A method for real-time indirect measurement of satellite structure concern response

Abstract

A real-time indirect measurement method for satellite structural response to points of interest is disclosed, belonging to the field of on-orbit spacecraft monitoring. The method includes: 1) deploying a limited number of sensors on the satellite structure and classifying the data acquired by the sensors; 2) acquiring satellite structural dynamic data; 3) constructing a finite element model of the satellite structure, using the excitation data measured in the dynamic experiment as input to obtain virtual data corresponding to the sensor positions on the satellite structure; 4) establishing the error distribution relationship between the measured data and virtual data of the points of interest using a neural network model, and then using the virtual data under different excitations as input to generate high-quality dynamic response data of the points of interest; 5) constructing a high-precision proxy model for the measurable point data and the data of the points of interest, and inputting the real-time measurable point data into the proxy model to achieve real-time measurement of the response to the points of interest. This method enables real-time monitoring of the response to the points of interest and also lays the foundation for health monitoring of satellite structures.

Description

Technical Field

[0001] This invention belongs to the field of spacecraft on-orbit monitoring, specifically relating to a real-time indirect measurement method for satellite structural concern responses. Background Technology

[0002] Satellites hold an extremely important position in modern society. With the continuous development of science and technology, the applications of satellites are becoming increasingly widespread, and humanity's dependence on them is growing ever stronger. Artificial satellites provide humanity with services such as global communication, weather forecasting, Earth observation, resource exploration, navigation, and scientific research. They also support development in fields such as transportation, environmental protection, disaster response, and national defense. Artificial satellites have become an important tool and support for human exploration and development, driving continuous progress and advancement.

[0003] With the increasing demands on the functions and performance of remote sensing satellites, the complexity of satellite structures also increases, potentially leading to insufficient ground testing and an increase in early failures after satellite launch. Statistics show that approximately 60% of serious failures occur in the first year after satellite launch, with many occurring shortly after launch. Furthermore, satellites in space are subjected to thermal loads such as radiation, convection, and conduction. These long-term, cyclical changes in thermal loads cause expansion and contraction of the satellite's main structure and critical components, accelerating damage to critical structures (such as load-bearing structures), shortening the on-orbit lifespan of remote sensing satellites, and reducing the quality of on-orbit service. Routine health management of spacecraft is typically achieved through a combination of two methods: (1) providing robust, damage-resistant designs and structural integrity margins to prevent potential failures; and (2) relying on regular maintenance and long-term inspections to determine necessary repairs. However, due to the complexity of the environment and loads, as well as the uncertainty of material properties and model assumptions, structural safety margins may lead to unnecessary heavy structures, and may not necessarily ensure satellite safety or mission success. Furthermore, because pre-determined maintenance plans cannot account for actual environmental and operational conditions, there is a risk of premature failure during flight. Adopting overly conservative approaches, such as increasing the frequency of scheduled maintenance, can effectively extend the satellite's on-orbit service life, but this also significantly increases maintenance costs. Overall, with technological advancements and increasing application demands, the requirements for the accuracy and stability of satellite platforms are constantly rising to ensure their successful execution and long-term reliable operation in various complex missions.

[0004] As a crucial component of satellite systems, satellite structures bear and protect precision instruments, ensuring the integrity and stability of the entire system and playing a vital role in satellite design and operation. The satellite structure and payload support may be dynamically coupled, generating significant vibration loads during launch and on-orbit operation. These vibrations not only affect the lifespan of equipment but also significantly reduce the measurement accuracy of precision instruments. However, in the aerospace field, existing vibration reduction and isolation methods are limited by poor damping performance of traditional materials, low efficiency of numerical simulation calculations, and a lack of systematic configuration optimization design methods, making it difficult to suppress the vibration response of areas where precision instruments are installed (points of interest). Due to space constraints, the dynamic response of these points of interest cannot be directly measured using sensors. Therefore, there is an urgent need for a method that can achieve real-time indirect measurement of the response of satellite structural points of interest, paving the way for research in satellite structural optimization design, health monitoring, and reliability assessment. Summary of the Invention

[0005] To overcome the problems existing in the prior art, the present invention provides a real-time indirect measurement method for satellite structure concern response, so as to obtain the concern response of satellite structure in real time and provide necessary data support for further research by relevant engineers.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for real-time indirect measurement of satellite structural concern response, the method comprising the following steps:

[0008] Step S1: Sensor installation and classification. A limited number of sensors are deployed on the satellite structure, and the data acquired by the sensors are divided into two categories: data from measurable points and data from points of interest.

[0009] Step S2: Acquisition of satellite structure dynamics data. Dynamic data of each measuring point of the satellite structure is obtained through dynamic experiments, and excitation data is obtained by sensors on the vibration table. During the experiment, each sensor needs to collect data synchronously.

[0010] Step S3: Construct a finite element model of the satellite structure. Use simulation software to build the finite element model of the satellite structure, and correct the key parameters in the model using experimental data. Use the measured excitation data from the dynamic experiment as input to obtain virtual data corresponding to the installation positions of each sensor on the satellite structure; specifically:

[0011] S31. Construct a three-dimensional model of the satellite structure and discretize it to determine the finite element nodes corresponding to the positions of the measurement points on the real satellite structure, and treat them as virtual measurement points;

[0012] S32. Conduct sinusoidal sweep frequency and free vibration decay experiments to obtain experimental values ​​of the structure's natural frequency and overall decay rate, which are used to correct key parameters of the model.

[0013] S33. Use optimization algorithms to correct key parameters of the satellite structure finite element model;

[0014] The key parameters of the finite element model are corrected by using the measured natural frequency values ​​and the overall free decay rate of the structure. The response surface method is used to correct the key parameters. The natural frequency values ​​of the satellite structure are used to correct the material properties in the model, including elastic modulus, Poisson's ratio and mass, while the free decay data are used to identify the overall damping of the structure.

[0015] S34. Using the collected vibration table response as input, calculate the virtual dynamic data of each corresponding measuring point using the corrected finite element model;

[0016] Step S4: Establish an error distribution model between the measured data and virtual data of the focus point using an LSTM-VAE neural network model. Then, use the virtual data of the focus point under different stimuli as input to generate high-quality dynamic response data of the focus point; specifically:

[0017] S41. Establish an LSTM-VAE neural network model;

[0018] The LSTM-VAE model replaces the feedforward neural network in VAE with LSTM. LSTM-VAE combines the local feature learning ability of VAE with the temporal modeling ability of LSTM neural network. The LSTM-VAE model can learn and capture complex temporal dependencies in time series data, thereby generating data with similar temporal dependencies.

[0019] S42. Using the measured attention response as the output and the virtual attention data calculated by the finite element model as the input, the generator model between the input and output is obtained by adjusting the hyperparameters of the LSTM-VAE neural network model.

[0020] The virtual response of the satellite structure calculated using the finite element model is defined as x, and the measured response is defined as y; according to the formula... Standardize the data;

[0021] The dataset is divided into training and validation sets, and the hyperparameters in the LSTM-VAE neural network model are tuned to minimize the difference between the output and the measured response during supervised training. The attention response generated by the LSTM-VAE model is defined as... ;

[0022] S43. Quantitatively evaluate the accuracy of the generator model. If the accuracy does not meet the requirements, return to S42 to reset the new hyperparameter combination and train.

[0023] The validation set is input into the constructed LSTM-VAE generator model to calculate the improved vibration response of the point of interest; the performance of the LSTM-VAE generator model is evaluated using correlation coefficient, root mean square error, standard root mean square error and cosine similarity metric.

[0024] Step S5: Construct a high-precision proxy model of the measured measurable point data and the generated data of the points of interest. Input the real-time measurable point data into the proxy model of the two to realize real-time measurement of the response of the points of interest.

[0025] Further, step S1 specifically includes:

[0026] S11. Select the measuring point and install the sensor;

[0027] S12. Classified by sensor installation location: measurable point sensors and interest point sensors; measurable point data is used as input to the surrogate model, while the measured interest point data is the output data.

[0028] Further, step S2 specifically includes:

[0029] S21. Use clamps to fix the satellite structure to the vibration table;

[0030] The satellite structure and the shaking table are connected by a fixture to establish a satellite structure dynamics experimental system, and the shaking table is used to input dynamic loads.

[0031] S22. An accelerometer is installed on the vibration table to collect excitation data applied to the satellite structure during random vibration.

[0032] Sensors are installed on a vibration table to collect the response data of the vibration table during random vibration; the vibration direction collected by the sensors is consistent with the vibration direction of the vibration table during the experiment.

[0033] S23. Conduct random vibration experiments and use a high-speed acquisition system to synchronously collect the dynamic response of each measuring point and the excitation data of the vibration table.

[0034] S24. Classify the data collected by the sensor into data of interest points and data of measurable points.

[0035] Furthermore, step S5 specifically includes:

[0036] S51. Construct an LSTM neural network model;

[0037] S52. Establish a proxy model of the measured measurable point response and the virtual concern point response using an LSTM neural network model;

[0038] S53. Quantitative evaluation of the surrogate model's accuracy. If the model's accuracy does not meet the requirements, return to S52 to reselect a new combination of LSTM neural network hyperparameters for training.

[0039] S54. Real-time measurable point response data is used as input to the surrogate model to achieve real-time indirect measurement of the response of the point of interest.

[0040] The beneficial effects of this invention are:

[0041] 1) Sensor Installation and Classification: A limited number of sensors are deployed on the satellite structure, and the data acquired by the sensors is divided into two categories: measurable data and data of interest. Due to space constraints in the satellite's load-bearing structure, the response of points of interest is often difficult to measure directly in actual operating conditions; sensor data installed at these locations are defined as data of interest. Data from locations where sensors can be installed and measured directly is defined as measurable data.

[0042] 2) Satellite structure dynamics data acquisition: Dynamic data of each measuring point of the satellite structure is obtained through dynamic experiments, and excitation data is obtained by sensors. During the experiment, each sensor needs to collect data synchronously.

[0043] 3) Construct a finite element model of the satellite structure. Use CAE simulation software (such as ANSYS) to construct a finite element model of the satellite structure. Correct the key parameters in the model using experimental data. Use the excitation data measured in the dynamic experiment as input to obtain virtual data corresponding to the positions of the satellite structure sensors.

[0044] 4) Utilize a Long Short-Term Memory Variational Autoencoder (LSTM-VAE) neural network model to establish the error distribution relationship between the measured data and virtual data of the focus. Then, use the virtual data under different stimuli as input to generate high-quality dynamic response data of the focus, so as to solve the problem of difficulty in obtaining the response of the focus.

[0045] 5) Construct a high-precision proxy model of measurable point data (actual measurement) and interest point data (generated), and input the real-time measurable point data into the proxy model of the two to realize real-time measurement of the response of interest points.

[0046] This scheme not only enables real-time monitoring of the response to interest, but also lays the foundation for health monitoring of satellite structures. This invention addresses the difficulty or inability to measure the response to interest in actual satellite structures in practical engineering scenarios, and can be applied to the indirect measurement of the response to interest in real-world satellite structures. Attached Figure Description

[0047] Figure 1 This is a flowchart of a real-time indirect measurement method for satellite structure concern response.

[0048] Figure 2 This is a schematic diagram of the satellite structure in an embodiment of the present invention.

[0049] Figure 3 This is a schematic diagram of the dynamic experiment of the satellite structure in an embodiment of the present invention.

[0050] Figure 4 This is a schematic diagram of the LSTM-VAE neural network model structure in an embodiment of the present invention.

[0051] Figure 5 These are the measured and virtual attention responses in this embodiment of the invention. In (a), the virtual attention response is directly calculated using a finite element model, and in (b), the virtual attention response is further improved using an LSTM-VAE model.

[0052] Figure 6 This is a schematic diagram of the LSTM neural network model structure in an embodiment of the present invention.

[0053] Figure 7 This is a time-history diagram of the concern response calculated by the proxy model in this embodiment of the invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0055] To illustrate the technical solution described in this invention, specific embodiments are described below.

[0056] like Figure 1 As shown, the real-time indirect measurement method for satellite structure concern response provided in this embodiment includes the following steps:

[0057] Step S1: Sensor installation and classification. A limited number of sensors are deployed on the satellite structure, and the data acquired by the sensors are divided into two categories: measurable point data and point of interest data.

[0058] This step completes the installation and classification of the sensors. The specific implementation process is as follows:

[0059] S11. Select the measuring point and install the sensor;

[0060] Determine the satellite structure based on engineering practice. Figure 2You can then place the sensors at the locations where you want to install precision instruments (the points of interest), and then install the sensors.

[0061] S12. Classify according to sensor installation location

[0062] Based on the sensor's installation location, the data collected by the sensor is divided into two categories: measurable point data and interest point data. Measurable point data is used as input to the surrogate model, while the measured interest point data is the output data.

[0063] Step S2: Acquisition of satellite structure dynamics data. Dynamic data of each measuring point of the satellite structure is obtained through dynamic experiments, and excitation data is obtained using sensors. During the experiment, each sensor needs to collect data synchronously.

[0064] This step completes the data collection for each measuring point. The specific implementation process is as follows:

[0065] S21. Use clamps to fix the satellite structure on the vibration table.

[0066] The satellite structure and shaking table are connected via a fixture to establish a satellite structural dynamics experimental system, using the shaking table to input dynamic loads. To avoid the influence of the experimental fixture on the structural vibration, the structural strength and natural frequency of the fixture are much greater than the relevant dynamic properties of the satellite structure.

[0067] S22. An accelerometer is installed on the vibration table to collect excitation data applied to the satellite structure during random vibration.

[0068] Sensors are installed on a vibration table to collect the table's response data during random vibration. The vibration direction collected by the sensors is consistent with the direction of vibration of the vibration table during the experiment. The collected vibration table response data can be used as the input load for S34.

[0069] S23. Conduct random vibration experiments and use a high-speed acquisition system to synchronously collect the dynamic response of each measuring point and the excitation data of the vibration table.

[0070] Four accelerometers were installed on the outer wall panel of the satellite structure, with an operating frequency of 20–9000 Hz. Lateral random vibration excitation was applied using a vibration table; the power spectral density of the random vibration settings is shown in Table 1.

[0071] Table 1 Power spectral density of satellite random vibration experiment settings

[0072]

[0073] Figure 3This is a simplified diagram of a satellite structural dynamics experiment. Three accelerometers are installed at the lower part of the structure, defined as measurable points, and one accelerometer is installed in the upper region of the structure, defined as the point of interest. Additionally, one accelerometer is mounted on a shaking table to measure the dynamic response of the shaking table. High-speed data acquisition equipment is used to synchronously acquire data from these sensors.

[0074] S24. Data classification: Classify the data collected by the sensor into data of interest points and data of measurable points.

[0075] The data collected during the S23 process are classified into three main categories: measurable point data, data of interest points, and vibration table data.

[0076] Step S3: Construct a finite element model of the satellite structure. Use CAE simulation software (such as ANSYS) to construct a finite element model of the satellite structure. Correct the key parameters in the model using experimental data. Use the excitation data measured in the dynamic experiment as input to obtain virtual data corresponding to the positions of the satellite structure sensors.

[0077] This step completes the construction of the satellite structure finite element model, parameter correction, and calculation of virtual data for the corresponding measurement points. The specific implementation process is as follows:

[0078] S31. Construct a three-dimensional model of the satellite structure and discretize it to determine the finite element nodes corresponding to the positions of the measurement points on the real satellite structure, and treat them as virtual measurement points;

[0079] A finite element model of the satellite is constructed and meshed. The resulting finite element model is then imported into the finite element simulation software ANSYS Workbench. It's important to note that satellite structural dynamics calculation software is not limited to ANSYS Workbench; other finite element software can also be used. Then, based on the locations of the measurement points on the satellite structure, the corresponding node positions in the discretized finite element model are determined.

[0080] S32. Conduct sinusoidal sweep frequency and free vibration decay experiments to obtain experimental values ​​of the structure's natural frequency and overall decay rate, which are used to correct key parameters of the model.

[0081] Frequency sweep analysis was performed using a shaking table with a frequency range of 20-100 Hz and an acceleration of 0.8 g. Furthermore, to obtain free decay response data for key points of interest in the load-bearing structure, instantaneous excitation was applied by hammer impact.

[0082] S33. Use optimization algorithms to correct key parameters of the satellite structure finite element model;

[0083] The key parameters of the finite element model are corrected using measured natural frequencies and the overall free decay rate of the structure. Considering that directly calling the satellite structure finite element model for key parameter correction is very time-consuming, a response surface methodology is used to address this issue. The natural frequencies of the satellite structure are used to correct the material properties in the model, including elastic modulus, Poisson's ratio, and mass, while the free decay data are used to identify the overall damping of the structure. Table 2 presents the measured and corrected natural frequencies of the finite element model of the satellite structure using sinusoidal sweep frequency analysis.

[0084] Table 2. Measured and corrected natural frequency values ​​of the satellite structure using sinusoidal frequency sweeping.

[0085]

[0086] S34. Using the vibration table response collected in S23 as input, calculate the virtual dynamic data of each corresponding measuring point using the corrected finite element model.

[0087] The transient dynamics calculation module in ANSYS Workbench was used to perform dynamic analysis on the satellite structure, with shaking table data acquired in S23 as the input to the model. After the calculation was completed, the dynamic response of the nodes corresponding to the positions of each measuring point on the satellite structure was extracted.

[0088] Step S4: Use the LSTM-VAE neural network model to establish the error distribution relationship between the measured data and virtual data of the focus, so as to generate high-quality dynamic response data of the focus;

[0089] This step utilizes an LSTM-VAE neural network to establish a generative model between measured and virtual data for measurable points. The specific implementation process is as follows:

[0090] S41. Establish an LSTM-VAE neural network model.

[0091] The structure of the LSTM-VAE model is as follows: Figure 4 As shown, the feedforward neural network in the VAE is replaced with an LSTM. LSTM-VAE combines the local feature learning capability of VAE with the temporal modeling capability of LSTM neural networks. Considering the temporal correlation between data, the LSTM-VAE model can learn and capture complex temporal dependencies in time series data, thereby generating data with similar temporal dependencies. The main purpose of establishing the LSTM-VAE neural network model is to narrow the gap between virtual data and measured data.

[0092] S42. Using the measured attention response as the output and the virtual attention data calculated by the finite element model as the input, the generator model between the input and output is obtained by adjusting the hyperparameters of the LSTM-VAE neural network model.

[0093] The virtual response of the satellite structure calculated using the finite element model is defined as x, and the measured response is defined as y. According to the formula... Data standardization is performed to eliminate distributional differences between measurement points. The dataset is divided into training and validation sets, and the hyperparameters of the LSTM-VAE neural network model are adjusted to minimize the difference between the output and the measured response during supervised training. The attention response generated by the LSTM-VAE model is defined as...

[0094] S43. Quantitatively evaluate the accuracy of the generator model. If the accuracy does not meet the requirements, return to S42 to reset the new hyperparameter combination and train.

[0095] The validation dataset partitioned in process S42 is input into the constructed LSTM-VAE generator model to calculate the improved vibration response of the point of interest. The performance of the LSTM-VAE generator model is evaluated using correlation coefficient, root mean square error, standard root mean square error, and cosine similarity metric. Figure 5 The responses are measured and virtual concerns, where (a) is the virtual concern response obtained directly through the finite element model, and (b) is the virtual concern response further improved through the LSTM-VAE model.

[0096] Table 3 Quantitative evaluation results of LSTM-VAE model performance

[0097]

[0098] Table 3 presents the quantitative indicators for the response to concerns, and the calculation of these indicators is as follows:

[0099] Correlation coefficient:

[0100] Root mean square error:

[0101] Standardized root mean square error:

[0102] Cosine similarity:

[0103] y i This represents the true value of the response of the focus point at time i. y represents the virtual value of the improved attention response at time i. y represents the measured attention response. The improved focus response virtual data has a sample length of n.

[0104] Step S5: Construct a high-precision proxy model of measurable point data (actual measurement) and interest point data (generated), and input the real-time measurable point data into the proxy model of the two to realize real-time measurement of the response of interest points.

[0105] This step utilizes an LSTM neural network to establish a proxy model between the measured data of measurable points and the virtual data of points of interest. The specific implementation process is as follows:

[0106] Constructing LSTM neural network model structures (such as...) Figure 6 As shown, measured data from measurable points are used as input, and the generated responses of the points of interest are used as output to establish a surrogate model between the two. Supervised training is performed by adjusting the hyperparameters in the LSTM neural network model to minimize the difference between the output and the responses of the points of interest, constructing a nonlinear surrogate model between the input and output. The measured responses from the measurable points are used as input to the surrogate model, and the dynamic response of the points of interest is calculated in a very short time. The time history of the responses of the points of interest calculated based on the LSTM surrogate model is shown below. Figure 7 As shown in Table 4, the quantitative evaluation results of the accuracy of the LSTM surrogate model are listed.

[0107] Table 4 Quantitative evaluation results of the accuracy of the LSTM surrogate model

[0108]

[0109] In summary, this invention aims to achieve real-time indirect measurement of satellite structural response to interest. By utilizing finite element modeling, model correction, generative modeling, and surrogate modeling techniques, it ultimately overcomes the challenge of directly measuring satellite structural response to interest, enabling real-time indirect measurement of satellite structural response to interest.

[0110] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for real-time indirect measurement of satellite structural concern response, characterized in that, The method includes the following steps: Step S1: Sensor installation and classification. A limited number of sensors are deployed on the satellite structure, and the data acquired by the sensors are divided into two categories: data from measurable points and data from points of interest. Step S2: Acquisition of satellite structure dynamics data. Dynamic data of each measuring point of the satellite structure is obtained through dynamic experiments, and excitation data is obtained by sensors on the vibration table. During the experiment, each sensor needs to collect data synchronously. Step S3: Construct a finite element model of the satellite structure. Use simulation software to build the finite element model of the satellite structure, and correct the key parameters in the model using experimental data. Use the measured excitation data from the dynamic experiment as input to obtain virtual data corresponding to the installation positions of each sensor on the satellite structure; specifically: S31. Construct a three-dimensional model of the satellite structure and discretize it to determine the finite element nodes corresponding to the positions of the measurement points on the real satellite structure, and treat them as virtual measurement points; S32. Conduct sinusoidal sweep frequency and free vibration decay experiments to obtain experimental values ​​of the structure's natural frequency and overall decay rate, which are used to correct key parameters of the model. S33. Use optimization algorithms to correct key parameters of the satellite structure finite element model; The key parameters of the finite element model are corrected by using the measured natural frequency value and the overall free decay rate of the structure. The response surface methodology was used to correct key parameters. The natural frequency values ​​of the satellite structure were used to correct the material properties in the model, including elastic modulus, Poisson's ratio, and mass, while the free decay data were used to identify the overall damping of the structure. S34. Using the collected vibration table response as input, calculate the virtual dynamic data of each corresponding measuring point using the corrected finite element model; Step S4: Establish an error distribution model between the measured data and virtual data of the focus point using an LSTM-VAE neural network model. Then, use the virtual data of the focus point under different stimuli as input to generate high-quality dynamic response data of the focus point; specifically: S41. Establish an LSTM-VAE neural network model; The LSTM-VAE model replaces the feedforward neural network in VAE with LSTM. LSTM-VAE combines the local feature learning ability of VAE with the temporal modeling ability of LSTM neural network. The LSTM-VAE model can learn and capture complex temporal dependencies in time series data, thereby generating data with similar temporal dependencies. S42. Using the measured attention response as the output and the virtual attention data calculated by the finite element model as the input, the generator model between the input and output is obtained by adjusting the hyperparameters of the LSTM-VAE neural network model. The virtual response of the satellite structure calculated using the finite element model is defined as x, and the measured response is defined as y; according to the formula... and Standardize the data; The dataset is divided into training and validation sets, and the hyperparameters in the LSTM-VAE neural network model are tuned to minimize the difference between the output and the measured response during supervised training. The attention response generated by the LSTM-VAE model is defined as... S43. Quantitatively evaluate the accuracy of the generator model. If the accuracy does not meet the requirements, return to S42 to reset the new hyperparameter combination and train. The validation set is input into the constructed LSTM-VAE generator model to calculate the improved vibration response of the point of interest; the performance of the LSTM-VAE generator model is evaluated using correlation coefficient, root mean square error, standard root mean square error and cosine similarity metric. Step S5: Construct a high-precision proxy model of the measured measurable point data and the generated data of the points of interest. Input the real-time measurable point data into the proxy model of the two to realize real-time measurement of the response of the points of interest.

2. The method for real-time indirect measurement of satellite structure concern response according to claim 1, characterized in that, Step S1 specifically includes: S11. Select the measuring point and install the sensor; S12. Classified by sensor installation location: measurable point sensors and interest point sensors; measurable point data is used as input to the surrogate model, while the measured interest point data is the output data.

3. The method for real-time indirect measurement of satellite structure concern response according to claim 1, characterized in that, Step S2 specifically includes: S21. Use clamps to fix the satellite structure to the vibration table; The satellite structure and the shaking table are connected by a fixture to establish a satellite structure dynamics experimental system, and the shaking table is used to input dynamic loads. S22. An accelerometer is installed on the vibration table to collect excitation data applied to the satellite structure during random vibration. Sensors are installed on a vibration table to collect the response data of the vibration table during random vibration; the vibration direction collected by the sensors is consistent with the vibration direction of the vibration table during the experiment. S23. Conduct random vibration experiments and use a high-speed acquisition system to synchronously collect the dynamic response of each measuring point and the excitation data of the vibration table. S24. Classify the data collected by the sensor into data of interest points and data of measurable points.

4. The method for real-time indirect measurement of satellite structure concern response according to claim 1, characterized in that, Step S5 specifically includes: S51. Construct an LSTM neural network model; S52. Establish a proxy model of the actual measurable point response and the virtual concern point response using an LSTM neural network model; S53. Quantitative evaluation of the surrogate model's accuracy. If the model's accuracy does not meet the requirements, return to S52 to reselect a new combination of LSTM neural network hyperparameters for training. S54. Real-time measurable point response data is used as input to the surrogate model to achieve real-time indirect measurement of the response of the point of interest.

Images