CN114781207B - Heat source layout temperature field prediction method based on uncertainty and semi-supervised learning - Google Patents
Heat source layout temperature field prediction method based on uncertainty and semi-supervised learning Download PDFInfo
- Publication number
- CN114781207B CN114781207B CN202210318240.0A CN202210318240A CN114781207B CN 114781207 B CN114781207 B CN 114781207B CN 202210318240 A CN202210318240 A CN 202210318240A CN 114781207 B CN114781207 B CN 114781207B
- Authority
- CN
- China
- Prior art keywords
- training data
- component layout
- temperature field
- model
- uncertainty
- 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
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000012549 training Methods 0.000 claims abstract description 211
- 238000013135 deep learning Methods 0.000 claims abstract description 73
- 238000003062 neural network model Methods 0.000 claims abstract description 65
- 230000008569 process Effects 0.000 claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 14
- 230000006870 function Effects 0.000 claims description 15
- 239000011159 matrix material Substances 0.000 claims description 14
- 238000011478 gradient descent method Methods 0.000 claims description 6
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 30
- 238000013461 design Methods 0.000 description 16
- 238000005457 optimization Methods 0.000 description 15
- 238000004088 simulation Methods 0.000 description 8
- 230000009471 action Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000005315 distribution function Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
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
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- 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)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a heat source layout temperature field prediction method based on uncertainty and semi-supervised learning, which comprises the following steps: acquiring first training data and second training data; preprocessing training data; constructing a deep learning neural network model; randomly initializing a deep learning neural network model, training the model by using first training data, and repeating the random initialization and training processes for a plurality of times to obtain a plurality of trained models; determining a model to be updated and an auxiliary model; inputting the component layout of the second training data into an auxiliary model, and calculating a predicted temperature field and a predicted uncertainty corresponding to the component layout; training the model to be updated by using the first training data, the second training data and the corresponding prediction temperature field and prediction uncertainty; and predicting the temperature field by using the model to be updated. The method can obtain the deep learning agent model with higher prediction precision by using a small amount of labeled training data, and reduces the calculation cost.
Description
Technical Field
The invention relates to the technical field of component layout optimization design, in particular to a heat source layout temperature field prediction method based on uncertainty and semi-supervised learning.
Background
To perform different functions and to perform various tasks, satellites often incorporate a large number of components therein, which generate a large amount of heat during normal operation, and may be considered a heat source. In order to meet the overall function of a satellite, it is generally considered in determining the layout of components within the satellite whether the temperature field under the current layout of components meets the design requirements, for example, whether the maximum temperature is too high, whether the temperature at a specific location is too high, etc., so as to avoid the degradation of the reliability of the components or the failure of the components due to the excessive temperature. Therefore, how to determine the temperature field corresponding to the component layout is an important problem to be solved when optimizing the design of the component layout.
Two methods are currently mainly used for how to determine the temperature field to which the component layout corresponds. The first method is to model the component layout, and then to perform simulation analysis on the component layout by adopting numerical calculation methods such as a finite element method, a finite difference method, a finite volume method and the like so as to determine a temperature field corresponding to the component layout. The second method is to construct a deep learning agent model, train the agent model by using a large amount of sample data with temperature field labels, and then predict the temperature field of the component layout by using the trained deep learning agent model.
However, since the optimization design of the component layout is a repeated iterative process, the existing finite element-based simulation calculation method determines that the single simulation calculation efficiency is not too high, and in the iterative optimization process, a large amount of calculation resources and calculation time are consumed by analyzing and calculating the component layout temperature field for multiple times, the calculation cost is high, the optimization efficiency is low, and the required calculation resources and calculation time are gradually increased along with the complexity of the component layout and the calculation precision of the component layout temperature field. The existing method for predicting the temperature field of the satellite component layout by using the deep learning proxy model needs to train the proxy model by using a large amount of sample data with temperature field labels, and because the real data of the temperature field corresponding to the satellite component layout is difficult to acquire, each sample data with the temperature field labels needs to be acquired by a simulation experiment and numerical calculation mode, and also needs to consume more calculation resources and calculation time, and the problems of high calculation cost and low optimization design efficiency of the component layout still exist.
Disclosure of Invention
In order to solve part or all of the technical problems in the prior art, the invention provides a heat source layout temperature field prediction method based on uncertainty and semi-supervised learning.
The technical scheme of the invention is as follows:
there is provided a heat source layout temperature field prediction method based on uncertainty and semi-supervised learning, the method comprising:
acquiring first training data and second training data, wherein the first training data comprises component layout and a corresponding temperature field thereof, and the second training data comprises the component layout;
preprocessing the first training data and the second training data;
constructing a deep learning neural network model;
randomly initializing a deep learning neural network model, training the deep learning neural network model by using the first training data to obtain a trained deep learning neural network model, and repeating the random initialization and training processes for a plurality of times to obtain a plurality of trained deep learning neural network models;
determining a model to be updated and an auxiliary model, wherein the model to be updated is one deep learning neural network model with optimal prediction performance in the plurality of trained deep learning neural network models, and the auxiliary model is other models except the model to be updated in the plurality of trained deep learning neural network models;
inputting the component layout of the second training data into the auxiliary model, and calculating a predicted temperature field and a predicted uncertainty corresponding to the input component layout;
training the model to be updated by utilizing the first training data, the second training data and the corresponding predicted temperature field and the prediction uncertainty thereof to update model parameters;
and predicting the component layout temperature field by using the trained model to be updated.
In some possible implementations, the acquiring the first training data and the second training data includes:
grid division is carried out on the component layout area, a corresponding number of grid placement components are selected randomly according to the number of the components to obtain a component layout, a temperature field corresponding to the component layout is simulated and calculated by using a finite element method to obtain first training data comprising the component layout and the temperature field corresponding to the component layout, and the random acquisition process is repeated for a plurality of times until the first training data of a first preset number is obtained;
and carrying out grid division on the component layout area, randomly selecting a corresponding number of grid placement components according to the number of the components to obtain a component layout, obtaining second training data comprising the component layout, and repeating the random acquisition process for a plurality of times until a second preset number of second training data are obtained.
In some possible implementations, the settings are: dividing a component layout area into M 1 ×M 2 Carrying out component layout on the grid;
the preprocessing the first training data and the second training data includes:
the dimension for the component layout of the first training data and the component layout of the second training data is M 1 ×M 2 The matrix elements corresponding to the grid positions with components are component powers and the matrix elements corresponding to the grid positions without components are 0.
In some possible implementations, the deep learning neural network model is a U-net convolutional neural network.
In some possible implementations, the training the deep learning neural network model using the first training data includes:
and taking the component layout of the first training data as input, taking a temperature field corresponding to the input component layout as output, and training the deep learning neural network model by a gradient descent method.
In some possible implementations, the predicted temperature field for the component layout is calculated using the following formula;
wherein,representing component clothThe prediction temperature field corresponding to office, Q represents the number of auxiliary models, T q Representing the predicted temperature field output by the q-th auxiliary model corresponding to the component layout.
In some possible implementations, the prediction uncertainty for the component layout is calculated using the following formula;
wherein T is std Representing the prediction uncertainty corresponding to the component layout.
In some possible implementations, the training the model to be updated with the first training data, the second training data, and their corresponding predicted temperature fields and predicted uncertainties to update model parameters includes:
and constructing a loss function by taking the component layout of the first training data and the component layout of the second training data as inputs, and training the model to be updated by a gradient descent method based on the constructed loss function according to the temperature field corresponding to the component layout of the first training data, the predicted temperature field and the predicted uncertainty corresponding to the component layout of the second training data and the predicted temperature field output by the model to be updated.
In some possible implementations, when training the model to be updated, the loss function is set to:
wherein alpha and beta represent preset weight coefficients, omega represents the number of elements of the temperature field, N represents the number of first training data input into the model to be updated, sum represents summing all elements in,and->Component layout and temperature field representing the i-th first training data after preprocessing, ++>A predicted temperature field corresponding to the component layout representing the ith first training data output by the model to be updated, N' representing the number of second training data input into the model to be updated, +.>Component layout representing the j-th second training data after preprocessing,/for the second training data>Representing a temperature field predicted value, W, corresponding to a component layout of the jth second training data output by the model to be updated j An uncertainty weight matrix representing a component layout correspondence of a jth second training data, +.>Representing a predicted temperature field corresponding to the component layout of the jth second training data.
In some possible implementations, an uncertainty weight matrix W j Determined using the following formula:
wherein T is std_j Representing the predicted uncertainty corresponding to the component layout of the jth second training data, and gamma representing the uncertainty threshold.
The technical scheme of the invention has the main advantages that:
the heat source layout temperature field prediction method based on uncertainty and semi-supervised learning can obtain the deep learning agent model with higher prediction precision by using a small amount of labeled training data, obviously reduces the demand of model training on labeled training data, effectively reduces the calculation time and calculation resources required for acquiring the training data, reduces the component layout optimization design cost, and improves the optimization efficiency.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a heat source layout temperature field prediction method based on uncertainty and semi-supervised learning according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a training process of a model to be updated according to an embodiment of the present 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 specific embodiments of the present invention and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The following describes in detail the technical scheme provided by the embodiment of the invention with reference to the accompanying drawings.
Referring to fig. 1, an embodiment of the present invention provides a heat source layout temperature field prediction method based on uncertainty and semi-supervised learning, the method comprising the steps of:
step S1, acquiring first training data and second training data, wherein the first training data comprises component layout and a corresponding temperature field thereof, and the second training data comprises the component layout;
step S2, preprocessing the first training data and the second training data;
s3, constructing a deep learning neural network model;
step S4, randomly initializing a deep learning neural network model, training the deep learning neural network model by using first training data to obtain a trained deep learning neural network model, and repeating the random initialization and training processes for a plurality of times to obtain a plurality of trained deep learning neural network models;
s5, determining a model to be updated and an auxiliary model, wherein the model to be updated is one deep learning neural network model with optimal prediction performance in a plurality of trained deep learning neural network models, and the auxiliary model is other models except the model to be updated in the plurality of trained deep learning neural network models;
s6, inputting the component layout of the second training data into an auxiliary model, and calculating a predicted temperature field and a predicted uncertainty corresponding to the input component layout;
step S7, training the model to be updated by using the first training data, the second training data and the corresponding prediction temperature field and prediction uncertainty thereof to update model parameters;
and S8, predicting a component layout temperature field by using the trained model to be updated.
According to the heat source layout temperature field prediction method based on uncertainty and semi-supervised learning, firstly, training of a plurality of deep learning neural network models is carried out by utilizing labeled training data, then integrated prediction of label-free training data is carried out by utilizing the trained plurality of deep learning neural network models, prediction temperature fields and prediction uncertainty corresponding to the label-free training data are determined, the model with optimal prediction performance in the trained plurality of deep learning neural network models is trained again by utilizing the labeled training data, the label-free training data and the corresponding prediction temperature fields and the prediction uncertainty thereof to update model parameters, a deep learning proxy model with higher prediction accuracy can be obtained by utilizing a small amount of labeled training data, the requirement of model training on the labeled training data is remarkably reduced, the calculation time and calculation resources required for acquiring the training data are effectively reduced, the layout optimization design cost of a component is reduced, and the optimization efficiency is improved.
In the component layout optimization design problem considering the temperature field index, each component can be simplified into one heat source. The component layout optimization design may be considered a heat source layout optimization design. When a certain number of components need to be placed in a certain layout area, the purpose of the component layout optimization design is to enhance the heat conduction efficiency of the layout area by optimizing the positions of the components to minimize the highest temperature of the layout area. Considering the heat source layout optimization design problem in a two-dimensional plane, the mathematical model can be expressed as:
wherein X represents a layout scheme of the heat source, (X) i ,y i ) Indicating the position coordinates of the ith heat source, N s The number of heat sources is represented, T represents the temperature field of the layout area, k represents the thermal conductivity of the layout area, (x, y) represents the position coordinates of any point of the layout area, phi (x, y) represents the intensity distribution function of the heat sources, T 0 Representing the temperature value at the isothermal boundary, n representing the normal direction at the boundary, h representing the surface heat transfer coefficient between the object and the surrounding fluid at the boundary.
The intensity distribution function phi (x, y) of the heat source is determined by the location of the heat source, specifically expressed as:
wherein phi is i Represents the intensity of the ith heat source Γ i Indicating the layout area covered by the heat source.
The following is a description of the steps and principles of the heat source layout temperature field prediction method based on uncertainty and semi-supervised learning according to an embodiment of the present invention, by taking the optimization design of the component layout in the two-dimensional plane as an example:
step S1, acquiring first training data and second training data.
In an embodiment of the present invention, the first training data and the second training data may be obtained by:
grid division is carried out on the component layout area, and a corresponding number of grid placement components are randomly selected according to the number of the components to obtain a component layout X m Temperature field T corresponding to component layout is simulated and calculated by using finite element method m Obtaining first training data (X) including a component layout and a temperature field corresponding to the component layout m ,T m ) The random acquisition process is repeated a plurality of times until a first preset number M of first training data { (X) is obtained m ,T m )|m＝1,2,…,M}；
Grid division is carried out on the component layout area, and a corresponding number of grid placement components are randomly selected according to the number of the components to obtain a component layout X' m′ A second training data (X 'including the layout of the components is obtained' m′ ) Repeating the random acquisition process a plurality of times until a second preset number M ' of second training data { (X ' is obtained ' m′ )|m′＝1,2,…,M′}。
The specific number of the first training data and the specific number of the second training data can be set according to the training efficiency and the accuracy requirement, for example, M can be 1000-5000 or other values, and M' can be 20000-50000 or other values. In order to improve the prediction accuracy of the finally obtained model and reduce the calculation cost, the number of the second training data is larger than that of the first training data.
In an embodiment of the present invention, the component layout area is determined according to the actual layout requirement, and the number of grids divided by the component layout area is also determined according to the actual situation. For example, when the component layout area is a square component layout area, the square component layout area may be divided into M 1 ×M 2 Grids, each of which can be provided with at most one heat source component, wherein the number of the heat source components is N s When the number of grids for placing the heat source components is selected to be N s And each.
Wherein the grid of component layout area divisionThe number may be determined according to the resolution and accuracy required by the component layout design, and when the required resolution and accuracy are higher, the larger the number of divided grids, i.e., the finer the grid division. For example, M can be 1 ×M 2 Set to 200 x 200.
Further, when calculating the temperature field corresponding to the component layout by using the finite element method, the component layout area can be divided into N 1 ×N 2 And performing simulation calculation on the finite element calculation grids to obtain a temperature field corresponding to the component layout. The specific values of the finite element calculation grids can be determined according to the resolution and the precision required by the layout design of the component, and when the required resolution and precision are higher, the more the finite element calculation grids are divided when the simulation calculation is performed, but correspondingly, the more the time and the resources required by the simulation calculation are. Optionally, to facilitate data processing, N 1 ×N 2 Can be equal to M 1 ×M 2 。
And step S2, preprocessing the first training data and the second training data.
Specifically, to facilitate training of a subsequently constructed deep learning neural network model with training data to divide a component layout region into M 1 ×M 2 For example, the grid performs component layout, and preprocesses the first training data and the second training data, including:
the dimension for the component layout of the first training data and the component layout of the second training data is M 1 ×M 2 The matrix elements corresponding to the grid positions with components are component powers and the matrix elements corresponding to the grid positions without components are 0.
Further, in order to facilitate data processing, the component layout area is divided into N 1 ×N 2 For example, the temperature field simulation calculation is performed by the finite element calculation grid, and the preprocessing of the first training data and the second training data may further include:
dimension N 1 ×N 2 Temperature field matrix T of (2) m The values of (1) are normalized to ensure that matrix elements all belong to [0,1 ]]。
And S3, constructing a deep learning neural network model.
In one embodiment of the invention, a U-Net convolutional neural network can be used as a deep learning neural network model to fit the mapping relationship between the component layout and the temperature field. By training the deep learning neural network model by using the obtained training data, the deep learning neural network model can learn the physical rule in the training data, so that a neural network with stronger generalization capability is obtained, and the rapid prediction of the temperature field corresponding to the component layout can be realized. In essence, the deep learning neural network model is a proxy model.
And S4, randomly initializing the deep learning neural network model, training the deep learning neural network model by using the first training data to obtain a trained deep learning neural network model, and repeating the random initialization and training processes for a plurality of times to obtain a plurality of trained deep learning neural network models.
In order to obtain a proxy model with higher prediction accuracy by using a small amount of labeled training data, in an embodiment of the present invention, a plurality of deep learning proxy models, that is, a plurality of trained deep learning neural network models, are obtained by using labeled training data through a random initialization and independent training method.
Specifically, when a deep learning agent model is obtained, the deep learning neural network model is randomly initialized to determine initial parameters of the model, and then the initialized deep learning neural network model is trained by utilizing first training data, so that a trained deep learning neural network model is obtained.
In an embodiment of the present invention, training a deep learning neural network model using first training data includes:
taking the component layout of the first training data as input, taking a temperature field corresponding to the input component layout as output, and training the deep learning neural network model by a gradient descent method.
Alternatively, when training the deep neural network model, the corresponding loss function may be set according to the actual situation, for example, an average absolute error (Mean Absolute Error, MAE) may be used as the loss function.
The more the number of the obtained deep learning neural network models is, the higher the prediction precision is when the temperature field prediction is performed by using a plurality of integrated deep learning neural network models, but the longer the corresponding training time and training period are. In an embodiment of the present invention, the number of deep learning neural network models may be set according to the prediction accuracy and training time of the actual requirements. For example, 4 to 8 may be provided.
And S5, determining a model to be updated and an auxiliary model, wherein the model to be updated is one deep learning neural network model with optimal prediction performance in the plurality of trained deep learning neural network models, and the auxiliary model is other models except the model to be updated in the plurality of trained deep learning neural network models.
In an embodiment of the present invention, the prediction performance of the deep learning neural network model may be determined by calculating the average absolute error (Mean Absolute Error, MAE) corresponding to the model, and when the MAE is lower, the prediction performance of the model is better.
Specifically, the best predictive performance among the plurality of trained deep-learning neural network models may be determined using the following:
selecting a certain amount of first training data, inputting the selected component layout of the first training data into a trained deep learning neural network model to obtain a predicted temperature field corresponding to the component layout output by the deep learning neural network model, calculating an average absolute error according to the temperature field of the first training data and the predicted temperature field output by the deep learning neural network model, and determining the deep learning neural network with the minimum average absolute error as a deep learning neural network model with optimal prediction performance.
And S6, inputting the component layout of the second training data into an auxiliary model, and calculating a predicted temperature field and a predicted uncertainty corresponding to the input component layout.
Referring to fig. 2, in fig. 2, the number of deep learning neural network models is 5, including the number of the uiet x, the number of the uiet 2, the number of the uiet 3, the number of the uiet 4 and the number of the auxiliary models are 4, including the number of the uiet 2, the number of the uiet 3, the number of the uiet 4 and the number of the uiet 5. In an embodiment of the present invention, the component layout of the second training data is input into each auxiliary model to obtain the temperature field output by each auxiliary model, and the predicted temperature field and the prediction uncertainty corresponding to the component layout are calculated according to the temperature fields output by all the auxiliary models.
In an embodiment of the present invention, the predicted temperature field corresponding to the component layout may be calculated and determined by using the following formula:
wherein,representing the predicted temperature field corresponding to the component layout, Q represents the number of auxiliary models, T q Representing the predicted temperature field output by the q-th auxiliary model corresponding to the component layout.
In an embodiment of the present invention, the prediction uncertainty corresponding to the component layout may be determined by the following formula calculation;
wherein T is std Representing the prediction uncertainty corresponding to the component layout.
According to the above calculation formula, by predicting the temperature field of the component layout of each second training data using a plurality of auxiliary models, the predicted temperature field and the prediction uncertainty corresponding to the component layout of each second training data can be determined.
In an embodiment of the present invention, the prediction accuracy of the temperature field can be further improved by performing the integrated prediction of the component layout of the second training data using a plurality of auxiliary models. And the degree of the predicted temperature field can be measured according to the obtained prediction uncertainty, and the smaller the prediction uncertainty is, the higher the reliability of the predicted temperature field is.
And S7, training the model to be updated by using the first training data, the second training data and the corresponding prediction temperature field and prediction uncertainty thereof so as to update model parameters.
In an embodiment of the invention, the model to be updated is trained by using the first training data, the second training data, the corresponding prediction temperature field and the prediction uncertainty, so that the problem of less quantity of the first training data can be solved, and the prediction precision of the model to be updated is further improved.
Referring to fig. 2, in an embodiment of the present invention, training a model to be updated to update model parameters by using first training data, second training data and corresponding predicted temperature fields and predicted uncertainties may include:
and constructing a loss function by taking the component layout of the first training data and the component layout of the second training data as inputs, and training the model to be updated by a gradient descent method based on the constructed loss function according to the temperature field corresponding to the component layout of the first training data, the prediction temperature field and the prediction uncertainty corresponding to the component layout of the second training data and the prediction temperature field output by the model to be updated.
Alternatively, when training the model to be updated, the loss function may be:
wherein alpha and beta represent preset weight coefficients,for the loss function portion corresponding to the first training data, Ω represents the number of elements of the temperature field, N represents the number of first training data input into the model to be updated, sum represents summing all elements in the sum,>and->Component layout and temperature field representing the i-th first training data after preprocessing, ++>Predicted temperature field corresponding to component layout representing ith first training data output by model to be updated, +.>For the loss function portion corresponding to the second training data, N' represents the number of second training data input to the model to be updated, < >>Component layout representing the j-th second training data after preprocessing,/for the second training data>Representing a temperature field predicted value, W, corresponding to a component layout of the jth second training data output by the model to be updated j An uncertainty weight matrix representing a component layout correspondence of a jth second training data, +.>Representing a predicted temperature field corresponding to the component layout of the jth second training data.
Uncertainty weight matrix W j The following formula may be used for determination:
wherein T is std_j Representing the predicted uncertainty corresponding to the component layout of the jth second training data, and gamma representing the uncertainty threshold.
Wherein the j th secondPredicted temperature field corresponding to component layout of training dataBy means of the above-mentioned predicted temperature field +.>Is determined by the calculation formula of the (j) th second training data, and the prediction uncertainty T corresponding to the component layout of the (j) th second training data std_j Using the above-described prediction uncertainty T std The uncertainty threshold gamma can be set in a practical case.
And S8, predicting a component layout temperature field by using the trained model to be updated.
Specifically, after training of the model to be updated is completed, the component layout of the temperature field to be calculated is input into the model to be updated, and then a temperature field predicted value corresponding to the component layout can be obtained, so that the optimal design of the component layout is assisted.
It should be noted that in this document, relational terms such as "first" and "second" and the like are 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. Moreover, 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 this context, "front", "rear", "left", "right", "upper" and "lower" are referred to with respect to the placement state shown in the drawings.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting thereof; although the 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 scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (8)
1. A heat source layout temperature field prediction method based on uncertainty and semi-supervised learning, comprising:
acquiring first training data and second training data, wherein the first training data comprises component layout and a corresponding temperature field thereof, and the second training data comprises the component layout;
preprocessing the first training data and the second training data;
constructing a deep learning neural network model;
randomly initializing a deep learning neural network model, training the deep learning neural network model by using the first training data to obtain a trained deep learning neural network model, and repeating the random initialization and training processes for a plurality of times to obtain a plurality of trained deep learning neural network models;
determining a model to be updated and an auxiliary model, wherein the model to be updated is one deep learning neural network model with optimal prediction performance in the plurality of trained deep learning neural network models, and the auxiliary model is other models except the model to be updated in the plurality of trained deep learning neural network models;
inputting the component layout of the second training data into the auxiliary model, and calculating a predicted temperature field and a predicted uncertainty corresponding to the input component layout;
training the model to be updated by utilizing the first training data, the second training data and the corresponding predicted temperature field and the prediction uncertainty thereof to update model parameters;
predicting a component layout temperature field by using the trained model to be updated;
the acquiring the first training data and the second training data includes:
grid division is carried out on the component layout area, a corresponding number of grid placement components are selected randomly according to the number of the components to obtain a component layout, a temperature field corresponding to the component layout is simulated and calculated by using a finite element method to obtain first training data comprising the component layout and the temperature field corresponding to the component layout, and the random acquisition process is repeated for a plurality of times until the first training data of a first preset number is obtained;
performing grid division on the component layout area, randomly selecting a corresponding number of grid placement components according to the number of the components to obtain a component layout, obtaining second training data comprising the component layout, and repeating the random acquisition process for a plurality of times until a second preset number of second training data is obtained;
setting: dividing a component layout area into M 1 ×M 2 Carrying out component layout on the grid;
the preprocessing the first training data and the second training data includes:
the dimension for the component layout of the first training data and the component layout of the second training data is M 1 ×M 2 The matrix elements corresponding to the grid positions with components are component powers and the matrix elements corresponding to the grid positions without components are 0.
2. The heat source layout temperature field prediction method based on uncertainty and semi-supervised learning of claim 1, wherein the deep learning neural network model is a U-net convolutional neural network.
3. The method of uncertainty and semi-supervised learning based heat source layout temperature field prediction as set forth in claim 1, wherein said training a deep learning neural network model with said first training data includes:
and taking the component layout of the first training data as input, taking a temperature field corresponding to the input component layout as output, and training the deep learning neural network model by a gradient descent method.
4. The heat source layout temperature field prediction method based on uncertainty and semi-supervised learning of claim 1, wherein the predicted temperature field for the component layout is calculated using the following formula;
wherein,representing the predicted temperature field corresponding to the component layout, Q represents the number of auxiliary models, T q Representing the predicted temperature field output by the q-th auxiliary model corresponding to the component layout.
5. The heat source layout temperature field prediction method based on uncertainty and semi-supervised learning of claim 4, wherein the prediction uncertainty for the component layout is calculated using the following formula;
wherein T is std Representing the prediction uncertainty corresponding to the component layout.
6. The heat source layout temperature field prediction method based on uncertainty and semi-supervised learning as set forth in any one of claims 1-5, wherein the training the model to be updated with the first training data, the second training data, and their corresponding predicted temperature fields and predicted uncertainties to update model parameters includes:
and constructing a loss function by taking the component layout of the first training data and the component layout of the second training data as inputs, and training the model to be updated by a gradient descent method based on the constructed loss function according to the temperature field corresponding to the component layout of the first training data, the predicted temperature field and the predicted uncertainty corresponding to the component layout of the second training data and the predicted temperature field output by the model to be updated.
7. The heat source layout temperature field prediction method based on uncertainty and semi-supervised learning of claim 6, wherein when training the model to be updated, a loss function is set to:
wherein alpha and beta represent preset weight coefficients, omega represents the number of elements of the temperature field, N represents the number of first training data input into the model to be updated, sum represents summing all elements in,and->Component layout and temperature field representing the i-th first training data after preprocessing, ++>A predicted temperature field corresponding to the component layout representing the ith first training data output by the model to be updated, N' representing the number of second training data input into the model to be updated, +.>Component layout representing the j-th second training data after preprocessing,/for the second training data>Component layout correspondence representing jth second training data of model output to be updatedIs a temperature field predicted value of W j An uncertainty weight matrix representing a component layout correspondence of a jth second training data, +.>Representing a predicted temperature field corresponding to the component layout of the jth second training data.
8. The heat source layout temperature field prediction method based on uncertainty and semi-supervised learning as set forth in claim 7, wherein the uncertainty weight matrix W j Determined using the following formula:
wherein T is std_j Representing the predicted uncertainty corresponding to the component layout of the jth second training data, and gamma representing the uncertainty threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210318240.0A CN114781207B (en) | 2022-03-29 | 2022-03-29 | Heat source layout temperature field prediction method based on uncertainty and semi-supervised learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210318240.0A CN114781207B (en) | 2022-03-29 | 2022-03-29 | Heat source layout temperature field prediction method based on uncertainty and semi-supervised learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114781207A CN114781207A (en) | 2022-07-22 |
| CN114781207B true CN114781207B (en) | 2024-04-12 |
Family
ID=82425779
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210318240.0A Active CN114781207B (en) | 2022-03-29 | 2022-03-29 | Heat source layout temperature field prediction method based on uncertainty and semi-supervised learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114781207B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115730509B (en) * | 2022-09-20 | 2024-08-20 | 中国人民解放军军事科学院国防科技创新研究院 | Spacecraft cabin temperature field reconstruction task research benchmark method based on machine learning |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111507032A (en) * | 2020-03-31 | 2020-08-07 | 中国人民解放军军事科学院国防科技创新研究院 | Deep learning technology-based component layout optimization design method for predicting temperature distribution |
| CN112733275A (en) * | 2021-01-19 | 2021-04-30 | 中国人民解放军军事科学院国防科技创新研究院 | Satellite assembly thermal layout temperature field prediction method based on semi-supervised learning |
| CN114036787A (en) * | 2021-10-09 | 2022-02-11 | 中国人民解放军军事科学院国防科技创新研究院 | Satellite component thermal layout temperature field prediction based on physical prior self-supervision learning |
| WO2022042002A1 (en) * | 2020-08-31 | 2022-03-03 | 华为技术有限公司 | Training method for semi-supervised learning model, image processing method, and device |
-
2022
- 2022-03-29 CN CN202210318240.0A patent/CN114781207B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111507032A (en) * | 2020-03-31 | 2020-08-07 | 中国人民解放军军事科学院国防科技创新研究院 | Deep learning technology-based component layout optimization design method for predicting temperature distribution |
| WO2022042002A1 (en) * | 2020-08-31 | 2022-03-03 | 华为技术有限公司 | Training method for semi-supervised learning model, image processing method, and device |
| CN112733275A (en) * | 2021-01-19 | 2021-04-30 | 中国人民解放军军事科学院国防科技创新研究院 | Satellite assembly thermal layout temperature field prediction method based on semi-supervised learning |
| CN114036787A (en) * | 2021-10-09 | 2022-02-11 | 中国人民解放军军事科学院国防科技创新研究院 | Satellite component thermal layout temperature field prediction based on physical prior self-supervision learning |
Non-Patent Citations (1)
| Title |
|---|
| 深度信念网络优化BP神经网络的交通流预测模型;孔繁辉;李健;;管理评论;20200331(03);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114781207A (en) | 2022-07-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US12086516B2 (en) | Generating integrated circuit floorplans using neural networks | |
| CN111507032B (en) | Deep learning technology-based component layout optimization design method for predicting temperature distribution | |
| 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 | |
| CN112733275B (en) | Satellite component thermal layout temperature field prediction method based on semi-supervised learning | |
| CN113792478B (en) | Heat source layout temperature field reliability analysis method based on integrated depth proxy model | |
| CN113591215B (en) | Abnormal satellite component layout detection method based on uncertainty | |
| CN114781207B (en) | Heat source layout temperature field prediction method based on uncertainty and semi-supervised learning | |
| CN115809705B (en) | Fluid dynamics computing system based on quantum computing and quantum computer | |
| CN114036787A (en) | Satellite component thermal layout temperature field prediction based on physical prior self-supervision learning | |
| CN112464567A (en) | Intelligent data assimilation method based on variational and assimilative framework | |
| CN114692347A (en) | Temperature field proxy model construction method based on neural network architecture search technology | |
| CN116822214A (en) | Multi-credibility data fusion method based on heteroscedastic Gaussian process regression | |
| CN113609763B (en) | Uncertainty-based satellite component layout temperature field prediction method | |
| CN114417653B (en) | Uncertainty-based satellite component circuit board thermal reliability analysis method | |
| CN114548526A (en) | Satellite component layout temperature field prediction method based on physical prior neural network | |
| CN114692348B (en) | Component layout temperature field prediction method based on multi-fidelity deep learning agent model | |
| CN113139334A (en) | Simulation optimization method based on bee colony | |
| CN116227339A (en) | Satellite cabin temperature field reconstruction method based on model uncertainty and semi-supervised learning | |
| CN118551668B (en) | High-speed aircraft characterization generalization method and device based on machine learning | |
| CN115577618B (en) | Construction method and prediction method of high-pressure converter valve hall environmental factor prediction model | |
| Rao et al. | Parallel solution of DDDAS variational inference problems | |
| CN117474128B (en) | Distributed online machine learning model construction method based on multi-cloud server | |
| US20230162010A1 (en) | Synthesizing Zero-Loss Low-Power Approximate DNN Accelerators With Large-Scale Search | |
| CN114880897B (en) | Unsteady flow field grid self-adaptive analysis method based on machine learning |
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 |