Satellite-ground integrated operation and control method for coaxial optical radar load satellites

Abstract

The invention relates to the technical field of directional control, and provides a coaxial optical radar load satellite satellite-ground integrated operation control method, which comprises the following steps: firstly, acquiring state parameters between a satellite and a ground station through a coaxial optical radar, and carrying out noise reduction processing on data to provide accurate input for subsequent state estimation; then, predicting the current state of the satellite by using Kalman filtering, and generating a state estimation value at the next moment; according to the estimated values, control instructions are generated and decomposed, and the control instructions of all dimensions are processed step by step and checked; then, state estimation and control instruction execution data are input into the deep reinforcement learning model, and a control instruction is optimized; in order to improve the Kalman filtering precision, a space-time correlation factor is introduced, and a state transition matrix is adjusted, so that the dynamic relationship between a satellite and a ground station is reflected more accurately; in addition, the deep reinforcement learning model further optimizes a control instruction through training of historical data and execution data.

Description

Technical Field

[0001] The present invention relates to the field of directional control technology, and more specifically, to a satellite-ground integrated operation and control method for a coaxial optical radar payload satellite. Background Art

[0002] Coaxial optical radar payload satellite is an advanced space measurement and control technology. It integrates ranging and tracking functions on a single optical axis through coaxial optical radar to achieve high-precision, real-time space positioning and trajectory control. The coaxial optical radar payload satellite satellite-ground integrated operation and control system combines ground control and satellite autonomous control to achieve accurate satellite positioning and efficient attitude adjustment, and can perform complex mission operations such as one station and multiple satellites under an integrated control framework. Through the satellite-ground integration approach, the complexity of ground control can be simplified, while improving the reliability and real-time performance of the system.

[0003] The coaxial optical radar integrated system has significant advantages in satellite-ground integrated control. According to the literature 1 (Hoge, F. (1974). Integrated Laser / Radar Satellite Ranging and Tracking System. Applied Optics , 13(12), 2356-2362.), The integrated laser and radar system can significantly improve the tracking and ranging accuracy of satellites through multi-function multiplexing, while reducing system costs and improving control reliability. This feature is of great significance for realizing efficient satellite-ground integrated control, enabling ground stations to operate flexibly in multiple mission modes, effectively expanding the application scope of the system, especially showing advantages in high-precision measurement and rapid response.

[0004] At present, satellite-ground integrated control technology has made significant progress in satellite orbit maintenance, attitude adjustment and data transmission. Traditional satellite-ground integrated control mostly relies on ground telemetry and control (TT&C) systems, which realize orbit and attitude control of satellites through measurement, calculation and command feedback from ground stations. Such systems usually share measurement and command data among multiple ground stations to ensure the continuity and accuracy of control. In addition, existing ground control technologies include advanced methods such as software-defined networks (SDN) and automated task scheduling, but such methods still have certain bottlenecks in hardware resource integration, response speed and data processing, and it is difficult to fully meet the needs of multi-satellite collaborative operations.

[0005] In recent years, the satellite-ground integrated control architecture based on cloud platform has gradually attracted attention. Reference 2 (Xia, Z., Li, W., & Zhang, H. (2023). Research on Satellite Operation Control System Based on Cloud Platform. Journal of Space Science&Technology , 24(5), 1234-1246.) proposed a cloud computing-based satellite operation control system architecture, which integrates the hardware and software resources of the ground control system to achieve "one station, multiple satellites" parallel control and automated operation to enhance the flexibility and real-time performance of the system. This architecture can achieve dynamic resource allocation through cloud computing, effectively improving the scalability and automation level of the system, and is particularly suitable for mission management and coordinated operation of multiple satellites.

[0006] Although the cloud-based satellite-ground integrated control system has significantly improved automation and flexibility, the technology still has some shortcomings. For example, the cloud platform architecture may affect the control accuracy due to delays when facing real-time and high-frequency data transmission. In addition, although the multi-tasking processing capability of cloud computing can improve the overall response efficiency of the system, it is difficult to fully meet the requirements of high-speed computing and high-precision control when dealing with tasks involving high-precision orbit and attitude adjustment.

[0007] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0008] In order to overcome the above-mentioned defects of the prior art, the present invention provides a coaxial optical radar payload satellite-ground integrated operation and control method, which optimizes the satellite orbit and attitude control by combining Kalman filtering with a deep reinforcement learning model; uses time-space correlation factors to improve the state transfer matrix, thereby improving the relative state estimation accuracy between the satellite and the ground station, and solving the problems of poor real-time performance and high-precision control in the existing satellite-ground integrated control.

[0009] To achieve the above object, the present invention provides the following technical solutions: A satellite-ground integrated operation and control method for a coaxial optical radar payload satellite comprises the following steps: Step S1, collecting the position parameters, attitude parameters, relative velocity parameters and acceleration parameters of the coaxial optical radar payload satellite, and performing noise reduction processing, and outputting the noise-reduced data as input for subsequent state estimation; Step S2, using Kalman filtering to analyze the denoised data, predicting the satellite state at the current moment based on the satellite state at the previous moment, and outputting a state estimate; Step S3, generating a control instruction based on the state estimation value, decomposing the control instruction, and generating control instructions for each dimension one by one according to the decomposed state data; Step S4, checking the generated control instructions for each dimension; Step S5, inputting the state estimation value and the control instruction execution data of each dimension into the state space of the trained deep reinforcement learning model to optimize the control instruction, wherein the reward function of the deep reinforcement learning model is: ; In the formula, is the reward value, is the total number of state parameters, For the The value of the target state parameter, For the The value of the actual execution status parameter, is the total number of control instructions, is the adjustment amount of the jth control instruction, For the The execution frequency of the control instructions, and is the weight coefficient.

[0010] As a further solution of the present invention, in combination with the spatiotemporal variation law of the relative states of the satellite and the ground station, a spatiotemporal correlation factor is introduced into the state transfer process of the Kalman filter to improve the update rule of the state transfer matrix, including the following steps: Step A1, extracting historical status data of relative position parameters, attitude parameters, relative velocity parameters and acceleration parameters between the satellite and the ground station; Step A2, performing differential analysis on the historical state data by a time series analysis method, and calculating the state change rate between adjacent time points, the formula is: Step A3, based on the change rate of the historical state, a linear regression model is constructed using a regression analysis method, and the historical data is fitted using the least squares method to solve the influence coefficient of the change rate of the historical state and the influence coefficient of the time interval; Step A4, defining a spatiotemporal correlation factor to reflect the spatiotemporal relationship of relative motion between the satellite and the ground station. The calculation formula of the spatiotemporal correlation factor is: ; In the formula, is the spatiotemporal correlation factor, For the moment The rate of state change, time The rate of state change, is the influence coefficient of the time interval, From the moment To time time difference; Step A5: In each time step, the state transfer matrix is ​​adjusted according to the currently calculated spatiotemporal correlation factor. The adjustment formula is: ; In the formula, is the state transfer matrix, is the traditional state transfer matrix, is the state transfer matrix at the previous moment, is the space-time correlation factor.

[0011] As a further scheme of the present invention, in step S5, the state estimation parameters after the state estimation value is decomposed and the actual state parameters after the control instruction of each dimension is executed are combined into complete state information, and the state information is input as input data into the state space of the trained deep reinforcement learning model to generate optimized position control instructions, posture control instructions, speed control instructions and acceleration control instructions; the state estimation parameters include estimated position parameters, estimated posture parameters, estimated speed parameters and estimated acceleration parameters, and the actual state parameters include actual position parameters, actual posture parameters, actual speed parameters and actual acceleration parameters; the position control instructions include position adjustment amount, adjustment direction and execution time information; the posture control instructions include angle adjustment amount, rotation axis and execution time information; the speed control instructions include speed adjustment amount, adjustment direction and execution time information; the acceleration control instructions include acceleration adjustment amount, adjustment direction and execution time information.

[0012] As a further solution of the present invention, training the deep reinforcement learning model includes the following steps: Step B1, constructing a state space of a deep reinforcement learning model, inputting the state estimation parameters after decomposing the state estimation value and the actual state parameters after executing the control instruction of each dimension into the state space of the deep reinforcement learning model; Step B2, defining the action space of the deep reinforcement learning model, and refining the control instructions into specific actions of attitude adjustment, track position adjustment, speed adjustment and acceleration adjustment as elements of the action space; Step B3, design a reward mechanism, set a reward function based on the control error and energy saving indicators, with reducing the control error as the main goal and saving energy as the secondary goal, and balance the relationship between accuracy and energy consumption; Step B4, using historical data and control instruction execution data to train the deep reinforcement learning model.

[0013] As a further solution of the present invention, the position parameter is the relative position between the satellite and the ground station, which is obtained by real-time acquisition of the relative position of the satellite and the ground station in the geographic coordinate system by a coaxial optical radar, and is expressed in Cartesian coordinates; the attitude parameter is the attitude angle of the satellite, including the satellite's pitch angle, roll angle and yaw angle, expressed in Euler angles; the relative velocity parameter is the relative velocity between the satellite and the ground station, including the relative horizontal velocity, the relative vertical velocity and the relative normal velocity; the acceleration parameter is the acceleration of the satellite in its orbit, including the horizontal acceleration component, the vertical acceleration component and the normal acceleration component.

[0014] As a further solution of the present invention, step S3 includes the following steps: Step S3.1, decomposing the state estimation value output by the Kalman filter into estimated position parameters, estimated attitude parameters, estimated relative velocity parameters and estimated acceleration parameters; Step S3.2, compare the current estimated position parameters with the target position parameters, calculate the required displacement adjustment amount, and decompose the total displacement adjustment amount according to the set fixed increment, calculate the required adjustment steps and the amount of each adjustment, and generate the posture. The position control instruction includes the position adjustment amount, adjustment direction and execution time information; Step S3.3, compare the current estimated attitude angle with the target attitude angle, calculate the required attitude angle adjustment, decompose the total attitude angle adjustment according to the set fixed increment, calculate the required adjustment steps and the angle increment for each adjustment, and generate an attitude control instruction, which includes the angle amount, the rotation axis and the execution time information; Step S3.4, compare the current estimated speed with the target speed, calculate the required speed adjustment, decompose the total speed adjustment according to the set fixed increment, calculate the required number of adjustment steps and the increment of each adjustment, and generate a speed control instruction, which includes the speed adjustment amount, adjustment direction and execution time information; Step S3.5, compare the current estimated acceleration with the target acceleration, calculate the required acceleration adjustment amount, decompose the total acceleration adjustment amount according to the set fixed increment, calculate the required number of adjustment steps and the increment of each adjustment, and generate an acceleration control instruction, which includes the acceleration adjustment amount, adjustment direction and execution time information; Step S3.6, schedule the decomposed control instructions according to the time series, and use the generated segmented control instruction set as the final control instruction.

[0015] As a further solution of the present invention, step S4 includes the following steps: Step S4.1, after executing the position control instruction, the current position parameters of the satellite are collected again through the coaxial optical radar, and the actual position after execution is compared with the target position, and the position deviation is calculated. When the position deviation is less than the preset error range, the calibration is passed; when the position deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.2, after executing the attitude control command, the current attitude parameters of the satellite are collected again through the coaxial optical radar, and the actual attitude after execution is compared with the target attitude, and the attitude angle deviation is calculated. When the attitude angle deviation is less than the preset error range, the verification is passed; when the attitude angle deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.3, after executing the speed control instruction, the current speed parameters of the satellite are collected again through the coaxial optical radar, and the actual speed after execution is compared with the target speed, and the speed deviation is calculated. When the speed deviation is less than the preset error range, the calibration is passed; when the speed deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.4, after executing the acceleration control instruction, the current acceleration parameters of the satellite are collected again through the coaxial optical radar, and the actual acceleration after execution is compared with the target acceleration, and the acceleration deviation is calculated. When the acceleration deviation is less than the preset error range, the calibration passes; when the acceleration deviation is greater than or equal to the preset error range, the error is recorded and marked as needing readjustment.

[0016] Compared with the prior art, the beneficial effects of the coaxial optical radar payload satellite-ground integrated operation and control method of the present invention are: The present invention realizes high-precision state estimation and control optimization between satellites and ground stations by combining Kalman filtering with deep reinforcement learning models. Traditional satellite-ground integrated control systems usually rely on ground telemetry and control (TT&C) systems, and have high requirements for real-time and high-precision control. However, the existing technology often faces delays when processing high-frequency data, which affects the control accuracy. In contrast, the present invention not only improves the accuracy of satellite control by introducing technical improvements of deep reinforcement learning and spatiotemporal correlation factors, but also solves the problems of real-time response and insufficient accuracy in the existing technology by dynamically optimizing control instructions.

[0017] The present invention improves the energy efficiency of the system by optimizing the control instruction generation process. Although Document 2 has made great improvements in task management and automated operation, it still faces the problem of high-precision orbit and attitude adjustment, and the delay of the cloud platform may cause the control accuracy to decrease. In contrast, the present invention balances the relationship between control accuracy and energy efficiency through the reward mechanism of the deep reinforcement learning model, reduces energy consumption, and improves the response speed and efficiency of satellite control. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 Architectural drawing of the AN / FPQ-6 C-band radar system and its supporting electronics.

[0019] Figure 2 A simplified schematic diagram of a laser receiver.

[0020] Figure 3 This is a block diagram of the electronic circuit of the laser / radar ranging system.

[0021] Figure 4 This is a schematic diagram of the cloud platform composition of the operation and control system in document 2.

[0022] Figure 5 The present invention is a schematic flow chart of a method for integrated satellite-ground operation and control of a coaxial optical radar payload satellite.

[0023] In the figure, ANODE OUTPUT: anode output; HV: high voltage; DAYLIGHT GATE: daylight gate; PM TUBE: photomultiplier tube; FOCAL PLANE: focal plane; SPATIAL FILTER: spatial filter; TELESCOPE BODY: telescope body; RADAR TV CAMERA: radar TV camera; ADDITIONAL COMPONENTS TO FORM LASERRECEIVER: additional components to form laser receiver; SPECTRAL FILTER: spectral filter; PRIMARY OPTICALSYSTEM: primary optical system; IN / OUT ACQUISITION AID SYSTEM FOR f / 5: input / output auxiliary acquisition system for f / 5; PRIMARY FOCAL PLANE f / 11 OR f / 5: primary focal plane f / 11 or f / 5; LASER TRANSMITORDER (TO LASER): laser transmission command (sent to the laser); RANGE RATE (FROM DRAM): range rate (from DRAM); SITE TIME OF DAY GENERATOR: Station Time Generator; COUNTER: Counter; STOREREGISTER: Storage Register; RANGE DESIGNATION FROM 4101: Range Designation from 4101; SHIFTREGISTER: Shift Register; RANGE COUNTER FWD / BKD BCD: Forward / Reverse BCD Range Counter; A / PCONV.: Analog / Pulse Converter; D / A CONV.: Digital-to-Analog Converter; 2ND. 1ST. INTEGRATOR: Second Order First Order Integrator; LASER TRANSMIT PULSE: Laser Transmit Pulse; LASER RETURN PULSE: Laser Return Pulse; HP5360ACOMPUTING COUNTER: HP5360A Computing Counter; TRANSFER GATES: Transmission Gate; GATE REGISTER: Gate Register; RANGE GATE GENERATOR BKW / FWD BCD COUNTER: Range Gate Generator for Forward / Reverse BCD Counter; TIME INTERVAL DATA TO 4101 COMPUTER: time interval data to 4101 computer; VALIDITYTEST: validity test; DATA VALID: data is valid; DECIMAL / BCD SWITCHES: decimal / BCD switch; Planning and scheduling: planning and scheduling;Simulation management: simulation management; Status monitoring: status monitoring; Telemetry and telecontrol management: telemetry and telecontrol management; Cloud platform management: cloud platform management; Operation control application visualization / client: operation control application visualization / client; RestFul: RestFul; BlazeDS: BlazeDS; WebSocket: WebSocket; XMPP: XMPP; WebService: WebService; Gateway service: gateway service; Telemetry and telecontrol: telemetry and telecontrol; Orbit calculation and prediction: orbit calculation and prediction; Monitoring and alerting: monitoring and alerting; Docker: Docker; Task planning and scheduling: task planning and scheduling; Control calculation: control calculation; Telemetry and telecontrol simulation: telemetry and telecontrol simulation; Cluster deployment: cluster deployment; Resource management: resource management; Cluster monitoring: cluster monitoring; Task scheduling: task scheduling; Load control: load control; Docker+Kubernetes: Docker+Kubernetes; server: server; storage: storage; switch: switch; Relational Database: relational database; Simulation model Database: simulation model database. ; DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] Example 1 A satellite-ground integrated operation and control method for a coaxial optical radar payload satellite comprises the following steps: Step S1, collecting the position parameters, attitude parameters, relative velocity parameters and acceleration parameters of the coaxial optical radar payload satellite, and performing noise reduction processing, and outputting the noise-reduced data as input for subsequent state estimation; Step S2, using Kalman filtering to analyze the denoised data, predicting the satellite state at the current moment based on the satellite state at the previous moment, and outputting a state estimate; Step S3, generating a control instruction based on the state estimation value, decomposing the control instruction, and generating control instructions for each dimension one by one according to the decomposed state data; Step S4, checking the generated control instructions for each dimension; Step S5: input the state estimation value and the control instruction execution data of each dimension into the state space of the trained deep reinforcement learning model to optimize the control instruction.

[0026] The position parameter in the embodiment of the present invention is the relative position between the satellite and the ground station, which is obtained by real-time acquisition of the relative position of the satellite and the ground station in the geographic coordinate system by a coaxial optical radar, and is expressed in Cartesian coordinates; the attitude parameter is the attitude angle of the satellite, including the pitch angle, roll angle and yaw angle of the satellite, expressed in Euler angles; the relative velocity parameter is the relative velocity between the satellite and the ground station, including the relative horizontal velocity, the relative vertical velocity and the relative normal velocity; the acceleration parameter is the acceleration of the satellite in its orbit, including the horizontal acceleration component, the vertical acceleration component and the normal acceleration component.

[0027] like Figure 2 The figure shows a simplified schematic diagram of a laser receiver, which is formed by using the auxiliary focal plane port of the existing AN / FPQ-6 radar aiming telescope (f / 11); the optical system from left to right in the figure includes the main optical system and the spectral filter, followed by optical elements such as the spatial filter, which is used to optimize the spatial selectivity of the received signal; at the rear end of the telescope body, the radar TV camera is used to assist in imaging, and the day baffle and the photomultiplier tube work together to ensure that the laser signal can be effectively detected both during the day and at night; high voltage (HV) provides power support for the photomultiplier tube; through these components, the laser receiver can accurately capture and process signals from the laser source for precise ranging and tracking of long-range targets.

[0028] In the embodiment of the present invention, the temporal and spatial correlation factors are introduced into the state transfer process of the Kalman filter in combination with the temporal and spatial variation law of the relative states of the satellite and the ground station, and the updating rule of the state transfer matrix is ​​improved, which includes the following steps: Step A1, extracting historical status data of relative position parameters, attitude parameters, relative velocity parameters and acceleration parameters between the satellite and the ground station; Step A2, performing differential analysis on the historical state data by a time series analysis method, and calculating the state change rate between adjacent time points, the formula is: Step A3, based on the change rate of the historical state, a linear regression model is constructed using a regression analysis method, and the historical data is fitted using the least squares method to solve the influence coefficient of the change rate of the historical state and the influence coefficient of the time interval; Step A4, defining a spatiotemporal correlation factor to reflect the spatiotemporal relationship of the relative motion between the satellite and the ground station. The calculation formula of the spatiotemporal correlation factor is: ; In the formula, is the spatiotemporal correlation factor, For the moment The rate of state change, time The rate of state change, is the influence coefficient of the time interval, From the moment To time time difference; Step A5: In each time step, the state transfer matrix is ​​adjusted according to the currently calculated spatiotemporal correlation factor. The adjustment formula is: ; In the formula, is the state transfer matrix, is the traditional state transfer matrix, is the state transfer matrix at the previous moment, is the space-time correlation factor.

[0029] Step S2 in the embodiment of the present invention includes the following steps: Step S2.1, construct a state transfer equation to predict the state based on the satellite's position parameters, attitude parameters, relative velocity parameters, and acceleration parameters at the previous moment, and calculate the satellite state at the current moment; the state transfer equation is: ; In the formula, is the predicted state vector at the current moment, is the state vector at the previous moment, is the control input, is the state transfer matrix, is the control input matrix; Step S2.2, the confidence in the satellite state prediction is expressed by calculating the prediction error covariance matrix, and the prediction error covariance matrix is ​​calculated as: ; In the formula, is the forecast error covariance matrix, is the process noise covariance matrix, is the state transfer matrix, is the transpose of the state transfer matrix, is the error covariance matrix of the previous moment; this formula shows that when predicting the state, the error covariance matrix will change with the state transfer. At the same time, considering the uncertainty of system modeling, the noise introduced by the process noise covariance matrix will also affect the error covariance matrix; Step S2.3, based on the relative measurement data between the satellite and the ground station, establish an observation equation to combine the measurement data with the prediction data. The observation equation is: ; In the formula, It is the real-time observation data collected by coaxial optical radar. is the observation matrix, is the observation noise, is the predicted state vector at the current moment; Step S2.4, by weighing the weights of the prediction error and the observation error, the Kalman gain is calculated and the estimated value of the satellite state is adjusted. The formula for calculating the Kalman gain is: ; In the formula, is the Kalman gain matrix, is the forecast error covariance matrix, is the transpose of the observation matrix, is the observation noise covariance, is the observation matrix; Step S2.5, according to the calculated Kalman gain, by adjusting the predicted value, an updated state estimate is obtained, and the state update formula is: ; In the formula, is the updated current moment prediction state vector, is the predicted state vector at the current moment, is the Kalman gain matrix, is the observed data vector, is the observation matrix, is the theoretical observation value calculated based on the predicted state Step S2.6, after the state is updated, the uncertainty of the system's state estimation is updated by adjusting the error covariance matrix. The updated error covariance matrix represents the system's confidence in the satellite state estimation. The update formula is: ; In the formula, is the updated error covariance matrix, is the identity matrix, is the Kalman gain matrix, is the observation matrix, is the prediction error covariance matrix.

[0030] Step S3 in the embodiment of the present invention includes the following steps: Step S3.1, decomposing the state estimation value output by the Kalman filter into estimated position parameters, estimated attitude parameters, estimated relative velocity parameters and estimated acceleration parameters; Step S3.2, compare the current estimated position parameters with the target position parameters, calculate the required displacement adjustment amount, and decompose the total displacement adjustment amount according to the set fixed increment, calculate the required adjustment steps and the amount of each adjustment, and generate the posture. The position control instruction includes the position adjustment amount, adjustment direction and execution time information; Step S3.3, compare the current estimated attitude angle with the target attitude angle, calculate the required attitude angle adjustment, decompose the total attitude angle adjustment according to the set fixed increment, calculate the required adjustment steps and the angle increment for each adjustment, and generate an attitude control instruction, which includes the angle adjustment amount, the rotation axis and the execution time information; Step S3.4, compare the current estimated speed with the target speed, calculate the required speed adjustment, decompose the total speed adjustment according to the set fixed increment, calculate the required number of adjustment steps and the increment of each adjustment, and generate a speed control instruction, which includes the speed adjustment amount, adjustment direction and execution time information; Step S3.5, compare the current estimated acceleration with the target acceleration, calculate the required acceleration adjustment amount, decompose the total acceleration adjustment amount according to the set fixed increment, calculate the required number of adjustment steps and the increment of each adjustment, and generate an acceleration control instruction, which includes the acceleration adjustment amount, adjustment direction and execution time information; Step S3.6, schedule the decomposed control instructions according to the time sequence, and use the generated segmented control instruction set as the final control instruction, ready for subsequent multi-layer verification and sending for execution.

[0031] The adjustment direction of the position control instruction in the embodiment of the present invention is determined according to the positive or negative of the displacement adjustment amount, and the execution time is allocated in sequence.

[0032] The rotation axes of the attitude control instructions in the embodiment of the present invention are the pitch axis, the roll axis, and the yaw axis, and the execution time is allocated in sequence.

[0033] The adjustment direction of the speed control instruction in the embodiment of the present invention is determined according to the positive or negative of the speed adjustment amount, and the execution time is allocated in sequence.

[0034] The adjustment direction of the acceleration control instruction in the embodiment of the present invention is determined according to the positive or negative acceleration adjustment amount, and the execution time is allocated in sequence.

[0035] Step S4 in the embodiment of the present invention includes the following steps: Step S4.1, after executing the position control instruction, the current position parameters of the satellite are collected again through the coaxial optical radar, and the actual position after execution is compared with the target position, and the position deviation is calculated. When the position deviation is less than the preset error range, the calibration is passed; when the position deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.2, after executing the attitude control command, the current attitude parameters of the satellite are collected again through the coaxial optical radar, and the actual attitude after execution is compared with the target attitude, and the attitude angle deviation is calculated. When the attitude angle deviation is less than the preset error range, the verification is passed; when the attitude angle deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.3, after executing the speed control instruction, the current speed parameters of the satellite are collected again through the coaxial optical radar, and the actual speed after execution is compared with the target speed, and the speed deviation is calculated. When the speed deviation is less than the preset error range, the calibration is passed; when the speed deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.4, after executing the acceleration control instruction, the current acceleration parameters of the satellite are collected again through the coaxial optical radar, and the actual acceleration after execution is compared with the target acceleration, and the acceleration deviation is calculated. When the acceleration deviation is less than the preset error range, the calibration passes; when the acceleration deviation is greater than or equal to the preset error range, the error is recorded and marked as needing readjustment.

[0036] In step S5 of the embodiment of the present invention, the state estimation parameters after the state estimation value is decomposed and the actual state parameters after the control instruction of each dimension is executed are combined into complete state information, and the state information is input as input data into the state space of the trained deep reinforcement learning model to generate optimized position control instructions, posture control instructions, speed control instructions and acceleration control instructions; the state estimation parameters include estimated position parameters, estimated posture parameters, estimated speed parameters and estimated acceleration parameters, and the actual state parameters include actual position parameters, actual posture parameters, actual speed parameters and actual acceleration parameters; the position control instruction includes position adjustment amount, adjustment direction and execution time information; the posture control instruction includes angle adjustment amount, rotation axis and execution time information; the speed control instruction includes speed adjustment amount, adjustment direction and execution time information; the acceleration control instruction includes acceleration adjustment amount, adjustment direction and execution time information.

[0037] The training of the deep reinforcement learning model in the embodiment of the present invention includes the following steps: Step B1, constructing a state space of a deep reinforcement learning model, inputting the state estimation parameters after decomposing the state estimation value and the actual state parameters after executing the control instruction of each dimension into the state space of the deep reinforcement learning model; Step B2, defining the action space of the deep reinforcement learning model, and refining the control instructions into specific actions of attitude adjustment, track position adjustment, speed adjustment and acceleration adjustment as elements of the action space; Step B3, designing a reward mechanism, setting a reward function according to the control error and energy saving index, taking reducing the control error as the main goal and saving energy consumption as the secondary goal, balancing the relationship between accuracy and energy consumption, wherein the reward function of the deep reinforcement learning model is: ; In the formula, is the reward value, is the total number of state parameters, For the The value of the target state parameter, For the The value of the actual execution status parameter, is the total number of control instructions, is the adjustment amount of the jth control instruction, For the The execution frequency of the control instructions, and is the weight coefficient; Step B4, using historical data and control instruction execution data to train the deep reinforcement learning model.

[0038] Example 2 In the process of training the deep reinforcement learning model, the state space of the model is first constructed, and the decomposed state estimation parameters and the actual state parameters after the execution of each dimensional control instruction are used as input to fill the state space of the deep reinforcement learning model. Next, the action space of the model is defined, and the control instructions are refined into specific actions of attitude adjustment, orbit position adjustment, speed adjustment and acceleration adjustment as the basic elements of the action space. Subsequently, a reward mechanism is designed to reduce the control error by setting the reward function, while taking into account the energy saving index, so as to balance the relationship between control accuracy and energy consumption. Finally, the deep reinforcement learning model is trained through historical data and control instruction execution data to improve its adaptability and control effect in different situations.

[0039] In the embodiment of the present invention, a deep reinforcement learning algorithm is used to train the model. During the training process, the model learns how to make optimal control decisions in different states by interacting with the virtual environment. In each training iteration, the model selects an action according to the current state, and then obtains the next state and the corresponding reward value through the result of simulated control. Specifically, the training process includes the following steps: Step C1, setting the initial parameters of the deep reinforcement learning model, including learning rate, discount factor, exploration rate, etc.; Step C2, the model interacts with the virtual environment in each training iteration and selects an action from the action space based on the current state; the action selection is implemented through a greedy strategy to ensure that the model can both explore the new state space and utilize the known optimal strategy; Step C3, executing the selected action in the virtual environment and recording the feedback of the environment, including the next state and reward value; Step C4, storing the current state, the selected action, the reward obtained, and the next state into the experience playback buffer; Step C5, randomly extract training samples from the experience replay buffer and perform batch update; experience replay helps to break the time correlation between samples and improve the stability of training; Step C6, using the extracted samples to update the policy network of the model, and improving the model's policy by minimizing the time difference error; Step C7, periodically copy the weights of the policy network to the target network to stabilize the training process; Step C8: As training progresses, the exploration rate is gradually reduced to reduce the selection of random actions, thereby better utilizing the learned strategy. Step C9: After each training iteration, evaluate the performance of the model. If the reward value continues to fail to meet expectations, further adjust the model parameters.

[0040] The following is a Python code example that trains a deep reinforcement learning model to output control instructions. Please note that this example is only a starting point and may need to be adjusted according to actual conditions and device interfaces in actual applications. ```Python import numpy as np import tensorflow as tf from tensorflow.keras import layers import random class DDPGAgent: def __init__(self, state_size, action_size): self.state_size = state_size self.action_size = action_size self.memory = [] self.gamma = 0.99 self.epsilon = 1.0 self.epsilon_min = 0.01 self.epsilon_decay = 0.995 self.learning_rate = 0.001 self.batch_size = 64 self.actor_model = self._build_actor_model() self.critic_model = self._build_critic_model() self.target_actor_model = self._build_actor_model() self.target_critic_model = self._build_critic_model() self.update_target_models() def _build_actor_model(self): model = tf.keras.Sequential([ layers.Dense(400, input_dim=self.state_size, activation='relu'), layers.Dense(300, activation='relu'), layers.Dense(self.action_size, activation='tanh') ]) model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=self.learning_rate)) return model def _build_critic_model(self): state_input = layers.Input(shape=(self.state_size,)) action_input = layers.Input(shape=(self.action_size,)) concat = layers.Concatenate()([layers.Dense(32, activation='relu')(state_input), layers.Dense(32, activation='relu')(action_input)]) out = layers.Dense(256, activation='relu')(concat) out = layers.Dense(1)(out) model = tf.keras.Model([state_input, action_input], out) model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=self.learning_rate), loss='mse') return model def update_target_models(self): self.target_actor_model.set_weights(self.actor_model.get_weights()) self.target_critic_model.set_weights(self.critic_model.get_weights()) def remember(self, state, action, reward, next_state, done): self.memory.append((state, action, reward, next_state, done)) def act(self, state): if np.random.rand()<= self.epsilon: return np.random.uniform(-1, 1, self.action_size) return self.actor_model.predict(state)[0] def replay(self): if len(self.memory)<self.batch_size: return minibatch = random.sample(self.memory, self.batch_size) for state, action, reward, next_state, done in minibatch: target = reward + self.gamma * (1 - done) * self.target_critic_model.predict([next_state, self.target_actor_model.predict(next_state)])[0] self.critic_model.train_on_batch([state, action], target) with tf.GradientTape() as tape: predicted_action = self.actor_model(state) actor_loss = -tf.reduce_mean(self.critic_model([state, predicted_action])) actor_grads = tape.gradient(actor_loss, self.actor_model.trainable_variables) tf.keras.optimizers.Adam(learning_rate=self.learning_rate).apply_gradients(zip(actor_grads, self.actor_model.trainable_variables)) if self.epsilon>self.epsilon_min: self.epsilon *= self.epsilon_decay # Example usage state_size = 4 action_size = 4 agent = DDPGAgent(state_size, action_size) # Training loop example for e in range(1000): state = np.reshape(np.random.rand(state_size), [1, state_size]) for time in range(500): action = agent.act(state) next_state = np.reshape(np.random.rand(state_size), [1, state_size]) reward = random.random() done = time == 499 agent.remember(state, action, reward, next_state, done) state = next_state if done: agent.update_target_models() print(f"Episode {e+1} / {1000} - Time: {time}, Epsilon:{agent.epsilon:.2}") break agent.replay() This code is only used as an example. In actual applications, appropriate modifications and adjustments need to be made according to specific circumstances.

[0041] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0042] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A satellite-ground integrated operation and control method for a coaxial optical radar payload satellite, characterized in that: The following steps are involved: Step S1, collecting the position parameters, attitude parameters, relative velocity parameters and acceleration parameters of the coaxial optical radar payload satellite, performing noise reduction processing, and outputting the noise-reduced data; Step S2, using Kalman filtering to analyze the denoised data, predicting the satellite state at the current moment based on the satellite state at the previous moment, and outputting a state estimate; Step S3, generating a control instruction based on the state estimation value, decomposing the control instruction, and generating control instructions for each dimension one by one according to the decomposed state data; Step S4, checking the generated control instructions for each dimension; Step S5, inputting the state estimation value and the control instruction execution data of each dimension into the state space of the trained deep reinforcement learning model to optimize the control instruction, wherein the reward function of the deep reinforcement learning model is: ； In the formula, is the reward value, is the total number of state parameters, For the The value of the target state parameter, For the The value of the actual execution status parameter, is the total number of control instructions, is the adjustment amount of the jth control instruction, For the The execution frequency of the control instructions, and is the weight coefficient.

2. The method for integrated satellite-ground operation and control of a coaxial optical radar payload satellite according to claim 1, characterized in that: Combined with the spatiotemporal variation law of the relative state between the satellite and the ground station, the spatiotemporal correlation factor is introduced into the state transfer process of the Kalman filter to improve the update rule of the state transfer matrix, including the following steps: Step A1, extracting historical status data of relative position parameters, attitude parameters, relative velocity parameters and acceleration parameters between the satellite and the ground station; Step A2, performing differential analysis on the historical state data by a time series analysis method, and calculating the state change rate between adjacent time points, the formula is: Step A3, based on the change rate of the historical state, a linear regression model is constructed using a regression analysis method, and the historical data is fitted using the least squares method to solve the influence coefficient of the change rate of the historical state and the influence coefficient of the time interval; Step A4, defining a spatiotemporal correlation factor to reflect the spatiotemporal relationship of the relative motion between the satellite and the ground station. The calculation formula of the spatiotemporal correlation factor is: ； In the formula, is the spatiotemporal correlation factor, For the moment The rate of state change, time The rate of state change, is the influence coefficient of the time interval, From the moment To time time difference; Step A5: In each time step, the state transfer matrix is ​​adjusted according to the currently calculated spatiotemporal correlation factor. The adjustment formula is: ； In the formula, is the state transfer matrix, is the traditional state transfer matrix, is the state transfer matrix at the previous moment, is the space-time correlation factor.

3. A satellite-to-ground integrated operation and control method for a coaxial optical radar payload according to claim 1, characterized in that: In step S5, the state estimation parameters after the state estimation value is decomposed and the actual state parameters after the control instruction of each dimension is executed are combined into complete state information, and the state information is input as input data into the state space of the trained deep reinforcement learning model to generate optimized position control instructions, posture control instructions, speed control instructions and acceleration control instructions; the state estimation parameters include estimated position parameters, estimated posture parameters, estimated speed parameters and estimated acceleration parameters, and the actual state parameters include actual position parameters, actual posture parameters, actual speed parameters and actual acceleration parameters; the position control instruction includes position adjustment amount, adjustment direction and execution time information; the posture control instruction includes angle adjustment amount, rotation axis and execution time information; the speed control instruction includes speed adjustment amount, adjustment direction and execution time information; the acceleration control instruction includes acceleration adjustment amount, adjustment direction and execution time information.

4. A satellite-ground integrated operation and control method for a coaxial optical radar payload according to claim 1 or 3, characterized in that: Training a deep reinforcement learning model consists of the following steps: Step B1, constructing a state space of a deep reinforcement learning model, inputting the state estimation parameters after decomposing the state estimation value and the actual state parameters after executing the control instruction of each dimension into the state space of the deep reinforcement learning model; Step B2, defining the action space of the deep reinforcement learning model, and refining the control instructions into specific actions of attitude adjustment, track position adjustment, speed adjustment and acceleration adjustment as elements of the action space; Step B3, design a reward mechanism, set a reward function based on the control error and energy saving indicators, with reducing the control error as the main goal and saving energy as the secondary goal, and balance the relationship between accuracy and energy consumption; Step B4, using historical data and control instruction execution data to train the deep reinforcement learning model.

5. The method for integrated satellite-ground operation and control of a coaxial optical radar payload satellite according to claim 1, characterized in that: The position parameter is the relative position between the satellite and the ground station, which is obtained by real-time acquisition of the relative position of the satellite and the ground station in the geographic coordinate system by coaxial optical radar, and is expressed in Cartesian coordinates; the attitude parameter is the attitude angle of the satellite, including the satellite's pitch angle, roll angle and yaw angle, expressed in Euler angles; the relative velocity parameter is the relative velocity between the satellite and the ground station, including the relative horizontal velocity, relative vertical velocity and relative normal velocity; the acceleration parameter is the acceleration of the satellite in its orbit, including the horizontal acceleration component, the vertical acceleration component and the normal acceleration component.

6. A satellite-to-ground integrated operation and control method for a coaxial optical radar payload according to claim 1, characterized in that: Step S3 includes the following steps: Step S3.1, decomposing the state estimation value output by the Kalman filter into estimated position parameters, estimated attitude parameters, estimated relative velocity parameters and estimated acceleration parameters; Step S3.2, comparing the current estimated position parameter with the target position parameter, calculating the required displacement adjustment amount, and decomposing the total displacement adjustment amount according to the set fixed increment, calculating the required number of adjustment steps and the amount of each adjustment, and generating a position control instruction, wherein the position control instruction includes the position adjustment amount, adjustment direction and execution time information; Step S3.3, compare the current estimated attitude angle with the target attitude angle, calculate the required attitude angle adjustment, decompose the total attitude angle adjustment according to the set fixed increment, calculate the required adjustment steps and the angle increment for each adjustment, and generate an attitude control instruction, wherein the attitude control instruction includes the angle adjustment amount, the rotation axis and the execution time information; Step S3.4, compare the current estimated speed with the target speed, calculate the required speed adjustment, decompose the total speed adjustment according to the set fixed increment, calculate the required number of adjustment steps and the increment of each adjustment, and generate a speed control instruction, which includes the speed adjustment amount, adjustment direction and execution time information; Step S3.5, compare the current estimated acceleration with the target acceleration, calculate the required acceleration adjustment amount, decompose the total acceleration adjustment amount according to the set fixed increment, calculate the required number of adjustment steps and the increment of each adjustment, and generate an acceleration control instruction, which includes the acceleration adjustment amount, adjustment direction and execution time information; Step S3.6, schedule the decomposed control instructions according to the time series, and use the generated segmented control instruction set as the final control instruction.

7. A satellite-to-ground integrated operation and control method for a coaxial optical radar payload according to claim 1, characterized in that: Step S4 includes the following steps: Step S4.1, after executing the position control instruction, the current position parameters of the satellite are collected again through the coaxial optical radar, and the actual position after execution is compared with the target position, and the position deviation is calculated. When the position deviation is less than the preset error range, the calibration is passed; when the position deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.2, after executing the attitude control command, the current attitude parameters of the satellite are collected again through the coaxial optical radar, and the actual attitude after execution is compared with the target attitude, and the attitude angle deviation is calculated. When the attitude angle deviation is less than the preset error range, the verification is passed; when the attitude angle deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.3, after executing the speed control instruction, the current speed parameters of the satellite are collected again through the coaxial optical radar, and the actual speed after execution is compared with the target speed, and the speed deviation is calculated. When the speed deviation is less than the preset error range, the calibration is passed; when the speed deviation is greater than or equal to the preset error range, the error is recorded and marked as needing to be readjusted; Step S4.4, after executing the acceleration control instruction, the current acceleration parameters of the satellite are collected again through the coaxial optical radar, and the actual acceleration after execution is compared with the target acceleration, and the acceleration deviation is calculated. When the acceleration deviation is less than the preset error range, the calibration passes; when the acceleration deviation is greater than or equal to the preset error range, the error is recorded and marked as needing readjustment.

Images