CN113221326B - Satellite component temperature field prediction method based on teacher-student agent model - Google Patents
Satellite component temperature field prediction method based on teacher-student agent model Download PDFInfo
- Publication number
- CN113221326B CN113221326B CN202110398966.5A CN202110398966A CN113221326B CN 113221326 B CN113221326 B CN 113221326B CN 202110398966 A CN202110398966 A CN 202110398966A CN 113221326 B CN113221326 B CN 113221326B
- Authority
- CN
- China
- Prior art keywords
- model
- student
- temperature field
- satellite
- training data
- 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
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a satellite component temperature field prediction method based on a teacher-student agent model, which comprises the following steps: building a structural model of the satellite assembly layout; generating a plurality of first training data and a plurality of second training data; constructing a teacher-student depth agent model; training the teacher model by using the second training data, and predicting a temperature field corresponding to the satellite component layout in the first training data by using the trained teacher model; training the student model by using the first training data, the temperature field corresponding to the first training data and the second training data; and using the trained student model as a temperature field prediction model to perform temperature field prediction of the satellite component layout. According to the method, the training of the student model is guided by using the teacher model, effective training of the deep proxy model can be realized by using a large number of training samples without marks and a small number of training samples with marks, accurate prediction of a satellite component layout temperature field is realized, computing resources can be saved, the design cost is reduced, and the design period is shortened.
Description
Technical Field
The invention relates to the technical field of satellite thermal layout, in particular to a satellite component temperature field prediction method based on a teacher-student agent model.
Background
The satellite technology plays an irreplaceable important role in the fields of communication, remote sensing, navigation, military reconnaissance and the like, and is a hot research topic in the current industrial field. The core function of the satellite is realized by different satellite components embedded inside, and the normal and stable operation of the satellite components is the guarantee for realizing the provision of various services by the satellite. Normally stable operation of satellite assembly needs certain temperature range because the space environment that the satellite was located is unusual abominable to the satellite assembly has higher power density usually, and the satellite assembly can produce a large amount of heats during normal operating, if the heat can't normally distribute away, can seriously influence satellite assembly operating performance and running life. Therefore, when designing a satellite, the layout of the satellite components needs to be considered, so as to improve the reliability and service life of the satellite.
The key to solving the thermal design of the satellite component layout is to predict the steady-state temperature fields corresponding to different satellite component layouts. At present, numerical calculation methods such as a finite difference method, a finite element method and the like and a deep learning method are mainly adopted to calculate temperature fields corresponding to different satellite component layouts. However, the conventional numerical calculation method needs to consume a large amount of calculation resources when calculating the temperature field corresponding to the satellite component layout, and the higher the accuracy of the temperature field is, the larger the calculation resources are consumed. The deep learning method regards the prediction of the satellite component layout mode to the temperature field as a graph-to-graph regression task, and the end-to-end temperature field prediction is realized by constructing a training deep proxy model by using a graph-to-graph regression model. However, in the training process of the deep proxy model, a large number of labeled training samples are required, that is, the training samples of different satellite component layouts should have the corresponding temperature field distributions as labels, and the true value samples of the satellite component layouts and the corresponding temperature fields are difficult to obtain.
Disclosure of Invention
In order to solve part or all technical problems in the prior art, the invention provides a satellite component temperature field prediction method based on a teacher-student agent model.
The invention discloses a satellite component temperature field prediction method based on a teacher-student agent model, which comprises the following steps:
carrying out approximate description on the structure of the satellite assembly, and establishing a structural model of the satellite assembly layout;
generating a plurality of first training data and a plurality of second training data, wherein the first training data comprises a satellite component layout, and the second training data comprises a satellite component layout and a temperature field corresponding to the satellite component layout;
constructing a teacher-student depth agent model;
training a teacher model by using the second training data to fit a mapping relation between the satellite component layout and the temperature field, and predicting the temperature field corresponding to the satellite component layout in the first training data by using the trained teacher model;
training a student model by using the first training data, the temperature field corresponding to the first training data and the second training data to fit a mapping relation between the satellite component layout and the temperature field;
and using the trained student model as a temperature field prediction model to perform temperature field prediction of the satellite component layout.
In some alternative embodiments, the structural model of the satellite component layout is built in the following way:
setting a satellite component layout as a square area, dividing the square area into n 1 ×n 1 A grid, wherein one of the four edges of the square region is provided with a small hole with a set length as a heat dissipation hole, and the temperature of the heat dissipation hole region is constant and is T 0 ；
The satellite assemblies are set to be in a square structure, one satellite assembly can be placed on one grid position, and different satellite assemblies are placed on different grid positions according to the specific position of each satellite assembly in the satellite assembly layout.
In some optional embodiments, the generating a plurality of first training data and a plurality of second training data comprises:
randomly selecting a plurality of grids in the square area, placing satellite components on the selected grids to obtain a satellite component layout, and repeating the random generation process for multiple times to obtain N first training data comprising the satellite component layout;
randomly selecting a plurality of grids in the square area, placing satellite components on the selected grids to obtain a satellite component layout, repeating the random generation process for multiple times to obtain M satellite component layouts, determining a temperature field corresponding to each satellite component layout, and acquiring M second training data including the satellite component layouts and the temperature fields corresponding to the satellite component layouts, wherein N and M are positive integers, and N is greater than M.
In some optional embodiments, the generating a plurality of first training data and a plurality of second training data further comprises: normalizing the temperature field data in the second training data to map the temperature field to a [0,1] interval.
In some optional embodiments, the building the teacher-student depth proxy model comprises:
two different depth proxy models are selected as a teacher model and a student model respectively.
In some optional embodiments, the teacher model employs one of a feature pyramid network model and a word sense segmentation network model, and the student model employs the other of the two models.
In some optional embodiments, an L1 norm loss function is adopted as a loss function of the teacher model and the student model during training.
In some optional embodiments, the training a student model by using the first training data and the temperature field and the second training data corresponding to the first training data includes:
and firstly, training the student model by using the first training data and the temperature field corresponding to the first training data, and then training the student model by using the second training data through a Fine-tune technology.
In some optional embodiments, the method further comprises:
taking the trained student model as a trained teacher model, taking the untrained teacher model as an untrained student model, predicting the temperature field corresponding to the satellite component layout in the first training data again, and training the student model again;
and repeating the interchange training process of the student model and the teacher model for multiple times until the prediction precision of the trained student model reaches a set requirement, or the interchange times of the student model and the teacher model reaches a preset threshold value.
In some optional embodiments, the method further comprises:
generating a plurality of test data, wherein the test data comprises a satellite component layout and a temperature field corresponding to the satellite component layout;
and evaluating the prediction accuracy of the trained student model by using a plurality of test data.
The technical scheme of the invention has the following main advantages:
according to the satellite component temperature field prediction method based on the teacher-student agent model, the teacher-student deep agent model is constructed, the teacher model is used for guiding the training of the student model, a large number of training samples without marks and a small number of training samples with marks can be used for realizing the effective training of the deep agent model, the accurate prediction of the satellite component layout temperature field is realized, the calculation resources can be saved, the calculation workload is reduced, the design cost is reduced, and the design period is shortened.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flowchart of a satellite component temperature field prediction method based on a teacher-student proxy model according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a structural model of a satellite component layout according to an embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the specific embodiments of the present invention and the accompanying drawings. It is to be understood that the described embodiments are merely a few embodiments of the invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
The technical scheme provided by the embodiment of the invention is described in detail below with reference to the accompanying drawings.
As shown in fig. 1, an embodiment of the present invention provides a satellite component temperature field prediction method based on a teacher-student agent model, which includes the following steps:
s1, approximately describing the structure of a satellite assembly, and establishing a structural model of the satellite assembly layout;
s2, generating a plurality of first training data and a plurality of second training data, wherein the first training data comprise satellite component layouts, and the second training data comprise satellite component layouts and temperature fields corresponding to the satellite component layouts;
s3, constructing a teacher-student depth proxy model;
s4, training the teacher model by using the second training data to fit a mapping relation between the satellite component layout and the temperature field, and predicting the temperature field corresponding to the satellite component layout in the first training data by using the trained teacher model;
s5, training the student model by using the first training data, the temperature field corresponding to the first training data and the second training data to fit the mapping relation between the satellite component layout and the temperature field;
and S6, taking the trained student model as a temperature field prediction model to predict the temperature field of the satellite component layout.
The following specifically describes each step and principle of the satellite component temperature field prediction method based on the teacher-student proxy model according to an embodiment of the present invention.
S1, approximately describing the structure of a satellite assembly, and establishing a structural model of the satellite assembly layout;
specifically, based on the structural characteristics and layout characteristics of the satellite components, the structure of the satellite components is approximately described, and a structural model of the satellite component layout is established.
In one embodiment of the present invention, the structural features and layout features of the satellite components are:
the satellite component layout is a set region layout, a certain number of satellite components with different sizes and different powers are arranged in the set region, and the satellite components with different powers continuously generate heat when working, so that the satellite components can be regarded as heat sources with different powers;
the heat dissipation mode of the satellite component adopts a heat conduction mode to dissipate heat, and obeys a heat conduction steady state differential equation, and the corresponding boundary conditions comprise Dirichlet boundary conditions and Neumann boundary conditions; a heat dissipation hole with a certain size is formed in one side of a satellite assembly layout area, the heat dissipation hole is under Dirichlet boundary conditions, namely, the temperature at the heat dissipation hole is constant, and Neumann boundary conditions are adopted except for the heat dissipation hole, namely, heat insulation is achieved.
Because of the special environment of the satellite, the heat dissipation of the satellite component is carried out in a heat conduction mode, and a heat conduction steady state differential equation is obeyed; wherein the steady state differential equation for heat transfer is expressed as:
where x and y are coordinates of a point in a two-dimensional plane, T is the temperature at the point, k represents the heat transfer coefficient, and φ (x, y) is the heat source intensity at the point.
The mapping relation between the satellite assembly layout and the temperature field meets a heat conduction steady-state differential equation and boundary conditions; the boundary condition is a heat dissipation condition at the edge of the layout area, and is used as an initial value condition for providing a heat conduction steady-state differential equation in the heat conduction steady-state differential equation. Boundary conditions can be generally classified into three types, namely, a first type of boundary condition (Dirichlet boundary condition), a second type of boundary condition (Neumann boundary condition), and a third type of boundary condition (Robin boundary condition).
As shown in fig. 2, based on the structural features and the layout features of the satellite components, a structural model of the satellite component layout is established in the following manner:
setting the satellite component layout as a square area, and meshing the square area to divide the squareDivision of the shape region into n 1 ×n 1 The satellite component adopts a natural heat dissipation mode to dissipate heat, a small hole with a set length is arranged on one of four edges of the square layout area to serve as a heat dissipation hole, and the temperature of the heat dissipation hole area is constant and is T 0 ；
The satellite assemblies are set to be in a square structure, one satellite assembly can be placed on one grid position, and different satellite assemblies are placed on different grid positions according to the specific position of each satellite assembly in the satellite assembly layout.
In an embodiment of the present invention, for different satellite component layouts, the structural models corresponding to the different satellite component layouts can be obtained by using the above method.
S2, generating a plurality of first training data and a plurality of second training data, wherein the first training data comprise satellite component layouts, and the second training data comprise satellite component layouts and temperature fields corresponding to the satellite component layouts;
based on the established structural model of the satellite component layout, generating a plurality of first training data and a plurality of second training data specifically includes:
randomly selecting a plurality of grids in the square area, placing satellite components on the selected grids to obtain a satellite component layout, and repeating the random generation process for multiple times to obtain N first training data comprising the satellite component layout;
randomly selecting a plurality of grids in the square area, placing satellite components on the selected grids to obtain a satellite component layout, repeating the random generation process for multiple times to obtain M satellite component layouts, determining a temperature field corresponding to each satellite component layout, and acquiring M second training data comprising the satellite component layouts and the temperature fields corresponding to the satellite component layouts, wherein N and M are positive integers, and N is greater than M.
Optionally, in an embodiment of the present invention, 50000 satellite component layouts with a grid size of 200 × 200 may be randomly generated by a computer to serve as the first training data; 100 satellite component layouts with the grid size of 200 × 200 can be randomly generated through a computer, and temperature fields corresponding to the 100 satellite component layouts are solved through fenics software to serve as second training data.
Furthermore, the temperature field is accurately described, so that the training of a proxy model is facilitated, and the prediction accuracy of the temperature field is improved; generating the plurality of first training data and the number of second training data may further include: the temperature field data in the second training data is normalized to map the temperature field to the [0,1] interval.
S3, constructing a teacher-student deep agent model;
specifically, the construction of the teacher-student deep proxy model comprises the following steps:
two different depth proxy models are selected as a teacher model and a student model respectively.
By selecting two different depth proxy models as the teacher model and the student model, complementary information of the different depth proxy models can be effectively utilized, and therefore the prediction accuracy of the depth proxy models in temperature field prediction is improved.
Optionally, the teacher model adopts one of a feature pyramid network model and a word sense segmentation network model, and the student model adopts the other of the two models.
The Feature Pyramid network structure (FPN) model comprises three parts, namely a down-sampling process from top to bottom, an up-sampling process from bottom to top and a middle connection, and multi-scale features are extracted by the Feature Pyramid network structure model through a Feature extraction model and can be fused through the multi-scale features, so that the satellite component layout can be mapped into a corresponding layout temperature field.
The word sense segmentation network model can be, for example, a SegNet model, which includes a top-down sampling process and an un-aggressive based top-down sampling process, and can map the satellite component layout to the corresponding layout temperature field through the down-sampling feature extraction and the un-aggressive based up-sampling prediction process.
Optionally, in an embodiment of the present invention, the teacher model adopts a SegNet model, and the student model adopts a feature pyramid network structure model.
S4, training the teacher model by using the second training data to fit a mapping relation between the satellite component layout and the temperature field, and predicting the temperature field corresponding to the satellite component layout in the first training data by using the trained teacher model;
setting: the first training data is represented as
Wherein, the first and the second end of the pipe are connected with each other,
representing a satellite component layout, N representing a first training data quantity; the second training data is represented as
Wherein xi represents the satellite component layout,
representing the temperature field to which the satellite component layout xi corresponds, and M represents the second training data amount.
Specifically, second training data comprising a satellite component layout and a temperature field corresponding to the satellite component layout are utilized
Training the teacher model to enable the teacher model to learn a mapping relation between the satellite component layout and the temperature field; wherein the training process takes an L1 norm loss function as a loss function.
Second training data
The satellite component layout xi in (1) is an input sample of the teacher model, and the temperature field corresponding to the satellite component layout xi
The label corresponding to the input sample.
After the training of the teacher model is completed, the first training data is used
Satellite module layout
Inputting the trained teacher model, and predicting the first training data by using the trained teacher model
Middle satellite assembly layout
Corresponding temperature field
S5, training the student model by using the first training data, the temperature field corresponding to the first training data and the second training data to fit the mapping relation between the satellite component layout and the temperature field;
specifically, training the student model by using the first training data, the temperature field corresponding to the first training data, and the second training data to fit the mapping relationship between the satellite component layout and the temperature field includes:
using the first training data
Temperature field corresponding to predicted first training data
Training the student model to utilize information trained by the teacher model;
the student model trained based on the first training data and the second training data are reused
By Fine-tune technique to stuTraining a dent model; wherein, the training process takes the L1 norm loss function as the loss function.
In one embodiment of the invention, a teacher model is trained by using a small amount of labeled training samples, namely second training data, and then the teacher model obtained by training the labeled training samples is used for guiding the training of the student model, so that a large amount of unlabeled training samples, namely first training data, are effectively used for training the student model, the efficient training and accurate prediction of the deep agent model are realized, and the problems of large consumption of computing resources, high design cost, long design period and the like caused by the need of a large amount of labeled training samples in the deep agent model training process can be avoided.
Further, the temperature field prediction accuracy of the trained student model is considered to possibly fail to meet the requirement; therefore, in an embodiment of the present invention, in order to further improve the temperature field prediction accuracy of the finally obtained student model, the training of the student model by using the first training data and the temperature field and the second training data corresponding to the first training data to fit the mapping relationship between the satellite component layout and the temperature field may further include:
the trained student model is used as a trained student model, the untrained student model is used as an untrained student model, the temperature field corresponding to the satellite component layout in the first training data is predicted again, and the student model is trained again;
and repeating the interchange training process of the student model and the teacher model for multiple times until the prediction precision of the trained student model meets the set requirement, or the interchange times of the student model and the teacher model exceeds a preset threshold value.
Optionally, the prediction accuracy requirement and the preset threshold are determined according to actual conditions, for example, the preset threshold may be 10 times, 15 times or 20 times.
Through the alternate interchange training of the teacher model and the student model, the information of the first training data can be fully utilized, and the prediction precision of the depth agent model is improved.
Further, in an embodiment of the present invention, the method for predicting the temperature field of the satellite assembly may further include:
generating a plurality of test data, wherein the test data comprises a satellite component layout and a temperature field corresponding to the satellite component layout;
and evaluating the prediction accuracy of the trained student model by using a plurality of test data.
By generating a plurality of test data, the prediction accuracy of the trained student model can be tested to confirm whether the prediction accuracy of the trained student model meets the set requirement.
S6, taking the trained student model as a temperature field prediction model to predict a temperature field of the satellite component layout;
and after training of the student model is completed, loading and storing the trained student model, and inputting the satellite component layout of the temperature field to be calculated into the student model to obtain the temperature field corresponding to the satellite component layout.
If the temperature field is normalized in the training data generation process, the normalized temperature field is obtained through the student model prediction, and the actual satellite component layout temperature field can be obtained through processing the normalized temperature field through the normalization process inverse process.
Therefore, according to the satellite component temperature field prediction method based on the teacher-student proxy model provided by the embodiment of the invention, the teacher-student deep proxy model is constructed, the teacher model is used for guiding the training of the student model, a large number of training samples without marks and a small number of training samples with marks can be used for realizing the effective training of the deep proxy model, the accurate prediction of the satellite component layout temperature field is realized, the calculation resources can be saved, the calculation workload is reduced, the design cost is reduced, and the design period is shortened.
It is noted that, in this document, relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. In addition, "front", "rear", "left", "right", "upper" and "lower" in this document are referred to the placement states shown in the drawings.
Finally, it should be noted that: the above examples are only for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A satellite component temperature field prediction method based on a teacher-student agent model is characterized by comprising the following steps:
approximately describing the structure of the satellite assembly, and establishing a structural model of the satellite assembly layout;
generating a plurality of first training data and a plurality of second training data, wherein the first training data comprise satellite component layouts, and the second training data comprise satellite component layouts and temperature fields corresponding to the satellite component layouts;
constructing a teacher-student depth agent model;
training a teacher model by using the second training data to fit a mapping relation between the satellite component layout and the temperature field, and predicting the temperature field corresponding to the satellite component layout in the first training data by using the trained teacher model;
training a student model by using the first training data, the temperature field corresponding to the first training data and the second training data to fit a mapping relation between the satellite component layout and the temperature field;
and using the trained student model as a temperature field prediction model to predict the temperature field of the satellite component layout.
2. The satellite component temperature field prediction method based on the teacher-student agent model as claimed in claim 1, wherein the structural model of the satellite component layout is established in the following manner:
setting a satellite component layout as a square area, dividing the square area into n 1 ×n 1 Each grid is provided with a small hole with a set length as a heat dissipation hole on one of the four edges of the square area, and the temperature of the heat dissipation hole area is constant and is T 0 ；
The satellite assemblies are set to be in a square structure, one satellite assembly can be placed on one grid position, and different satellite assemblies are placed on different grid positions according to the specific position of each satellite assembly in the satellite assembly layout.
3. The method of claim 2, wherein the generating a plurality of first training data and a plurality of second training data comprises:
randomly selecting a plurality of grids in the square area, placing satellite components on the selected grids to obtain a satellite component layout, and repeating the random generation process for multiple times to obtain N first training data comprising the satellite component layout;
randomly selecting a plurality of grids in the square area, placing satellite components on the selected grids to obtain a satellite component layout, repeating the random generation process for multiple times to obtain M satellite component layouts, determining a temperature field corresponding to each satellite component layout, and acquiring M second training data including the satellite component layouts and the temperature fields corresponding to the satellite component layouts, wherein N and M are positive integers, and N is greater than M.
4. The method of predicting a temperature field of a satellite component based on a teacher-student agent model as claimed in claim 3, wherein the generating a plurality of first training data and a plurality of second training data further comprises: normalizing the temperature field data in the second training data to map the temperature field to a [0,1] interval.
5. The method for predicting the temperature field of the satellite component based on the teacher-student proxy model as claimed in claim 1, wherein the constructing the teacher-student depth proxy model comprises:
two different depth proxy models are selected as a teacher model and a student model respectively.
6. The method for predicting the temperature field of the satellite component based on the teacher-student agent model as claimed in claim 5, wherein the teacher model adopts one of a feature pyramid network model and a word sense division network model, and the student model adopts the other of the two models.
7. The satellite component temperature field prediction method based on the teacher-student surrogate model as claimed in claim 6, wherein an L1 norm loss function is adopted as a loss function of the teacher model and the student model during training.
8. The method for predicting the temperature field of the satellite component based on the teacher-student agent model as claimed in claim 1, wherein the training of the student model by using the first training data, the temperature field corresponding to the first training data, and the second training data comprises:
and firstly, training the student model by using the first training data and the temperature field corresponding to the first training data, and then training the student model by using the second training data through a Fine-tune technology.
9. The method for predicting the temperature field of a satellite component based on a teacher-student surrogate model according to any one of claims 1 to 8, further comprising:
using the trained student model as a trained student model, using the untrained student model as an untrained student model, re-predicting a temperature field corresponding to the satellite component layout in the first training data, and re-training the student model;
and repeating the interchange training process of the student model and the teacher model for multiple times until the prediction precision of the trained student model reaches a set requirement, or the interchange times of the student model and the teacher model reaches a preset threshold value.
10. The method for predicting the temperature field of a satellite component based on a teacher-student surrogate model according to any one of claims 1 to 9, further comprising:
generating a plurality of test data, wherein the test data comprises a satellite component layout and a temperature field corresponding to the satellite component layout;
and evaluating the prediction accuracy of the trained student model by using a plurality of test data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110398966.5A CN113221326B (en) | 2021-04-14 | 2021-04-14 | Satellite component temperature field prediction method based on teacher-student agent model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110398966.5A CN113221326B (en) | 2021-04-14 | 2021-04-14 | Satellite component temperature field prediction method based on teacher-student agent model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113221326A CN113221326A (en) | 2021-08-06 |
| CN113221326B true CN113221326B (en) | 2022-11-11 |
Family
ID=77087126
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110398966.5A Active CN113221326B (en) | 2021-04-14 | 2021-04-14 | Satellite component temperature field prediction method based on teacher-student agent model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113221326B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114417495A (en) * | 2021-12-15 | 2022-04-29 | 中国人民解放军军事科学院国防科技创新研究院 | Uncertainty-based satellite component layout reliability analysis method |
| CN114417499B (en) * | 2021-12-27 | 2023-02-28 | 中国人民解放军军事科学院国防科技创新研究院 | Neural network agent model-based component thermal layout optimization design method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111783331A (en) * | 2020-06-04 | 2020-10-16 | 中国人民解放军军事科学院国防科技创新研究院 | Satellite component temperature field determination method based on transfer learning |
| CN112257321A (en) * | 2020-10-27 | 2021-01-22 | 中国人民解放军军事科学院国防科技创新研究院 | Unsupervised learning-based satellite component temperature field determination method |
-
2021
- 2021-04-14 CN CN202110398966.5A patent/CN113221326B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111783331A (en) * | 2020-06-04 | 2020-10-16 | 中国人民解放军军事科学院国防科技创新研究院 | Satellite component temperature field determination method based on transfer learning |
| CN112257321A (en) * | 2020-10-27 | 2021-01-22 | 中国人民解放军军事科学院国防科技创新研究院 | Unsupervised learning-based satellite component temperature field determination method |
Non-Patent Citations (1)
| Title |
|---|
| 基于TWSVM和模糊的木质采暖地板蓄热温度预测模型;曹正彬等;《林业科学》;20181115(第11期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113221326A (en) | 2021-08-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10699043B2 (en) | Generating integrated circuit floorplans using neural networks | |
| US20200110913A1 (en) | RC Tool Accuracy Time Reduction | |
| CN111507032B (en) | Deep learning technology-based component layout optimization design method for predicting temperature distribution | |
| CN113221326B (en) | Satellite component temperature field prediction method based on teacher-student agent model | |
| CN113177290B (en) | Satellite component temperature field prediction method based on depth agent model normalization | |
| CN112257321B (en) | Unsupervised learning-based satellite component temperature field determination method | |
| CN114036787A (en) | Satellite component thermal layout temperature field prediction based on physical prior self-supervision learning | |
| CN112733275B (en) | Satellite component thermal layout temperature field prediction method based on semi-supervised learning | |
| CN113591215A (en) | Abnormal satellite component layout detection method based on uncertainty | |
| US11914931B2 (en) | Predicting on chip transient thermal response in a multi-chip system using an RNN-based predictor | |
| JP2011253279A (en) | Method, device, and program for generating model type | |
| Zhang et al. | Parallelized genetic identification of the thermal-electrochemical model for lithium-ion battery | |
| Li et al. | CAD/CAE associative features for cyclic fluid control effect modeling | |
| US11366947B2 (en) | Systems and methods for machine learning based fast static thermal solver | |
| CN114781207B (en) | Heat source layout temperature field prediction method based on uncertainty and semi-supervised learning | |
| Raghavan et al. | Implicit constraint handling for shape optimisation with pod-morphing | |
| CN110825952A (en) | Army comment detection method and system based on nested LSTM | |
| CN105989372A (en) | System, device and method for identifying significant and consumable-insensitive trace features | |
| Meng et al. | Classification of customer service tickets in power system based on character and word level semantic understanding | |
| US11669667B2 (en) | Automatic test pattern generation (ATPG) for parametric faults | |
| Yun et al. | Quality evaluation and satisfaction analysis of online learning of college students based on artificial intelligence | |
| US11354475B2 (en) | Systems and methods for accurate voltage impact on integrated timing simulation | |
| US11842130B1 (en) | Model-based simulation result predictor for circuit design | |
| Wang et al. | Mine water inrush source discrimination model based on KPCA-ISSA-KELM | |
| Jing et al. | CBLA_PM: an improved ann-based power consumption prediction algorithm for multi-type jobs on heterogeneous computing server |
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 |