method using the temperature data of inside configuration temperature point and obtains corresponding be heated
The heat flow density of surface point, process are as follows:
For one-dimensional and unsteady state heat conduction problem, the governing equation of temperature T (x, t), primary condition and boundary condition can be indicated
Are as follows:
T (x, 0)=T0(x)
In formula, t indicates the time, and material initial temperature is distributed as T0(x), length L, one end load hot-fluid q, other end insulation, ρ
It is the density of material, CpFor specific heat capacity, k (T) is the thermal coefficient varied with temperature；
According to the physical process of heat transfer, hot-fluid is estimated sequentially in time, that is, passes through tM,tM+1,…,tM+r-1When total r future
Temperature measured by spacer step estimates tMThe heat flow value at moment；
Measuring point xmThe temperature measuring data of place's temperature sensor can indicate are as follows:
In formula: xmIt is measuring point position, v (t) is measurement noise；
With the minimum objective function of accumulated error of measuring point temperature calculations and temperature measured value, and assume tMMoment hot-fluid is only capable of
Influence tMTo tM+r-1Between measuring point temperature, and the discrete of time in numerical solution is considered, then as identification tMThe hot-fluid q at moment
(tM) when, the objective function to be optimized are as follows:
It is solved using Newton-Raphson algorithm；
Establish tM+iMoment hot-fluid qM+iWith tMMoment hot-fluid qMRelationship, it is assumed that hot-fluid is linear change during this period, works as sampling
When interval is fixed, have:
qM+1=qM+(qM-qM-1)
qM+n=qM+n(qM-qM-1)
Then the iterated revision formula of hot-fluid may be expressed as:
γ is the iterated revision step-length of Newton-Raphson method, i.e. relaxation factor；K is the number of iterations, is wanted until meeting precision
It asks, the heat flow value being calculated at this time is exactly tMThe identifier at moment.
3. surface heat flux three-dismensional effect modification method neural network based according to claim 1, feature exist
In will make after heat flow data normalized after the heat flow value for being finally inversed by corresponding generating surface point by one-dimensional discrimination method
For the input of neural network, if the number of nodes of neural network input layer is n, the number of hidden nodes is set as l, and output layer number of nodes is m,
List entries is expressed as vector X, and output sequence is then expressed as Y；According to input X, each node of input layer and hidden node are defined
Between weight be ωi,j, hidden layer threshold value is denoted as a, and after summation unit and activation primitive effect, hidden layer exports H can be with table
It is shown as:
Wherein f is the activation primitive of hidden layer, and note output layer threshold value is b, and using hidden layer result as the input of output layer, hidden layer is each
Weight between node and output node layer is ωj,k, the reality output O of output layer can equally be calculated:
The then error of this wheel training are as follows:
ek=Yk-OkK=1,2 ..., m
Update is iterated to weight using gradient descent method, for input layer and hidden layer, right value update formula respectively can be with
It indicates are as follows:
ωj,k=ωj,k+ηHjekK=1,2 ..., m；J=1,2 ..., l
Wherein η is the iteration step length in optimization process, i.e. learning rate；
Similarly, in gradient descent method network threshold more new formula are as follows:
bk=bk+ekK=1,2 ..., m.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910294868.XA CN110083895A (en) | 2019-04-12 | 2019-04-12 | A kind of surface heat flux three-dismensional effect modification method neural network based |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910294868.XA CN110083895A (en) | 2019-04-12 | 2019-04-12 | A kind of surface heat flux three-dismensional effect modification method neural network based |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110083895A true CN110083895A (en) | 2019-08-02 |
Family
ID=67414903
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910294868.XA Pending CN110083895A (en) | 2019-04-12 | 2019-04-12 | A kind of surface heat flux three-dismensional effect modification method neural network based |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110083895A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111307481A (en) * | 2020-02-24 | 2020-06-19 | 中国空气动力研究与发展中心超高速空气动力研究所 | Dynamic hot wall heat flow inverse estimation method based on nonlinear artificial neural network |
| CN112990442A (en) * | 2021-04-21 | 2021-06-18 | 北京瑞莱智慧科技有限公司 | Data determination method and device based on spatial position and electronic equipment |
| CN113065275A (en) * | 2021-03-05 | 2021-07-02 | 浙江大学 | Online predication method for stagnation point heat flow in flying process of aircraft |
| CN113255181A (en) * | 2021-04-27 | 2021-08-13 | 北京航空航天大学 | Heat transfer inverse problem identification method and device based on deep learning |
| CN113553779A (en) * | 2021-09-22 | 2021-10-26 | 中国航天空气动力技术研究院 | Mars entering device stagnation point heat flow prediction method and device, electronic equipment and medium |
| CN116935985A (en) * | 2023-07-17 | 2023-10-24 | 中国地质调查局油气资源调查中心 | Sensitivity analysis method for experimental parameter change in coal gasification process |
| CN117910348A (en) * | 2024-01-11 | 2024-04-19 | 哈尔滨工业大学 | Heat flow identification method based on multisource information fusion dynamic Bayesian network |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103559537A (en) * | 2013-10-30 | 2014-02-05 | 南京邮电大学 | Template matching method based on error back propagation in out-of-order data streams |
-
2019
- 2019-04-12 CN CN201910294868.XA patent/CN110083895A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103559537A (en) * | 2013-10-30 | 2014-02-05 | 南京邮电大学 | Template matching method based on error back propagation in out-of-order data streams |
Non-Patent Citations (1)
| Title |
|---|
| 彭玉酌: "高超声速飞行环境参数辨识方法研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111307481A (en) * | 2020-02-24 | 2020-06-19 | 中国空气动力研究与发展中心超高速空气动力研究所 | Dynamic hot wall heat flow inverse estimation method based on nonlinear artificial neural network |
| CN111307481B (en) * | 2020-02-24 | 2021-09-07 | 中国空气动力研究与发展中心超高速空气动力研究所 | Dynamic hot wall heat flow inverse estimation method based on nonlinear artificial neural network |
| CN113065275A (en) * | 2021-03-05 | 2021-07-02 | 浙江大学 | Online predication method for stagnation point heat flow in flying process of aircraft |
| CN113065275B (en) * | 2021-03-05 | 2022-05-24 | 浙江大学 | Online predication method for stagnation point heat flow in flight process of aircraft |
| CN112990442A (en) * | 2021-04-21 | 2021-06-18 | 北京瑞莱智慧科技有限公司 | Data determination method and device based on spatial position and electronic equipment |
| CN112990442B (en) * | 2021-04-21 | 2021-08-06 | 北京瑞莱智慧科技有限公司 | Data determination method and device based on spatial position and electronic equipment |
| CN113255181A (en) * | 2021-04-27 | 2021-08-13 | 北京航空航天大学 | Heat transfer inverse problem identification method and device based on deep learning |
| CN113553779A (en) * | 2021-09-22 | 2021-10-26 | 中国航天空气动力技术研究院 | Mars entering device stagnation point heat flow prediction method and device, electronic equipment and medium |
| CN116935985A (en) * | 2023-07-17 | 2023-10-24 | 中国地质调查局油气资源调查中心 | Sensitivity analysis method for experimental parameter change in coal gasification process |
| CN116935985B (en) * | 2023-07-17 | 2024-03-15 | 中国地质调查局油气资源调查中心 | Sensitivity analysis method for experimental parameter change in coal gasification process |
| CN117910348A (en) * | 2024-01-11 | 2024-04-19 | 哈尔滨工业大学 | Heat flow identification method based on multisource information fusion dynamic Bayesian network |
| CN117910348B (en) * | 2024-01-11 | 2024-08-23 | 哈尔滨工业大学 | Heat flow identification method based on multisource information fusion dynamic Bayesian network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110083895A (en) | A kind of surface heat flux three-dismensional effect modification method neural network based | |
| Zare et al. | Frequency-based design of a free piston Stirling engine using genetic algorithm | |
| Goudarzi et al. | Applying artificial neural networks (ANN) to the estimation of thermal contact conductance in the exhaust valve of internal combustion engine | |
| Parwani et al. | Estimation of boundary heat flux using experimental temperature data in turbulent forced convection flow | |
| Nie et al. | Inversing heat flux boundary conditions based on precise integration FEM without iteration and estimation of thermal stress in FGMs | |
| Chen et al. | Identification of boundary conditions for non-Fourier heat conduction problems by differential transformation DRBEM and improved cuckoo search algorithm | |
| Wen et al. | Prediction of the coupled heat radiation and conduction parameters and boundary condition using the unscented Kalman filter | |
| Vikulov et al. | Intensification of mathematical models of the heat exchange in space vehicles | |
| CN106407562A (en) | A determining method for the heat insulation effect of a thermal protective layer in thermal vacuum environment | |
| Zhou et al. | Fluid-thermal modeling of hypersonic vehicles via a gas-kinetic BGK scheme-based integrated algorithm | |
| Du et al. | Interior ballistic characteristics of electromagnetic rail launcher considering the dynamic characteristics of real launcher | |
| CN113065275B (en) | Online predication method for stagnation point heat flow in flight process of aircraft | |
| Haghighi | Estimation of heat flux in variable thickness functionally graded annular fin | |
| Cortés et al. | Artificial neural networks for inverse heat transfer problems | |
| Wang et al. | Design and application of temperature control system based on fuzzy PID algorithm | |
| Abdelhamid et al. | Reconstruction of the heat transfer coefficients and heat fluxes in heat conduction problems | |
| Shi et al. | Inverse problem of estimating space and time dependent hot surface heat flux in transient transpiration cooling process | |
| Yang et al. | An iterative regularization method in simultaneously estimating the inlet temperature and heat-transfer rate in a forced-convection pipe | |
| Abdelhamid et al. | Prediction of multi-parameters in the inverse heat conduction problems | |
| Mifsud et al. | A case study on the aerodynamic heating of a hypersonic vehicle | |
| Huang et al. | An experimental inverse problem in determining spatially varying surface heat flux using measured temperature readings | |
| Li | A Novel Heat-Proof Clothing Design Algorithm Based on Heat Conduction Theory | |
| Zhao et al. | Estimating Internal Temperature of Slow Wave Structure Based on RBF Neural Network and Designing Test Model | |
| Guangyue et al. | Prediction to Temperature Field of Heating Efficiency Detecting Cavity for Wood Heating Floor based on BP Network | |
| Czél et al. | Simultaneous measurement of temperature dependent thermophysical properties |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190802 |
|
| WD01 | Invention patent application deemed withdrawn after publication |