A real-time monitoring method and system for deformation of antenna main reflector

Abstract

The present invention relates to a real-time monitoring method and system for the deformation of an antenna's main reflector. By establishing a finite element model of a parabolic antenna and applying gravity and temperature loads, the deformation of the antenna's main reflector under multi-factor loads is analyzed and fitted with an ideal antenna parabola, thereby obtaining a normal deformation directly corresponding to the required displacement of the actuator. A mapping relationship between local temperature and global temperature is established based on a machine learning algorithm. A relationship between the antenna's posture, the main reflector's temperature, and the surface deformation of the antenna's main reflector is established based on a deep learning network model. The temperature mapping model is combined with a deformation proxy model to calculate the deformation of the antenna's main reflector. The present invention can replace the finite element method within an effective accuracy range, greatly improving the calculation speed of the main reflector's deformation, compensating for the defects of the finite element method, and resolving the limitation of being unable to obtain the global temperature.

Description

Technical Field

[0001] The present invention relates to the field of deformation measurement of a parabolic antenna main reflector surface, and in particular to a real-time monitoring method and system for deformation of an antenna main reflector surface. Background Art

[0002] The surface accuracy of a parabolic antenna's primary reflector determines its performance and observation efficiency. Ideally, the primary reflector should be a perfect parabola. However, due to mechanical errors from machining and assembly, as well as systematic errors caused by gravity, temperature, wind load, and other dynamic loads, large-aperture reflectors can exhibit surface errors. Mechanical errors from machining and assembly can be reduced by improving manufacturing and assembly accuracy. However, the constant changes in the antenna's operating posture and variations in external climate make dynamic loads such as gravity, temperature, and wind load unavoidable. Of these loads, gravity has the greatest impact on the antenna structure and the surface accuracy of the primary reflector, followed by temperature.

[0003] To study the impact of dynamic loads on antenna surface errors, various observatories create finite element models of their radio telescopes and use finite element analysis to study the deformation and surface errors of the main reflector under different loads. Furthermore, to perform real-time corrections, large radio telescopes at home and abroad, such as the TM-65m (Tianma Radio Telescope, Shanghai Astronomical Observatory), the GBT-110m (Green Bank Telescope, National Astronomical Observatory), and the Effelsberg-100m (Effelsberg Telescope, European Astronomical Observatory), have installed actuators and other devices between the main reflector and the backing structure to form an active surface system to compensate for and correct deformation of the antenna reflector. Currently, active surface technology, which performs finite element analysis on antennas under a single load and uses the analysis results to correct the shape of the antenna's main reflector, is relatively mature. However, the shortcomings of multi-factor load coupling analysis and the inherent shortcomings of finite element technology have not yet been resolved, and there are still many deficiencies.

[0004] For the deformation analysis of large radio telescopes caused by dynamic loads, the most widely studied method is the single load analysis based on the finite element method. That is, a single dynamic load is applied to the antenna finite element model in the finite element software. For example, gravity is applied to study the deformation of the main reflector under different postures, or different temperatures are applied to study the deformation of the main reflector caused by different temperatures. However, in real environments, the deformation of the antenna main reflector is affected by many factors.

[0005] Whether it is thermal analysis or structural analysis, the finite element method itself has the disadvantage of a relatively long solution time. In addition, for structural deformation calculation, the finite element method requires a complex post-processing process after solution to finally calculate the displacement that needs to be adjusted for the main reflector. In addition, there must be errors between the finite element model and the actual antenna structure. The error between the deformation calculated using the finite element method and the actual deformation cannot be eliminated. Summary of the Invention

[0006] The purpose of the present invention is to provide a real-time monitoring method and system for the deformation of the main reflector surface of an antenna, which replaces the traditional finite element method, and performs real-time monitoring, correction and compensation for the deformation of the main reflector surface of an antenna caused by two main factors, gravity and temperature. While ensuring high accuracy, it has the advantages of short calculation time and high calculation speed.

[0007] To achieve the above object, the present invention provides the following solutions:

[0008] A method for real-time monitoring of deformation of an antenna main reflector surface, comprising:

[0009] Construct a three-dimensional finite element model of the antenna based on the actual antenna structure;

[0010] Setting different thermal environments for the three-dimensional finite element model of the antenna to obtain local and global temperatures under each of the thermal environments, and setting different postures for the three-dimensional finite element model of the antenna under each of the thermal environments to obtain displacement data of all nodes of the antenna under the influence of different combinations of gravity and temperature values;

[0011] Training a machine learning model based on the local temperature and the global temperature under different thermal environments to obtain a local-global temperature mapping model;

[0012] An ideal antenna parabola is fitted using the least squares method according to the displacement data, and the normal error value of each node between the ideal antenna parabola and the main reflector of the loaded deformed antenna is calculated to obtain calculated deformation data of the main reflector of the loaded deformed antenna under different numerical combinations; the main reflector of the loaded deformed antenna is a parabola that is deformed in the three-dimensional finite element model of the antenna; the ideal antenna parabola is a two-parameter ideal antenna parabola, a five-parameter ideal antenna parabola, or a six-parameter ideal antenna parabola;

[0013] The vertex of the main reflector when the pitch angle of the designed antenna parabola is 90 degrees is used as the origin o, the xoy coordinate plane is parallel to the parabola aperture plane when the pitch angle is 90 degrees, and the z coordinate axis is perpendicular to the aperture plane, and a global coordinate system of the main reflector is established; based on the global coordinate system of the main reflector, the y coordinate and z coordinate of each node under different postures are obtained;

[0014] Taking the y-coordinate and z-coordinate of each node and the global temperature as input and the calculated deformation data of the corresponding loaded deformed surface as output to train a deep learning network model to obtain a deformation proxy model of the main reflective surface;

[0015] The local-global temperature mapping model and the main reflector deformation proxy model are used to perform deformation analysis of the main reflector surface.

[0016] A real-time monitoring system for deformation of an antenna main reflector surface, comprising:

[0017] Finite element model building module, used to build a three-dimensional finite element model of the antenna based on the actual antenna structure;

[0018] a node displacement data acquisition module, configured to set different thermal environments for the antenna three-dimensional finite element model to obtain local and global temperatures under each of the thermal environments, and to set different postures for the antenna three-dimensional finite element model under each of the thermal environments to obtain displacement data for all nodes of the antenna under the influence of different combinations of gravity and temperature values;

[0019] a temperature mapping model establishment module, configured to train a machine learning model based on the local temperature and global temperature under different thermal environments to obtain a local-global temperature mapping model;

[0020] a deformation data calculation value acquisition module, configured to fit an ideal antenna parabola using the least squares method based on the displacement data, and calculate the normal error value of each node between the ideal antenna parabola and the main reflector of the loaded deformed antenna, thereby obtaining calculated deformation data of the main reflector of the loaded deformed antenna under different numerical value combinations; the main reflector of the loaded deformed antenna is the parabola that is deformed in the three-dimensional finite element model of the antenna; and the ideal antenna parabola is a two-parameter ideal parabola, a five-parameter ideal parabola, or a six-parameter ideal parabola;

[0021] A main reflector deformation proxy model establishment module is used to establish a global coordinate system for the main reflector with the vertex of the main reflector when the pitch angle of the designed antenna parabola is 90 degrees as the origin o, the xoy coordinate plane is parallel to the parabola aperture plane when the pitch angle is 90 degrees, and the z coordinate axis is perpendicular to the aperture plane; based on the global coordinate system of the main reflector, the y coordinate and z coordinate of each node under different postures are obtained; the deep learning network model is trained using the y coordinate and z coordinate of each node and the global temperature as input and the calculated deformation data of the corresponding loaded deformed surface as output to obtain the main reflector deformation proxy model;

[0022] The main reflector deformation calculation module is used to perform main reflector deformation analysis using the local-global temperature mapping model and the main reflector deformation proxy model.

[0023] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0024] The present invention relates to a real-time monitoring method and system for the deformation of an antenna's main reflector. By establishing a finite element model of a parabolic antenna and applying gravity and temperature loads, the deformation of the antenna's main reflector under multi-factor loads is analyzed, and the analysis results are used to fit an ideal antenna parabola to obtain a normal deformation directly corresponding to the displacement required by the actuator. A mapping relationship between the local temperature and the global temperature of a temperature feature point is established based on a machine learning algorithm. A relationship between the antenna's posture and the temperature of the main reflector and the surface deformation of the antenna's main reflector is established based on a deep learning network model. The temperature mapping model is combined with a deformation proxy model to calculate the deformation of the antenna's main reflector. The present invention can replace the finite element method within an effective accuracy range, greatly improving the calculation speed of the main reflector's deformation and overcoming the limitation of being unable to obtain the global temperature. The deformation of the antenna's main reflector can be quickly calculated with high accuracy using only the local temperature and the antenna's posture. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0026] Figure 1 A flow chart of a method for real-time monitoring of deformation of an antenna main reflector provided in Example 1 of the present invention;

[0027] Figure 2 This is a framework diagram of a method for real-time monitoring of deformation of the main reflector surface of an antenna provided in Example 1 of the present invention;

[0028] Figure 3 A schematic diagram of the basic structure of the TM-65m antenna provided in Example 1 of the present invention;

[0029] Figure 4 The temperature characteristic point distribution diagram provided in Example 1 of the present invention;

[0030] Figure 5 Schematic diagram of the positional relationship between the main reflector, the ideal antenna parabola, and the designed antenna parabola after being deformed by load provided in Example 1 of the present invention;

[0031] Figure 6 Schematic diagram of the six-parameter fitting method for an ideal parabola provided in Example 1 of the present invention;

[0032] Figure 7A schematic diagram of the residual block structure in the deep learning network provided in Example 1 of the present invention;

[0033] Figure 8 This is a deformation comparison diagram of the finite element method provided in Example 1 of the present invention and the deformation calculation model provided in the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] Studying the deformation of the main reflector caused by dynamic loads requires acquiring temperature data. However, for large-aperture reflectors, it is not possible to place temperature sensors in every area to obtain the global temperature. Only local temperatures can be obtained, representing the temperature of the entire reflector. This inevitably leads to large temperature measurement errors and affects the deformation analysis of the main reflector. Using the finite element method for global thermal analysis of the antenna cannot avoid the shortcomings of the finite element method. Currently, the study of the temperature of the main reflector of large-aperture antennas is still in the numerical analysis stage.

[0036] The purpose of this invention is to provide a real-time monitoring method and system for deformation of an antenna's main reflector. This method, which replaces the traditional finite element method, calculates deformation caused by gravity and temperature, two primary factors, and performs real-time correction and compensation. While ensuring high accuracy, it offers the advantages of short computation time and high speed. Furthermore, it eliminates the need to input the temperature of every point on the reflector, resolving the issue of global temperature access. Furthermore, it leverages the transfer learning capabilities of deep learning networks, allowing real-world data to be used to modify proxy models.

[0037] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Example 1

[0039] like Figure 1 and 2 As shown, this embodiment provides a method for real-time monitoring of deformation of an antenna main reflector surface, including:

[0040] S1: Construct a three-dimensional finite element model of the antenna based on the actual antenna structure.

[0041] This example uses the TM-65m (Tianma radio telescope) as an example. Based on the actual antenna structure, finite element software is used to build a 3D model of the radio telescope. Based on the actual antenna data, material parameters, finite element unit types, and real parameters are set. The finite element mesh is then generated to create the finite element model.

[0042] To facilitate understanding of the present invention, the basic structure of the parabolic antenna and the influence of dynamic load on the surface accuracy of the main reflector of the antenna are briefly introduced by taking the antenna TM-65m (Tianma radio telescope) as an example.

[0043] like Figure 3 (a) Figure 3 As shown in (b), Figure 3 (a) is a schematic diagram of the finite element structure of a large parabolic antenna with a pitch angle of 90 degrees in an embodiment of the present invention. Figure 3 (b) is a schematic diagram of the finite element structure of a large parabolic antenna with a 45-degree elevation angle in an embodiment of the present invention. The basic structure of the TM-65m consists of a back frame structure, a primary reflector, a secondary reflector, legs, an elevation structure, and an azimuth mount.

[0044] The antenna's back frame utilizes a spatial truss structure, with members primarily connected by tube-to-tube welding and reinforced ribs welded at the joints. The deformation and surface accuracy of the main reflector determine the antenna's operating efficiency. The main reflector is designed to be divided into 14 circles, comprising 1,008 panels. The panels average 3.3 square meters in area. Between the main reflector and the back frame lies an active surface system composed primarily of actuators. A total of 1,104 motor-driven actuators enable fine-tuning of each panel on the antenna's main reflector to ensure antenna efficiency. The pitch mechanism primarily consists of a pitch platform, a sector gear, and a pitch axis. The pitch platform directly supports the back frame structure; the pitch axis is connected to the azimuth mount and transfers the weight of the pitch mechanism and back frame structure to the azimuth mount. The sector gear, connected to a drive rod, enables the pitch mechanism to rotate the back frame structure around the pitch axis, thereby changing the antenna's pitch angle. A counterweight is installed at the lower end of the sector gear to ensure the center of gravity of the back frame and pitch mechanism is centered on the pitch axis. The azimuth mount is supported by six rollers at the bottom, including four driving rollers and two driven rollers. The driving rollers drive the mount to achieve horizontal rotation around the central pivot.

[0045] Antenna reflector surface deformation is primarily caused by machining and assembly errors, gravity, temperature, wind load, and other dynamic loads. Mechanical errors such as machining and assembly can be reduced by improving manufacturing and assembly precision. However, the constant changes in the antenna's operating attitude and climatic variations make dynamic loads such as gravity, temperature, and wind load unavoidable. Of these loads, gravity has the greatest impact on the antenna structure and the surface accuracy of the primary reflector, followed by temperature. Considering only deformation due to gravity, the maximum RMS deviation occurs at the extremes of pointing up and pointing horizontally, approximately 1mm, when the antenna's elevation angle varies from 5 to 90 degrees. Considering temperature unevenness and temperature differences, the maximum RMS deviation is approximately 0.5mm for a 40-degree temperature difference. When gravity and temperature are combined, the maximum RMS deviation is approximately 2mm, resulting in observation efficiency that is only half of the ideal efficiency. Because the antenna's operating attitude requires adjustment and the ambient temperature fluctuates rapidly, real-time measurement of the antenna reflector surface deformation is necessary to ensure antenna efficiency.

[0046] S2: Setting different thermal environments for the three-dimensional finite element model of the antenna to obtain the local temperature and global temperature under each thermal environment, and setting different postures for the three-dimensional finite element model of the antenna under each thermal environment to obtain the displacement data of all nodes of the antenna under the influence of different numerical combinations of gravity and temperature.

[0047] Different types of loads can be applied to the established three-dimensional finite element model of the radio telescope, including gravity and temperature, which have the greatest impact on the accuracy of the radio telescope's main reflector. When the antenna's operating attitude (pitch angle) changes, the main reflector and back frame structure move up and down, causing the weight distribution of the entire antenna to change. This deformation affects the surface accuracy and operating efficiency. Therefore, the antenna pitch angle is used as a variable to control the antenna's gravity distribution. In addition, thermal deformation caused by temperature changes also affects the surface accuracy of the main reflector. The deformation caused by gravity and temperature does not superimpose linearly, so it is necessary to comprehensively consider the effects of gravity (pitch angle) and temperature and perform a thermal-structural coupling analysis.

[0048] The antenna's attitude, or weight distribution, is determined by modifying the pitch angle of the antenna's three-dimensional finite element model. Finite element software can be used to change the thermal environment of the three-dimensional finite element model. The temperature distribution under this thermal environment can be obtained through finite element thermal analysis, and the temperatures of all nodes in the model are then stored in the temperature load file. Therefore, in the operation of applying gravity loads and temperature loads, applying global gravity and temperature loads to the finite element model can be done by setting different thermal environments for the finite element model and performing temperature simulation. By changing the thermal environment, the temperature of the antenna is simulated. After the finite element thermal analysis is solved, the temperature of the entire antenna can be obtained and the temperature data can be exported. At the same time, the pitch angle, or attitude, of the antenna is changed under different thermal environments to change the distribution of gravity, thereby completing the application of gravity loads and thermal loads. Finally, the finite element structural analysis solver is used for solution, and the result of the finite element solution is the displacement of each node when the antenna is loaded.

[0049] In view of the many difficulties in obtaining the temperature of the main reflector of large parabolic antennas and the defect that the global temperature cannot be accurately and quickly obtained, a clustering algorithm or the manual arrangement of the positions of temperature feature points is used. The temperatures of a small number of temperature feature points and the machine learning - XGBoost (Extreme GradientBoosting) method are used to achieve the mapping of local temperature to global temperature while meeting the accuracy requirements.

[0050] In this embodiment, setting different thermal environments for the antenna three-dimensional finite element model in step S2 to obtain the local temperature under each thermal environment specifically includes:

[0051] (1) With the goal of minimizing the sum of the squares of the distances between each node on the main reflection surface of the antenna three-dimensional finite element model and the central node of the region to which it belongs, a clustering algorithm is used to evenly divide the main reflection surface of the antenna three-dimensional finite element model into multiple regions.

[0052] like Figure 4 As shown in the figure, the main reflector of the antenna is divided into 120 uniform areas (determined by the number of temperature sensors that can be actually arranged) through the clustering algorithm. The clustering goal is to minimize the sum of the squares of the distances between each node on the reflector and the center point of the area to which it belongs. Calculate, where x i is the i-th node, c i is x i The area to which it belongs, is the center point of the region, M is the total number of main reflection surface nodes. The center point of each cluster is selected as a temperature feature point.

[0053] (2) Obtain the temperature of the central node (temperature characteristic point) of each of the regions to obtain the local temperature.

[0054] In addition to the aforementioned method for selecting temperature characteristic points, temperature sensors can also be manually positioned as temperature characteristic points based on the antenna's structure. Specifically, multiple temperature sensors are evenly arranged on the main reflective surface of the antenna's three-dimensional finite element model, and the temperature of each temperature sensor is obtained, which is the local temperature.

[0055] S3: training a machine learning model according to the local temperature and the global temperature under different thermal environments to obtain a local-global temperature mapping model;

[0056] The temperature of the temperature feature point in step 2 is used as the input of the model, and the global temperature is used as the output to train the machine learning model - XGBoost. XGBoost integrates multiple weak learners (gradient boosting regression tree model) according to the ensemble learning idea and combines them into a strong learner with better accuracy and robustness. The objective function of XGBoost can be expressed as Indicates that is the representation of the linear space of the objective function, y i is the actual value, Represents the i-th sample x i The calculated value of , where K is the number of weak learners, Is the loss function, which represents the error between the calculated value and the actual value. k ω(f k ) is a regularization term, where ω(f k ) is the complexity of each weak learner, that is, the weight of the weak learner, and the regularization term is used to prevent overfitting.

[0057] S4: Fitting an ideal antenna parabola using the least squares method according to the displacement data, and calculating the normal error value of each node between the ideal antenna parabola and the main reflector of the loaded deformed antenna, to obtain the calculated deformation data of the main reflector of the loaded deformed antenna under different numerical combinations; the main reflector of the loaded deformed antenna is the parabola that is deformed in the three-dimensional finite element model of the antenna; the ideal antenna parabola is a two-parameter ideal antenna parabola, a five-parameter ideal antenna parabola or a six-parameter ideal antenna parabola.

[0058] The finite element solution results in the displacement of each node when the antenna is loaded. For an adjustable main reflector with an active surface system, compensating for the main reflector's deformation using the displacements calculated by the finite element solver can restore the deformed surface to the designed antenna parabola. However, for large antennas, adjusting the main reflector from a deformed surface to the designed antenna parabola is prohibitively laborious. Therefore, simply adjusting the main reflector to the ideal antenna parabola is sufficient, significantly reducing the workload on the actuators.

[0059] The initial position coordinates and displacement data of all nodes on the main reflector of the antenna (1104 nodes corresponding to 1104 actuators) are exported to fit the ideal antenna parabola and calculate the deformation that needs to be compensated for the main reflector. After the load deformation, the ideal antenna parabola no longer coincides with the initial design antenna parabola, see Figure 5 , we need to use the least square method to obtain the unknown parameters of the ideal antenna parabola. Specifically, Figure 6 As shown, in step S4, fitting the ideal antenna parabola using the least squares method according to the displacement data specifically includes:

[0060] (1) The objective function is to minimize the sum of squares of the normal errors of the ideal antenna parabola and the loaded deformed surface at each node. The sum of squares of the normal errors can be calculated according to Calculation, where M is the number of nodes, Δn i is the normal error of the node, i.e. the required displacement of the actuator.

[0061] (2) fitting the six-parameter ideal antenna parabola using the least squares method according to the displacement data; the expression of the six-parameter ideal antenna parabola is:

[0062]

[0063] Among them, (x, y, z) is the node coordinate of the main reflector of the antenna when it is not loaded, which is a known parameter; f is the focal length of the main reflector of the antenna when it is not loaded, which is a known parameter; (u a ,v a ,w a ) is the displacement of the vertex of the ideal antenna parabola after deformation under load, and is the parameter to be solved; is the rotation angle of the ideal antenna parabola around the x and y axes after the load deformation, which is the parameter to be solved; Δf is the change in the focal length of the ideal antenna parabola corresponding to the main reflection surface of the loaded deformed antenna, which is the parameter to be solved.

[0064] According to the extreme value theorem, we can Obtain the six parameters when the objective function takes the minimum value.

[0065] The ideal antenna parabola to be fitted can also be a two-parameter ideal antenna parabola or a five-parameter ideal antenna parabola. The parameter fitting method can be used Represents a two-parameter ideal antenna parabola; a five-parameter fitting method can be used represents the five-parameter ideal antenna parabola;

[0066] Based on the above expression for the ideal antenna parabola and the locations of the nodes on the deformed surface, the least squares method can be used to solve for the unknown parameters in the expression and fit the ideal antenna parabola. The deformation that needs to be compensated for on the main reflector is the normal difference between the deformed surface after loading and the fitted ideal antenna parabola.

[0067] In step S4, the calculation of the normal error value of each node between the ideal antenna parabola and the main reflector of the loaded deformed antenna specifically includes:

[0068] Calculating the z-coordinate difference between the main reflective surface of the loaded deformed antenna and the nodes in the ideal antenna parabola;

[0069] Calculate the direction cosine value of the normal vector according to the focal length of the main reflector of the antenna when it is not loaded and the horizontal and vertical coordinate values ​​of each node of the main reflector of the antenna when it is not loaded;

[0070] The normal error value of each node is calculated according to the z coordinate difference of each node and the direction cosine value of the normal vector.

[0071] The expression of the normal error value is:

[0072] Δn=Δzcosα;

[0073] Among them, Δz is the axial error in the z-axis direction, which is calculated based on the z-coordinate difference between the main reflector of the loaded deformed antenna and the ideal antenna parabola. cosα is the direction cosine of the normal vector. Calculate, f is the focal length of the main reflector of the antenna when it is unloaded; x and y are the horizontal and vertical coordinates of each node of the main reflector of the antenna when it is unloaded, obtained in the local coordinate system.

[0074] After obtaining the normal error value, the required adjustment amount of the actuator of the main reflector of the parabolic antenna is obtained. The adjustment amount is the deformation that needs to be compensated for the main reflector, that is, the actual deformation of the main reflector of the large antenna, which is used as a data set for subsequent deep learning network model training.

[0075] S5: Establishing a global coordinate system for the main reflector with the vertex of the main reflector when the pitch angle of the designed antenna parabola is 90 degrees as the origin o, an xoy coordinate plane parallel to the parabola aperture plane when the pitch angle is 90 degrees, and a z coordinate axis perpendicular to the aperture plane; and obtaining the y coordinate and z coordinate of each node at different postures based on the global coordinate system of the main reflector. The designed antenna parabola is the ideal antenna parabola when unloaded.

[0076] According to the main factors affecting deformation - gravity and temperature, and the deformation caused by gravity is mainly caused by the change of the antenna pitch angle, and the change of the pitch angle can be fully represented by the y and z coordinates of each node on the main reflector of the antenna. The input of the deep learning network is determined to be the y coordinate, z coordinate and global temperature of each node on the main reflector of the antenna in the global coordinate system in the current state. The output is determined to be the normal error between the main reflector and the ideal antenna parabola after loading. That is, the y coordinate and z coordinate of each node and the global temperature are used as input, and the calculated deformation data of the corresponding loaded deformation surface are used as output to train the deep learning network model to obtain the main reflector deformation proxy model;

[0077] The y and z coordinates of nodes on the main reflector are directly obtained from the 3D finite element model. The global temperature is obtained through finite element thermal simulation. The normal error between the main reflector and the ideal antenna parabola is calculated using finite element structural simulation and least squares calculations. Using interpolation, the input and output data are converted from 1104 node data to 128*128 image data. The size of each pixel in the image represents the information of that point, that is, the input dimension is 128*128*3, and the output dimension is 128*128*1. The value of points outside the circular boundary of the antenna main reflector is set to 0.

[0078] All data are preprocessed and normalized to eliminate the dimensional differences between different types of data. Scale the range of each type of data to [0,1]. For one type of data (y coordinate, z coordinate, temperature, deformation), x o represents unnormalized data, x new Represents the normalized data, x min Represents the minimum value of this type of data, x max Represents the maximum value of this type of data. For example, when normalizing temperature data, the x value is the temperature data, and x max That is the maximum value in the temperature data; when the y coordinate data is normalized, the x value is the y coordinate data, x max This is the maximum value in the y-coordinate data.

[0079] For the determination of the deep learning network model: set the network structure, network depth, and network parameters (learning rate, optimizer) of the deep learning network.

[0080] The network structure is built using tensorflow2.0. According to the characteristics of the input and output data, a U-shaped deep network is selected, which includes upsampling and downsampling parts. Each sampling layer contains a convolution layer and a normalization layer to prevent gradient explosion and gradient disappearance while extracting and calculating features.

[0081] Based on the above structure, a residual block is added between downsampling and upsampling to increase the network depth and prevent the model effect from decreasing. Figure 7 The residual block adds skip connections to the original network structure, so that the output of the residual block contains both shallow features and deep features after convolution, further improving the calculation accuracy and robustness of the deep learning network.

[0082] Preferably, for the above optimization goals, the Adam optimization algorithm is used to iteratively train the deep network model to obtain the optimal network connection parameters, and as the number of training times increases, the learning rate is automatically adjusted and continuously reduced, and finally the objective function loss converges to obtain deformation calculation within effective accuracy.

[0083] Construct the loss function of the model loss = α*mse+β*(1-ssim), making it the optimization target of the deep learning network model, where represents the mean square error between the deformation calculated by the deep learning model and the true deformation, where Indicates the overall similarity between the output deformation image and the true deformation image (0≤ssim≤1). The closer it is to 1, the more similar the two images are. Indicates the brightness similarity of the two deformed images, represents the contrast similarity of the two deformed images, It represents the structural similarity of the two deformed images, and α and β represent the weights of the mean square error and the image similarity.

[0084] In this embodiment, after obtaining the main reflector deformation proxy model, in order to ensure the accuracy of the model calculation, this embodiment also involves adjusting the model parameters using real antenna deformation data. Using a transfer learning method, the antenna main reflector data obtained from actual measurements is used to correct the real-time deformation calculation model. Specifically, the real posture data of the real antenna structure, the local temperature, and the actual deformation data at the same time are obtained.

[0085] Inputting the local temperature data into the local-global temperature mapping model to obtain the corresponding global temperature;

[0086] Inputting the true value of the y coordinate and the true value of the z coordinate of each node obtained based on the true posture data into the main reflection deformation proxy model in combination with the global temperature to obtain calculated deformation data;

[0087] Adjusting the parameters of the main reflection deformation proxy model according to the error between the actual deformation data and the calculated deformation data to obtain a parameter-adjusted main reflection deformation proxy model;

[0088] The deformation analysis of the main reflector surface is performed using the local-global temperature mapping model and the adjusted main reflector deformation proxy model.

[0089] S6: Perform deformation analysis of the main reflector surface using the local-global temperature mapping model and the main reflector deformation proxy model.

[0090] Combine the main reflector deformation proxy model with the temperature mapping model, obtain the y and z coordinates of all nodes on the reflector surface according to the antenna posture, obtain the global temperature data within the effective accuracy according to the temperature of the temperature feature point and the temperature mapping model, take the y, z coordinates and the mapped global temperature as input, and use the deformable proxy model trained and verified in step 5 to obtain the main reflector deformation distribution within the effective accuracy, thus realizing a complete real-time deformation calculation model. See the effect diagram for details. Figure 8 , Figure 8 (a) is the deformation diagram calculated by the finite element method. Figure 8 (b) is a diagram showing the test effect of the deformation calculation model provided by an embodiment of the present invention.

[0091] Compared with the prior art, the present invention has the following beneficial effects:

[0092] 1. The present invention considers the combined effects of gravity and temperature on the surface accuracy of the main reflector of a parabolic antenna. Based on the principle, the present invention analyzes these two main factors that cause deformation of the main reflector of the antenna and calculates the surface deformation.

[0093] 2. The present invention only needs to use the current position of the antenna and a small amount of local temperature data of the temperature characteristic points to map the global temperature and the deformation of the entire main reflecting surface with high accuracy. The deformation data directly corresponds to the adjustment amount required by the actuator.

[0094] 3. The present invention utilizes deep learning and machine learning methods, which is less time-consuming than the finite element calculation method. It can calculate the deformation of the main reflector in real time according to the current position and temperature of the antenna, thereby making timely compensation and correction.

[0095] 4. The present invention utilizes the transfer learning method to compensate for the calculation error caused by the error between the finite element model and the actual antenna structure, and the calculation model can be adjusted according to the actual antenna data.

[0096] This embodiment can replace the finite element method within the effective accuracy range, greatly improving the calculation speed of the deformation of the main reflector, compensating for the defects of the finite element method, and solving the limitation of being unable to obtain the global temperature. The deformation of the main reflector surface of the antenna can be quickly calculated with high accuracy only through the local temperature and antenna posture.

[0097] Example 2

[0098] This embodiment provides a real-time monitoring system for deformation of an antenna main reflector, including:

[0099] The finite element model building module M1 is used to build a three-dimensional finite element model of the antenna according to the actual antenna structure.

[0100] The node displacement data acquisition module M2 is used to set different thermal environments for the antenna three-dimensional finite element model to obtain the local temperature and global temperature in each thermal environment, and to set different postures for the antenna three-dimensional finite element model in each thermal environment to obtain the displacement data of all nodes of the antenna under the influence of different numerical combinations of gravity and temperature.

[0101] The temperature mapping model establishment module M3 is used to train a machine learning model according to the local temperature and global temperature under different thermal environments to obtain a local-global temperature mapping model.

[0102] The deformation data calculation value acquisition module M4 is used to fit the ideal antenna parabola using the least squares method based on the displacement data, and calculate the normal error value of each node between the ideal antenna parabola and the main reflector of the loaded deformed antenna, so as to obtain the calculated deformation data of the main reflector of the loaded deformed antenna under different numerical combinations; the main reflector of the loaded deformed antenna is the parabola that has been deformed in the three-dimensional finite element model of the antenna; the ideal antenna parabola is a two-parameter ideal parabola, a five-parameter ideal parabola or a six-parameter ideal parabola.

[0103] Wherein, the deformation data calculation value acquisition module includes an ideal antenna parabola fitting submodule;

[0104] The ideal antenna parabola fitting submodule is used to take the minimum sum of squares of normal errors of each node between the ideal antenna parabola and the loaded deformed surface as the objective function; and to fit the six-parameter ideal antenna parabola using the least squares method based on the displacement data. The expression of the six-parameter ideal antenna parabola is:

[0105]

[0106] Where (x, y, z) is the node coordinate of the main reflector of the antenna when it is not loaded; f is the focal length of the main reflector of the antenna when it is not loaded; (u a ,v a ,w a ) is the displacement of the apex of the ideal antenna parabola after deformation due to load; is the rotation angle of the ideal antenna parabola around the x and y axes after the load deformation; Δf is the change in the focal length of the ideal antenna parabola corresponding to the main reflection surface of the loaded deformed antenna.

[0107] The main reflector deformation proxy model establishment module M5 is used to establish a global coordinate system for the main reflector with the vertex of the main reflector when the pitch angle of the designed antenna parabola is 90 degrees as the origin o, the xoy coordinate plane is parallel to the parabola aperture plane when the pitch angle is 90 degrees, and the z coordinate axis is perpendicular to the aperture plane; based on the global coordinate system of the main reflector, the y coordinate and z coordinate of each node under different postures are obtained; the y coordinate and z coordinate of each node and the global temperature are used as input, and the calculated deformation data of the corresponding loaded deformation surface is used as output to train a deep learning network model to obtain the main reflector deformation proxy model.

[0108] The main reflector deformation calculation module M6 is used to perform main reflector deformation analysis using the local-global temperature mapping model and the main reflector deformation proxy model.

[0109] As for the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0110] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for real-time monitoring of deformation of the main reflector surface of an antenna, characterized in that: include: Construct a three-dimensional finite element model of the antenna based on the actual antenna structure; Setting different thermal environments for the three-dimensional finite element model of the antenna to obtain local and global temperatures under each of the thermal environments, and setting different postures for the three-dimensional finite element model of the antenna under each of the thermal environments to obtain displacement data of all nodes of the antenna under the influence of different combinations of gravity and temperature values; Training a machine learning model based on the local temperature and global temperature under different thermal environments to obtain a local-global temperature mapping model; An ideal antenna parabola is fitted using the least squares method according to the displacement data, and the normal error value of each node between the ideal antenna parabola and the main reflector of the loaded deformed antenna is calculated to obtain calculated deformation data of the main reflector of the loaded deformed antenna under different numerical combinations; the main reflector of the loaded deformed antenna is a parabola that is deformed in the three-dimensional finite element model of the antenna; the ideal antenna parabola is a two-parameter ideal antenna parabola, a five-parameter ideal antenna parabola, or a six-parameter ideal antenna parabola; The vertex of the main reflector when the elevation angle of the designed antenna parabola is 90 degrees is taken as the origin o, the xoy coordinate plane is parallel to the parabola aperture when the elevation angle is 90 degrees, and the z coordinate axis is perpendicular to the aperture plane, so as to establish the global coordinate system of the main reflector; The y coordinate and the z coordinate of each node under different postures are obtained based on the global coordinate system of the main reflector; the designed antenna parabola is the ideal antenna parabola when it is not loaded; Taking the y-coordinate and z-coordinate of each node and the global temperature as input, and taking the calculated deformation data of the corresponding main reflector of the loaded deformed antenna as output, a deep learning network model is trained to obtain a main reflector deformation proxy model; Performing deformation analysis of the main reflector surface using the local-global temperature mapping model and the main reflector surface deformation proxy model; The calculating of the normal error value of each node between the ideal antenna parabola and the main reflector of the loaded deformed antenna specifically includes: Calculating the z-coordinate difference between the main reflective surface of the loaded deformed antenna and the nodes in the ideal antenna parabola; Calculate the direction cosine value of the normal vector according to the focal length of the main reflector of the antenna when it is not loaded and the horizontal and vertical coordinate values ​​of each node of the main reflector of the antenna when it is not loaded; The normal error value of each node is calculated according to the z coordinate difference of each node and the direction cosine value of the normal vector.

2. The method according to claim 1, characterized in that The step of setting different thermal environments for the antenna three-dimensional finite element model to obtain the local temperature under each thermal environment specifically includes: The main reflection surface of the three-dimensional finite element model of the antenna is evenly divided into multiple regions using a clustering algorithm with the goal of minimizing the sum of the squares of the distances between each node and the central node of the region to which it belongs; The temperature of the central node of each of the regions is acquired to obtain the local temperature.

3. The method according to claim 1, characterized in that The step of setting different thermal environments for the antenna three-dimensional finite element model to obtain the local temperature under each thermal environment specifically includes: A plurality of temperature sensors are evenly arranged on the main reflection surface of the three-dimensional finite element model of the antenna, and the temperature of each temperature sensor is obtained, that is, the local temperature.

4. The method according to claim 1, wherein The fitting of the ideal antenna parabola by the least square method according to the displacement data specifically includes: The objective function is to minimize the sum of squares of normal errors of each node between the ideal antenna parabola and the loaded deformed surface; The six-parameter ideal antenna parabola is fitted using the least squares method according to the displacement data; the expression of the six-parameter ideal antenna parabola is: ； in, is the node coordinate of the main reflector surface of the antenna when it is not loaded; is the focal length of the main reflector of the antenna when it is unloaded; is the displacement of the apex of the ideal antenna parabola after deformation due to load; The ideal parabola of the antenna after deformation under load The angle of rotation of the axis; is the focal length change of the ideal antenna parabola corresponding to the main reflection surface of the loaded deformed antenna.

5. The method according to claim 4, characterized in that The expression of the two-parameter ideal antenna parabola is: ； The expression of the five-parameter ideal antenna parabola is: 。 6. The method according to claim 1, wherein The expression of the normal error value is: ； in, for The axial error in the axis direction is calculated based on the difference in the z coordinates of the nodes before and after loading. is the direction cosine of the normal vector, according to calculate, is the focal length of the main reflector of the antenna when it is unloaded; x and y are the horizontal and vertical coordinates of each node of the main reflector of the antenna when it is unloaded.

7. The method according to claim 1, characterized in that Before performing the main reflector deformation analysis using the local-global temperature mapping model and the main reflector deformation proxy model, the method further includes: Acquiring the real attitude data, the local temperature and the actual deformation data of the real antenna structure at the same moment; Inputting the local temperature data into the local-global temperature mapping model to obtain the corresponding global temperature; Inputting the true value of the y coordinate and the true value of the z coordinate of each node obtained based on the true posture data into the main reflective surface deformation proxy model in combination with the global temperature to obtain calculated deformation data; Adjusting the parameters of the main reflector deformation proxy model according to the error between the actual deformation data and the calculated deformation data to obtain a parameter-adjusted main reflector deformation proxy model; The deformation analysis of the main reflector surface is performed using the local-global temperature mapping model and the adjusted main reflector surface deformation proxy model.

8. A system based on the method according to any one of claims 1 to 7, characterized in that: include: Finite element model building module, used to build a three-dimensional finite element model of the antenna based on the actual antenna structure; a node displacement data acquisition module, configured to set different thermal environments for the antenna three-dimensional finite element model to obtain local and global temperatures under each of the thermal environments, and to set different postures for the antenna three-dimensional finite element model under each of the thermal environments to obtain displacement data for all nodes of the antenna under the influence of different combinations of gravity and temperature values; a temperature mapping model establishment module, configured to train a machine learning model based on the local temperature and global temperature under different thermal environments to obtain a local-global temperature mapping model; a deformation data calculation value acquisition module, configured to fit an ideal antenna parabola using a least squares method based on the displacement data, and calculate the normal error value of each node between the ideal antenna parabola and the main reflector of the loaded deformed antenna, thereby obtaining calculated deformation data of the main reflector of the loaded deformed antenna under different numerical value combinations; the main reflector of the loaded deformed antenna is the parabola that is deformed in the three-dimensional finite element model of the antenna; and the ideal antenna parabola is a two-parameter ideal antenna parabola, a five-parameter ideal antenna parabola, or a six-parameter ideal antenna parabola; The module for establishing the deformation proxy model of the main reflector is used to establish the global coordinate system of the main reflector with the vertex of the main reflector when the pitch angle of the designed antenna parabola is 90 degrees as the origin o, the xoy coordinate plane being parallel to the parabola aperture plane when the pitch angle is 90 degrees, and the z coordinate axis being perpendicular to the aperture plane; The y coordinate and z coordinate of each node under different postures are obtained based on the global coordinate system of the main reflector; the y coordinate and z coordinate of each node and the global temperature are used as input, and the calculated deformation data of the corresponding loaded deformed main reflector of the antenna are used as output to train a deep learning network model to obtain a main reflector deformation proxy model; the designed antenna parabola is the ideal antenna parabola when it is unloaded; The main reflective surface deformation calculation module is used to perform main reflective surface deformation analysis using the local-global temperature mapping model and the main reflective surface deformation proxy model.

9. The system according to claim 8, characterized in that The deformation data calculation value acquisition module includes an ideal antenna parabola fitting submodule; The ideal antenna parabola fitting submodule is used to take the minimum sum of squares of normal errors of each node between the ideal antenna parabola and the loaded deformed surface as the objective function; and to fit the six-parameter ideal antenna parabola using the least squares method based on the displacement data. The expression of the six-parameter ideal antenna parabola is: ； in, is the node coordinate of the main reflector surface of the antenna when it is not loaded; is the focal length of the main reflector of the antenna when it is unloaded; is the displacement of the apex of the ideal antenna parabola after deformation due to load; The ideal parabola of the antenna after deformation under load The angle of rotation of the axis; is the focal length change of the ideal antenna parabola corresponding to the main reflection surface of the loaded deformed antenna.

Images