Multi-physics field coupling agent model construction method for microstrip array antenna product

Abstract

The invention discloses a multi-physics field coupling agent model construction method for a microstrip array antenna product, and the method couples the comprehensive effects of force, heat and electromagnetic environments together through a structure displacement information transmission principle, thereby carrying out the simulation of an antenna under a working condition which is closer to the actual use condition of the antenna, and achieving the simulation of the antenna. The antenna simulation working condition and the real working condition can be consistent as much as possible; the used training data does not need to be subjected to large-scale physical tests, time and labor are wasted, traversal batch simulation is only carried out on the data by using a parameterized scanning technical means, and the data serve as the training data for constructing the proxy model, so that the cost and time for constructing the proxy model are saved; a black box proxy model is packaged into an FMU format and is used in system simulation modeling, and the input and output response relationship of the microstrip array antenna in a multi-physical field environment is embedded in system simulation in a component form, that is, when the microstrip array antenna or a system containing the antenna is simulated in the future, only the component in the FMU format needs to be called, and the input and output response relationship of the microstrip array antenna in the multi-physical field environment can be simulated. Therefore, the output response of the specific input can be quickly obtained, and computing resources are greatly saved.

Description

Technical Field

[0001] This invention relates to the field of digital simulation modeling technology, specifically to a method for constructing a multi-physics coupling proxy model for microstrip array antenna products. Background Technology

[0002] Existing techniques for analyzing the multi-physics coupling performance of microstrip array antennas mainly focus on methods that superimpose radiation patterns under different error factors, or convert mechanical and thermal deformation displacement data into phase change data to obtain radiation patterns after processing amplitude and phase distribution data, and then compare the changes in the shape and direction of the radiation patterns under different influencing factors. These methods for studying the performance of antennas under multi-physics coupling are essentially superpositions of the individual effects of each physical field, failing to reflect the "coupling" effect. Furthermore, they have high hardware requirements, as they require acquiring array surface data in the mechanical, electrical, and thermal dimensions of the microstrip array antenna, placing high demands on both hardware and data processing. Therefore, in the study of the performance of microstrip array antennas under multi-physics coupling, there is currently a lack of a simple, practical simulation analysis method that can reflect the coupling effects between various physical fields.

[0003] In the construction of surrogate models for microstrip array antennas, existing techniques mostly involve selecting several sets of design parameters from the antenna design space as input samples, inputting these samples into the initial antenna model to obtain the antenna model response, and then using neural network methods to simulate the mapping relationship to obtain the surrogate model. This method is a traditional surrogate model construction approach, i.e., obtaining data and then training. The surrogate model constructed by this method does not take into account multi-physics coupling factors, making it unsuitable for surrogate modeling of antennas in complex environments. Furthermore, data obtained from the design space cannot truly reflect all operating conditions of the antenna, limiting its performance analysis scope.

[0004] In addition, there is currently no surrogate model construction technique for microstrip array antennas that combines multiphysics coupling technology with surrogate model construction technology. These two key technologies are often studied separately, making it impossible to simultaneously achieve rapid and efficient antenna performance analysis while also considering the impact of multiphysics coupling.

[0005] Patent applications CN202310693856, entitled "Method and System for Analyzing the Influence of Phased Array SAR Antenna Array Deformation on Radiation Pattern," and CN201810729086, entitled "Novel Antenna Design Method," correspond to the multiphysics coupling part and the surrogate model construction part of this invention, respectively. However, both have the following drawbacks:

[0006] To obtain a radiation pattern, actual data is required, which means that physical prototypes or physical experiments are necessary. However, in reality, during the design phase, only a virtual geometric model of the microstrip array antenna is often available.

[0007] The method of superimposing amplitude and phase data cannot reflect the coupling relationship between multiple physical fields; it is essentially still the effect of a single physical field.

[0008] The method of selecting design parameters from the microstrip array antenna design space as input samples has significant limitations because it ignores the influence of external environmental factors and cannot perform multi-physics coupling analysis.

[0009] In existing technologies, simulation and surrogate model construction are separate processes, and there is no complete technical route from multiphysics coupled simulation to surrogate model construction. Summary of the Invention

[0010] In view of this, the present invention provides a method for constructing a multi-physics coupling proxy model for microstrip array antenna products.

[0011] The specific technical solution adopted in this invention is as follows:

[0012] Methods for constructing multi-physics coupling proxy models for microstrip array antenna products include:

[0013] Step S1: Determine the environmental variables and design variables of the antenna, and select the variables to be studied as input data; select the performance indicators to be studied as output data;

[0014] Step S2: Design the input and output data selected in step S1 into several sets of simulation conditions;

[0015] Step S3: Construct the geometric model of the antenna and import it into the multiphysics simulation software;

[0016] Step S4: First, set up the simulation preprocessing and re-add the physical fields corresponding to the variables selected in step S1 in the multiphysics simulation software;

[0017] Step S5: Based on each set of simulation conditions designed in step S2, set the simulation sequence according to the actual occurrence order of the physical field; when simulating in sequence, save the solution results of the previous step, that is, retain the structural displacement information data, and then perform the simulation of the next physical field.

[0018] Step S6: Use multiphysics simulation software to solve for the performance indicators output under the current simulation conditions;

[0019] Step S7: For the simulation model constructed in steps S3 to S6, traverse all simulation conditions designed in step S2 and retain all simulation input and output data;

[0020] Step S8: For the simulation data obtained in step S7, select the data points that best represent the antenna performance as sample data;

[0021] Step S9: Import the sample data into the selected surrogate model training software, train it, and obtain the antenna surrogate model;

[0022] Step S10: Encapsulate the obtained proxy model into a usable format file.

[0023] Preferably, in step S1, environmental variables include external force and temperature; design variables include input voltage, geometric parameters, and operating frequency; and antenna performance indicators include gain, electric field strength, and radiated power.

[0024] Preferably, in step S3, the multiphysics simulation software includes COMSOL.

[0025] Preferably, the physical field includes mechanical, thermal, and electromagnetic fields.

[0026] Preferably, in step S5, the mechanical and thermal physical fields are solved simultaneously before the electromagnetic field is solved.

[0027] Ideally, the mechanical and thermal physical fields are coupled through thermal expansion.

[0028] Preferably, in step S5, for software that cannot retain structural displacement information data, mesh remodeling technology can be used to re-mesh the deformed structure so that it is suitable for simulation of the next physics field.

[0029] Preferably, artificial neural networks, kriging methods, or multinomial fitting methods are selected for training the surrogate model to obtain the microstrip array antenna surrogate model.

[0030] Ideally, when choosing from multiple surrogate model training algorithms, the surrogate model training algorithm with the best output performance index should be selected.

[0031] Preferably, step S8 also includes the following data processing: the output index data in each task profile and its corresponding input variable data are stored in the same data subset library; outliers in the data are removed; and the data is stored in a format that facilitates the training of the surrogate model.

[0032] The present invention has the following beneficial effects:

[0033] (1) Multi-physics coupling technology: In the past, antenna simulation modeling often only focused on the performance simulation of the antenna under a single environment (such as under the action of force). This invention uses the principle of structural displacement information transmission to couple the combined effects of force, heat and electromagnetic environment together, so as to simulate the antenna under conditions that are closer to the actual use of the antenna, and can ensure that the antenna simulation conditions are as consistent as possible with the actual working conditions.

[0034] (2) Proxy model construction technology: This invention starts with data acquisition. The training data used does not require large-scale physical experiments, which is time-consuming and laborious. Only parametric scanning technology is needed to perform batch simulations. The simulation data is used as training data for building proxy models, which greatly saves the cost and time of building proxy models.

[0035] (3) Component Encapsulation Application Technology: The surrogate model itself cannot be directly applied to engineering problems. This invention encapsulates the black-box surrogate model into an FMU format and uses it in system simulation modeling (e.g., MWORKS). The input-output response relationship of the microstrip array antenna in a multi-physics environment is embedded in the system simulation as a component. This means that when simulating microstrip array antennas or systems containing antennas, there is no need for physical experiments or extensive simulations; simply calling the FMU format component will quickly obtain the output response of a specific input, greatly saving computational resources. For commercial use, pre-packaged microstrip array antenna FMU components can be sold directly to manufacturers.

[0036] (4) This invention covers the entire process of building a microstrip array antenna proxy model, from designing the microstrip array antenna operating conditions to multiphysics finite element simulation, then extracting simulation data for surrogate model training, and finally encapsulating and applying the surrogate model. Users can follow the steps to fully implement the entire process of building a microstrip array antenna proxy model. Attached Figure Description

[0037] Figure 1 This is the overall technical roadmap of the present invention.

[0038] Figure 2 This is a schematic diagram illustrating the implementation principle of the multiphysics coupling technology in this invention. Detailed Implementation

[0039] This invention provides a method for constructing a multi-physics coupling proxy model for microstrip array antenna products, comprising:

[0040] Step S1: Determine the antenna's operating conditions and design variables. Determine environmental variables such as external force and temperature, and design variables such as input voltage, geometric parameters, and operating frequency. Determine the antenna's performance indicators, such as gain, electric field strength, and radiated power. Based on the existing set of variables and indicators, select the inputs and outputs to be studied. For example, select external force and temperature as inputs, and select gain, electric field strength, and radiated power as outputs.

[0041] Step S2: Design the input and output selected in Step S1 into several sets of simulation conditions. This involves designing specific simulation input values ​​and selecting an experimental design method, such as orthogonal array method, central composite experiment method, Monte Carlo sampling method, etc., to design several sets of simulation conditions. For example, a force of 10kN, a temperature of 45℃, an operating frequency of 1.5GHz, and a voltage of 100V constitute one set of conditions. When designing the conditions, the following principles should be followed:

[0042] (1) Cover the operating conditions within the extreme operating condition envelope as much as possible;

[0043] (2) The working conditions should be distributed as evenly as possible;

[0044] (3) The number of design working conditions is determined according to the accuracy, and is generally several hundred sets.

[0045] Step S3: Build a high-precision geometric model of the antenna and import it into multiphysics simulation software, such as COMSOL.

[0046] Step S4: First, set up the simulation preprocessing, paying special attention to adding physical fields. The added physical fields should correspond to the variables selected in Step S1. For example, if force, temperature, and electrical parameters are selected, then mechanics, thermodynamics, and electromagnetic fields should be added. If a single-discipline simulation software is used and physical fields cannot be added, then the specialized software for that discipline must be selected. For example, ABAQUS can be selected for mechanics, Fluent for fluid dynamics, and HFSS for electromagnetism.

[0047] Step S5: Set the simulation sequence according to the actual occurrence order of the physical fields. For example, if force and heat act simultaneously, causing deformation of the array structure and thus affecting electrical performance, then when setting the solution steps, mechanical and thermal physical fields should be solved simultaneously first, followed by electromagnetic field solutions. During the solution process, special attention should be paid to selecting to save the solution results from the previous step, i.e., retaining the structural displacement information data. For software that cannot perform this operation, mesh reconstruction technology can be used to re-mesh the deformed structure, making it suitable for the simulation of the next physical field.

[0048] Step S6: Solve and perform post-processing to obtain the antenna output under a certain operating condition, such as gain, electric field strength, and radiated power.

[0049] Step S7: For the simulation model constructed in steps S3 to S6, perform a traversal simulation using parametric scanning or scripts. The simulation conditions must traverse all the conditions designed in step S2. All simulation input and output data must be retained.

[0050] Step S8: Process the simulation data, which can be done according to the following principles:

[0051] (1) The output index data and the corresponding input variable data in each task profile are stored in the same data subset library;

[0052] (2) Remove outliers from the data (values ​​that are significantly higher or lower than other data);

[0053] (3) Select data points that best represent the antenna performance;

[0054] (4) Data is stored in a format that facilitates the training of the agent model.

[0055] Step S9: Import the prepared simulation dataset into the surrogate model training software. Select one or more surrogate model training algorithms (e.g., Kriging method, artificial neural network method, or multinomial fitting method), and use the sample data in the dataset to train the antenna surrogate model according to the selected algorithm. When selecting multiple surrogate model training algorithms, choose the one with the best output performance index.

[0056] Step S10: Encapsulate the obtained proxy model into a usable FMU format, GUI program, DLL program, or other format file.

[0057] Example:

[0058] A microstrip patch array antenna consists of a substrate, radiating patches, and cables. During use, its center point experiences concentrated stress, and the surrounding ambient temperature undergoes significant changes. To verify how the antenna's gain, radiated power, and electric field strength are affected under these conditions, a surrogate model of the antenna is established, and its response is analyzed.

[0059] Step S1: Determine the operating conditions and design variables of the microstrip patch array antenna. Set external force and temperature as simulation inputs, gain, electric field strength, and radiated power as simulation outputs. Other parameters such as voltage, geometric parameters, and operating frequency are set according to the specific antenna conditions and remain constant.

[0060] Step S2: Design several sets of simulation conditions using a traversal iterative design method. First, select 20 sets of external force values: 50, 55, 60, 65...140, 145, 150 kN. Then, select 25 sets of temperature values: -70, -65, -60...45, 50, 55℃. Combining these values ​​with each force corresponding to one temperature, create 500 sets of conditions, namely "50 kN, -70℃", "50 kN, -65℃", ..., "50 kN, 55℃", "55 kN, -70℃", ..., "150 kN, 55℃".

[0061] Step S3: Construct the geometric model of the 8*4 array microstrip patch antenna. Any modeling software can be used to construct it. Then import the model into COMSOL software for preprocessing. Note that global variables should be defined.

[0062] Step S4: Perform simulation preprocessing settings, see Table 1 for details. Add three physical fields: solid mechanics, solid heat transfer, and electromagnetic wave frequency domain. Add thermal expansion for physical field coupling, that is, the coupling between mechanics and thermodynamics is achieved through thermal expansion. Then, generate an adaptive mesh.

[0063] Table 1 Preprocessing settings for the embodiment

[0064]

[0065] Step S5: Set up the solver and add a study, which consists of two steps: Step 1, solve for the mechanical-thermal coupling, disabling electromagnetic physics parameters; Step 2, check "Save Solution" to transfer the deformation displacement information data obtained in Step 1 to the initial values ​​of the solution in Step 2. Step 2 disables mechanical and thermal parameters and enables electromagnetic physics parameters.

[0066] Step S6: After solving, the antenna far-field radiation pattern, gain, electric field distribution, radiated power radiation pattern, deformation pattern, temperature distribution, etc. are obtained. The results are processed and exported as a CSV table, with the data format output as: azimuth angle-force-temperature-gain-electric field strength-radiated power.

[0067] Step S7: Based on the 500 sets of working conditions designed in Step S2, perform parametric scanning and traversal simulations, scanning the azimuth parameters and automatically changing the two global variables of force and temperature in the model. A total of 500 simulations are performed, and the input and output data of each set of working conditions are automatically processed using a secondary development script, finally obtaining 500 sets of antenna input and output response datasets under each working condition.

[0068] Step S8: Process the simulation data and select the data with the maximum gain azimuth angle as the training data for building the surrogate model.

[0069] Step S9: Import the prepared simulation dataset into the surrogate model training software. Select artificial neural network, Kriging method, and multinomial fitting method for training to obtain the microstrip array antenna surrogate model.

[0070] Step S10: Encapsulate the obtained proxy model into a usable FMU format, GUI program, DLL program, or other format file.

[0071] The following summarizes the innovative aspects of this invention:

[0072] (1) The training data of the proxy model of the present invention are all derived from the finite element simulation input and output data of the microstrip array antenna, that is, it can still be carried out without the antenna physical prototype and test equipment.

[0073] (2) The multi-physics coupling method used in this invention involves the displacement of the antenna array structure caused by physical fields such as force and heat. The resulting displacement information data can be transmitted to the electromagnetic physical field simulation, thereby obtaining the antenna's performance output response values ​​before and after deformation. Therefore, it can better reflect the antenna performance simulation analysis under the combined action of multiple physical fields.

[0074] (3) The proxy model constructed in this invention is used to quantitatively evaluate the performance of microstrip array antennas. It can obtain accurate data of various performance indicators (such as electric field strength, gain, radiated power, etc.) of microstrip array antennas before and after deformation, and evaluate the performance of microstrip array antennas objectively and quantitatively.

[0075] (4) The training data for the proxy model constructed in this invention are all from the previous simulation data. The simulation data includes the influencing factors of each physical field, such as force, temperature, voltage, etc. Therefore, the proxy model constructed in this way can be used to analyze the influence weight of multi-physical field influencing factors on the performance of microstrip array antennas and predict the influence trend of factors.

[0076] (5) This invention provides a complete methodology from multiphysics simulation to surrogate model construction, which combines finite element level 3D performance simulation with surrogate model capability simulation by using simulation inputs and outputs as training data for the surrogate model. Referring to this invention, full-cycle design optimization research of microstrip array antennas can be conducted from the design stage to the performance evaluation and verification stage. It is a complete multiphysics simulation and modeling method for microstrip array antennas.

[0077] The specific embodiments described above only illustrate the design principles of the present invention. The shapes and names of the components in this description may differ and are not limited. Therefore, those skilled in the art can modify or make equivalent substitutions to the technical solutions described in the foregoing embodiments; and these modifications and substitutions do not depart from the inventive spirit and technical solutions of the present invention, and should all fall within the protection scope of the present invention.

Claims

1. A method for constructing a multi-physics coupling surrogate model for microstrip array antenna products, characterized in that, include: Step S1: Determine the environmental variables and design variables of the microstrip array antenna, and select the variables to be studied as input data; Select the performance indicators to be studied as output data; Step S2: Design the input and output data selected in step S1 into multiple sets of simulation conditions; Step S3: Construct the geometric model of the microstrip array antenna and import it into the multiphysics simulation software; Step S4: First, set up the simulation preprocessing and re-add the physical fields corresponding to the variables selected in step S1 in the multiphysics simulation software; Step S5: Based on each set of simulation conditions designed in step S2, set the simulation sequence according to the actual occurrence order of the physical field; when simulating in sequence, save the solution results of the previous step, that is, retain the structural displacement information data, and then perform the simulation of the next physical field. Step S6: Use multiphysics simulation software to solve for the performance indicators output under the current simulation conditions; Step S7: For the simulation model constructed in steps S3 to S6, traverse all simulation conditions designed in step S2 and retain all simulation input and output data; Step S8: For the simulation data obtained in step S7, select the data points that best represent the antenna performance as sample data; Step S9: Import the sample data into the selected surrogate model training software, train it, and obtain the antenna surrogate model; Step S10: Encapsulate the obtained proxy model into a usable format file.

2. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 1, characterized in that, In step S1, environmental variables include external force and temperature; design variables include input voltage, geometric parameters, and operating frequency; and antenna performance indicators include gain, electric field strength, and radiated power.

3. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 2, characterized in that, In step S3, the multiphysics simulation software includes COMSOL.

4. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 2, characterized in that, The physical fields include mechanical, thermal, and electromagnetic fields.

5. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 4, characterized in that, In step S5, mechanical and thermal physical fields are solved simultaneously first, and then electromagnetic field is solved.

6. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 4, characterized in that, Mechanical and thermal physical fields are coupled through thermal expansion.

7. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 6, characterized in that, In step S5, for software that cannot retain structural displacement information data, mesh remodeling technology can be used to re-mesh the deformed structure so that it is suitable for simulation of the next physics field.

8. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 1, characterized in that, The surrogate model is trained by selecting artificial neural networks, Kriging methods, or polynomial fitting methods to obtain the surrogate model of the microstrip array antenna.

9. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 1, characterized in that, When choosing from multiple surrogate model training algorithms, select the surrogate model training algorithm with the best output performance index.

10. The method for constructing a multi-physics coupling proxy model for microstrip array antenna products as described in claim 1, characterized in that, Step S8 also includes the following data processing: the output index data in each task profile and its corresponding input variable data are stored in the same data subset library; outliers in the data are removed; and the data is stored in a format that facilitates the training of the surrogate model.

Images