A synthetic aperture radiometer array optimization method and related equipment

Abstract

The present invention provides a method for optimizing an aperture-synthetic radiometer array and related equipment. The method includes obtaining an action for adjusting an initial aperture-synthetic radiometer array based on a strategy network, and adjusting to obtain multiple intermediate aperture-synthetic radiometer arrays; designing a reward function based on the difference between the initial aperture-synthetic radiometer array and the intermediate aperture-synthetic radiometer array, and calculating the reward value of the action; constructing an array optimization model; training the array optimization model according to the action, the reward value corresponding to each action, and the intermediate aperture-synthetic radiometer array; inputting the aperture-synthetic radiometer array to be optimized into the trained array optimization model, determining the action corresponding to the maximum Q value in the output result of the array optimization model as the optimization strategy of the aperture-synthetic radiometer array to be optimized, and optimizing the aperture-synthetic radiometer array to be optimized according to the optimization strategy. The present invention can improve the effect of the aperture-synthetic radiometer array.

Description

Technical Field

[0001] The invention belongs to the field of computer and communication technology, and in particular relates to a synthetic aperture radiometer array optimization method and related equipment. Background Art

[0002] Passive microwave remote sensing technology has made great progress in recent years, but low spatial resolution is still the main problem of low-frequency passive microwave remote sensing. Technicians in related fields have come up with the idea of ​​combining multiple small antenna arrays into a large antenna array - Aperture Synthesis Radiometer (ASR). This combined antenna array has higher spatial resolution and lower system complexity.

[0003] As a key technology of ASR, the arrangement of antenna arrays has an important impact on the performance of the system. Generally, different array arrangements refer to different numbers, spacings, and spatial distributions of antennas, and these different factors correspond to different baseline combinations. The baseline combination will affect the characteristics of array factors such as the sidelobe peak and main beam width of the ASR system, thereby affecting the system's angular resolution and sensitivity and other performance indicators. In practical applications, the non-uniform arrangement of antennas will lead to a decrease in the spatial frequency domain sampling coverage of the system, which in turn will lead to a decrease in the quality of the reconstructed brightness temperature image. Therefore, it is necessary to improve the spatial frequency domain sampling coverage of the system by optimizing the non-uniform array.

[0004] At present, the optimization algorithms commonly used in the ASR array arrangement optimization problem include particle swarm optimization algorithm, genetic algorithm, ant colony algorithm, simulated annealing algorithm and taboo search, etc. However, these methods ignore the system spatial frequency domain sampling coverage and the quality of the reconstructed brightness temperature image, resulting in poor results of the generated synthetic aperture radiometer array. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a synthetic aperture radiometer array optimization method and related equipment to improve the effect of the synthetic aperture radiometer array.

[0006] In a first aspect, the present invention provides a method for optimizing a synthetic aperture radiometer array, the method comprising the following steps:

[0007] Randomly initialize the positions of multiple aperture synthesis radiometers to obtain an initial aperture synthesis radiometer array;

[0008] Based on the strategy network, a plurality of actions for adjusting the positions of the plurality of aperture synthesis radiometers are obtained, and the initial aperture synthesis radiometer array is adjusted according to each action to obtain a plurality of intermediate aperture synthesis radiometer arrays; wherein the action is set as the position change amount of each aperture synthesis radiometer, and the plurality of actions correspond to the plurality of intermediate aperture synthesis radiometer arrays one by one;

[0009] For each intermediate aperture radiometer array, a reward function is designed based on the difference between the initial aperture radiometer array and the intermediate aperture radiometer array, and the reward value of the action corresponding to the intermediate aperture radiometer array is calculated according to the reward function; the difference includes the difference in spatial frequency domain sampling coverage and the difference in brightness temperature image. The spatial frequency domain sampling coverage refers to the proportion of grids containing actual sampling frequency points after the rectangular area formed within the maximum sampling frequency range of the radiometer system is gridded in the aperture radiometer.

[0010] Based on the Q-Learning algorithm, the array optimization model is constructed with the initial aperture synthesis radiometer array as input and the reward value obtained by adjusting the positions of multiple aperture synthesis radiometers as output.

[0011] The array optimization model is trained according to a plurality of actions, a reward value corresponding to each action, and an intermediate synthetic aperture radiometer array corresponding to each action until a loss value of the array optimization model is less than a preset loss threshold, thereby obtaining a trained array optimization model;

[0012] The integrated aperture radiometer array to be optimized is input into the trained array optimization model, the action corresponding to the maximum Q value in the output result of the array optimization model is determined as the optimization strategy of the integrated aperture radiometer array to be optimized, and the integrated aperture radiometer array to be optimized is optimized according to the optimization strategy.

[0013] Optionally, the policy network is an ε-greedy policy network.

[0014] Optionally, the reward function can be expressed as:

[0015]

[0016] in, Indicates execution of an action The corresponding reward value is represents the weight, Indicates the difference in spatial frequency domain sampling coverage, , Indicates action The corresponding intermediate aperture radiometer array The spatial frequency domain sampling coverage, represents the spatial frequency domain sampling coverage of the initial synthetic aperture radiometer array, express The corresponding brightness temperature image and The mean square error between the corresponding brightness temperature images.

[0017] Optionally, the loss value calculation expression of the array optimization model is:

[0018]

[0019] in, The model parameters are The loss value of the array optimization model, Expressing hope, represents the reward at time step t, represents the discount factor, Indicates the next state The optimal action that maximizes the Q value is represents the target Q value, represents the parameters of the target Q network, Indicates the Q value of the current network.

[0020] Optionally, use the Adam optimization algorithm to optimize the model parameters of the array optimization model to update.

[0021] Optionally, the expression for spatial frequency domain sampling coverage is ,in, represents the number of grids containing sampling frequency points, Indicates the total number of grids after meshing.

[0022] In a second aspect, the present invention provides a synthetic aperture radiometer array optimization device, comprising:

[0023] An initialization module, used for randomly initializing the positions of multiple aperture synthesis radiometers to obtain an initial aperture synthesis radiometer array;

[0024] A position adjustment module, used for obtaining a plurality of actions for adjusting the positions of a plurality of comprehensive aperture radiometers based on a policy network, and adjusting the initial comprehensive aperture radiometer array according to each action to obtain a plurality of intermediate comprehensive aperture radiometer arrays; wherein the action is set as the position change amount of each comprehensive aperture radiometer, and the plurality of actions correspond one-to-one to the plurality of intermediate comprehensive aperture radiometer arrays;

[0025] A reward function module is used to design a reward function for each intermediate aperture radiometer array based on the difference between the initial aperture radiometer array and the intermediate aperture radiometer array, and calculate the reward value of the action corresponding to the intermediate aperture radiometer array according to the reward function; the difference includes the spatial frequency domain sampling coverage difference and the brightness temperature image difference. The spatial frequency domain sampling coverage refers to the proportion of the grid containing the actual sampling frequency points after the rectangular area formed within the maximum sampling frequency range of the radiometer system is gridded in the aperture radiometer.

[0026] An array optimization model building module is used to build an array optimization model based on a Q-Learning algorithm with an initial aperture synthesis radiometer array as input and a reward value obtained by adjusting the positions of multiple aperture synthesis radiometers as output;

[0027] An array optimization model training module is used to train the array optimization model according to multiple actions, the reward value corresponding to each action, and the intermediate synthetic aperture radiometer array corresponding to each action, until the loss value of the array optimization model is less than a preset loss threshold, thereby obtaining a trained array optimization model;

[0028] The array optimization module is used to input the integrated aperture radiometer array to be optimized into the trained array optimization model, determine the action corresponding to the maximum Q value in the output result of the array optimization model as the optimization strategy of the integrated aperture radiometer array to be optimized, and optimize the integrated aperture radiometer array to be optimized according to the optimization strategy.

[0029] In a third aspect, the present invention provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned comprehensive aperture radiometer array optimization method when executing the computer program.

[0030] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, which implements the above-mentioned synthetic aperture radiometer array optimization method when executed by a processor.

[0031] The beneficial effects of the present invention are:

[0032] The aperture synthesis radiometer array optimization method provided by the present invention designs a reward function based on the difference in spatial frequency domain sampling coverage and brightness temperature image difference between the initial aperture synthesis radiometer array and the intermediate aperture synthesis radiometer array, thereby constructing an array optimization model, and determining the action corresponding to the maximum Q value as the optimization strategy of the aperture synthesis radiometer array to be optimized. The optimization strategy can obtain an aperture synthesis radiometer array with higher spatial frequency domain sampling coverage and smaller mean square error of brightness temperature image, which is beneficial to improving the effect of the aperture synthesis radiometer array. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a schematic diagram of the working principle of a comprehensive aperture radiometer in one embodiment of the present application;

[0034] Figure 2 This is a flow chart of a method for optimizing a synthetic aperture radiometer array in one embodiment of the present application;

[0035] Figure 3 This is a schematic diagram of the structure of an array optimization model in one embodiment of the present application;

[0036] Figure 4a A brightness temperature image generated by a comprehensive aperture radiometer array optimization method in one embodiment of the present application;

[0037] Figure 4b A brightness temperature image generated by a particle swarm optimization algorithm in one embodiment of the present application;

[0038] Figure 5 This is a schematic diagram of the structure of a comprehensive aperture radiometer array optimization device in one embodiment of the present application;

[0039] Figure 6 This is a schematic diagram of the structure of a terminal device in one embodiment of the present application. DETAILED DESCRIPTION

[0040] In order to solve the problem that traditional methods ignore the spatial frequency domain sampling coverage of the system and the quality of the reconstructed brightness temperature image, resulting in poor effect of the generated comprehensive aperture radiometer array, the present invention provides a comprehensive aperture radiometer array optimization method and related equipment. The comprehensive aperture radiometer array optimization method designs a reward function based on the spatial frequency domain sampling coverage difference and brightness temperature image difference between the initial comprehensive aperture radiometer array and the intermediate comprehensive aperture radiometer array, thereby constructing an array optimization model, and determining the action corresponding to the maximum Q value as the optimization strategy of the comprehensive aperture radiometer array to be optimized. The optimization strategy can obtain a comprehensive aperture radiometer array with higher spatial frequency domain sampling coverage and smaller brightness temperature image mean square error, which is beneficial to improving the effect of the comprehensive aperture radiometer array.

[0041] For ease of explanation, the working principle of the synthetic aperture radiometer is described below.

[0042] like Figure 1 As shown, any two points on the OXY plane and Two receiving antennas (antenna i, antenna j) are placed at the same position to receive the radiation source. The radiation is obtained by transmitting the received signal to two receiving channels (receiver i and receiver j). The channel output signals are then complex correlated to obtain the visibility function, and finally the brightness temperature image is reconstructed.

[0043] The synthetic aperture radiometer array optimization method provided by the present invention is described below.

[0044] like Figure 2 As shown, the comprehensive aperture radiometer array optimization method provided by the present invention comprises the following steps:

[0045] Step 21, randomly initialize the positions of multiple aperture synthesis radiometers to obtain an initial aperture synthesis radiometer array.

[0046] In an embodiment of the present invention, the aperture synthesis radiometer array is composed of a plurality of aperture synthesis radiometers and their positions. For example, the initial aperture synthesis radiometer array can be expressed as , Indicates The location of the aperture radiometer , , Represents the total number of aperture radiometers.

[0047] It should be noted that the present invention has the best effect on the array optimization of the non-uniform aperture synthesis radiometer.

[0048] Step 22, based on the strategy network, multiple actions for adjusting the positions of multiple aperture synthesis radiometers are obtained, and the initial aperture synthesis radiometer array is adjusted according to each action to obtain multiple intermediate aperture synthesis radiometer arrays.

[0049] In an embodiment of the present invention, the action is set as the position change amount of each aperture synthesis radiometer, and multiple actions correspond one-to-one to multiple intermediate aperture synthesis radiometer arrays.

[0050] In some embodiments of the present invention, the above-mentioned policy network may be an ε-greedy policy network.

[0051] Specifically, the above actions can be expressed as , Indicates The position change of a synthetic aperture radiometer.

[0052] Step 23, for each intermediate comprehensive aperture radiometer array, a reward function is designed based on the difference between the initial comprehensive aperture radiometer array and the intermediate comprehensive aperture radiometer array, and the reward value of the action corresponding to the intermediate comprehensive aperture radiometer array is calculated according to the reward function.

[0053] The differences include differences in spatial frequency domain sampling coverage and brightness temperature images. The spatial frequency domain sampling coverage refers to the proportion of grids containing actual sampling frequency points after the rectangular area constructed within the maximum sampling frequency range of the radiometer system is gridded in the comprehensive aperture radiometer. This coverage reflects the sampling density and distribution uniformity of the system in the spatial frequency domain, and determines the spatial resolution of the imaging quality.

[0054] Specifically, in an embodiment of the present invention, the expression of the reward function is:

[0055]

[0056] in, Indicates execution of an action The corresponding reward value is represents the weight, Indicates the difference in spatial frequency domain sampling coverage, , Indicates action The corresponding intermediate aperture radiometer array The spatial frequency domain sampling coverage, represents the spatial frequency domain sampling coverage of the initial synthetic aperture radiometer array, express The corresponding brightness temperature image and The mean square error between the corresponding brightness temperature images.

[0057] The non-uniform arrangement of antennas will lead to a decrease in the spatial frequency domain sampling coverage of the system, which in turn leads to a decrease in the quality of the reconstructed brightness temperature image. Therefore, the optimization effect of the array can be evaluated by comparing the imaging quality of the scene brightness temperature image obtained by the intermediate comprehensive aperture radiometer array and the initial comprehensive aperture radiometer array using the same inversion method for the same target scene. First, the intermediate comprehensive aperture radiometer array and the initial comprehensive aperture radiometer array are used to receive the signal of the same radiation source, and the antenna pairs are respectively subjected to complex correlation processing to obtain spatial frequency domain sampling of different interference baselines within the field of view, and the brightness temperature image is obtained by performing an inverse Fourier transform on the visibility function.

[0058] It is worth mentioning that since the spatial frequency domain sampling coverage and the mean square error of the brightness temperature image are taken into consideration when designing the reward function, it is beneficial to subsequently select actions with higher spatial frequency domain sampling coverage and smaller mean square error of the brightness temperature image to optimize the synthetic aperture radiometer array and improve the effect of the synthetic aperture radiometer array.

[0059] Step 24, based on the Q-Learning algorithm, an array optimization model is constructed with the initial aperture synthesis radiometer array as input and the reward value obtained by adjusting the positions of multiple aperture synthesis radiometers as output.

[0060] The purpose of constructing the array optimization model is to obtain the optimization strategy of the aperture synthesis radiometer array. The optimization strategy is the position change of each aperture synthesis radiometer.

[0061] Specifically, in an embodiment of the present invention, the model structure of the array optimization model is as follows: Figure 3 The specific processing process is as follows:

[0062] enter: ;

[0063] The first convolution plus activation function: ;

[0064] The second convolution plus activation function: ;

[0065] Enter the residual module: ;

[0066] ;

[0067] Enter the dropout layer: ;

[0068] Finally, after full connection, the estimated Q value of each action in the current state is obtained: ;

[0069] It should be noted that the traditional Q network model is generally composed of a combination of convolutional layers and fully connected layers. The present invention also adds a residual structure, and the obtained Q network can maintain a deeper network structure, improve the ability to learn the relationship between complex states and actions, and achieve better optimization effect.

[0070] Step 25, training the array optimization model according to the multiple actions, the reward value corresponding to each action, and the intermediate comprehensive aperture radiometer array corresponding to each action, until the loss value of the array optimization model is less than a preset loss threshold, thereby obtaining a trained array optimization model.

[0071] The training process of the array optimization model is described below, which specifically includes steps I to III.

[0072] Step I: Use the Adam optimization algorithm to optimize the model parameters of the array optimization model Update and obtain a new array optimization model.

[0073] Step II, calculate the loss value of the new array optimization model.

[0074] Specifically, the loss value calculation expression of the array optimization model is:

[0075]

[0076] in, The model parameters are The loss value of the array optimization model, Expressing hope, represents the reward at time step t, represents the discount factor, Indicates the next state The optimal action that maximizes the Q value is represents the target Q value, represents the parameters of the target Q network, Indicates the Q value of the current network.

[0077] Step III: if the loss value is less than the preset loss threshold, the new array optimization model is used as the trained array optimization model; otherwise, the new array optimization model is used as the array optimization model in step I, and the process returns to step I.

[0078] In other embodiments of the present invention, the number of iterations may be limited, and the array optimization model corresponding to the maximum number of iterations may be used as the trained array optimization model.

[0079] In another embodiment of the present invention, in order to improve efficiency, an experience pool can also be used to train the array optimization model. Specifically, the initial comprehensive aperture radiometer array, each action, the reward value corresponding to each action, and the intermediate comprehensive aperture radiometer array corresponding to each action are stored in the experience pool D as a set of data experience. When the data stored in the experience pool D reaches a certain amount, a small batch of data is randomly extracted as training data to train the array optimization model.

[0080] Step 26, input the integrated aperture radiometer array to be optimized into the trained array optimization model, determine the action corresponding to the maximum Q value in the output result of the array optimization model as the optimization strategy of the integrated aperture radiometer array to be optimized, and optimize the integrated aperture radiometer array to be optimized according to the optimization strategy.

[0081] The spatial frequency domain sampling coverage of the optimized synthetic aperture radiometer array based on the optimization strategy is higher and the mean square error of the brightness temperature image is smaller.

[0082] In order to further illustrate that the comprehensive aperture radiometer array optimization method provided by the embodiment of the present invention can achieve higher quality image reconstruction of the target scene, in another embodiment of the present invention, the comprehensive aperture radiometer array optimization method provided by the present invention is compared with the particle swarm optimization algorithm, and the generated brightness temperature images are as follows: Figure 4a and Figure 4b As shown, the horizontal axis represents the spatial domain The vertical coordinate is the brightness temperature of the scene in the spatial domain. Figure 4a and Figure 4b It can be seen that the integrated aperture radiometer array obtained by the integrated aperture radiometer array optimization method provided by the present invention has a smaller sidelobe value. The smaller the sidelobe value means that when the antenna receives the signal, the energy distribution in the sidelobe direction is less, which reduces the noise signal from the non-target direction, and the corresponding reconstructed brightness temperature image has a higher quality. This shows that the integrated aperture radiometer array optimization method provided by the present invention can improve the effect of the integrated aperture radiometer array.

[0083] From the above content, it can be seen that the comprehensive aperture radiometer array optimization method provided by the present invention designs a reward function based on the difference in spatial frequency domain sampling coverage and brightness temperature image difference between the initial comprehensive aperture radiometer array and the intermediate comprehensive aperture radiometer array, so as to construct an array optimization model, and determine the action corresponding to the maximum Q value as the optimization strategy of the comprehensive aperture radiometer array to be optimized. The optimization strategy can obtain a comprehensive aperture radiation array with higher spatial frequency domain sampling coverage and smaller mean square error of brightness temperature image, which is beneficial to improving the effect of the comprehensive aperture radiometer array.

[0084] The comprehensive aperture radiometer array optimization device provided by the present invention is described below.

[0085] like Figure 5 As shown, the comprehensive aperture radiometer array optimization device 500 includes:

[0086] An initialization module 501 is used to randomly initialize the positions of multiple aperture synthesis radiometers to obtain an initial aperture synthesis radiometer array;

[0087] The position adjustment module 502 is used to obtain multiple actions for adjusting the positions of multiple aperture synthesis radiometers based on the policy network, and adjust the initial aperture synthesis radiometer array according to each action to obtain multiple intermediate aperture synthesis radiometer arrays; wherein the action is set as the position change amount of each aperture synthesis radiometer, and the multiple actions correspond to the multiple intermediate aperture synthesis radiometer arrays one by one;

[0088] The reward function module 503 is used to design a reward function for each intermediate aperture radiometer array based on the difference between the initial aperture radiometer array and the intermediate aperture radiometer array, and calculate the reward value of the action corresponding to the intermediate aperture radiometer array according to the reward function; the difference includes the difference in spatial frequency domain sampling coverage and the difference in brightness temperature image, the spatial frequency domain sampling coverage refers to the proportion of grids containing sampling frequency points after the rectangular area formed by the maximum sampling frequency of the system is gridded; the difference includes the difference in spatial frequency domain sampling coverage and the difference in brightness temperature image, the spatial frequency domain sampling coverage refers to the proportion of grids containing actual sampling frequency points after the rectangular area formed within the maximum sampling frequency range of the radiometer system is gridded in the aperture radiometer;

[0089] An array optimization model building module 504 is used to build an array optimization model based on a Q-Learning algorithm with an initial aperture synthesis radiometer array as input and a reward value obtained by adjusting the positions of multiple aperture synthesis radiometers as output;

[0090] An array optimization model training module 505 is used to train the array optimization model according to multiple actions, the reward value corresponding to each action, and the intermediate synthetic aperture radiometer array corresponding to each action, until the loss value of the array optimization model is less than a preset loss threshold, thereby obtaining a trained array optimization model;

[0091] The array optimization module 506 is used to input the integrated aperture radiometer array to be optimized into the trained array optimization model, determine the action corresponding to the maximum Q value in the output result of the array optimization model as the optimization strategy of the integrated aperture radiometer array to be optimized, and optimize the integrated aperture radiometer array to be optimized according to the optimization strategy.

[0092] It should be noted that the information interaction, execution process and other contents between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. The specific functions and technical effects brought about by them can be specifically referred to the method embodiment part, which will not be repeated here. The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit, and the above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0093] like Figure 6 As shown, an embodiment of the present invention provides a terminal device, such as Figure 6 As shown, the terminal device D10 of this embodiment includes: at least one processor D100 ( Figure 6 Only one processor is shown in the figure), a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, wherein the processor D100 implements the steps of any of the above-mentioned method embodiments when executing the computer program D102.

[0094] Specifically, when the processor D100 executes the computer program D102, it randomly initializes the positions of multiple aperture-integrated radiometers to obtain an initial aperture-integrated radiometer array; based on a policy network, it obtains multiple actions for adjusting the positions of multiple aperture-integrated radiometers, and adjusts the initial aperture-integrated radiometer array according to each action to obtain multiple intermediate aperture-integrated radiometer arrays; for each intermediate aperture-integrated radiometer array, a reward function is designed based on the difference between the initial aperture-integrated radiometer array and the intermediate aperture-integrated radiometer array, and the reward value of the action corresponding to the intermediate aperture-integrated radiometer array is calculated according to the reward function; based on the Q-Learning algorithm The array optimization model is constructed by taking the initial aperture radiometer array as input and the reward value obtained by adjusting the positions of multiple aperture radiometers as output; the array optimization model is trained according to multiple actions, the reward value corresponding to each action, and the intermediate aperture radiometer array corresponding to each action until the loss value of the array optimization model is less than the preset loss threshold, and the trained array optimization model is obtained; the aperture radiometer array to be optimized is input into the trained array optimization model, and the action corresponding to the maximum Q value in the output result of the array optimization model is determined as the optimization strategy of the aperture radiometer array to be optimized, and the aperture radiometer array to be optimized is optimized according to the optimization strategy. Among them, the reward function is designed based on the difference in spatial frequency domain sampling coverage and brightness temperature image between the initial aperture radiometer array and the intermediate aperture radiometer array, so as to construct the array optimization model, and the action corresponding to the maximum Q value is determined as the optimization strategy of the aperture radiometer array to be optimized. The optimization strategy can obtain an aperture radiometer array with higher spatial frequency domain sampling coverage and smaller mean square error of brightness temperature image, which is conducive to improving the effect of the aperture radiometer array.

[0095] The processor D100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0096] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may also be an external storage device of the terminal device D10, such as a plug-in hard disk, a smart memory card (SMC, SmartMedia Card), a secure digital (SD, Secure Digital) card, a flash card (Flash Card), etc. equipped on the terminal device D10. Further, the memory D101 may also include both an internal storage unit of the terminal device D10 and an external storage device. The memory D101 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program, etc. The memory D101 may also be used to temporarily store data that has been output or is to be output.

[0097] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0098] An embodiment of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0099] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of protection of the present application is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0100] One or more embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present application should be included in the protection scope of the present application.

Claims

1. A method for optimizing a synthetic aperture radiometer array, characterized in that: include: Randomly initialize the positions of multiple aperture synthesis radiometers to obtain an initial aperture synthesis radiometer array; Based on the strategy network, a plurality of actions for adjusting the positions of the plurality of integrated aperture radiometers are obtained, and the initial integrated aperture radiometer array is adjusted according to each of the actions to obtain a plurality of intermediate integrated aperture radiometer arrays; wherein the action is set as the position change amount of each of the integrated aperture radiometers, and the plurality of actions correspond one-to-one to the plurality of intermediate integrated aperture radiometer arrays; For each of the intermediate comprehensive aperture radiometer arrays, a reward function is designed based on the difference between the initial comprehensive aperture radiometer array and the intermediate comprehensive aperture radiometer array, and a reward value of the action corresponding to the intermediate comprehensive aperture radiometer array is calculated according to the reward function; the difference includes a difference in spatial frequency domain sampling coverage and a difference in brightness temperature image, the spatial frequency domain sampling coverage refers to the proportion of grids containing actual sampling frequency points after a rectangular area formed within the maximum sampling frequency range of the radiometer system is gridded in the comprehensive aperture radiometer, and the coverage reflects the sampling density and distribution uniformity of the system in the spatial frequency domain, and determines the spatial resolution of the imaging quality; Based on the Q-Learning algorithm, the initial aperture synthesis radiometer array is used as input, and the reward value obtained by adjusting the positions of the multiple aperture synthesis radiometers is used as output to construct an array optimization model; The array optimization model is trained according to the plurality of actions, the reward value corresponding to each action, and the intermediate synthetic aperture radiometer array corresponding to each action, until the loss value of the array optimization model is less than a preset loss threshold, thereby obtaining a trained array optimization model; The integrated aperture radiometer array to be optimized is input into the trained array optimization model, the action corresponding to the maximum Q value in the output result of the array optimization model is determined as the optimization strategy of the integrated aperture radiometer array to be optimized, and the integrated aperture radiometer array to be optimized is optimized according to the optimization strategy.

2. The method for optimizing a synthetic aperture radiometer array according to claim 1, characterized in that: The policy network is an ε-greedy policy network.

3. The method for optimizing a synthetic aperture radiometer array according to claim 1, characterized in that: The expression of the reward function is: in, Indicates execution of an action The corresponding reward value is represents the weight, Indicates the difference in spatial frequency domain sampling coverage, , Indicates action The corresponding intermediate aperture radiometer array The spatial frequency domain sampling coverage, represents the spatial frequency domain sampling coverage of the initial synthetic aperture radiometer array, express The corresponding brightness temperature image and The mean square error between the corresponding brightness temperature images.

4. The method for optimizing a synthetic aperture radiometer array according to claim 3, characterized in that: The loss value calculation expression of the array optimization model is: in, The model parameters are The loss value of the array optimization model, Expressing hope, represents the reward at time step t, represents the discount factor, Indicates the next state The optimal action that maximizes the Q value is represents the target Q value, represents the parameters of the target Q network, Indicates the Q value of the current network.

5. The method for optimizing a synthetic aperture radiometer array according to claim 4, characterized in that: The Adam optimization algorithm is used to optimize the model parameters of the array optimization model. to update.

6. The method for optimizing a synthetic aperture radiometer array according to claim 5, characterized in that: The expression of the spatial frequency domain sampling coverage is: ,in, represents the number of grids containing sampling frequency points, Indicates the total number of grids after meshing.

7. A synthetic aperture radiometer array optimization device, characterized in that: include: An initialization module, used for randomly initializing the positions of multiple aperture synthesis radiometers to obtain an initial aperture synthesis radiometer array; A position adjustment module, used for acquiring a plurality of actions for adjusting the positions of the plurality of comprehensive aperture radiometers based on a policy network, and adjusting the initial comprehensive aperture radiometer array according to each of the actions to obtain a plurality of intermediate comprehensive aperture radiometer arrays; wherein the action is set as the position change amount of each of the comprehensive aperture radiometers, and the plurality of actions correspond one-to-one to the plurality of intermediate comprehensive aperture radiometer arrays; A reward function module is used to design a reward function for each of the intermediate comprehensive aperture radiometer arrays based on the difference between the initial comprehensive aperture radiometer array and the intermediate comprehensive aperture radiometer array, and calculate the reward value of the action corresponding to the intermediate comprehensive aperture radiometer array according to the reward function; the difference includes the difference in spatial frequency domain sampling coverage and the difference in brightness temperature image, and the spatial frequency domain sampling coverage refers to the proportion of grids containing actual sampling frequency points after the rectangular area formed within the maximum sampling frequency range of the radiometer system is gridded in the comprehensive aperture radiometer; An array optimization model building module is used to build an array optimization model based on a Q-Learning algorithm with the initial aperture synthesis radiometer array as input and a reward value obtained by adjusting the positions of the multiple aperture synthesis radiometers as output; An array optimization model training module is used to train the array optimization model according to the plurality of actions, the reward value corresponding to each action, and the intermediate synthetic aperture radiometer array corresponding to each action, until the loss value of the array optimization model is less than a preset loss threshold, thereby obtaining a trained array optimization model; The array optimization module is used to input the integrated aperture radiometer array to be optimized into the trained array optimization model, determine the action corresponding to the maximum Q value in the output result of the array optimization model as the optimization strategy of the integrated aperture radiometer array to be optimized, and optimize the integrated aperture radiometer array to be optimized according to the optimization strategy.

8. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the synthetic aperture radiometer array optimization method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the synthetic aperture radiometer array optimization method according to any one of claims 1 to 6 is implemented.

Images