Method, device and equipment for optimizing UV coverage rate of synthetic aperture radiometer

Abstract

The invention provides a method, a device and equipment for optimizing the UV coverage rate of a synthetic aperture radiometer. The method comprises the following steps of: determining an initial antenna position of an antenna array in the synthetic aperture radiometer; determining a first initial optimal basic step length; selecting a target antenna to be moved, and moving the target antenna in one direction based on the first initial optimal basic step length; calculating a second step length based on the target antenna with the changed position and the first initial optimal basic step length, and moving the target antenna with the changed position in one direction based on the second step length so as to change the second step length and the position of the target antenna; when the previous step is repeated to the target iteration number, Q values of the target antenna at different positions are determined; and performing position optimization of the antenna array based on the position of the target antenna corresponding to the maximum Q value. The method provided by the invention is used for optimizing the non-uniform array arrangement so as to improve the UV coverage rate of the synthetic aperture radiometer.

Description

Technical Field

[0001] The present invention belongs to the technical field of microwave remote sensing and detection, and particularly relates to a method, device and equipment for optimizing the UV coverage rate of a synthetic aperture radiometer. Background Art

[0002] A synthetic aperture radiometer (ASR) is a high-resolution radiation measurement instrument. Based on the principle of interferometry, it uses an antenna array composed of multiple small-aperture antennas to equivalent the function of a large-aperture antenna, thereby realizing high-resolution radiation measurement of the target scene.

[0003] The arrangement of the antenna array is one of the core elements of the synthetic aperture radiometer and plays a decisive role in its performance. For different array arrangements, changes in the number, spacing, and spatial distribution of antennas will result in different baseline combinations. In terms of UV coverage, the arrangement method of the antenna array directly determines the distribution of sampling points on the UV plane. Different antenna spacings and relative positions will generate different baseline vectors (the distance vector between two antennas), and these baseline vectors correspond to different sampling points on the UV plane. A reasonable arrangement can improve the UV coverage rate. Thus, it guarantees the high-resolution imaging of the synthetic aperture radiometer, reduces the data inversion error, enables the synthetic aperture radiometer to obtain more complete spatial frequency information, and then generates a clearer and more detailed image based on the Fourier transform principle.

[0004] Currently, in non-uniform array arrangements, some common methods for adjusting the spacing and position between antennas include genetic algorithms, particle swarm optimization algorithms, ant colony algorithms, simulated annealing algorithms, and tabu search, etc. However, there is still much room for improvement in the arrangement effect and complexity reduction. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method, device and equipment for optimizing the UV coverage rate of a synthetic aperture radiometer, which is used to move the selected antennas in a variable step size manner to achieve the effect of optimizing the non-uniform array arrangement, and thereby improve and optimize the UV coverage rate of the synthetic aperture radiometer.

[0006] The content of the present invention includes a method for optimizing the UV coverage rate of a synthetic aperture radiometer, comprising:

[0007] Determine the initial antenna positions of the antenna array in the synthetic aperture radiometer;

[0008] Calculate the basic step size for the initial antenna movement based on the initial antenna positions, and different initial antenna positions correspond to different basic step sizes;

[0009] Select a target antenna to be moved from the antenna array;

[0010] Take the basic step corresponding to the target antenna as the first initial optimal basic step, and move the target antenna in a certain direction based on the first initial optimal basic step to change the first initial optimal basic step and the position of the target antenna;

[0011] Calculate and determine the second step corresponding to its current state based on the changed first initial optimal basic step and the target antenna, and move the target antenna at the changed position in a certain direction based on the second step to change the second step and the position of the target antenna;

[0012] Repeat the above step until the target number of iterations, and determine the Q values of the target antenna at different positions;

[0013] Optimize the position of the antenna array based on the position of the target antenna corresponding to the maximum Q value. Different position distributions of the antenna array correspond to different UV coverage rates.

[0014] In one embodiment, calculating the basic step of the initial antenna movement based on the initial antenna position includes:

[0015] Use the hyperopt function to calculate the initial antenna position to obtain the basic step of the initial antenna movement. Different positions of the initial antenna correspond to different basic steps.

[0016] In one embodiment, selecting the target antenna to be moved from the antenna array includes:

[0017] Remove each movable antenna in the antenna array in turn, and calculate the UV coverage rate of the remaining antennas after removing one antenna each time;

[0018] Select and determine the target antenna to be moved based on the UV coverage rate.

[0019] In one embodiment, selecting and determining the target antenna to be moved based on the UV coverage rate includes:

[0020] Determine the maximum value of the UV coverage rate;

[0021] Take the removed antenna corresponding to the maximum value of the UV coverage rate as the target antenna to be moved.

[0022] In one embodiment, moving the target antenna in a certain direction based on the first initial optimal basic step includes:

[0023] Determine the target action based on the current position of the target antenna and the first initial optimal basic step;

[0024] Starting from the current position of the target antenna, perform a target action on the target antenna based on the first initial optimal basic step size, so that it moves in a specified direction to change the position of the target antenna.

[0025] In one embodiment, the target action is any one of the following actions:

[0026] Performing different actions in different directions includes increasing the step size for upward movement, decreasing the step size for upward movement, decreasing the step size for downward movement, increasing the step size for downward movement, decreasing the step size for leftward movement, increasing the step size for leftward movement, decreasing the step size for rightward movement, and increasing the step size for rightward movement.

[0027] In one embodiment, the method further includes:

[0028] Calculate the UV coverage rate after each optimization of the antenna array;

[0029] Construct an optimization model for iterative calculation of the initial optimal basic step size and the position of the target antenna using a reinforcement learning algorithm;

[0030] Taking the antenna position and the corresponding initial optimal basic step size as the state, taking the movement executed in different directions as the action, and taking the improvement of the UV coverage rate as the reward, record the state, action, and reward of each round. The movement of the target antenna after the target iteration several times constitutes a round of training, and different rounds select corresponding target antennas;

[0031] Control the optimization model to learn the state, action, and reward data of each round recorded, calculate the corresponding loss function, and perform backpropagation;

[0032] Execute the gradient descent algorithm through an optimizer to update the parameters of the optimization model, so that the optimization model can directly determine the position of the target antenna with the largest Q value.

[0033] In one embodiment, the method further includes:

[0034] Based on the UV coverage rate of each round, draw a line chart of the change of the UV coverage rate with the number of rounds.

[0035] Another embodiment of the present invention also provides a device for optimizing the UV coverage rate of a synthetic aperture radiometer, including:

[0036] A determination module for determining the initial antenna position of the antenna array in the synthetic aperture radiometer;

[0037] A first calculation module for calculating the basic step size of the initial antenna movement based on the initial antenna position, and the corresponding basic step size is different for different positions of the initial antenna;

[0038] A selection module, configured to select a target antenna to be moved from the antenna array;

[0039] A movement module, configured to use the basic step corresponding to the target antenna as the first initial optimal basic step, and move the target antenna in a certain direction based on the first initial optimal basic step to change the first initial optimal basic step and the position of the target antenna;

[0040] A second calculation module, configured to calculate and determine a second step corresponding to its current state based on the changed first initial optimal basic step and the target antenna, and move the target antenna with the changed position in a certain direction to change the second step and the position of the target antenna;

[0041] An iterative calculation module, configured to repeat the above step until the target number of iterations, and determine the Q value corresponding to the calculation result of each generation;

[0042] An optimization module, configured to optimize the position of the antenna array based on the position of the target antenna corresponding to the maximum Q value, and different position distributions of the antenna array correspond to different UV coverage rates.

[0043] Another embodiment of the present invention further provides an electronic device, including;

[0044] One or more processors;

[0045] A memory, configured to store one or more programs;

[0046] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for optimizing the UV coverage rate of the synthetic aperture radiometer as described in any one of the above.

[0047] The beneficial effects of the present invention include selecting the antenna to be moved in the antenna array of the synthetic aperture radiometer, and then moving the selected antenna in a variable step manner to achieve the effect of optimizing the non-uniform array arrangement, so as to improve the UV coverage rate of the synthetic aperture radiometer, which can not only improve the integrity of spatial frequency sampling, but also improve the imaging quality and enhance the target detection and recognition ability.

[0048] Other features and advantages of the present application will be described in the following specification, and some of them will become obvious from the specification, or be understood by implementing the present application. The objectives and other advantages of the present application can be achieved and obtained through the structures specifically pointed out in the written specification, claims, and drawings.

[0049] The technical solutions of the present application will be further described in detail below with reference to the drawings and embodiments. Description of the Drawings

[0050] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0051] Figure 1 It is a schematic flowchart of the variable step-size movement of the target antenna in the method for optimizing the UV coverage rate of the synthetic aperture radiometer in the embodiment of the present invention.

[0052] Figure 2 It is a schematic flowchart of selecting the target antenna in the method for optimizing the UV coverage rate of the synthetic aperture radiometer in the embodiment of the present invention.

[0053] Figure 3 It is a schematic diagram of the application result of the method for optimizing the UV coverage rate of the synthetic aperture radiometer in the embodiment of the present invention.

[0054] Figure 4 It is a structural block diagram of the device for optimizing the UV coverage rate of the synthetic aperture radiometer in the embodiment of the present invention. Specific Embodiments

[0055] Next, specific embodiments of the present invention will be described in detail with reference to the drawings, but it is not a limitation of the present invention.

[0056] It should be understood that various modifications can be made to the embodiments disclosed herein. Therefore, the following description should not be regarded as a limitation, but only as an example of the embodiments. Those skilled in the art will think of other modifications within the scope of the present disclosure.

[0057] The drawings included in the specification and constituting a part of the specification illustrate the embodiments of the present disclosure, and together with the general description of the present disclosure given above and the detailed description of the embodiments given below, are used to explain the principles of the present disclosure.

[0058] By the following description of the preferred forms of the embodiments given as non-limiting examples with reference to the drawings, these and other characteristics of the present invention will become apparent.

[0059] It should also be understood that although the present invention has been described with reference to some specific examples, those skilled in the art can surely implement many other equivalent forms of the present invention, which have the features as described in the claims and thus are all within the protection scope defined thereby.

[0060] When combined with the drawings, in view of the following detailed description, the above and other aspects, features and advantages of the present disclosure will become more apparent.

[0061] Specific embodiments of the present disclosure will hereinafter be described with reference to the accompanying drawings; however, it should be understood that the disclosed embodiments are merely examples of the present disclosure, which can be implemented in various ways. Well-known and / or repetitive functions and structures are not described in detail to avoid obscuring the present disclosure with unnecessary or redundant details. Therefore, the specific structural and functional details disclosed herein are not intended to be limiting, but are merely used as a basis and representative basis for the claims to teach those skilled in the art to use the present disclosure in substantially any suitable detailed structure in a variety of ways.

[0062] This specification may use the phrases "in one embodiment", "in another embodiment", "in yet another embodiment", or "in other embodiments", each of which may refer to one or more of the same or different embodiments according to the present disclosure.

[0063] Next, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0064] Synthetic aperture microwave radiometry is a method for indirectly measuring the microwave brightness temperature image of a scene. It measures the components (spatial frequency components) of the Fourier transform of the scene radiation, and then obtains the brightness temperature image of the scene through Fourier transform. This is the basic principle of synthetic aperture microwave radiometry.

[0065] Performing complex correlation operations using antenna pairs separated by a certain distance can measure the spatial frequency components of the scene radiation. The spatial distance and direction between the antenna pairs are called the baseline vector, abbreviated as the baseline. The baseline vector can be a one-dimensional vector (such as a one-dimensional synthetic aperture) or a two-dimensional vector (such as a two-dimensional synthetic aperture). For a two-dimensional baseline vector, it is usually represented by plane coordinates, with the horizontal axis of the coordinates represented by u and the vertical axis represented by v. In this way, a baseline corresponds to a point on the (u, v) plane. Placing an antenna at the starting point and the ending point of a baseline vector forms an antenna pair, and the complex correlation output of this antenna pair is called the visibility. Since the visibility is a function of the baseline, that is, a function of (u, v), it is called the visibility function V(u, v). The visibility function is related to the orientation and length of the baseline, and is independent of the absolute position of a single antenna.

[0066] According to the principle of interferometry, any baseline vector formed by an antenna pair corresponds to a (u, v) sampling point on the UV plane. Then, all the baselines formed by an antenna array composed of N units form a (u, v) sampling distribution on the UV plane, which is called the UV coverage.

[0067] Since the non-uniform arrangement of antennas will lead to a decrease in the UV coverage rate of the system, and further lead to a decrease in the quality of the reconstructed brightness temperature image. Therefore, it is very important to optimize the arrangement of the non-uniform antenna array to improve the UV coverage rate. To solve this problem, as Figure 1As shown in the figure, the present invention provides a method for optimizing the UV coverage rate of a synthetic aperture radiometer, including:

[0068] S1: Determine the initial antenna positions of the antenna array in the synthetic aperture radiometer;

[0069] S2: Calculate the basic step length for the initial antenna movement based on the initial antenna positions. Different initial antenna positions correspond to different basic step lengths;

[0070] S3: Select a target antenna to be moved from the antenna array;

[0071] S4: Take the basic step length corresponding to the target antenna as the first initial optimal basic step length, and move the target antenna in a certain direction based on the first initial optimal basic step length to change the first initial optimal basic step length and the position of the target antenna;

[0072] S5: Calculate and determine the second step length corresponding to its current state based on the changed first initial optimal basic step length and the target antenna, and move the target antenna at the changed position in a certain direction based on the second step length to change the second step length and the position of the target antenna;

[0073] S6: When repeating the above step S6 until the target number of iterations, determine the Q values of the target antenna at different positions;

[0074] S7: Optimize the position of the antenna array based on the position of the target antenna corresponding to the maximum Q value. Different position distributions of the antenna array correspond to different UV coverage rates.

[0075] The optimization method of this embodiment actually realizes the optimization of the UV coverage rate of the synthetic aperture radiometer by means of variable step length based on the policy gradient algorithm. The method in this embodiment selects the target antenna to be moved, continuously adjusts the position of the current target antenna in a variable step length movement manner, then selects the position of the target antenna with the maximum Q value to update the array positions of all antennas in the antenna array, calculates the UV coverage rate of the updated antenna array, and records various relevant parameters of this process. Repeatedly execute the above process, select different target antennas each time, and use the reinforcement learning algorithm of policy gradient descent to optimize the optimal position for selecting and determining the target antenna through the results obtained from multiple rounds of updates and learning, and accordingly optimize the parameters of the optimization model for the distribution of each antenna in the antenna array. In this parameter optimization process, the method used in this embodiment is to directly enhance or weaken the probability of the selection behavior using the reward, so that the model (agent) can better select the movement direction of the target antenna and the increase or decrease of the step length, optimize the antenna arrangement method, and achieve the purpose of improving the UV coverage rate of the synthetic aperture radiometer.

[0076] The method further includes:

[0077] S8: Calculate the UV coverage rate after each optimization of the antenna array;

[0078] S9: Construct an optimization model for iterative calculation of the initial optimal basic step size and the target antenna position using a reinforcement learning algorithm;

[0079] S10: Using the antenna position and the corresponding initial optimal basic step size as the state, the movement executed in different directions as the action, and the improvement of the UV coverage rate as the reward, record the state, action, and reward of each round. The movement of the target antenna through the target number of iterations constitutes one round of training, and different rounds select corresponding target antennas;

[0080] S11: Control the optimization model to learn the state, action, and reward data of each round recorded, calculate the corresponding loss function, and perform backpropagation;

[0081] S12: Execute the gradient descent algorithm through an optimizer to update the parameters of the optimization model, so that the optimization model can directly determine the position of the target antenna with the largest Q value.

[0082] Specifically, the state in this embodiment: The state is essentially a set of information that can reflect the current antenna position and the moving step size. In a non-uniform antenna element array in a two-dimensional region, after selecting the antenna (target antenna) to be moved according to the method, the current antenna position coordinates and the current step size are used as the state at this time: state = [x, y, step].

[0083] Action: Adjust the position of the selected antenna element (target antenna), combine the moving direction with the increase and decrease of the moving step size. There are eight actions in the action space, namely moving up with an increasing step size, moving up with a decreasing step size, moving down with a decreasing step size, moving down with an increasing step size, moving left with a decreasing step size, moving left with an increasing step size, moving right with a decreasing step size, and moving right with an increasing step size.

[0084] Reward: The purpose is to improve the UV coverage rate of the ASR. Set the reward to be positively correlated with the UV coverage rate. In order to make the reward more sensitive to the increase in the UV coverage rate, the form of an inverse proportional function is adopted. The UV coverage rate is represented by the result variable, and the reward expression is:

[0085] In the initialization stage of the program corresponding to the solution of this embodiment, the antenna array, key parameters of the model, and global variables are configured to facilitate subsequent reinforcement learning model training and data analysis. First, define the layout range of the antennas, that is, the grid size max baseline where they are located, and the number of antennas with adjustable positions array num. Then, set the core parameters in the reinforcement learning algorithm, including the number of neurons in the hidden layer n hiddens, learning factor learning rate, and discount factor gamma. To track the performance and decision-making process of the model, create some lists. The return list is used to record the reward for each episode, and the max q value list is used to store the maximum action value for each step. These are all key data for evaluating and optimizing the model.

[0086] After completing the parameter configuration, define the classes related to the network structure. First, the TransformerNet class is defined and inherits from nn.Module of PyTorch. In its constructor init, build a complex network structure, including a linear embedding layer embedding for input mapping, an encoder transformer encoder composed of multiple stacked Transformer encoder layers encoder layer, and a fully connected layer fc for output. In the forward method, implement the forward propagation logic of the data, so that after the input data is processed by these layers, the probability distribution of each action is output through the torch.softmax function. Secondly, define the PolicyGradient class, which represents a reinforcement learning agent. In its constructor, initialize basic attributes such as the number of states and actions, and call the build_net method to build a TransformerNet as the policy network. At the same time, also initialize some lists to record key information such as the state, action, reward, and loss of the agent during the learning process. These information are crucial for the learning and performance optimization of the agent.

[0087] Among them, the TransformerNet in this embodiment mainly uses the Transformer encoding layer to map and extract features from the input information. The input is the current state, which includes the coordinate position of the current antenna unit (target antenna) and the current step size. The output is the probability distribution of eight actions, that is, moving up with an increasing step size, moving up with a decreasing step size, moving down with a decreasing step size, moving down with an increasing step size, moving left with a decreasing step size, moving left with an increasing step size, moving right with a decreasing step size, and moving right with an increasing step size, which can realize variable step size movement to optimize the array position.

[0088] Furthermore, the linear embedding layer maps the input state from the n states dimension to the d model dimension space. In a Transformer, the embedding layer is used to convert the input sequence into a continuous vector representation. This process can extract and transform the features of the input data to prepare for subsequent processing. In the Transformer encoder layer, the multi-head attention mechanism is one of the core components of the Transformer layer. It divides the input d model-dimensional feature vector into nhead heads, and each head calculates the attention distribution separately. The multi-head attention mechanism allows the model to focus on different parts of the input sequence in different subspaces, thus capturing the relationships between features more comprehensively. The feed-forward neural network is specified by the dim_feedforward parameter for its dimension. It usually contains two linear layers and an activation function (such as ReLU) to perform further non-linear transformations on the features at each position. This component can increase the expressive power of the model, further process the features output by the multi-head attention mechanism, and extract more complex feature information. Residual connection and layer normalization In the Transformer encoder layer, the output of each sub-layer (multi-head attention layer and feed-forward neural network layer) is added to the input through a residual connection and then layer normalization is performed. Layer normalization normalizes each sample in the feature dimension, which helps to accelerate the training process and improve the stability of the model. Through residual connection and layer normalization, the model can better handle the vanishing gradient problem in deep networks and make the network easier to train.

[0089] In the Transformer architecture of this embodiment, stacking (represented by "Nx") plays an important role. Stacking multiple Transformer encoder layers forms a Transformer encoder, and the encoder layers with N layers process the data in sequence, so that more advanced and abstract features can be gradually extracted. In the output layer, it is set as a linear layer to map the output of the Transformer encoder from the d_model dimension to the n_actions dimension, where n_actions refers to the size of the output action space of the model. After the output layer, the softmax function is used to convert the output into the probability distribution of actions.

[0090] In one embodiment, calculating the basic step size of the initial antenna movement based on the initial antenna position includes:

[0091] S13: Use the hyperopt function to calculate the initial antenna position to obtain the basic step length for the initial antenna movement. Different positions of the initial antenna correspond to different basic step lengths.

[0092] For example, in a two-dimensional plane area, generate the initial antenna position, fix the positions of the four vertices and randomly generate the coordinates of the remaining movable antennas, calculate the initial coverage rate, and add it to the result list.

[0093] For the initial antenna position, use the hyperopt function to obtain the initial optimal basic step length. Different initial antenna positions have different initial optimal basic step lengths. The initial optimal basic step length is a hyperparameter that affects the moving distance of the antenna in each step. By optimizing this parameter, the performance of the model can be improved. Based on the initial optimal basic step length, in the subsequent reinforcement learning training, move the antenna position with variable step lengths.

[0094] As Figure 2 shown, the selection of the target antenna to be moved from the antenna array includes:

[0095] S14: Remove each movable antenna in the antenna array in sequence, and calculate the UV coverage rate of the remaining antennas each time one antenna is removed;

[0096] S15: Select and determine the target antenna to be moved based on the UV coverage rate.

[0097] Among them, the selection and determination of the target antenna to be moved based on the UV coverage rate includes:

[0098] S16: Determine the maximum value of the UV coverage rate;

[0099] S17: Take the removed antenna corresponding to the maximum value of the UV coverage rate as the target antenna to be moved.

[0100] Exemplarily, in each round of training of the model, in order to effectively reduce the complexity of optimizing the antenna array layout, first perform the following operations for each movable antenna: remove it in sequence, then calculate the UV coverage rate of the remaining antenna layout, and store all these coverage rate results in a list. Then, find the unit antenna corresponding to the maximum value of the UV coverage rate in this list as the target antenna. This antenna is the unit antenna that contributes the least to the UV coverage rate of the current antenna array layout. Obtain its coordinates, and use the position of this unit antenna as the basis for starting to adjust the antenna array layout in this round.

[0101] In one embodiment, the movement of the target antenna in one direction based on the first initial optimal basic step length includes:

[0102] S18: Determine a target action based on the current position of the target antenna and the first initial optimal basic step size;

[0103] S19: Starting from the current position of the target antenna, perform the target action on the target antenna based on the first initial optimal basic step size to move it in a certain direction to change the position of the target antenna.

[0104] Among them, the target action is any one of the following actions:

[0105] Performing different actions in different directions includes increasing the step size when moving upward, decreasing the step size when moving upward, decreasing the step size when moving downward, increasing the step size when moving downward, decreasing the step size when moving left, increasing the step size when moving left, decreasing the step size when moving right, and increasing the step size when moving right.

[0106] For example, the optimization model (also called the agent) controls the position and movement step size of the antenna through its take action function. At the beginning of each iteration, the agent selects an action according to the current state (composed of the current position of the antenna and the corresponding initial optimal basic step size). These actions include increasing the step size when the target antenna moves upward, decreasing the step size when the target antenna moves upward, decreasing the step size when the target antenna moves downward, increasing the step size when the target antenna moves downward, decreasing the step size when the target antenna moves left, increasing the step size when the target antenna moves left, decreasing the step size when the target antenna moves right, and increasing the step size when the target antenna moves right, a total of eight possible actions. After each action is selected, the movement step size of the antenna will change accordingly. After performing the selected action on the target antenna and moving its position, the new coordinate position and the current step size of the target antenna form a new state. The agent will record the current state, the executed action, and the obtained reward into three lists: stateslist, actions list, and rewards list respectively. Through N times of reinforcement learning iteration training, all the new positions that the antenna unit has moved to are recorded. Then, the agent will select the new position with the maximum Q value from these N new positions, and then move the selected target antenna to this new position for optimizing the arrangement of the antenna array. In this way, the agent can learn how to adjust the position and step size of the antenna to achieve the best antenna array arrangement.

[0107] After each round of training, call the learn method of the agent, traverse the reward, state, and action information recorded in this round from back to front, calculate the loss function loss for each step at the same time, and perform backpropagation. The optimizer is used to perform gradient descent to update the parameters of the policy network in the optimization model.

[0108] In order to make the change of UV coverage rate more clearly reflected and determine the optimization effect of UV coverage rate, the method further includes:

[0109] S20: Plot a line graph showing the variation of the UV coverage rate with the number of rounds based on the UV coverage rate of each round.

[0110] For example, record the UV coverage rate of the optimized corresponding M antenna array positions after M rounds, and plot a line graph showing the variation of the UV coverage rate with the number of rounds.

[0111] To further illustrate that the method for optimizing the UV coverage rate of a synthetic aperture radiometer with variable step size based on the policy gradient algorithm in the embodiments of the present invention can achieve an improvement in the UV coverage rate of a non-uniform antenna array, the following is a detailed description in combination with specific Figure 3 The experimental results shown are as follows:

[0112] Optimize the array arrangement of 20 antennas in a 20*20 area. Fix 4 antennas at the four top corners of the area. Among the remaining 16 antennas, select antennas and move the antennas with variable step sizes. Figure 3 (a) and Figure 3 (b) are respectively line graphs of the optimization results of the UV coverage rate for different initial non-uniform antenna array arrangements. It can be intuitively seen from the figure that the present embodiment has an optimization effect on the UV coverage rate of different non-uniform antenna array arrangements, and the UV coverage rate has increased a lot, and the optimization effect is obvious.

[0113] As Figure 4 shown, another embodiment of the present invention also provides a device 100 for optimizing the UV coverage rate of a synthetic aperture radiometer, including:

[0114] A determination module, configured to determine the initial antenna positions of the antenna array in the synthetic aperture radiometer;

[0115] A first calculation module, configured to calculate a basic step size for the initial antenna movement based on the initial antenna positions, where the basic step sizes corresponding to different positions of the initial antenna are different;

[0116] A selection module, configured to select a target antenna to be moved from the antenna array;

[0117] A movement module, configured to use the basic step size corresponding to the target antenna as the first initial optimal basic step size, and move the target antenna in a certain direction based on the first initial optimal basic step size to change the first initial optimal basic step size and the position of the target antenna;

[0118] A second calculation module, configured to calculate and determine a second step size corresponding to its current state based on the changed first initial optimal basic step size and the target antenna, and move the target antenna with the changed position in a certain direction based on the second step size to change the second step size and the position of the target antenna;

[0119] An iterative calculation module, configured to determine the Q value corresponding to the calculation result of each generation when repeating the previous step until the target number of iterations is reached;

[0120] An optimization module, configured to perform position optimization of the antenna array based on the position of the target antenna corresponding to the maximum Q value. Different position distributions of the antenna array correspond to different UV coverage rates.

[0121] In one embodiment, calculating the basic step length of the initial antenna movement based on the initial antenna position includes:

[0122] Using the hyperopt function to calculate the initial antenna position to obtain the basic step length of the initial antenna movement. Different positions of the initial antenna correspond to different basic step lengths.

[0123] In one embodiment, selecting the target antenna to be moved from the antenna array includes:

[0124] Sequentially removing each movable antenna in the antenna array, and calculating the UV coverage rate of the remaining antennas each time one antenna is removed;

[0125] Based on the UV coverage rate, select and determine the target antenna to be moved.

[0126] In one embodiment, selecting and determining the target antenna to be moved based on the UV coverage rate includes:

[0127] Determine the maximum value of the UV coverage rate;

[0128] Take the removed antenna corresponding to the maximum value of the UV coverage rate as the target antenna to be moved.

[0129] In one embodiment, moving the target antenna in one direction based on the first initial optimal basic step length includes:

[0130] Determine the target action based on the current position of the target antenna and the first initial optimal basic step length;

[0131] Starting from the current position of the target antenna, perform the target action on the target antenna based on the first initial optimal basic step length to move it in a certain direction to change the position of the target antenna.

[0132] In one embodiment, the target action is any one of the following actions:

[0133] Performing different actions in different directions includes increasing the step length when moving upward, decreasing the step length when moving upward, decreasing the step length when moving downward, increasing the step length when moving downward, decreasing the step length when moving left, increasing the step length when moving left, decreasing the step length when moving right, and increasing the step length when moving right.

[0134] In one embodiment, the device further comprises:

[0135] A third calculation module, configured to calculate the UV coverage rate after each optimization of the antenna array;

[0136] A construction module, configured to construct an optimization model for iteratively calculating the initial optimal basic step length and the target antenna position by using a reinforcement learning algorithm;

[0137] A recording module, configured to record the state, action, and reward of each round with the antenna position and the corresponding initial optimal basic step length as the state, the movement executed in different directions as the action, and the improvement of the UV coverage rate as the reward. The movement of the target antenna through the target number of iterations constitutes one round of training, and different rounds select corresponding target antennas;

[0138] A control module, configured to control the optimization model to learn the state, action, and reward data of each round recorded, calculate the corresponding loss function, and perform backpropagation;

[0139] A parameter optimization module, configured to update the parameters of the optimization model by executing a gradient descent algorithm through an optimizer, so that the optimization model can directly determine the position of the target antenna with the largest Q value.

[0140] In one embodiment, the device further comprises:

[0141] A plotting module, configured to plot a line graph of the change of the UV coverage rate with the number of rounds based on the UV coverage rate of each round.

[0142] Another embodiment of the present invention further provides an electronic device, comprising:

[0143] One or more processors;

[0144] A memory, configured to store one or more programs;

[0145] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for optimizing the UV coverage rate of a synthetic aperture radiometer as described in any one of the above.

[0146] Furthermore, an embodiment of the present invention further provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, the method for optimizing the UV coverage rate of a synthetic aperture radiometer as described above is implemented. It should be understood that each solution in this embodiment has the corresponding technical effects in the above method embodiment, and will not be elaborated herein.

[0147] Furthermore, an embodiment of the present invention also provides a computer program product, which is tangibly stored on a computer-readable medium and includes computer-readable instructions. When the computer-executable instructions are executed, at least one processor is caused to execute the synthetic aperture radiometer UV coverage rate optimization method in the embodiment described above.

[0148] It should be noted that the computer storage medium of the present invention can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. The computer-readable medium can, for example but not limited to, be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access storage medium (RAM), a read-only storage medium (ROM), an erasable programmable read-only storage medium (EPROM or flash memory), an optical fiber, a portable compact disk read-only storage medium (CD-ROM), an optical storage medium, a magnetic storage medium, or any suitable combination of the above. In the present invention, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device. In the present invention, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, in which the computer-readable program code is carried. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program configured to be used by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wireless, antenna, optical cable, RF, etc., or any suitable combination of the above.

[0149] In addition, those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0150] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing device generate a system for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0151] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction system that implements the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0152] Those of ordinary skill in the art should understand that the discussion of any of the above embodiments is exemplary only and is not intended to imply that the scope of the present application is limited to these examples; under the concept of the present application, the technical features in the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations in different aspects of one or more embodiments of the present application as described above, and they are not provided in detail for the sake of brevity.

Claims

1. A method for optimizing UV coverage of a comprehensive aperture radiometer, characterized in that: include: Determine initial antenna positions for antenna arrays in a synthetic aperture radiometer; A basic step length of the initial antenna movement is calculated based on the initial antenna position, wherein different positions of the initial antenna correspond to different basic step lengths; Selecting a target antenna to be moved from the antenna array; Taking the basic step length corresponding to the target antenna as a first initial optimal basic step length, and moving the target antenna in one direction based on the first initial optimal basic step length to change the first initial optimal basic step length and the position of the target antenna; Calculating and determining a second step length corresponding to the current state of the target antenna based on the changed first initial optimal basic step length and the target antenna, and moving the target antenna after the position is changed in a certain direction based on the second step length to change the second step length and the position of the target antenna; Repeat the previous step until the target iteration number is reached, and determine the Q value of the target antenna at different positions; The position of the antenna array is optimized based on the position of the target antenna corresponding to the maximum Q value. Different position distributions of the antenna array have different corresponding UV coverage rates.

2. The UV coverage optimization method of the synthetic aperture radiometer according to claim 1, characterized in that: The calculating, based on the initial antenna position, a basic step length of initial antenna movement comprises: The initial antenna position is calculated using the hyperopt function to obtain a basic step length of the initial antenna movement. Different positions of the initial antenna correspond to different basic step lengths.

3. The UV coverage optimization method of the synthetic aperture radiometer according to claim 1, characterized in that: The selecting a target antenna to be moved from the antenna array comprises: Removing each movable antenna in the antenna array in turn, and calculating the UV coverage of the remaining antennas after each removal of an antenna; The target antenna to be moved is selected based on the UV coverage.

4. The UV coverage optimization method of the synthetic aperture radiometer according to claim 3, characterized in that: The step of selecting and determining the target antenna to be moved based on the UV coverage rate includes: Determining a maximum value of the UV coverage; The removed antenna corresponding to the maximum value of the UV coverage rate is used as the target antenna to be moved.

5. The UV coverage optimization method of the synthetic aperture radiometer according to claim 1, characterized in that: The moving the target antenna in a certain direction based on the first initial optimal basic step size includes: Determining a target action based on a current position of the target antenna and the first initial optimal basic step size; Starting from the current position of the target antenna, a target action is performed on the target antenna based on the first initial optimal basic step length, and the target antenna moves in a certain direction to change the position of the target antenna.

6. The method for optimizing UV coverage of a synthetic aperture radiometer according to claim 5, characterized in that: The target action is any one of the following actions: The performing different actions in different directions include increasing the upward moving step length, decreasing the upward moving step length, decreasing the downward moving step length, increasing the downward moving step length, decreasing the leftward moving step length, increasing the leftward moving step length, decreasing the rightward moving step length, and increasing the rightward moving step length.

7. The UV coverage optimization method of the synthetic aperture radiometer according to claim 5, characterized in that: The method further comprises: Calculating the UV coverage of the antenna array after each optimization; Constructing an optimization model for iterative calculation of the initial optimal basic step size and target antenna position using a reinforcement learning algorithm; The antenna position and the corresponding initial optimal basic step length are used as the initial state, the movement performed in different directions is used as the action, and the improvement of UV coverage is used as the reward. The state, action, and reward of each round are recorded. The movement of the target antenna after the target iteration several times constitutes a round of training, and the corresponding target antenna is selected in different rounds; Control the optimization model to learn the recorded state, action, and reward data of each round, calculate the corresponding loss function, and perform back propagation; The optimizer executes a gradient descent algorithm to update the parameters of the optimization model, so that the optimization model can directly determine the position of the target antenna with the maximum Q value.

8. The method for optimizing UV coverage of a synthetic aperture radiometer according to claim 7, characterized in that: The method further comprises: Based on the UV coverage rate in each round, a line graph showing the variation of UV coverage rate with the number of rounds is drawn.

9. A UV coverage optimization device for a comprehensive aperture radiometer, characterized in that: include: A determination module, used for determining an initial antenna position of an antenna array in a synthetic aperture radiometer; A first calculation module, configured to calculate a basic step length of initial antenna movement based on the initial antenna position, wherein different positions of the initial antenna correspond to different basic step lengths; A selection module, used for selecting a target antenna to be moved from the antenna array; A moving module, configured to use the basic step length corresponding to the target antenna as a first initial optimal basic step length, and move the target antenna in a certain direction based on the first initial optimal basic step length to change the first initial optimal basic step length and the position of the target antenna; A second calculation module is used to calculate and determine a second step length corresponding to the current state of the target antenna based on the changed first initial optimal basic step length and the target antenna, and to move the target antenna after the position is changed in a certain direction based on the second step length to change the second step length and the position of the target antenna; An iterative calculation module is used to determine the Q value corresponding to each generation of calculation results when repeating the previous step to the target iteration number; The optimization module is used to optimize the position of the antenna array based on the position of the target antenna corresponding to the maximum Q value, and the position distribution of the antenna array is different, and the corresponding UV coverage is different.

10. An electronic device, characterized in that: include; one or more processors; a memory configured to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the synthetic aperture radiometer UV coverage optimization method according to any one of claims 1 to 8.

Images