Thermophysical parameter identification method, system, medium and equipment
By selecting appropriate identification methods and optimization algorithms in the thermal property parameter recognition method, the optimization range of thermal property parameters is narrowed, and the problems of large errors and low efficiency in the prior art are solved, and higher recognition accuracy and efficiency are achieved.
Patent Information
- Application Number
- CN202411705130.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-26
- Publication Date
- 2025-03-14
AI Technical Summary
The existing thermal property parameter identification methods have problems with large errors and low efficiency, especially when the thermal property parameter range is large.
By selecting different recognition methods, including deep learning models and traditional analytical solution fitting algorithms, the optimization range of thermal physical properties parameters is narrowed, and the optimization algorithm is used to optimize and identify small interval parameters to obtain the optimal value of thermal physical properties parameters.
It effectively narrows the optimization range of thermal properties parameters and improves the accuracy and efficiency of parameter identification.
Smart Images
Figure CN119623278A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of thermophysical property parameter measurement, and in particular to a thermophysical property parameter identification method, system, medium and equipment. Background Art
[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.
[0003] Accurate measurement of thermophysical parameters can change the mode and rate of heat transfer, and is highly valued in the fields of architecture, electronic chips, aerospace, etc. The traditional analytical solution of the current thermophysical property measurement method is obtained based on an ideal model, but there are differences between the ideal model and the actual model, resulting in errors in the thermophysical parameters obtained by fitting the analytical solution and the test data. Therefore, existing research focuses on correcting the analytical solution, but due to the approximate treatment of the correction term, the error cannot be completely eliminated. Some other studies use existing optimization models to identify thermophysical parameters. However, when using the optimization model to identify the thermophysical parameters, the range of the thermophysical parameters is large, resulting in large parameter identification errors and low efficiency.
[0004] In summary, the existing thermophysical property parameter identification methods still have shortcomings. How to effectively narrow the range of thermophysical property parameters and thus improve the accuracy of thermophysical property parameter identification has become a technical problem that needs to be solved in the existing technology. Summary of the invention
[0005] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method, system, medium and equipment for identifying thermal physical property parameters, which can select different identification methods according to different parameter conditions, effectively narrow the optimization range of thermal physical property parameters, and improve the identification accuracy of thermal physical property parameters.
[0006] In order to achieve the above object, the present invention is implemented through the following technical solutions:
[0007] The first aspect of the present invention provides a method for identifying thermophysical property parameters, comprising the following steps:
[0008] Obtaining the thermophysical property parameters to be identified, and selecting a first range reduction model according to the known number of samples;
[0009] Using the first range reduction model to reduce the range of the thermophysical property parameters, a first range of the thermophysical property parameters is obtained;
[0010] A second range reduction model is constructed, and the first range of the thermophysical property parameter is reduced twice by a single parameter range in steps using the second range reduction model to obtain a second range of the thermophysical property parameter;
[0011] The optimization algorithm is used to perform small interval parameter optimization identification on the second range of the thermophysical property parameters to obtain the optimal value of the thermophysical property parameters.
[0012] Furthermore, the specific steps of selecting the first range reduction model according to the known sample size are:
[0013] Set a threshold with a known number of samples;
[0014] If the number of known samples exceeds the threshold, the deep learning model is used as the first range reduction model;
[0015] If the number of known samples does not exceed the threshold, the traditional analytical solution fitting algorithm is used to construct the first range reduction model.
[0016] Furthermore, when the deep learning model is used as the first range-narrowing model, the training set and the test set are divided using known samples, the real thermophysical properties are labeled, the deep learning model is trained using the training set, and the trained deep learning model is tested using the test set to obtain the optimal value of the thermal properties of the material, taking into account the training error, and obtaining the first range of the thermophysical property parameters.
[0017] Furthermore, when the first range reduction model is constructed using the traditional analytical solution fitting algorithm, the traditional analytical solution fitting algorithm of the thermophysical property measurement method is used to obtain the optimal value of the material's thermophysical property, and the first range of the thermophysical property parameters is obtained by taking the fitting error into consideration.
[0018] Furthermore, the specific steps of constructing the second scope reduction model are:
[0019] Construct physical and mathematical heat conduction models of thermophysical property parameters based on thermophysical property measurement methods;
[0020] According to the physical and mathematical heat conduction model, an optimized numerical model of the inverse problem of heat conduction of thermal physical property parameters is established as the second range reduction model.
[0021] Furthermore, the specific steps of using the second range reduction model to perform a second step-by-step reduction of the first range of the thermophysical property parameters by a single parameter range are as follows:
[0022] The optimization algorithm is used to cycle through the single parameters of the thermophysical properties to be identified in turn to find the best.
[0023] Taking the optimization error into consideration, the range of each parameter is narrowed down twice through a cycle, and the range of each thermophysical property parameter is obtained as the second range of the thermophysical property parameter.
[0024] Furthermore, an optimization algorithm is used to perform small interval parameter optimization identification on the second range of the thermophysical property parameters. The optimization algorithm is a parallel optimization algorithm of one or several combined algorithms such as a gradient algorithm, a bionic algorithm, and a probability algorithm.
[0025] A second aspect of the present invention provides a thermal physical property parameter identification system, comprising:
[0026] A data acquisition module is configured to acquire the thermophysical property parameters to be identified and select a first range reduction model according to a known number of samples;
[0027] A first range module is configured to reduce the range of the thermophysical property parameter using a first range reduction model to obtain a first range of the thermophysical property parameter;
[0028] A second range module is configured to construct a second range reduction model, and use the second range reduction model to perform a single parameter range step-by-step secondary reduction on the first range of the thermophysical property parameter to obtain a second range of the thermophysical property parameter;
[0029] The optimization identification module is configured to use an optimization algorithm to perform small interval parameter optimization identification on the second range of the thermophysical property parameters to obtain the optimal value of the thermophysical property parameters.
[0030] The third aspect of the present invention provides a medium having a program stored thereon, which, when executed by a processor, implements the steps in the method for identifying thermal physical property parameters as described in the first aspect of the present invention.
[0031] The fourth aspect of the present invention provides a device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for identifying thermal physical property parameters as described in the first aspect of the present invention are implemented.
[0032] One or more of the above technical solutions have the following beneficial effects:
[0033] The present invention discloses a method, system, medium and equipment for identifying thermophysical property parameters. The method uses a traditional analytical solution fitting algorithm of a thermophysical property measurement method or uses the test data of the thermophysical property measurement method and the real thermophysical property of the material to train a deep learning model, obtains a better value of the thermophysical property of the material, considers the fitting error or the training error, and constructs a smaller initial range of the thermophysical property parameters. Based on the physical and mathematical heat conduction model of the thermophysical property measurement method, an optimized numerical model of the heat conduction inverse problem of the thermophysical property parameters is established, and the optimization algorithm is used to sequentially cycle the single thermophysical parameter to be identified, consider the optimization error, and achieve a secondary reduction in the range of each parameter through one or more cycles. On the basis of the secondary reduction in the range of the thermophysical property parameters, the optimization algorithm is used to obtain the optimal value of the thermophysical property parameters. The present invention can effectively reduce the optimization range of the thermophysical property parameters and improve the efficiency and accuracy of parameter identification. The method described in the present invention is applicable to various optimization algorithms and effectively improves the efficiency and accuracy of parameter identification.
[0034] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.
[0036] Figure 1 This is a flow chart of a method for identifying thermal physical property parameters in Embodiment 1 of the present invention;
[0037] Figure 2 This is a comparison chart of test data obtained based on actual values of thermal physical property parameters of silica aerogel and test data obtained based on optimized values in Example 1 of the present invention;
[0038] Figure 3 It is a comparison chart of the test data obtained based on the actual values of the thermal physical property parameters of glass wool in Example 1 of the present invention and the test data obtained based on the optimized values. DETAILED DESCRIPTION
[0039] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.
[0040] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "include" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or their combinations;
[0041] Embodiment 1:
[0042] Embodiment 1 of the present invention provides a method for identifying thermal physical property parameters, such as Figure 1 As shown, the following steps are included:
[0043] Step 1: Obtain the thermophysical property parameters to be identified and select the first range reduction model based on the known number of samples.
[0044] In this embodiment, the thermal physical property parameters include thermal conductivity, specific heat capacity, thermal diffusivity, and contact thermal resistance.
[0045] Step 1.1: Set a threshold with known number of samples.
[0046] Step 1.2: If the number of known samples exceeds the threshold, the deep learning model is used as the first range reduction model.
[0047] If the number of known samples does not exceed the threshold, the traditional analytical solution fitting algorithm is used to construct the first range reduction model.
[0048] This can avoid the problem of overfitting of the deep learning model due to insufficient sample size. This embodiment can select a range reduction method according to different sample sizes, and is therefore applicable to all scenarios.
[0049] Step 2: Use the first range reduction model to reduce the range of the thermophysical property parameters to obtain the first range of the thermophysical property parameters.
[0050] In a specific embodiment, when a deep learning model is used as the first range reduction model, known samples are used to divide the training set and the test set, the real thermal properties are labeled, the deep learning model is trained with the training set, and the trained deep learning model is tested with the test set to obtain the optimal value of the thermal properties of the material. Taking the training error into consideration, a smaller initial range of the thermal property parameters is constructed to obtain the first range of the thermal property parameters.
[0051] The training error is the relative error between the optimal value and the actual value based on the optimal value. The upper and lower limits of the first range are the optimal value plus or minus the training error multiplied by the optimal value multiplied by the correlation coefficient. When the training error is less than 20%, the correlation coefficient is 2; when the training error is greater than 20%, the correlation coefficient is 1.5; when the training error is greater than 50%, the correlation coefficient is 1.2, which can avoid missing the actual value in the parameter range. If the lower limit is less than 0, it is replaced by 0.
[0052] In a specific embodiment, when a traditional analytical solution fitting algorithm is used to construct a first range reduction model, a traditional analytical solution fitting algorithm of a thermophysical property measurement method is used to obtain a better value of the thermophysical property of the material, and a smaller initial range of the thermophysical property parameters is constructed taking into account the fitting error to obtain a first range of the thermophysical property parameters.
[0053] Among them, the traditional analytical solution fitting algorithm is suitable for calculating thermal conductivity, specific heat, thermal diffusivity and other thermophysical parameters. The specific steps of the traditional analytical solution fitting algorithm are: using the analytical solution of the TPS method as the fitting formula, iterating the thermal diffusivity, fitting the temperature rise and dimensionless function, obtaining the thermal diffusivity and the corresponding thermal conductivity with the best linear correlation, and then calculating the specific heat. The use of the traditional analytical solution fitting algorithm to calculate thermophysical parameters is a conventional technical means in this field, and the detailed steps will not be repeated here.
[0054] It should be noted that the optimal value in step 2 is the prediction result calculated by using the first range reduction model. The calculated optimal value takes into account the training error or the fitting error and expands the upper and lower limits of different ranges according to the error size to obtain the first range. Among them, the smaller the error, the larger the correlation coefficient. This embodiment gradually reduces the error range in this way to achieve high-precision thermal physical property parameter identification.
[0055] Step 3: Construct a second range reduction model, and use the second range reduction model to perform a second step-by-step reduction of the first range of the thermophysical property parameters by a single parameter range to obtain a second range of the thermophysical property parameters.
[0056] Step 3.1: Construct a physical and mathematical heat conduction model of the thermophysical property parameters according to the thermophysical property measurement method; wherein the physical model is a two-dimensional axisymmetric model, including two samples and a probe between them. The mathematical heat conduction model is a two-dimensional axisymmetric heat conduction differential equation.
[0057] Step 3.2: Based on the physical and mathematical heat conduction models, an optimized numerical model of the heat conduction inverse problem of the thermal property parameters is established as the second range reduction model.
[0058] In a specific embodiment, the optimization numerical model of the inverse problem of thermal conductivity of thermophysical property parameters is constructed in the same way as the first range, and the trained model is used for prediction to obtain the second range, specifically: the optimization algorithm is applied to identify parameters, and the TPS thermal conductivity physical model is used as a constraint condition. The actual value of the parameter is substituted to obtain the average temperature rise of the heat source of the benchmark probe, and the predicted value of the parameter is substituted to obtain the average temperature rise of the heat source of the predicted probe. The benchmark temperature rise and the predicted temperature rise are compared as the objective function, and the parameter prediction value is changed to obtain the minimum value of the objective function. The corresponding parameter prediction value is the optimal value for the current step.
[0059] Step 3.2.1: Use the optimization algorithm to cycle through the single parameters of the thermophysical properties that need to be identified to find the best result.
[0060] Among them, the single thermophysical parameter refers to a single thermophysical parameter, that is, one of the thermal conductivity, specific heat capacity, and contact thermal resistance. Single parameter independent optimization means that only one parameter is predicted, and the other parameters are substituted as fixed values.
[0061] The specific steps are as follows: according to the constructed second range reduction model, by fixing two parameters, inverting the third parameter, realizing single parameter optimization, using the parameter prediction value to narrow the parameter range after each inversion, inverting the three parameters once respectively, and inverting the contact thermal resistance, specific heat capacity and thermal conductivity in turn.
[0062] Step 3.2.2: Considering the optimization error, the range of each parameter is reduced twice through one or more cycles to obtain the range of each thermophysical property parameter as the second range of the thermophysical property parameter.
[0063] Step 4: Use the optimization algorithm to optimize and identify the small interval parameters in the second range of the thermophysical property parameters to obtain the optimal value of the thermophysical property parameters. Unlike the previous step of using the same optimization algorithm for single parameter inversion, this step uses the same optimization algorithm to invert multiple parameters (such as thermal conductivity, specific heat capacity, contact thermal resistance), and finally outputs the optimal value of each parameter.
[0064] An optimization algorithm is used to perform small interval parameter optimization identification on the second range of the thermophysical property parameters. The optimization algorithm is a parallel optimization algorithm of one or several combined gradient algorithms, bionic algorithms, and probability algorithms.
[0065] In order to better illustrate the method described in this embodiment, the following experiment is performed:
[0066] Experiment 1:
[0067] In this experiment, the measured material is silica aerogel, with a thermal conductivity of 0.02 W / (m·K) and a specific heat capacity of 1200 J / (kg·K). The transient plane heat source method is used for the thermal property measurement, and the contact thermal resistance is 1.57×10 -3 m2·K / W; the method for constructing a smaller initial range of thermophysical property parameters selects the traditional analytical solution fitting algorithm; the optimization algorithm selects the Bayesian optimization algorithm.
[0068] In this experiment, the traditional analytical solution fitting results show that the measurement error of thermal conductivity is 7.38%, and the measurement error of specific heat capacity is 23.90%.
[0069] In this experiment, when the thermophysical property identification method described in this paper was not applied, the initial range of thermal conductivity was 0.001-1W / (m·K), and the initial range of specific heat capacity was 10-2500J / (kg·K). The optimization results of the Bayesian optimization algorithm showed that the measurement error of thermal conductivity was -6.68%, the measurement error of specific heat capacity was 11.29%, and the optimization time was 21925.49s.
[0070] In this experiment, after applying the thermophysical property parameter identification method described in this embodiment, the initial range of thermal conductivity is 0.0185-0.0244 W / (m·K), and the initial range of specific heat capacity is 1056.61-1916.94 J / (kg·K). After the range of each parameter is narrowed three times, the sequential optimization results of the Bayesian optimization algorithm show that the measurement error of thermal conductivity is -2.47%, the measurement error of specific heat capacity is 3.66%, and the optimized value of contact thermal resistance is 1.68×10 -3 m 2 K / W, the optimization time is 2344.16s. The optimization time of the Bayesian optimization algorithm for 16 points in parallel is 1347.55s. Figure 2Comparison of test data obtained based on actual values of silica aerogel thermophysical properties and test data obtained based on optimized values
[0071] Experiment 2:
[0072] In this experiment, the measured material is glass wool, with a thermal conductivity of 0.0318W / (m·K) and a specific heat capacity of 850J / (kg·K). The transient plane heat source method is selected as the thermal property measurement method; the traditional analytical solution fitting algorithm is selected as the method for constructing a smaller initial range of thermal property parameters; and the Bayesian optimization algorithm is selected as the optimization algorithm.
[0073] In this experiment, the traditional analytical solution fitting results show that the measurement error of thermal conductivity is -26.41%, and the measurement error of specific heat capacity is -38.37%.
[0074] In this experiment, when the thermophysical property identification method described in this article was not applied, the initial range of thermal conductivity was 0.001~1W / (m·K), and the initial range of specific heat capacity was 10-2500J / (kg·K). The optimization results of the Bayesian optimization algorithm showed that the measurement error of thermal conductivity was 0.12%, the measurement error of specific heat capacity was -0.07%, and the optimization time was 41208.01s.
[0075] In this experiment, after applying the thermophysical property parameter identification method described in this embodiment, the initial range of thermal conductivity is 0.0056-0.0448 W / (m·K), and the initial range of specific heat capacity is 128.65-1414.27 J / (kg·K). After the range of each parameter is narrowed three times, the sequential optimization results of the Bayesian optimization algorithm show that the measurement error of thermal conductivity is -0.31%, the measurement error of specific heat capacity is 1.63%, and the optimized value of contact thermal resistance is 1.73×10 -3 m 2 K / W, the optimization time is 9974.55s; the optimization time of the Bayesian optimization algorithm for 16 points in parallel is 3478.24s. Figure 3 This is a comparison between the test data obtained based on the actual values of the thermal physical properties of glass wool and the test data obtained based on the optimized values.
[0076] Embodiment 2:
[0077] Embodiment 2 of the present invention provides a thermal physical property parameter identification system, including:
[0078] A data acquisition module is configured to acquire the thermophysical property parameters to be identified and select a first range reduction model according to a known number of samples;
[0079] A first range module is configured to reduce the range of the thermophysical property parameter using a first range reduction model to obtain a first range of the thermophysical property parameter;
[0080] A second range module is configured to construct a second range reduction model, and use the second range reduction model to perform a single parameter range step-by-step secondary reduction on the first range of the thermophysical property parameter to obtain a second range of the thermophysical property parameter;
[0081] The optimization identification module is configured to use an optimization algorithm to perform small interval parameter optimization identification on the second range of the thermophysical property parameters to obtain the optimal value of the thermophysical property parameters.
[0082] Embodiment three:
[0083] The third embodiment of the present invention provides a medium on which a program is stored. When the program is executed by a processor, the steps in the method for identifying thermal physical property parameters as described in the first embodiment of the present invention are implemented.
[0084] Embodiment 4:
[0085] Embodiment 4 of the present invention provides a device, including a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in the method for identifying thermal physical property parameters as described in Embodiment 1 of the present invention are implemented.
[0086] The steps involved in the above embodiments 2, 3 and 4 correspond to the method embodiment 1. For the specific implementation methods, please refer to the relevant description part of embodiment 1.
[0087] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.
[0088] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.
Claims
1. A method for identifying thermal physical property parameters, characterized in that: The following steps are involved: Obtaining the thermophysical property parameters to be identified, and selecting a first range reduction model according to the known number of samples; Using the first range reduction model to reduce the range of the thermophysical property parameters, a first range of the thermophysical property parameters is obtained; A second range reduction model is constructed, and the first range of the thermophysical property parameter is reduced twice by a single parameter range in steps using the second range reduction model to obtain a second range of the thermophysical property parameter; The optimization algorithm is used to perform small interval parameter optimization identification on the second range of the thermophysical property parameters to obtain the optimal value of the thermophysical property parameters.
2. The method for identifying thermal physical property parameters according to claim 1, characterized in that: The specific steps for selecting the first range reduction model based on the known sample size are: Set a threshold with a known number of samples; If the number of known samples exceeds the threshold, the deep learning model is used as the first range reduction model; If the number of known samples does not exceed the threshold, the traditional analytical solution fitting algorithm is used to construct the first range reduction model.
3. The method for identifying thermal physical property parameters according to claim 2, characterized in that: When the deep learning model is used as the first range reduction model, the known samples are used to divide the training set and the test set, the real thermophysical properties are labeled, the training set is used to train the deep learning model, and the trained deep learning model is tested with the test set to obtain the optimal value of the thermal properties of the material. Considering the training error, the first range of the thermophysical property parameters is obtained.
4. The method for identifying thermal physical property parameters according to claim 2, characterized in that: When the first range reduction model is constructed using the traditional analytical solution fitting algorithm, the traditional analytical solution fitting algorithm of the thermophysical property measurement method is used to obtain the optimal value of the material's thermophysical property, and the first range of the thermophysical property parameters is obtained by taking the fitting error into consideration.
5. The method for identifying thermal physical property parameters according to claim 1, characterized in that: The specific steps to construct the second scope reduction model are: Construct physical and mathematical heat conduction models of thermophysical property parameters based on thermophysical property measurement methods; According to the physical and mathematical heat conduction model, an optimized numerical model of the inverse problem of heat conduction of thermal physical property parameters is established as the second range reduction model.
6. The method for identifying thermal physical property parameters according to claim 1, characterized in that: The specific steps of using the second range reduction model to perform a single parameter range secondary reduction on the first range of the thermophysical property parameters are as follows: The optimization algorithm is used to cycle through the single parameters of the thermophysical properties to be identified in turn to find the best. Taking the optimization error into consideration, the range of each parameter is narrowed down twice through a cycle, and the range of each thermophysical property parameter is obtained as the second range of the thermophysical property parameter.
7. The method for identifying thermal physical property parameters according to claim 1, characterized in that: An optimization algorithm is used to perform small interval parameter optimization identification on the second range of the thermophysical property parameters. The optimization algorithm is a parallel optimization algorithm of one or several combined gradient algorithms, bionic algorithms, and probability algorithms.
8. A thermal property parameter identification system, characterized in that: include: A data acquisition module is configured to acquire the thermophysical property parameters to be identified and select a first range reduction model according to a known number of samples; A first range module is configured to reduce the range of the thermophysical property parameter using a first range reduction model to obtain a first range of the thermophysical property parameter; A second range module is configured to construct a second range reduction model, and use the second range reduction model to perform a single parameter range step-by-step secondary reduction on the first range of the thermophysical property parameter to obtain a second range of the thermophysical property parameter; The optimization identification module is configured to use an optimization algorithm to perform small interval parameter optimization identification on the second range of the thermophysical property parameters to obtain the optimal value of the thermophysical property parameters.
9. A computer-readable storage medium, characterized in that: A plurality of instructions are stored therein, and the instructions are suitable for being loaded by a processor of a terminal device and executing the method for identifying thermal physical property parameters as described in any one of claims 1-7.
10. A terminal device, characterized in that: It includes a processor and a computer-readable storage medium, the processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, and the instructions are suitable for being loaded by the processor and executing the thermal physical property parameter identification method described in any one of claims 1-7.