Control method and system for intelligent networking astronomical telescope of synchronous base

Abstract

The invention provides an intelligent networking astronomical telescope control method and system for a synchronous base, and relates to the technical field of telescope control, and the method comprises the steps: obtaining historical observation data of an astronomical telescope and initial star map data uploaded in real time; performing star map matching and attitude estimation of the astronomical telescope based on the initial star map data, and performing coordinate transformation on data obtained by attitude estimation to obtain initial pointing data of the astronomical telescope; calculating a star motion error based on the initial pointing data, and inputting the star motion error into a preset control strategy determination model for processing to obtain an initial control strategy of the astronomical telescope; and controlling the astronomical telescope to obtain astronomical image data based on the initial control strategy, and optimizing the initial control strategy based on the astronomical image data and historical observation data to obtain a final control strategy of the astronomical telescope. According to the method, the problems of low astronomical data storage efficiency and insufficient long-term observation strategy optimization capability are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of telescope control, and in particular, to an intelligent networking astronomical telescope control method and system for a synchronous base. Background Art

[0002] In the field of modern astronomical observations, high-precision automated astronomical telescopes have become important tools for obtaining astronomical data. Traditional astronomical telescope control methods mainly rely on fixed observation plans and manual adjustments, including rough alignment based on stellar navigation, preset trajectory tracking, and single error compensation techniques. However, these methods have various limitations in practical applications. First, since traditional star chart matching methods are mainly based on template matching or simple feature point comparison, they are easily affected by star chart distortion, light pollution, and noise interference, resulting in low matching accuracy and thus affecting the pointing accuracy of the telescope. Second, existing astronomical telescope control systems usually adopt static error compensation methods and cannot effectively cope with complex dynamic observation environments, leading to the accumulation of pointing errors. In addition, the traditional control strategy optimization method mainly relies on preset observation plans and lacks the ability of real-time task allocation and multi-telescope collaborative control, resulting in low observation efficiency and difficulty in meeting the requirements of high-precision, long-term, and multi-target observations.

[0003] Therefore, there is an urgent need for an intelligent networking astronomical telescope control method and system for a synchronous base to solve the above problems. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent networking astronomical telescope control method and system for a synchronous base to improve the above problems. To achieve the above purpose, the technical solutions adopted by the present invention are as follows: In a first aspect, the present application provides an intelligent networking astronomical telescope control method for a synchronous base, including: Obtaining historical observation data of the astronomical telescope and initial star chart data uploaded in real time; Performing star chart matching and attitude estimation of the astronomical telescope based on the initial star chart data, and performing coordinate transformation on the data obtained by the attitude estimation to obtain initial pointing data of the astronomical telescope; Calculating the stellar motion error based on the initial pointing data, and inputting the stellar motion error into a preset control strategy determination model for processing to obtain an initial control strategy of the astronomical telescope; Controlling the astronomical telescope to obtain astronomical image data based on the initial control strategy, and optimizing the initial control strategy based on the astronomical image data and historical observation data to obtain a final control strategy of the astronomical telescope.

[0005] In a second aspect, the present application also provides an intelligent networking astronomical telescope control system for a synchronous base, including: An acquisition unit for acquiring historical observation data of the astronomical telescope and initial star chart data uploaded in real time; A processing unit for performing star chart matching and attitude estimation of the astronomical telescope based on the initial star chart data, and performing coordinate transformation on the data obtained from the attitude estimation to obtain initial pointing data of the astronomical telescope; A calculation unit for calculating the stellar motion error based on the initial pointing data, and inputting the stellar motion error into a preset control strategy determination model for processing to obtain an initial control strategy of the astronomical telescope; An optimization unit for controlling the astronomical telescope to acquire astronomical image data based on the initial control strategy, and optimizing the initial control strategy based on the astronomical image data and historical observation data to obtain a final control strategy of the astronomical telescope.

[0006] The beneficial effects of the present invention are as follows: First, based on the star chart matching optimization network and adaptive particle filter, the present invention realizes high-precision attitude estimation and coordinate transformation of the astronomical telescope, thereby improving the pointing accuracy of the telescope. Secondly, the generalized sparse optical flow estimation algorithm is used to analyze the stellar motion error, and combined with the dual adaptive Kalman filter and adaptive sliding mode observer for error compensation, thereby optimizing the control strategy of the telescope and improving the pointing stability in the dynamic observation environment. At the same time, this method introduces an entropy-regularized deep reinforcement learning model and a dynamic graph task allocation algorithm to realize the task collaborative optimization of multiple telescopes and improve the observation efficiency.

[0007] In terms of data processing, the present invention extracts stellar spectral features based on the probability tensor decomposition method, and combines the hybrid greedy search to optimize the calculation task, improving the intelligent level of data processing. In addition, to solve the problems of low astronomical data storage efficiency and insufficient long-term observation strategy optimization ability, the present invention realizes data storage optimization and dynamic adjustment of the long-term observation strategy, thereby ensuring the high efficiency of data storage and the accuracy of long-term observation tasks.

[0008] Other features and advantages of the present invention will be described in the subsequent specification, and part of them will become obvious from the specification, or be understood by implementing the embodiments of the present invention. The objectives and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in the written specification, claims, and drawings. Description of the Drawings

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0010] Figure 1 Schematic diagram of the control method for the intelligent networked astronomical telescope of the synchronous base described in the embodiments of the present invention; Figure 2 Schematic diagram of the control system structure for the intelligent networked astronomical telescope of the synchronous base described in the embodiments of the present invention.

[0011] In the figure: 701, acquisition unit; 702, processing unit; 703, calculation unit; 704, optimization unit. Detailed implementation manners

[0012] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and shown in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed present invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0013] It should be noted that: similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present invention, terms such as "first" and "second" are only used for differential description and cannot be understood as indicating or implying relative importance. Embodiment 1:

[0014] This embodiment provides a control method for an intelligent networked astronomical telescope of a synchronous base.

[0015] See Figure 1 , the figure shows that this method includes step S1, step S2, step S3, and step S4.

[0016] Step S1, acquire the historical observation data of the astronomical telescope and the initial star chart data uploaded in real time: It is understandable that in this step, the historical observation data includes the past observation records of the telescope, the identified star data, the equipment operation parameters, and the environmental conditions (such as atmospheric disturbances, temperature, humidity, etc.). These data can not only provide reference for the current observation, but also optimize the control strategy of the telescope through data-driven methods. For example, historical data can be used to analyze the pointing deviation of the astronomical telescope in different environments, and then an error compensation model can be formed.

[0017] The initially uploaded star chart data is mainly obtained by the star chart acquisition system carried by the telescope, including the starry sky images of the current observation area and the corresponding time and coordinate information. After preprocessing, these star chart data can be matched with the historical star catalog to correct the initial pointing error of the telescope. By constructing a data cache and an edge computing module, rapid retrieval of historical data is realized, and the accuracy of real-time star chart data is improved by using adaptive data fusion technology, thus laying a foundation for subsequent star chart matching and attitude estimation.

[0018] Step S2: Perform star chart matching and attitude estimation of the astronomical telescope based on the initially uploaded star chart data, and perform coordinate transformation on the data obtained from the attitude estimation to obtain the initial pointing data of the astronomical telescope; It is understandable that in this step, through a regularized star chart matching network, factors such as light pollution and imaging distortion are effectively suppressed, and the matching accuracy is improved. Through adaptive particle filtering, complex noise environments can be processed more precisely, and the attitude estimation accuracy of the telescope is improved. By using the Lie group optimal alignment algorithm, cumulative errors are reduced, ensuring high-precision conversion of the telescope pointing data and providing reliable input data for subsequent error compensation and observation task execution. In this step, step S2 includes Step S21: Perform optimized processing of star chart matching according to the initially uploaded star chart data and the preset standard star positions. Among them, star chart features are extracted and star chart matching is performed by constructing a regularized star chart matching network to obtain an optimized star chart matching result; It is understandable that in actual observations, the initially uploaded star chart data is usually interfered by factors such as light pollution, atmospheric turbulence, and imaging distortion, resulting in blurred star points and position offsets, which affect the accuracy of matching. Therefore, in this step, a regularized star chart matching network is constructed to optimize the star chart matching process, improving the matching accuracy and robustness.

[0019] Among them, due to the low brightness and poor contrast of weak signal star points, they are often difficult to be stably detected by traditional methods. Therefore, a local contrast enhancement method is adopted to optimize the visibility of star points. Specifically, the adaptive histogram equalization method is used to equalize the gray level of the local area of the star map, so that the dark star points can still be distinguishable under different background brightnesses. In addition, the Laplace-Poisson equation is introduced to calculate the local gradient field to balance the brightness distribution in the high dynamic range star map and avoid the excessive amplification of background noise during the enhancement process. Then, the histogram of oriented gradients method is used to extract the feature points of the star map. After the feature points are extracted, the regularized star map matching network is used for matching optimization. Since weak star points are often vulnerable to noise interference during the matching process, a multi-scale residual regularization loss function is introduced to enhance the matching weight of weak star points and suppress background noise at the same time, obtaining a set of matching points. Among them, the defined loss function is as follows: ;

[0020] Among them, represents the matching degree between two data or images, represents the -th element in the data set, , and are weight coefficients used to adjust the contribution of each loss function to the total loss function, represents the element in the -th data set, represents the -th element in the second data set, represents the gradient, represents the Gaussian filtering function.

[0021] The transformation parameters are calculated by randomly sampling the set of matching points, and the projection error is used to screen out the mismatched points, ultimately improving the credibility of the matching data. The whole process ensures the effective matching of weak signal star points and improves the attitude estimation accuracy and stability of the astronomical telescope.

[0022] Step S22: According to the adaptive particle filter and the optimized star map matching result, perform the attitude estimation of the astronomical telescope to obtain the attitude estimation data of the telescope; After obtaining the optimized star map matching result, it is necessary to calculate the preliminary pointing angle of the telescope based on the optical center and celestial coordinate transformation model. Assume that the pixel coordinates of the matching star points are , and its corresponding celestial equatorial coordinates are , then the optical axis direction of the telescope can be solved through the optical imaging model. Assume that the imaging system of the telescope can be approximated as an ideal pinhole camera model, and its projection transformation can be expressed as: ;

[0023] Among them, is the camera internal parameter matrix, is the rotation matrix of the astronomical telescope, is the three-dimensional celestial coordinate of the star point. is the pixel coordinate of the matched star point.

[0024] By directly linear transformation to solve the rotation matrix of the telescope, the preliminary pointing angle of the telescope can be obtained, and further the attitude matrix of the telescope can be constructed through quaternion rotation transformation: ;

[0025] Among them, represents the rotation matrix, represents the real part of the quaternion, , and represent the imaginary part of the quaternion.

[0026] Then, set the initial particle set (i.e., multiple possible attitude states), calculate the weight of each particle through importance sampling, obtain the particle distribution of each particle, and further obtain the attitude estimation data of the telescope.

[0027] Step S23: According to the attitude estimation data, align the telescope coordinate system with the celestial coordinate system through the Lie group optimal alignment algorithm to obtain the initial pointing data of the astronomical telescope.

[0028] After star map matching, we obtain the set of star point coordinates observed by the telescope and the set of standard star coordinates in the star catalog. The goal of coordinate transformation is to find the best rotation matrix and translation vector so that the star points in the telescope coordinate system can be mapped to the astronomical coordinate system as accurately as possible. For this purpose, first perform a rotation transformation through the quaternion rotation matrix, and then set the star point coordinates in the standard star catalog. The goal is to minimize the transformation error between the observed coordinates and the standard star coordinates, and construct an optimization objective function: ;

[0029] Among them, represents the optimization objective function, is the rotation matrix, is the observed star point coordinate, is the corresponding standard coordinate in the star catalog, represents the coordinate translation amount.

[0030] During the optimization process, the gradient descent method is used to adjust the rotation vector to minimize the error, and the new rotation matrix is calculated through exponential mapping. This method can ensure that no numerical drift occurs and improves the accuracy and stability of star map alignment.

[0031] Step S3: Calculate the stellar motion error based on the initial pointing data, and input the stellar motion error into a preset control strategy determination model for processing to obtain the initial control strategy of the astronomical telescope; It can be understood that in this step, the stellar motion error is calculated based on the initial pointing data, and this error is input into a preset control strategy determination model for processing to determine the initial control strategy of the telescope, so that it can track the target celestial body more accurately. In this step, the telescope's target pointing is corrected through error calculation to make it closer to the true position of the target celestial body. In this step, step S3 includes step S31, step S32, and step S33.

[0032] Step S31: Conduct an analysis of the stellar motion error based on the initial pointing data. Among them, the offset information of the stellar motion is estimated through generalized sparse optical flow to obtain the motion error data of the astronomical telescope; It can be understood that in this step, first, the initial pointing information is obtained through the observation data of the astronomical telescope, including the positions of the stars calibrated in the star map. Each star has a known initial position in the observation (given by the preset star catalog data). These data are the initial observation positions of the stars in the telescope coordinate system.

[0033] Among them, the position of a star in the field of view will shift due to various factors (such as the rotation of the Earth, the orbital motion of celestial bodies, atmospheric effects, etc.). This position change is time - related, that is, the trajectory of the star in the telescope's field of view will no longer be stationary over time. The goal is to calculate the stellar offset and its corresponding error by quantifying these motions.

[0034] Through the optical flow constraint equation, the movement information of each star point can be deduced. Assuming that the movement of the star map image is smooth and continuous, the optical flow constraint equation can be expressed as: ;

[0035] Among them, is the image intensity with respect to the spatial coordinate partial derivative, representing the rate of change of the image intensity in the horizontal direction, is the image intensity with respect to the spatial coordinate partial derivative, representing the rate of change of the image intensity in the vertical direction, is the image intensity with respect to time partial derivative, representing the change of the image intensity over time. This term reflects the change of image pixels over time. and are the components of the optical flow, representing the motion speed of the pixels in the image.

[0036] After calculating the motion offset of each celestial body, using this offset information, we can determine the motion error of the telescope.

[0037] Step S32: Perform error compensation processing on the motion error data according to the dual adaptive Kalman filtering algorithm to obtain the pointing data after error compensation. It can be understood that in this step, for the observation error of the astronomical telescope, the changes in system noise and observation noise are unpredictable, especially in high-precision applications. Therefore, the dual adaptive Kalman filtering algorithm dynamically adjusts the noise estimation value, enabling the filter to update the noise covariance matrix in real time according to the changes in the observation data. By adaptively adjusting the Kalman gain, we can optimize the error compensation process.

[0038] After each observation by the dual adaptive Kalman filtering algorithm, the Kalman gain is adjusted according to the actual observation error to ensure that the gain of the filter adapts to the current noise environment. The update formula for the Kalman gain in the dual adaptive Kalman filtering algorithm is: ;

[0039] where represents the Kalman gain, represents the error covariance matrix of the system at time , represents the observation matrix, represents the measurement noise covariance matrix, represents the transpose of the observation matrix .

[0040] Using the dual processing ability of the dual adaptive Kalman filter, we perform more precise compensation for the motion error. The first step is to correct the predicted motion error, and the second step is to gradually optimize according to the dynamic observation data to obtain the pointing data after compensation. Through the correction process, the error caused by external disturbances (such as wind speed, temperature change, etc.) can be eliminated. Among them, through the update process of the Kalman filter, the system state is estimated more precisely, and the error is effectively compensated. The adaptive mechanism enables the filter to dynamically adjust and adapt to various environmental noises and system errors. Due to the gradual optimization of the filtering algorithm, the pointing data is more stable, reducing the error caused by short-term fluctuations or external factors.

[0041] Step S33: Perform synchronous base control according to the pointing data after error compensation, and optimize the control strategy through an adaptive sliding mode observer to obtain the control strategy of the astronomical telescope.

[0042] It is understandable that this step ensures that the base of the astronomical telescope can respond quickly and accurately to the pointing data after error compensation, so that the lens of the telescope remains on the target star, avoiding the deviation or bias of the observation data caused by inconsistent movements.

[0043] In this step, first, according to the pointing data after error compensation, the adjustment angle required for the telescope base is calculated. The goal is to ensure that the movement of the telescope is completely synchronized with the movement of the star. Then, through the PID control algorithm, precise adjustment of the telescope base is achieved. Specifically, the PID control algorithm adjusts the angular velocity and acceleration of the base to ensure that the base can respond to changes in the pointing data within the shortest time and reach the predetermined angle. The calculation formula for the target angle is as follows: ;

[0044] where, represents the reference signal or control signal output by the controller, that is, the target angle, represents the proportional gain, which is the proportional part of the controller; represents the error between the target angle and the current angle, represents the derivative gain, which is the derivative part of the controller, represents the rate of change of the error (i.e., the derivative of the error), represents the integral gain, which is the integral part of the controller, represents the integral of the error.

[0045] In this step, after performing the synchronous base control, a sliding mode surface is set. When the system state is on the sliding mode surface, the control system can eliminate the error and force the system to slide along this surface, ensuring that the system can resist external disturbances and quickly stabilize. The sliding mode surface can be defined as a weighted combination of the error and the rate of change of the error, as shown below: ;

[0046] where, represents the control signal or the state error function, is the weight coefficient, representing the contribution degree of the angle error to the control signal, represents the current angle of the synchronous base, represents the target angle of the synchronous base, is another weight coefficient, representing the contribution degree of the speed error to the control signal, is the current angular velocity of the synchronous base, represents the target angular velocity of the synchronous base.

[0047] This step aims to overcome the problems in traditional sliding mode control by introducing an adaptive adjustment mechanism. Based on the real-time feedback information of the system, the gain parameters of the sliding mode controller are dynamically adjusted to obtain the control strategy of the astronomical telescope, so that the system can cope with various external disturbances and uncertainties. Among them, the adaptive control law of the adaptive sliding mode observer is as follows: ;

[0048] Among them, denotes the control signal. It is the output signal of the control system, used to adjust the state of the controlled object, denotes the adaptive gain matrix, denotes the difference between the current state and the target state of the system, which is composed of position error and velocity error.

[0049] Through the combination of sliding mode control and adaptive mechanism in this step, the system can automatically adapt to environmental changes and internal errors, providing strong robustness and ensuring stable control performance even in an environment with large noise or interference.

[0050] Step S4: Control the astronomical telescope based on the initial control strategy to obtain astronomical image data, and optimize the initial control strategy based on the astronomical image data and historical observation data to obtain the final control strategy of the astronomical telescope.

[0051] It can be understood that this step uses the obtained astronomical image data and historical observation data to optimize the initial control strategy. This optimization process based on data feedback enables the control strategy to have stronger adaptive capabilities, capable of making real-time adjustments for different environmental conditions, target celestial bodies, and observation tasks. In this way, the telescope can dynamically respond to different observation scenarios, enhancing the robustness and stability of the system. In this step, step S4 includes step S41, step S42, and step S43.

[0052] Step S41: Input the control strategy of the astronomical telescope and the preset observation requirements into the entropy-regularized deep reinforcement learning model for optimizing the task allocation strategy, and obtain the preliminarily optimized task allocation strategy; It can be understood that in this step, the control strategy of the astronomical telescope includes attitude adjustment, base control, etc. These strategies are the results based on the motion error compensation and control strategy optimization of the telescope. It defines the actions and behaviors of the telescope under different observation conditions. The preset observation requirements include parameters such as the observation target of a specific celestial body, the observation time window, the image resolution, and the exposure time.

[0053] It can be understood that in this step, the control strategies of the telescope, such as attitude control and motion control of the telescope, are combined with preset observation requirements, such as the observation of specific celestial bodies, exposure duration, image clarity, etc., and input into the deep reinforcement learning model. Through interaction with the environment, the model learns how to dynamically adjust task allocation under the given control strategies and task requirements to ensure the optimal utilization of telescope resources (such as field of view, exposure time, etc.). The entropy regularization mechanism plays an important role here. By penalizing the uncertainty of the strategy, it promotes a more stable and predictable task allocation. Through this optimization process, the initially optimized task allocation strategy can more precisely guide the resource allocation and task execution of the telescope under different observation requirements, ensuring that resource conflicts are reduced and observation efficiency is improved when multiple tasks are carried out simultaneously. The technical effect of this step is to adaptively optimize the task scheduling of the telescope through deep reinforcement learning, avoiding the deficiencies of traditional static strategies and improving the system response ability and the accuracy of task execution. Among them, the optimization objectives of task allocation include not only the maximization of rewards, but also factors such as task priorities, resource allocation, and the execution efficiency of control strategies. Therefore, the reward function for task allocation needs to comprehensively consider the actual task requirements and control strategies of the astronomical telescope. The specific task allocation reward function is as follows: ;

[0054] Among them, denotes the task allocation reward function, denotes the action the task execution efficiency in a given state, denotes the priority of the current task, denotes the clarity of the astronomical image and the accuracy of celestial body observation, 、 and respectively denote 、 and the index weights of.

[0055] Step S42: Through the dynamic graph task allocation algorithm, perform multi-telescope task collaborative optimization on the initially optimized task allocation strategy to obtain the task allocation strategy after collaborative optimization; It can be understood that the collaborative optimization of multi-telescope tasks needs to consider constraints such as the resources, observation requirements, and operation space of multiple telescopes. The dynamic graph task allocation algorithm optimizes by updating the task status and observation data of each telescope in real time. In the dynamic graph task allocation algorithm, it is first necessary to model the task requirements, status information, and respective execution strategies of multiple telescopes. The task allocation strategy for each telescope should not only consider its own status (such as position, available resources, observation targets, etc.), but also consider the collaborative task requirements with other telescopes. This task collaborative optimization problem can be modeled by the dynamic flow algorithm in graph theory. Among them, during the task allocation process, the dynamic graph algorithm will dynamically adjust the allocation of each task according to real-time observation data and task priorities. For example, if the execution time of a certain task is delayed or a telescope cannot complete the task due to a fault, the corresponding task execution order is adjusted or assigned to other telescopes through dynamic flow. After obtaining the priority of each task, a scheduling algorithm is used for dynamic scheduling and allocation of tasks. In the collaborative task optimization of multiple telescopes, task allocation not only considers the execution ability of each telescope, but also ensures the collaborative cooperation of multiple telescopes in observation targets, avoiding conflicts and waste of resources in task execution. Through collaborative optimization, the comprehensive task benefits of all telescopes are maximized, and the working efficiency of the overall system is optimized.

[0056] Step S43: Predict the optimal observation time window according to the task allocation strategy after collaborative optimization and historical observation data. Among them, a control strategy prediction model is established through a hierarchical memory network, and a control strategy prediction is performed to obtain the final control strategy of the astronomical telescope.

[0057] It can be understood that in this step, the hierarchical memory network includes multiple processing levels, and each level stores and processes information through different memory units. This network can capture both short-term and long-term dependencies, so that under multi-dimensional data input, the most representative features are extracted for task prediction. For the control strategy prediction of the astronomical telescope, the input data includes: historical observation data: such as the targets observed by the astronomical telescope in the past, task execution situations, timestamps, etc. Task allocation strategy: The task allocation scheme after collaborative optimization, which describes the tasks executed by each telescope at different time points and their priorities. Telescope status data: including information such as the operating status and resource availability of each telescope. Through multiple levels of the hierarchical memory network, the model can dynamically learn the correlation of this information and predict the future observation time window combined with past experience.

[0058] During the control strategy prediction process, the model not only focuses on the priorities and resource allocations of the current tasks but also needs to predict the optimal switching times between different observation tasks. This process requires deep learning on the past task execution situations to identify which observation tasks can be efficiently executed during which time periods. Specifically, the prediction model analyzes historical data to predict which tasks have high execution benefits at the current moment and calculates the most appropriate observation time window. Among them, the control strategy prediction formula can be expressed as: ;

[0059] Among them, represents the optimal observation time window at time , represents the task execution efficiency, represents the priority of the task, represents selecting the task that maximizes the product of efficiency and priority, represents the task or operation.

[0060] It can be understood that after step S4, steps S5, S6, and S7 are also included.

[0061] Step S5: Control the astronomical telescope for observation based on the final control strategy, and perform image denoising processing on the obtained astronomical image data to obtain the denoised astronomical image data; It can be understood that after the astronomical telescope completes the observation, the obtained image data will contain the optical images of celestial bodies. However, due to the complexity of the observation environment (such as factors like atmospheric disturbances and telescope movement), the obtained images are often accompanied by noise. The noise may come from various sources, such as the noise of optical devices, atmospheric interference, errors in signal transmission, etc. These noises not only affect the clarity of the images but may also interfere with subsequent astronomical analysis and research. For the astronomical telescope system, denoising processing can ensure that the obtained data is more real and reliable. Especially in the case of long-exposure or high-noise environments, it can effectively improve the signal-to-noise ratio of the images and enhance the effect of celestial body observation. In addition, the introduction of deep learning methods makes the denoising process not only highly adaptable but also able to be continuously optimized as the data volume increases in practical applications, further improving the accuracy and efficiency of image processing.

[0062] Step S6: Perform probability tensor decomposition on the denoised astronomical image data, and extract and classify the spectral features of the stars to obtain star identification data; It can be understood that in this step, a tensor is an extension of a high-dimensional array and can naturally represent multi-dimensional data. In the scenario of astronomical images, we can represent the features of multiple dimensions such as the spatial position of the image (such as row and column pixels), spectral information (such as image data in different bands), etc. as tensors.

[0063] In this step, the goal of probabilistic tensor decomposition is to extract the underlying low-dimensional structure from the tensor, thereby revealing the hidden patterns in the image. In this process, the tensor is decomposed through a probabilistic model to generate a set of latent factors (factor matrices), which contain the main information in the image. The factor matrices obtained through probabilistic tensor decomposition contain information such as space and spectrum in the image. By further analyzing these factor matrices, the spectral characteristics of each star can be extracted. Then, through K-means clustering, celestial bodies are classified into multiple categories based on the similarity of spectral characteristics, and star identification data is obtained, including: the position of each star (i.e., spatial coordinates), the spectral characteristics of each star (such as spectral type, temperature, redshift, etc.), and the classification label of the star (such as star, planet, nebula, etc.).

[0064] Step S7: Optimally allocate the computing tasks according to the hybrid greedy search optimization algorithm and the star identification data to obtain the optimized astronomical computing tasks.

[0065] In this step, the hybrid greedy search optimization algorithm initializes the task list and the computing resource pool according to the input star identification data and computing task information. Each computing task is labeled with a different priority according to the star identification data. Tasks with high priority may be important star feature extraction or complex spectral analysis tasks. The computing task pool is sorted according to the complexity of the tasks, and the tasks are divided into two categories: "simple tasks" and "complex tasks".

[0066] Then, the tasks are sorted according to factors such as the priority of the stars and the complexity of the tasks. For example, tasks with high priority and high computational complexity will be allocated to high-performance computing resources. Resource allocation: Select appropriate resources from the computing resource pool to process the sorted tasks. At this time, the greedy algorithm selects the current optimal resource to process the current task without considering future task allocation. For example, if the first task requires more computing resources (such as time, memory), and the second task is relatively simple, the greedy algorithm will give priority to allocating more computing resources to the first task and less resources to the second task.

[0067] Among them, although the greedy algorithm can quickly obtain a preliminary solution, since it may be locally optimal, local search is used for further optimization, including: adjusting the task order and resource reallocation. Adjusting the task order includes that local search will adjust the allocation order of tasks and try different task scheduling to find a more optimized solution. For example, some computationally intensive tasks may require more computing resources, but may also be assigned to inefficient resources due to their low priority. Local search can avoid this kind of resource waste by adjusting the matching of priorities and resources. Resource reallocation includes that local search will also adjust resource allocation according to the actual execution situation of tasks (such as execution time, resource consumption, etc.) to optimize the utilization efficiency of computing resources.

[0068] It can be understood that after step S7, steps S8 and S9 are also included.

[0069] Step S8: Perform local cache and cloud storage optimization based on the denoised astronomical image data and star identification data. Among them, the data is structurally stored through a hierarchical storage management model to obtain optimized storage data; It can be understood that in this step, first, the denoised astronomical image data is classified according to the priority of the stars or the needs of scientific research. For example, the image data of key stars may be marked as "high priority", while the image data of ordinary background stars may be marked as "low priority". Classification of star identification data: The identification data is structurally classified according to the characteristics or classification labels of the stars. Some stars with special scientific value (such as supernovae or black holes, etc.) may be marked as high-priority data, while ordinary stars may be marked as low-priority data. Then, according to the data classification, the high-priority astronomical image data and star identification data are allocated to the local cache. Since local storage has a faster access speed, it can ensure the quick response of high-priority tasks and ensure the real-time processing and analysis of high-priority tasks by the astronomical telescope.

[0070] Step S9: Optimize the observation strategy for a preset time period based on the optimized storage data. Among them, the observation strategy for the preset time period is dynamically adjusted through a cyclic Bayesian update algorithm to obtain the observation strategy for the preset time period; It can be understood that in this step, the optimized storage data and the observation strategy for the preset time period are used as input data. The optimized storage data includes data such as astronomical image data and star identification data that have been optimized through storage, and these data may reflect the observation characteristics in different time periods. The observation strategy for the preset time period is the initially set observation time period and strategy plan, including the observation activities and targets that should be carried out within a specific time period.

[0071] Then, for the observation strategy of each time period, set its initial estimate (prior distribution). For example, the initial observation time period may be set based on the average of historical data or empirical rules.

[0072] Next, update the likelihood function. As the observation data accumulates, each round of new observation data will be used to update the likelihood function. These new data reflect the observation effect of the current time period, and may include the specific performance of celestial bodies, the clarity of images, the usage of observation resources, etc.

[0073] Furthermore, calculate the posterior distribution. According to Bayes' formula, by combining the likelihood function with the prior distribution, the posterior distribution is calculated. This posterior distribution represents the best estimate of the future observation strategy under the current observation data.

[0074] Finally, make strategy adjustments. Among them, extract the optimal strategy parameters (such as the best observation time period, target selection, etc.) from the posterior distribution, and adjust the future observation strategy according to these parameters. Embodiment 2:

[0075] As Figure 2 shown, this embodiment provides an intelligent networked astronomical telescope control system for a synchronous base. Refer to Figure 2 The system includes an acquisition unit 701, a processing unit 702, a calculation unit 703, and an optimization unit 704.

[0076] The acquisition unit 701 is used to acquire the historical observation data of the astronomical telescope and the initial star chart data uploaded in real time; The processing unit 702 is used to perform star chart matching and attitude estimation of the astronomical telescope based on the initial star chart data, and perform coordinate transformation on the data obtained from the attitude estimation to obtain the initial pointing data of the astronomical telescope; The calculation unit 703 is used to calculate the stellar motion error based on the initial pointing data, and input the stellar motion error into a preset control strategy determination model for processing to obtain the initial control strategy of the astronomical telescope; The optimization unit 704 is used to control the astronomical telescope to acquire astronomical image data based on the initial control strategy, and optimize the initial control strategy based on the astronomical image data and historical observation data to obtain the final control strategy of the astronomical telescope.

[0077] It should be noted that regarding the system in the above embodiments, the specific manners in which each module performs operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0078] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0079] As described above, the above are only the specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or replacements, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A control method for an intelligent networked astronomical telescope with a synchronous base, characterized in that: include: Obtain historical observation data from astronomical telescopes and initial star map data uploaded in real time; Performing star map matching and attitude estimation of the astronomical telescope based on the initial star map data, and performing coordinate conversion on the data obtained by the attitude estimation to obtain initial pointing data of the astronomical telescope; Calculating the star motion error based on the initial pointing data, and inputting the star motion error into a preset control strategy determination model for processing to obtain an initial control strategy for the astronomical telescope; The astronomical telescope is controlled to acquire astronomical image data based on the initial control strategy, and the initial control strategy is optimized based on the astronomical image data and historical observation data to obtain a final control strategy for the astronomical telescope.

2. The control method of the intelligent networked astronomical telescope of the synchronous base according to claim 1, characterized in that: Star map matching and attitude estimation of the astronomical telescope are performed based on the initial star map data, and coordinate conversion is performed on the data obtained by the attitude estimation, including: According to the initial star map data and the preset standard star positions, star map matching optimization processing is performed, wherein the star map features and star map matching are extracted by constructing a regularized star map matching network to obtain an optimized star map matching result; According to the adaptive particle filter and the optimized star map matching results, the attitude estimation of the astronomical telescope is performed to obtain the attitude estimation data of the telescope; According to the attitude estimation data, the telescope coordinate system is aligned with the celestial sphere coordinate system through a Lie group optimal alignment algorithm to obtain initial pointing data of the astronomical telescope.

3. The control method of the intelligent networked astronomical telescope of the synchronous base according to claim 1, characterized in that: Calculating the star motion error based on the initial pointing data and inputting the star motion error into a preset control strategy determination model for processing, including: Performing star motion error analysis based on the initial pointing data, wherein the offset information of star motion is calculated by generalized sparse optical flow estimation to obtain motion error data of the astronomical telescope; Performing error compensation processing on the motion error data according to a dual adaptive Kalman filter algorithm to obtain error-compensated pointing data; The synchronous base control is performed according to the pointing data after error compensation, and the control strategy is optimized through an adaptive sliding mode observer to obtain the control strategy of the astronomical telescope.

4. The control method of the intelligent networked astronomical telescope of the synchronous base according to claim 1, characterized in that: And optimizing the initial control strategy based on the astronomical image data and historical observation data, including: The control strategy and preset observation requirements of the astronomical telescope are input into the entropy regularized deep reinforcement learning model to optimize the task allocation strategy and obtain a preliminary optimized task allocation strategy. The task allocation strategy initially optimized is optimized through the dynamic graph task allocation algorithm to coordinate the multi-telescope tasks and obtain the task allocation strategy after coordination optimization. The optimal observation time window is predicted based on the collaboratively optimized task allocation strategy and historical observation data. A control strategy prediction model is established through a hierarchical memory network, and the control strategy prediction is performed to obtain the final control strategy of the astronomical telescope.

5. The control method of the intelligent networked astronomical telescope of the synchronous base according to claim 1, characterized in that: After obtaining the final control strategy of the astronomical telescope, it also includes: Based on the final control strategy, the astronomical telescope is controlled to perform observation, and image noise reduction processing is performed based on the astronomical image data obtained by the observation to obtain the astronomical image data after noise reduction; Performing probability tensor decomposition on the astronomical image data after noise reduction, and extracting spectral features and classification of stars to obtain star identification data; The computing tasks are optimally allocated according to the hybrid greedy search optimization algorithm and the star recognition data to obtain optimized astronomical computing tasks.

6. An intelligent networked astronomical telescope control system with a synchronous base, characterized in that: include: An acquisition unit is used to acquire historical observation data of the astronomical telescope and initial star map data uploaded in real time; A processing unit, configured to perform star map matching and attitude estimation of the astronomical telescope based on the initial star map data, and perform coordinate conversion on the data obtained by the attitude estimation to obtain initial pointing data of the astronomical telescope; A calculation unit, used for calculating the star motion error based on the initial pointing data, and inputting the star motion error into a preset control strategy determination model for processing to obtain an initial control strategy for the astronomical telescope; The optimization unit is used to control the astronomical telescope to obtain astronomical image data based on the initial control strategy, and optimize the initial control strategy based on the astronomical image data and historical observation data to obtain a final control strategy for the astronomical telescope.

7. The intelligent networked astronomical telescope control system of the synchronous base according to claim 6, characterized in that: The processing unit comprises: The first processing subunit is used to perform star map matching optimization processing according to the initial star map data and the preset standard star positions, wherein the star map features and star map matching are extracted by constructing a regularized star map matching network to obtain an optimized star map matching result; The second processing subunit is used to perform attitude estimation of the astronomical telescope according to the adaptive particle filtering and the optimized star map matching results to obtain attitude estimation data of the telescope; The third processing subunit is used to align the telescope coordinate system with the celestial sphere coordinate system through a Lie group optimal alignment algorithm according to the attitude estimation data, so as to obtain initial pointing data of the astronomical telescope.

8. The intelligent networked astronomical telescope control system of the synchronous base according to claim 6, characterized in that: The computing unit comprises: A first calculation subunit is used to perform star motion error analysis based on the initial pointing data, wherein the offset information of the star motion is calculated by generalized sparse optical flow estimation to obtain the motion error data of the astronomical telescope; A second calculation subunit is used to perform error compensation processing on the motion error data according to a dual adaptive Kalman filter algorithm to obtain error-compensated pointing data; The third computing subunit is used to perform synchronous base control according to the pointing data after error compensation, and optimize the control strategy through an adaptive sliding mode observer to obtain the control strategy of the astronomical telescope.

9. The intelligent networked astronomical telescope control system of the synchronous base according to claim 6, characterized in that: The optimization unit comprises: The first optimization subunit is used to input the control strategy of the astronomical telescope and the preset observation requirements into the entropy regularized deep reinforcement learning model to optimize the task allocation strategy and obtain a preliminary optimized task allocation strategy; The second optimization subunit is used to perform multi-telescope task collaborative optimization on the initially optimized task allocation strategy through a dynamic graph task allocation algorithm to obtain a collaboratively optimized task allocation strategy; The third optimization subunit is used to predict the optimal observation time window according to the collaboratively optimized task allocation strategy and historical observation data. A control strategy prediction model is established through a hierarchical memory network, and control strategy prediction is performed to obtain the final control strategy of the astronomical telescope.

10. The intelligent networked astronomical telescope control system of the synchronous base according to claim 6, characterized in that: After the optimization unit, it also includes: A fourth processing subunit is used to control the astronomical telescope to perform observation based on the final control strategy, and perform image noise reduction processing based on the astronomical image data obtained by observation to obtain the astronomical image data after noise reduction; a fifth processing subunit, configured to perform probability tensor decomposition on the astronomical image data after noise reduction, and extract spectral features and classification of stars to obtain star identification data; The fourth computing subunit is used to optimize the allocation of computing tasks according to the hybrid greedy search optimization algorithm and the star recognition data to obtain optimized astronomical computing tasks.

Images