Orbit control management system for short wave satellite

Abstract

The application provides a kind of short wave satellite orbit control management system, it is related to short wave satellite application technical field, including: data acquisition module, for collecting the position data of mobile ship, motion state data and environmental disturbance data, form dynamic tracking input set;Trajectory prediction module, for the dynamic tracking input set is input into pre-trained neural network model and is handled to obtain the prediction result data set including ship future trajectory and position offset;Attitude pointing module, for generating attitude adjustment instruction autonomously according to the prediction result data set, and calculating the initial pointing expected value of beam;Satellite-borne actuator adjusts satellite attitude in real time according to attitude adjustment instruction, to obtain the initial pointing state of beam and corresponding actual pointing deviation data.The application realizes accurate beam pointing and continuous stable energy transmission, and guarantees the auxiliary power supply stability of key equipment.

Description

Technical Field

[0001] This invention relates to the field of shortwave satellite application technology, and in particular to an orbit control and management system for shortwave satellites. Background Technology

[0002] When ocean-going cargo ships are sailing across the ocean, traditional shortwave satellite orbit control and management systems mostly rely on fixed-rule disturbance correction schemes for trajectory prediction of moving ships. They do not fully integrate real-time and complex multi-source environmental disturbance data, nor do they have the ability to deeply adapt to the dynamic changes in the ship's motion state. When cargo ships encounter the combined effects of sudden strong ocean currents and strong high-altitude winds, their actual sailing speed and heading will show non-uniform dynamic deviations. It is impossible to collect and integrate dynamic data such as the impact force of ocean currents and the yaw angle of the ship in real time, which are transmitted back by the cargo ship, resulting in a deviation between the predicted trajectory and the actual trajectory.

[0003] Furthermore, traditional shortwave satellite orbit control and management systems mostly use fixed-parameter filtering algorithms for beam pointing deviation compensation, which cannot be dynamically adjusted according to real-time changes in the ship's environment. This results in insufficient timeliness and accuracy of deviation compensation. When a cargo ship enters a magnetically disturbed region near the equator from a temperate sea area, changes in the Earth's magnetic field can increase the measurement error of the satellite attitude sensor, leading to sudden fluctuations in the actual beam pointing deviation. Compensation for beam pointing deviation will be delayed, and it may also cause instability in the auxiliary power supply of the cargo ship's navigation system. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an orbit control and management system for shortwave satellites to achieve continuous and stable microwave energy transmission to moving ships, thereby improving charging efficiency and navigation requirements.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, an orbit control and management system for shortwave satellites includes:

[0007] The data acquisition module is used to collect the position data, motion status data, and environmental disturbance data of the moving vessel to form a dynamic tracking input set;

[0008] The trajectory prediction module is used to input the dynamic tracking input set into a pre-trained neural network model for processing, so as to obtain a dataset of prediction results including the ship's future trajectory and position offset.

[0009] The attitude pointing module is used to generate attitude adjustment commands autonomously by the satellite based on the prediction result dataset and to calculate the initial expected value of the beam pointing. The onboard actuators adjust the satellite attitude in real time according to the attitude adjustment commands to obtain the initial pointing state of the beam and the corresponding actual pointing deviation data.

[0010] The deviation analysis module is used to divide the beam coverage area corresponding to the actual pointing deviation data into multiple spatial sampling units. Based on the statistical characteristics of the deviation data in multiple spatial sampling units, an adaptive filtering algorithm is used for dynamic evaluation and compensation to generate the primary adjustment amount of beam pointing.

[0011] The adjustment generation module is used to verify the primary adjustment, calculate the compensation residual, and perform a secondary correction on the primary adjustment based on the compensation residual, ultimately obtaining the adjusted beam pointing parameters.

[0012] The control execution module is used to drive the onboard beam transmitter based on the adjusted beam pointing parameters and provide real-time feedback on beam transmission performance indicators, forming a closed-loop onboard autonomous decision-making and control process to achieve stable energy transmission to moving ships.

[0013] In a second aspect, a computing device includes:

[0014] One or more processors;

[0015] A storage device for storing one or more programs that, when executed by one or more processors, enable the one or more processors to implement the system.

[0016] Thirdly, a computer-readable storage medium storing a program that, when executed by a processor, implements the system.

[0017] The above-described solution of the present invention has at least the following beneficial effects:

[0018] By integrating the position data, motion state data, and environmental disturbance data of moving vessels, a pre-trained neural network model is driven to achieve multi-step prediction of the vessel's future trajectory. This not only avoids control lag errors caused by vessel movement in advance but also effectively avoids the risk of energy transmission interruption due to the dynamic displacement of the vessel. The beam coverage area is divided into independent spatial sampling units, and key parameters such as process noise covariance and measurement noise covariance of adaptive Kalman filtering are dynamically configured based on the statistical mean and variance of the deviation within each unit. This can accurately filter out random deviations caused by environmental disturbances and improve the precision of beam pointing compensation.

[0019] By acquiring real-time energy reception efficiency data from beam transmission, the parameters of the neural network trajectory prediction model and the adaptive Kalman filter algorithm are corrected in reverse to construct a complete closed-loop control loop. This not only reduces the bandwidth consumption of ground communication links and decreases reliance on ground intervention, but also maintains control continuity in complex space environments, improving the system's reliability in emergency scenarios such as ship emergency communication and ocean rescue. To adapt to dynamic scene changes, environmental disturbance data can be updated in real time to ensure accurate matching between ship trajectory prediction results and actual motion states. Furthermore, the granularity of spatial sampling units is dynamically adjusted based on deviation distribution characteristics, and noise covariance parameters are optimized in real time to flexibly adapt to deviation changes under different sea states and weather conditions, thereby improving energy utilization efficiency. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating an orbit control and management system for shortwave satellites provided in an embodiment of the present invention.

[0021] Figure 2 This is a flowchart illustrating the process of dividing the beam coverage area corresponding to the actual pointing deviation data into multiple spatial sampling units, as provided in an embodiment of the present invention. Detailed Implementation

[0022] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0023] like Figure 1 As shown, an embodiment of the present invention proposes an orbit control and management system for shortwave satellites, comprising:

[0024] The data acquisition module is used to collect the position data, motion status data, and environmental disturbance data of the moving vessel to form a dynamic tracking input set;

[0025] The trajectory prediction module is used to input the dynamic tracking input set into a pre-trained neural network model for processing, so as to obtain a dataset of prediction results including the ship's future trajectory and position offset.

[0026] The attitude pointing module is used to generate attitude adjustment commands autonomously by the satellite based on the prediction result dataset and to calculate the initial expected value of the beam pointing. The onboard actuators adjust the satellite attitude in real time according to the attitude adjustment commands to obtain the initial pointing state of the beam and the corresponding actual pointing deviation data.

[0027] The deviation analysis module is used to divide the beam coverage area corresponding to the actual pointing deviation data into multiple spatial sampling units. Based on the statistical characteristics of the deviation data in multiple spatial sampling units, an adaptive filtering algorithm is used for dynamic evaluation and compensation to generate the primary adjustment amount of beam pointing.

[0028] The adjustment generation module is used to verify the primary adjustment, calculate the compensation residual, and perform a secondary correction on the primary adjustment based on the compensation residual, ultimately obtaining the adjusted beam pointing parameters.

[0029] The control execution module is used to drive the onboard beam transmitter based on the adjusted beam pointing parameters and provide real-time feedback on beam transmission performance indicators, forming a closed-loop onboard autonomous decision-making and control process to achieve stable energy transmission to moving ships.

[0030] In this embodiment of the invention, by integrating the position data, motion state data, and environmental disturbance data of the moving vessel, a pre-trained neural network model is driven to achieve multi-step prediction of the vessel's future trajectory. This not only avoids control lag errors caused by vessel movement in advance but also effectively avoids the risk of energy transmission interruption due to the vessel's dynamic displacement. The beam coverage area is divided into independent spatial sampling units, and based on the statistical mean and variance of the deviation within each unit, key parameters such as the process noise covariance and measurement noise covariance of the adaptive Kalman filter are dynamically configured. This can accurately filter out random deviations caused by environmental disturbances and improve the precision of beam pointing compensation.

[0031] By acquiring real-time energy reception efficiency data from beam transmission, the parameters of the neural network trajectory prediction model and the adaptive Kalman filter algorithm are corrected in reverse to construct a complete closed-loop control loop. This not only reduces the bandwidth consumption of ground communication links and decreases reliance on ground intervention, but also maintains control continuity in complex space environments, improving the system's reliability in emergency scenarios such as ship emergency communication and ocean rescue. To adapt to dynamic scene changes, environmental disturbance data can be updated in real time to ensure accurate matching between ship trajectory prediction results and actual motion states. Furthermore, the granularity of spatial sampling units is dynamically adjusted based on deviation distribution characteristics, and noise covariance parameters are optimized in real time to flexibly adapt to deviation changes under different sea states and weather conditions, thereby improving energy utilization efficiency.

[0032] In a preferred embodiment of the present invention, the dynamic tracking input set is input into a pre-trained neural network model for processing to obtain a prediction result dataset including the ship's future trajectory and position offset, which may include:

[0033] In this embodiment of the invention, based on a dynamic tracking input set, a pre-trained neural network model is used to predict the ship's trajectory in multiple steps, generating a trajectory sequence of the ship within a preset time period in the future, including the ship's position coordinates at multiple predicted times. Specifically, the input layer of the neural network model needs to incorporate three types of feature data strongly correlated with the ship's motion to comprehensively capture the key factors affecting the ship's trajectory. The first type is the ship's historical position data, which selects the ship's longitude, latitude, and altitude data collected every 5 minutes within 1 hour before the current time, for a total of 12 sets of data, each set containing 3 dimensions, corresponding to 36 feature nodes in the input layer; the second type is the ship's... The historical motion state data also selects ship speed, acceleration, and heading angle data collected every 5 minutes within the hour before the current time, totaling 12 sets of data. Each set contains 3 dimensions, corresponding to 36 feature nodes in the input layer. The third type is historical environmental disturbance data, which selects ocean current speed, ocean current direction, wind speed, wind direction, and wave level data collected every 5 minutes within the hour before the current time, totaling 12 sets of data. Each set contains 5 dimensions, corresponding to 60 feature nodes in the input layer. In summary, the total number of nodes in the input layer is 36 + 36 + 60 = 132, ensuring complete coverage of the ship's own motion attributes and external environmental disturbance factors.

[0034] The neural network model is designed with two hidden layers to achieve progressive feature extraction and deep fusion. The first hidden layer has 128 nodes, and its main function is to initially extract basic features and simple correlation features of a single type of data, such as identifying the linear relationship between changes in ship longitude and sailing speed, or the matching relationship between ocean current direction and ship heading angle. The second hidden layer has 64 nodes, and its core function is to deeply fuse the basic features extracted by the first layer. For example, it combines ocean current speed and sailing acceleration to analyze the combined effect of the two on changes in ship position, or combines wind speed and heading angle to determine the synergistic effect of wind on ship steering. This allows the neural network model to learn the ship motion patterns under the interaction of multiple factors. The output layer of the neural network model needs to match the trajectory prediction target. In combination with the response requirements of satellite attitude adjustment in the ocean cargo ship navigation scenario, the future preset time period is set to 30 minutes, and a prediction time is set every 5 minutes, for a total of 6 prediction times. At each prediction time, the ship's longitude, latitude, and altitude coordinates need to be output to determine its position. Therefore, the output layer is set with 6 sets of nodes, each set containing 3 nodes, for a total of 18 output nodes. Each set of nodes corresponds to the complete position coordinates of a prediction time, ensuring that the prediction results can directly reflect the future position change sequence of the ship.

[0035] First, a dynamic tracking input set is obtained. This dataset must perfectly match the features of the neural network model's input layer. It includes real-time data from the hour preceding the current moment, specifically ship position data (longitude, latitude, altitude), ship motion status data (speed, acceleration, heading angle), and environmental disturbance data (ocean current speed, current direction, wind speed, wind direction, wave level) collected every 5 minutes. This ensures the data accurately reflects the ship's current and recent motion status and its surrounding environment. The dynamic tracking input set is then input into the pre-trained neural network model. The model performs multi-step predictions through forward propagation. The input layer first passes 132 feature data points to the first hidden layer, which performs preliminary feature extraction on these data, such as extracting features from changes in ship speed and historical longitude. The influence of speed on longitude shift is characterized by extracting the driving characteristics of ocean currents on ship position from ocean current speed and direction. Then, the first hidden layer passes the extracted basic features to the second hidden layer, which performs deep fusion of these basic features. For example, it combines ocean current speed driving characteristics with ship acceleration characteristics to calculate the trend of ship position change under the combined effect of the two, and combines wind speed characteristics with heading angle characteristics to analyze the trend of wind force's influence on the ship's position after turning. Finally, the second hidden layer passes the fused features to the output layer. The output layer calculates the ship's longitude, latitude, and altitude at a predicted time of 5 minutes within the next 30 minutes based on the motion laws learned during model training. The position coordinates of the 6 predicted times are arranged in chronological order to form a sequence of the ship's future trajectory.

[0036] Based on the trajectory sequence and combined with real-time acquired environmental disturbance data, the estimated position offset of the ship relative to the current satellite orbit at each predicted time is calculated. Specifically, this includes: first, acquiring the satellite's current orbital data, which includes orbital parameters such as the satellite's semi-major axis, orbital eccentricity, orbital inclination, right ascension of the ascending node, argument of perigee, and mean anomaly. Based on these parameters, the satellite's orbital position coordinates (longitude, latitude, and altitude) at each predicted time can be determined. More specifically, the distance from the satellite to the Earth's center is calculated using the orbital semi-major axis and orbital eccentricity; the satellite's azimuth in space is determined by combining the orbital inclination and right ascension of the ascending node; and then, the specific position of the satellite in orbit is determined by combining the argument of perigee and mean anomaly. Finally, the estimated position is obtained. The system first obtains the precise position coordinates of the satellite at each predicted time. Then, it extracts the ship's position coordinates (longitude, latitude, and altitude) at each predicted time from the generated trajectory sequence. Simultaneously, it acquires real-time environmental disturbance data for each predicted time. This real-time environmental disturbance data is collected jointly by the ship's onboard sensors and the satellite's onboard sensors. The ship's onboard sensors collect ocean current speed, ocean current direction, and wave level, while the satellite's onboard sensors collect wind speed, wind direction, and the Earth's magnetic field strength (especially for the equatorial magnetic disturbance region, real-time magnetic field strength data needs to be collected in detail to correct positioning errors). The data is collected within 5 minutes before and after each predicted time, and the average value is taken as the real-time environmental disturbance data for that predicted time to ensure that it reflects the actual environmental state of the ship at that time.

[0037] For each predicted time, the basic position difference between the ship and the satellite orbit in three dimensions—longitude, latitude, and altitude—is calculated. The basic position difference in the longitude direction is equal to the predicted longitude of the ship at the predicted time minus the longitude of the satellite orbit at the predicted time; the basic position difference in the latitude direction is equal to the predicted latitude of the ship at the predicted time minus the latitude of the satellite orbit at the predicted time; and the basic position difference in the altitude direction is equal to the predicted altitude of the ship at the predicted time minus the altitude of the satellite orbit at the predicted time. This step of calculating the basic position difference represents the initial position difference between the ship and the satellite orbit without considering the influence of additional environmental disturbances.

[0038] To calculate the additional longitude offset of ocean currents, first determine the time interval between the predicted time and the current time. Then, multiply the ocean current speed by this time interval to obtain the distance the ocean current propels the ship per unit time. Next, multiply this by the cosine of the angle between the ocean current direction and the ship's longitude direction (the angle is determined by the difference between the ocean current direction and the ship's longitude direction, with true north as 0° and the ocean current direction calculated clockwise, and east longitude as positive and west longitude as negative to determine the ship's longitude direction). Finally, the additional longitude offset of ocean currents is obtained. To calculate the additional latitude offset of ocean currents, multiply the ocean current speed by the time interval, and then multiply this by the sine of the angle between the ocean current direction and the ship's latitude direction (the ship's latitude direction is determined with north latitude as positive and south latitude as negative) to obtain the additional latitude offset of ocean currents. Ocean currents have no significant effect on altitude, so the additional altitude offset of ocean currents is not calculated.

[0039] The calculation logic is consistent with the additional offset of ocean currents, except that ocean current speed is replaced with wind speed and ocean current direction is replaced with wind direction. The additional offset of wind speed with respect to longitude is equal to wind speed multiplied by the time interval, and then multiplied by the cosine of the angle between the wind direction and the ship's longitude direction; the additional offset of wind speed with respect to latitude is equal to wind speed multiplied by the time interval, and then multiplied by the sine of the angle between the wind direction and the ship's latitude direction; wind speed has no significant effect on altitude, so the additional offset of wind speed in the altitude direction is not calculated.

[0040] When a ship is in the equatorial magnetic disturbance region, there is a difference between the real-time magnetic field strength and the standard magnetic field strength. This difference can cause positioning errors in ship positioning sensors (such as GPS). The calculation first involves subtracting the standard magnetic field strength from the real-time magnetic field strength. This difference is then multiplied by a sensor error coefficient (calibrated through numerous experiments, representing the average sensor positioning error corresponding to different magnetic field strength differences) to obtain the total positioning error caused by the magnetic field. Based on the distribution ratio of the positioning error across longitude, latitude, and altitude, it is then decomposed into the additional offset of the magnetic field relative to longitude, the magnetic field error, and the magnetic field error itself. Additional offsets for latitude and magnetic field for altitude; the additional offsets of each environmental disturbance factor in the same dimension at the same prediction time are added together to obtain the total additional offset for that dimension. Specifically, the total additional longitude offset equals the ocean current longitude offset plus the wind speed longitude offset plus the magnetic field longitude offset; the total additional latitude offset equals the ocean current latitude offset plus the wind speed latitude offset plus the magnetic field latitude offset; the total additional altitude offset equals the magnetic field altitude offset (since ocean currents and wind speed have no significant effect on altitude, only the magnetic field offset is included).

[0041] The estimated position offset for each prediction time is obtained by adding the base position difference for each dimension to the corresponding total additional offset. The estimated longitude offset is equal to the base longitude position difference plus the total additional longitude offset; the estimated latitude offset is equal to the base latitude position difference plus the total additional latitude offset; and the estimated altitude offset is equal to the base altitude position difference plus the total additional altitude offset. Each prediction time corresponds to a set of estimated position offset data containing the three dimensions of longitude, latitude, and altitude, which fully reflects the actual position offset of the ship relative to the satellite orbit at that time.

[0042] The trajectory sequence is associated and integrated with the corresponding position offset, and timestamp information is added to construct a structured prediction result dataset. Specifically, this includes: first, establishing the correspondence between prediction time and data. In the trajectory sequence, each prediction time corresponds to a set of ship position coordinates (longitude, latitude, and altitude); in the estimated position offset, each prediction time corresponds to a set of estimated position offset data (estimated longitude offset, estimated latitude offset, and estimated altitude offset); pairing the ship position coordinates and estimated position offset data at the same prediction time to form a basic data group for the prediction time. Each basic data group contains six core data items: ship longitude, ship latitude, ship altitude, estimated longitude offset, estimated latitude offset, and estimated altitude offset, ensuring that the position information and offset information at the same time are completely correlated.

[0043] Two types of timestamps are added to each basic data set to clarify the time attribute of the data. The first type is the prediction time timestamp, which records the specific time of the future prediction corresponding to the basic data set. For example, if the current time is 14:00:00 on a certain day, the first prediction time is 14:05:00, the second prediction time is 14:10:00, and so on, with each prediction time determined at 5-minute intervals and used as a timestamp. The second type is the raw data acquisition timestamp, which records the acquisition time of the latest set of data in the dynamic tracking input set. For example, if the last time the dynamic tracking input set was acquired was 13:55:00, this time is added as the raw data acquisition timestamp to all basic data sets for use in... The timeliness of the original data on which the prediction operation is based is traced; according to the order of the prediction time, all the basic data groups after adding timestamps are arranged in sequence to form a structured prediction result dataset. Each entry in this dataset corresponds to a prediction time. The data in each entry (ship position coordinates, estimated position offset, prediction time timestamp, and original data collection timestamp) are completely related and have a unified format. For example, the first entry corresponds to the prediction time of 14:05:00, which includes the ship's longitude, latitude, and altitude at that time, the estimated longitude offset, estimated latitude offset, estimated altitude offset, the prediction time timestamp of 14:05:00, and the original data collection timestamp of 13:55:00.

[0044] By constructing a pre-trained neural network model covering multiple features, it is possible to fully learn the correlation between ship motion and environmental disturbances in different sea areas and sea conditions, accurately capture trajectory change features, reduce the deviation between predicted and actual trajectories, and the structured prediction result dataset is designed with clear timestamps and data associations to facilitate the extraction of target data, reduce data processing time, and avoid control errors caused by data chaos or unclear timeliness, thus providing a data foundation for stable energy transmission from satellites to moving ships.

[0045] In a preferred embodiment of the present invention, based on the prediction result dataset, the satellite autonomously generates attitude adjustment commands and calculates the initial expected pointing value of the beam, which may include:

[0046] In this embodiment of the invention, based on the ship's future trajectory information and satellite orbital parameters in the prediction result dataset, the three-axis attitude adjustment required for the satellite to establish and maintain energy transmission beam coverage over the target ship is calculated. Specifically, this includes: first, extracting the ship's future trajectory information from the prediction result dataset. This information contains the ship's specific position coordinates at multiple consecutive prediction times, each coordinate consisting of longitude, latitude, and altitude. Simultaneously, the satellite's current orbital parameters are obtained, including the satellite's orbital inclination, semi-major axis, perigee, argument, ascending node, right ascension, and orbital period. Next, calculating the relative positional relationship between the satellite and the target ship at each prediction time. During the calculation, the ship's geographical coordinates are first converted to coordinates in a Cartesian coordinate system, and then the satellite's orbital parameters are converted to satellite position coordinates in the same Cartesian coordinate system. Then, the relative position vector between the two is obtained by subtracting the ship's position coordinates from the satellite's position coordinates. In this process, the potential impact of sudden strong ocean currents and high-altitude strong winds on the ship must be fully considered. The ship's position offset caused by these environmental disturbances is superimposed on the ship's basic position coordinates to ensure that the calculation of the relative positional relationship is accurate. Based on the actual motion state of the ship, and then according to the coverage parameters of the satellite beam, including the horizontal half-power angle and vertical half-power angle of the beam, the specific changes in the roll, pitch, and yaw angles required by the satellite to ensure complete coverage of the target ship at each predicted moment are determined. The roll angle adjustment is calculated based on the current roll angle of the satellite, combined with the ratio of the ship's lateral offset relative to the satellite to the horizontal coverage range of the beam; the larger the lateral offset, the larger the roll angle adjustment. Similarly, the pitch angle adjustment is calculated based on the current pitch angle of the satellite, combined with the ratio of the ship's longitudinal offset relative to the satellite to the vertical coverage range of the beam; the larger the longitudinal offset, the larger the pitch angle adjustment. The yaw angle adjustment is calculated based on the angle between the satellite orbital plane and the ship's direction of motion; the larger the angle, the larger the yaw angle adjustment. These three angle changes together constitute the three-axis attitude adjustment. During the calculation process, changes in the ship's motion state are continuously monitored. If the ship experiences non-uniform dynamic offset due to environmental disturbances, the relative position vector is updated in real time and the three-axis attitude adjustment is recalculated to ensure that the adjustment always adapts to the ship's motion state.

[0047] The three-axis attitude adjustment values ​​are converted into a standard command format to drive the satellite's attitude maneuvers, generating specific attitude adjustment commands. This process involves: first, after obtaining the three-axis attitude adjustment values, determining the command format standard that the satellite's onboard actuators can recognize. This standard is pre-set by the satellite control system and includes parameters such as the binary encoding rules of the commands, the baud rate verification method for data transmission, and the actuator's response time. Then, the angle values ​​of the three-axis attitude adjustment values ​​are converted into digital signals conforming to this standard. During the conversion, the roll, pitch, and yaw angle values ​​are multiplied by their corresponding conversion coefficients. These conversion coefficients are determined based on the mechanical transmission ratio of the actuators. For example, the roll angle conversion coefficient corresponds to how many motor rotation pulses per degree, the pitch angle conversion coefficient corresponds to how many millimeters of hydraulic device extension / retraction per degree, and the yaw angle conversion coefficient corresponds to how many electrical pulses per degree. The power supply pulses to the magnetic coil are used to calculate the actuator control parameters corresponding to each angle adjustment. These control parameters are then encoded into binary instruction codes according to standard instruction coding rules. Each instruction code includes an instruction type identifier, actuator number, control parameter value, and check code. A real-time verification mechanism is added during the conversion process. Each set of instructions is compared with the latest three-axis attitude adjustment. If a deviation is found, the conversion is immediately restarted to ensure that the converted instructions accurately reflect the latest adjustment requirements. The final specific attitude adjustment instructions must also include the execution time sequence, i.e., the order of roll, pitch, and yaw angle adjustments, as well as the amplitude and duration of each adjustment action. The amplitude is determined by the control parameter value, and the duration is determined by the response speed of the actuator, so that the onboard actuator can accurately and orderly complete the attitude adjustment according to the instructions.

[0048] Based on the satellite's expected attitude determined by the attitude adjustment command, and combined with the real-time relative geometric relationship between the satellite and the target ship, the initial expected beam pointing value for aligning the communication beam center with the target ship is calculated. Specifically, this includes: first, determining the satellite's expected attitude after adjustment based on the attitude adjustment command. This expected attitude includes the specific values ​​of the satellite's azimuth, pitch, and roll angles in space. These values ​​are defined in the satellite's own coordinate system, with the azimuth angle based on the satellite's forward direction, the pitch angle based on the satellite's lateral centerline, and the roll angle based on the satellite's longitudinal centerline. Next, the real-time position coordinates of the target ship are acquired using onboard measurement equipment, such as microwave radar and a GPS receiver. Simultaneously, the satellite's own real-time orbital position parameters are acquired using its onboard sensors. Then, the real-time relative geometric relationship between the satellite and the target ship is calculated. During the calculation, the ship's real-time position coordinates and the satellite's orbital position parameters are first converted to the same inertial coordinate system. Then, the straight-line distance between them is calculated, which is obtained by subtracting each component of the satellite's position coordinates from the ship's position coordinates, summing the squares, and then taking the square root. Simultaneously, the line connecting the satellite to the ship and the axes of the inertial coordinate system are calculated. The included angles are the azimuth angle and the pitch angle. The azimuth angle is determined by the angle between the projection of the line connecting the two axes onto the horizontal plane and the X-axis of the coordinate system. The pitch angle is determined by the angle between the line connecting the two axes and the horizontal plane. Then, based on the expected attitude of the satellite and the aforementioned relative geometric relationships, the azimuth and pitch angles that the communication beam center needs to point to are determined. During the calculation, the predicted position of the ship's future trajectory is used as a reference. The initial direction of the beam under the expected attitude of the satellite is compared with the direction of the line connecting the satellite and the ship. If there is a deviation, the required adjustment angle is calculated so that the beam center can be aligned with the ship's predicted position. The system measures the ship's position. Considering potential positional shifts due to sudden strong ocean currents or winds, a certain adjustment margin is reserved when calculating the initial expected value of the beam pointing, based on the maximum deviation of the ship under similar conditions in historical navigation data. This margin is typically 1.2 times the maximum deviation. Specifically, the angles between the line connecting the satellite to the predicted ship position and each axis of the satellite's coordinate system are first calculated. Then, combined with the beam divergence angle parameters, these angles are converted into the initial expected value of the beam pointing. This ensures that the expected value allows the beam center to be precisely aligned with the target ship, laying a solid foundation for deviation compensation.

[0049] By fully integrating the ship's future trajectory information and satellite orbit parameters, and combining the impact of environmental disturbances on the ship's motion, the three-axis attitude adjustment amount is calculated. This allows the adjustment amount to accurately adapt to the ship's dynamic motion state. The three-axis attitude adjustment amount is converted into a standard command format, ensuring that the onboard actuators can accurately receive and execute the adjustment commands, improving the timeliness of attitude adjustment, and avoiding adjustment lag caused by command mismatch.

[0050] In a preferred embodiment of the present invention, the onboard actuator adjusts the satellite attitude in real time according to attitude adjustment commands to obtain the initial pointing state of the beam and the corresponding actual pointing deviation data, which may include:

[0051] In this embodiment of the invention, the onboard actuator receives attitude adjustment commands and drives the satellite's attitude control mechanism to perform corresponding roll, pitch, and yaw three-axis adjustments. Specifically, the onboard actuator consists of an onboard control unit, a drive circuit, and attitude control components (roll axis momentum wheel, pitch axis torque motor, yaw axis magnetic torque generator, etc.). It first receives attitude adjustment commands transmitted from the attitude pointing module. These commands include the target adjustment angle of the roll axis (e.g., ±5°), the target adjustment angle of the pitch axis (e.g., ±3°), the target adjustment angle of the yaw axis (e.g., ±2°), and the order of the three-axis adjustments (usually first the roll axis, then the pitch axis, and finally the yaw axis to avoid inter-axis interference).

[0052] The onboard control unit drives the rolling shaft attitude control component according to the target adjustment angle of the rolling shaft in the command. If a momentum wheel is used, the preset torque coefficient of the momentum wheel is first retrieved, such as 0.8 N·m / (r / min), meaning that every 1 r / min change in rotational speed corresponds to a torque output of 0.8 N·m. Then, the rotational speed adjustment is calculated. The rotational speed adjustment is equal to the target adjustment angle of the rolling shaft multiplied by the torque coefficient, and then divided by the moment of inertia of the momentum wheel (such as 0.5 kg·m²). Finally, the required rotational speed value is obtained. For example, when the target adjustment angle of the rolling shaft is 3°, the rotational speed adjustment... The total torque is calculated as 3° × 0.8 N·m / (r / min) ÷ 0.5 kg·m² = 4.8 r / min. The drive circuit controls the momentum wheel to accelerate or decelerate based on this rotational speed, generating a control torque in the rolling direction to drive the satellite's rolling axis to the target angle. If a jet propulsion system is used, the torque-duration conversion coefficient of the propulsion system is selected, such as 0.5 N·m / s, meaning that 0.5 N·m of torque is output every second. The jet propulsion duration is calculated as the target adjustment angle of the rolling axis × torque coefficient ÷ conversion coefficient. Angle adjustment is achieved by controlling the opening and closing time of the jet valve.

[0053] The control unit drives the pitch axis torque motor. First, it obtains the motor's gear ratio, such as 1:100, meaning 100 motor rotations correspond to 1° of satellite pitch axis rotation. Then, it calculates the number of rotations: the number of rotations equals the target pitch axis adjustment angle multiplied by the gear ratio. For example, when the target pitch axis adjustment angle is 2°, the number of rotations = 2° × 100 = 200 rotations. Simultaneously, it retrieves the motor's rated speed (e.g., 100 revolutions per minute) and calculates the rotation time = number of rotations ÷ rated speed, ensuring the motor completes rotation within the specified time to avoid overshoot. During adjustment, the motor's built-in encoder provides real-time feedback on the number of rotations. The control unit compares the feedback value with the calculated value. If there is a discrepancy, such as only 198 rotations when the feedback is 200, the corresponding number of rotations is added to ensure precise pitch axis positioning.

[0054] Considering the possibility that the cargo ship might enter a magnetically disturbed region near the equator, the control unit first retrieves real-time magnetic field strength data (e.g., 35000 nT) collected by the onboard magnetometer, then retrieves historical average magnetic field data for the satellite's current orbital region (e.g., 34500 nT), calculating the magnetic field deviation as: real-time magnetic field strength data - historical average magnetic field data (i.e., 500 nT). Next, the output coefficient of the torque converter is calibrated based on the magnetic field deviation. The calibrated output coefficient is calculated as: original output coefficient (e.g., 0.3 N·m / (1000 nT)) × (historical average magnetic field data ÷ real-time magnetic field strength data), ensuring that the torque converter's output torque is unaffected by magnetic field disturbances. Finally, the calculation... The output torque of the magnetic torque generator = yaw axis target adjustment angle × calibrated output coefficient, drives the magnetic torque generator to generate yaw direction torque, and completes the yaw axis adjustment; during the entire three-axis adjustment process, the onboard control unit receives the angle feedback signal of each axis every 0.5 seconds. If it is found that the adjustment progress of a certain axis is lagging behind (e.g., the rolling axis should rotate 3° but only rotates 2.5°), the drive power is immediately increased (e.g., increasing the momentum wheel speed or extending the jet duration) to ensure that the three axes are adjusted synchronously according to the instructions, adapting to the lateral or longitudinal offset of the ship caused by sudden strong ocean currents (e.g., current speed 3m / s) and strong winds at high altitudes (e.g., wind speed 20m / s), and avoiding the attitude adjustment from being out of sync with the ship's motion trend.

[0055] After the three-axis adjustment is completed, the actual attitude data representing the satellite's position in space is acquired through onboard sensors. Specifically, this includes: when the roll, pitch, and yaw adjustments are all completed, and the difference between the angle feedback value of each axis and the target adjustment angle is less than 0.1° as the judgment criterion, the onboard control unit triggers the onboard multi-sensor collaborative acquisition process, with the acquisition duration set to 10 seconds. Fiber optic gyroscope data acquisition and processing are then initiated, activating the fiber optic gyroscopes on the three axes of the satellite, acquiring angular velocity data for the corresponding axis every 0.1 seconds, for a total of 100 sets of data within 10 seconds. The 100 sets of angular velocity data for each axis are processed, first removing outliers that exceed the normal range, and then calculating the average value of the remaining valid data to obtain the average angular velocity for each axis. Next, the angle change for each axis is calculated: angle change = average angular velocity × acquisition duration. This value is the angular offset of the satellite relative to its initial attitude after adjustment.

[0056] The star sensor is activated, capturing images of stars in space (1024×1024 pixels resolution) through its optical lens. One image is captured every 2 seconds, for a total of 5 images within 10 seconds. The positions of stars in each image are compared with the star sensor's built-in stellar database (containing the standard inertial coordinates of 100,000 stars). First, the top 20 brightest stars in the image are identified, their pixel coordinates are extracted, and then converted to right ascension and declination in an inertial coordinate system. These coordinates are then compared with the standard right ascension and declination of the corresponding stars in the database to calculate the star sensor's attitude measurement values ​​(including zenith). (Angle, azimuth angle); take the average of the attitude measurements from 5 images as the final star sensor attitude data. Then, use this data to calibrate the angle change of the fiber optic gyroscope. The calibrated roll angle = roll angle measured by the star sensor - roll axis angle change calculated by the fiber optic gyroscope; the calibrated pitch angle = pitch angle measured by the star sensor - pitch axis angle change calculated by the fiber optic gyroscope; the calibrated yaw angle = yaw angle measured by the star sensor - yaw axis angle change calculated by the fiber optic gyroscope. This corrects the drift error of the gyroscope caused by long-term operation.

[0057] The onboard magnetometer is activated to collect magnetic field strength data at the current location, once per second, for a total of 10 sets of data within 10 seconds. The average value is calculated as the real-time magnetic field data. Simultaneously, the historical average magnetic field data for the current orbital region is retrieved from the onboard data storage unit. This data is calculated using magnetic field data collected by the satellite in this region over the past 3 months. The magnetic field deviation is calculated as: Real-time magnetic field data - Historical average magnetic field data. The magnetic field deviation is used to correct the magnetometer's attitude measurement value. The corrected magnetometer yaw angle is: Original yaw angle measured by the magnetometer - (Magnetic field deviation × Magnetic disturbance correction coefficient), where the magnetic disturbance correction coefficient is a preset value, thus offsetting the influence of equatorial magnetic disturbance on the magnetometer.

[0058] The calibrated fiber optic gyroscope attitude data (roll angle, pitch angle), star sensor attitude data (zenith angle, azimuth angle), and corrected magnetometer attitude data (yaw angle) are fused according to weights. The star sensor has the highest accuracy, with a weight of 50%; the fiber optic gyroscope is next, with a weight of 30%; and the magnetometer provides supplementary information, with a weight of 20%. Specifically, the final roll angle is calculated as follows: final roll angle = (calibrated gyroscope roll angle × 30%) + (roll angle derived from star sensor × 50%) + (roll angle associated with corrected magnetometer × 20%). The final pitch angle and yaw angle are calculated using the same logic, ultimately yielding the actual satellite attitude data containing specific values.

[0059] Based on actual attitude data and the geometric relationship between the satellite and the target ship, the actual pointing state of the communication beam is calculated. Specifically, the onboard control unit first retrieves the satellite's current orbital parameters from the onboard orbit calculations, including orbital radius, orbital inclination, and right ascension of the ascending node. Then, combining this with the obtained actual attitude data (final roll angle, pitch angle, and yaw angle), the satellite's body coordinate system is converted to an inertial coordinate system (J2000 inertial system). The satellite's orbital radius, orbital inclination, and right ascension of the ascending node are then converted into satellite position coordinates (X-satellite, Y-satellite, Z-satellite) in the inertial coordinate system. For each satellite, X = orbital radius × cos(right ascension of ascending node) × cos(orbital inclination), Y = orbital radius × sin(right ascension of ascending node) × cos(orbital inclination), and Z = orbital radius × sin(orbital inclination). Then, based on the roll angle, pitch angle, and yaw angle from the actual attitude data, the attitude transformation matrix is ​​calculated. Using Euler angle transformation logic, the coordinate axes of the body coordinate system are rotated to align with the inertial coordinate system, mapping the satellite's attitude parameters to the inertial coordinate system to obtain the satellite's attitude orientation in the inertial coordinate system, such as the X-axis direction and Y-axis direction.

[0060] Next, a detection signal is sent to the target ship via an onboard microwave radar (with a detection range of up to 500 km). After receiving the signal reflected by the ship, the real-time geographical coordinates of the ship (longitude, latitude, and altitude, with altitude taken as 0m above sea level) are analyzed and converted into the ship's position coordinates (X ship, Y ship, Z ship) in an inertial coordinate system. First, the ship's longitude and latitude are converted into horizontal angles in a geocentric coordinate system (longitude corresponds to geocentric longitude angle, and latitude corresponds to geocentric latitude angle); X ship = (Earth's radius + ship's altitude) × cos(geocentric latitude angle) × cos(geocentric longitude angle), Y ship = (Earth's radius + The coordinates are calculated as follows: (Satellite altitude) × cos(geocentric latitude angle) × sin(geocentric longitude angle), Zship = (Earth's radius + ship's altitude) × sin(geocentric latitude angle); Then, the relative geometric relationship between the satellite and the ship is calculated: relative position vector = (Xsatellite - Xship, Ysatellite - Yship, Zsatellite - Zship), which is obtained by subtracting the corresponding components of the ship's inertial coordinates from the X, Y, and Z components of the satellite's inertial coordinates to obtain the relative distance components in the inertial coordinate system; azimuth base angle = arctan[(Ysatellite - Yship) / (Xsatellite - Xship)], pitch base angle = arctan[(Zsatellite - Zship) / ... Then, the installation parameters of the communication beam on the satellite's coordinate system are retrieved, including the offset angle of the beam's center axis relative to the satellite's roll axis (e.g., 0.5°, meaning the beam is offset 0.5° in the positive direction of the roll axis) and the offset angle relative to the satellite's pitch axis (e.g., 0.3°, meaning the beam is offset 0.3° in the positive direction of the pitch axis). Combined with the positional shift of the ship caused by sudden strong ocean currents and high-altitude winds, the correction is calculated. Data on sudden ocean currents in the area over the past year is retrieved from the satellite's historical database to statistically determine the maximum lateral offset angle caused by such currents, which is used as the azimuth correction. Similarly, data on high-altitude winds during the same period are retrieved to statistically determine the maximum longitudinal offset angle caused by such winds, which is used as the pitch correction.

[0061] Finally, the actual pointing state of the communication beam is calculated. The actual azimuth angle = actual satellite roll angle + beam offset angle relative to roll axis + azimuth base angle + azimuth correction, which is the sum of satellite roll attitude angle, beam installation offset angle, relative azimuth base angle, and ocean current offset correction angle. The actual pitch angle = actual satellite pitch angle + beam offset angle relative to pitch axis + pitch base angle + pitch correction, which is the sum of satellite pitch attitude angle, beam installation offset angle, relative pitch base angle, and strong wind offset correction angle. Through the above calculations, the actual pointing state of the communication beam containing specific values ​​is obtained.

[0062] The actual pointing state is compared with the initial expected beam pointing value to calculate the azimuth angle deviation and the elevation angle deviation, forming the actual pointing deviation data. Specifically, the onboard control unit first extracts the previously generated initial expected beam pointing value from the historical attitude pointing calculation results. This value includes the expected azimuth angle and the expected elevation angle. At the same time, the actual pointing state of the communication beam is extracted. The azimuth angle deviation is calculated by subtracting the expected azimuth angle from the actual azimuth angle. If the result is positive, it indicates that the actual azimuth angle is larger than the expected angle; if it is negative, it indicates that the actual azimuth angle is smaller than the expected angle. The elevation angle deviation is calculated by subtracting the expected elevation angle from the actual elevation angle. Similarly, the direction of deviation is determined by the positive or negative value.

[0063] During the calculation process, the onboard control unit monitors the changing trend of the deviation in real time. If the cargo ship enters the equatorial magnetic disturbance region, the deviation fluctuation recording mode is activated, recording the values ​​of the azimuth and pitch deviations every 0.2 seconds, and statistically analyzing the duration, amplitude, and frequency of the fluctuations. It also marks the satellite orbit position, ship geographical location, and other related information when the fluctuations occur. Finally, the azimuth and pitch angle deviations, along with the duration, amplitude, frequency, and related location information of the fluctuations, are integrated to form structured actual pointing deviation data. This data is first stored in the onboard non-volatile memory and then transmitted to the deviation analysis in real time through the satellite-to-ground communication link. At the same time, a backup copy of the data is kept for fault backtracking.

[0064] When driving attitude control via onboard actuators, the system incorporates environmental data such as magnetic disturbances and ship dynamic offsets to calibrate actions, reducing attitude adjustment deviations caused by ignoring environmental disturbances. By utilizing multi-sensor collaborative acquisition and fusion of attitude data, the system effectively corrects the impact of magnetic disturbances on the sensors. By accurately calculating the relative geometric relationship between the satellite and the ship to obtain the actual beam pointing, the system can adapt in real time to the dynamic offsets caused by sudden ocean currents and strong winds. Finally, through detailed comparison, the system obtains actual pointing deviation data containing wave information, reducing the lag in deviation compensation and thus ensuring the stability of the auxiliary power supply for the cargo ship navigation system.

[0065] like Figure 2 As shown, in a preferred embodiment of the present invention, the beam coverage area corresponding to the actual pointing deviation data is divided into regions to form multiple spatial sampling units, which may include:

[0066] In this embodiment of the invention, based on the statistical distribution characteristics of the actual pointing deviation data, the key areas requiring refined compensation within the beam coverage area are determined. Specifically, this includes: First, extracting the azimuth angle deviation, pitch angle deviation, duration of deviation fluctuation, amplitude of deviation fluctuation, and the satellite's orbital position and the target ship's real-time geographical location at the time of the deviation from the previously acquired actual pointing deviation data; then, conducting statistical distribution analysis on these deviation data. The azimuth angle deviation is divided into multiple continuous deviation intervals at 0.1-degree intervals, for example, from -1.2 degrees to -1.1 degrees, -1.1 degrees to -1.0 degrees... 0.9 degrees to 1.0 degrees, 1.0 degrees to 1.1 degrees; similarly, the pitch angle deviation is divided into corresponding deviation intervals at 0.1-degree intervals. Then, the frequency of data occurrence within each deviation interval is statistically analyzed. Specifically, the number of deviation data entries recorded in the interval is divided by the total number of all deviation data entries, and then multiplied by 100 to obtain the percentage of deviation occurrence frequency for each interval.

[0067] If the frequency percentage of a certain deviation range exceeds 35%, and the absolute value of the deviation within that range is greater than 0.3 degrees, then the beam coverage angle range corresponding to that deviation range is a critical area requiring fine-grained compensation. For example, when a cargo ship enters the equatorial magnetic disturbance region, if the onboard sensors detect a 40% frequency percentage of pitch angle deviations between 0.3 and 0.5 degrees, and this deviation range causes the beam to fail to accurately cover the ship, then the beam coverage area corresponding to this pitch angle range is considered a critical area. When a cargo ship is subjected to sudden... Strong ocean currents cause azimuth angle deviations of -0.5 to -0.3 degrees to occur in 38% of cases, and this deviation can reduce energy transmission efficiency by more than 15%. Therefore, the beam coverage area corresponding to this azimuth angle range should also be included in the critical area. At the same time, it is also necessary to combine the ship's historical navigation data to supplement the critical area with the range of azimuth and pitch angles where the ship may experience non-uniform dynamic deviations under the influence of strong winds. This ensures that the critical area can fully cover the beam area with concentrated deviations and significant impacts, providing targeted targets for refined compensation.

[0068] Based on the boundary range of the key area, the granularity of spatial division is defined, and the azimuth and elevation angle spans of each sampling unit are determined. Specifically, this includes: firstly, calculating the angular boundaries of the beam coverage area based on the determined deviation range of the key area. For example, the deviation range corresponding to the azimuth of the key area is -0.5 degrees to -0.3 degrees. Combining this with the expected azimuth of 120.0 degrees for the initial beam pointing, the angular boundaries of the key area are calculated as 120.0 degrees minus 0.5 degrees equals 119.5 degrees and 120.0 degrees minus 0.3 degrees equals 119.7 degrees. The boundary range of the region is 119.5 degrees to 119.7 degrees, and the boundary difference is 119.7 degrees minus 119.5 degrees, which equals 0.2 degrees. The deviation range of the pitch towards the critical region is 0.3 degrees to 0.5 degrees. Combining the expected pitch angle of 30.0 degrees for the initial beam pointing, the angle boundary of the pitch towards the critical region is calculated to be 30.0 degrees plus 0.3 degrees, which equals 30.3 degrees, and 30.0 degrees plus 0.5 degrees, which equals 30.5 degrees. That is, the boundary range of the pitch towards the critical region is 30.3 degrees to 30.5 degrees, and the boundary difference is 30.5 degrees minus 30.3 degrees, which equals 0.2 degrees.

[0069] Based on the compensation accuracy requirements and deviation fluctuation frequency, the granularity of spatial division is defined. Critical areas, due to frequent deviation fluctuations and high compensation accuracy requirements, require finer granularity. Non-critical areas, with gentler deviation fluctuations and lower compensation accuracy requirements, can be divided into coarser granularities. For critical areas, the number of sampling units is determined according to the deviation fluctuation frequency. If the deviation fluctuates 3 times every 8 seconds, at least 6 sampling units are needed to fully capture the deviation details of each fluctuation. Therefore, the 0.2-degree boundary difference in azimuth to the critical area is divided into 6 sampling units, with each sampling unit having an azimuth angle span of 0.2 degrees divided by 6, rounded to three decimal places, yielding approximately 0.033 degrees. Similarly, the 0.2-degree boundary difference in pitch to the critical area is also divided into 6 sampling units, with each sampling unit having a pitch angle span of approximately 0.033 degrees.

[0070] For non-critical areas, such as the azimuth boundary range of 118.0 degrees to 119.5 degrees, the boundary difference is 119.5 degrees minus 118.0 degrees, which equals 1.5 degrees. Because the deviation fluctuation is gentle, only 3 sampling units are needed to meet the compensation requirements. The azimuth angle span of each sampling unit is 1.5 degrees divided by 3, which equals 0.5 degrees. The elevation boundary range of 28.0 degrees to 30.3 degrees, the boundary difference is 30.3 degrees minus 28.0 degrees, which equals 2.3 degrees. Five sampling units are set, and the elevation angle span of each sampling unit is 2.3 degrees divided by 5, which equals 0.46 degrees. Through this differentiated division, it is ensured that the critical areas can accurately capture deviation changes through fine-grained sampling, while avoiding the waste of onboard computing resources due to over-sampling in non-critical areas.

[0071] Based on the azimuth and elevation spans, the beam coverage area is divided into grids in the spatial angular domain. Specifically, this involves: first, determining the overall angular domain of the beam coverage area; then, considering the relative positions of the satellite and the ship, determining the overall azimuth coverage area to be 118.0 degrees to 121.0 degrees, and the overall elevation coverage area to be 28.0 degrees to 31.0 degrees. Starting from the initial azimuth angle of 118.0 degrees, grid lines are drawn according to the determined spans of different regions. In the non-critical region from 118.0 degrees to 119.5 degrees, grid lines are drawn by successively adding 0.5 degrees to the azimuth span, starting from 118.0 degrees. That is, 118.0 degrees plus 0.5 degrees equals 118.5 degrees, 118.5 degrees plus 0.5 degrees equals 119.0 degrees, and 119.0 degrees plus 0.5 degrees equals 119.5 degrees. These grid lines divide the non-critical region into three preliminary... The azimuth sub-interval is defined; after entering the critical area from 119.5 degrees to 119.7 degrees, grid lines are drawn by successively adding 0.033 degrees starting from 119.5 degrees, i.e., 119.5 degrees plus 0.033 degrees equals 119.533 degrees, 119.533 degrees plus 0.033 degrees equals 119.566 degrees, and so on, until 119.7 degrees is reached. These grid lines divide the critical area. The azimuth sub-intervals are divided into 6 fine azimuth sub-intervals; then, entering the non-critical area from 119.7 degrees to 121.0 degrees, grid lines are drawn by successively adding 0.5 degrees from 119.7 degrees, i.e., 119.7 degrees plus 0.5 degrees equals 120.2 degrees, 120.2 degrees plus 0.5 degrees equals 120.7 degrees, and 120.7 degrees plus 0.3 degrees equals 121.0 degrees, thus completing the grid line division of the entire azimuth direction.

[0072] The grid line division logic for the pitch direction is completely consistent with that for the azimuth direction. Starting from the initial angle of 28.0 degrees, the non-critical area from 28.0 degrees to 30.3 degrees is divided into 5 pitch sub-intervals by successively adding 0.46 degrees to the pitch angle span, i.e., 28.0 degrees plus 0.46 degrees equals 28.46 degrees, 28.46 degrees plus 0.46 degrees equals 28.92 degrees, and so on up to 30.3 degrees. The critical area from 30.3 degrees to 30.5 degrees is divided into 6 fine pitch sub-intervals by successively adding 0.033 degrees to the pitch angle span, and the non-critical area from 30.5 degrees to 31.0 degrees is divided into 6 fine pitch sub-intervals by successively adding 0.46 degrees to the pitch angle span, thus completing the pitch grid line division.

[0073] These azimuth and elevation grid lines intersect perpendicularly within the spatial angular domain, dividing the entire beam coverage area into multiple independent rectangular grids. For example, the sub-interval of azimuth from 118.0 to 118.5 degrees intersects with the sub-interval of elevation from 28.0 to 28.46 degrees, forming a grid for a non-critical region; the sub-interval of azimuth from 119.5 to 119.533 degrees intersects with the sub-interval of elevation from 30.3 to 30.333 degrees, forming a fine grid for a critical region. The grids in the critical region are densely distributed due to their small spans, while the grids in the non-critical region are sparsely distributed due to their large spans, ensuring that the grid division can accurately match the deviation compensation requirements of different regions.

[0074] Each grid cell is assigned an independent identifier, and a mapping relationship with the physical space is established to form multiple spatial sampling units. Specifically, when assigning an independent identifier to each grid cell, a combination of azimuth and elevation sub-interval numbers is used for encoding. First, all azimuth sub-intervals are sequentially numbered from left to right (azimuth angle from smallest to largest) as integers. For example, the azimuth sub-interval from 118.0° to 118.5° is numbered 1, 118.5° to 119.0° as number 2, ..., 119.5° to 119.533° as number 6, 119.533° to 119.566° as number 7, ..., 120.7° to 121.0° as number 15. Then, all elevation sub-intervals are sequentially numbered from bottom to top (elevation angle from smallest to largest) as integers. The grid is numbered with integers. For example, the sub-intervals from 28.0 degrees to 28.46 degrees in elevation are numbered 1, from 28.46 degrees to 28.92 degrees in elevation are numbered 2, and so on, from 30.3 degrees to 30.333 degrees in elevation are numbered 8, and from 30.5 degrees to 31.0 degrees in elevation are numbered 14. Each grid is uniquely identified by multiplying the azimuth sub-interval number by 100 and then adding the elevation sub-interval number. For example, the grid formed by the intersection of the azimuth sub-interval 1 and the elevation sub-interval 1 is identified as 1 multiplied by 100 plus 1, which equals 101. The grid formed by the intersection of the azimuth sub-interval 7 (within the critical area) and the elevation sub-interval 8 (within the critical area) is identified as 7 multiplied by 100 plus 8, which equals 708. This encoding method ensures that each grid has a unique identifier and can be quickly associated with the corresponding azimuth and elevation sub-intervals.

[0075] When establishing the mapping relationship between the grid and physical space, the current orbital parameters of the satellite are first obtained from the on-board orbital calculation unit, including orbital altitude (e.g., 350 km), orbital inclination (e.g., 6 degrees), and nadir latitude and longitude (e.g., 0.5 degrees North latitude, 110.0 degrees East longitude). For each grid, the beam center pointing angle corresponding to the grid is calculated based on the median values ​​of the azimuth and elevation angle ranges. Then, the tangent of the median elevation angle is multiplied by the satellite orbital altitude to obtain the horizontal distance from the satellite nadir to the edge of the grid's coverage area. For example, multiplying the orbital altitude of 350 km by the tangent of the elevation angle of 30.3165 degrees yields a horizontal distance of approximately 200 km. Finally, the corresponding horizontal distance is determined by combining the median azimuth angle value. The ground orientation is determined by factors such as the azimuth angle of 119.5165 degrees, which corresponds to the southeast direction. Finally, using the latitude and longitude of the satellite's nadir point as a reference, the ground latitude and longitude range of the area covered by the grid is calculated based on the horizontal distance and orientation. For example, if the nadir point is at 0.5 degrees north latitude and 110.0 degrees east longitude, 200 kilometers southeast corresponds to a ground latitude and longitude range of 0.3 to 0.4 degrees north latitude and 110.2 to 110.3 degrees east longitude. Through this calculation, each grid can accurately correspond to a specific physical area on the Earth's surface. When a cargo ship enters a certain physical area, the on-board system can directly associate the ship's real-time geographical location with the corresponding grid, providing a clear physical location basis for regional deviation compensation.

[0076] By combining the statistical distribution characteristics of actual pointing deviations, key areas were identified, and the beam regions where deviations were concentrated due to equatorial magnetic disturbances, sudden strong ocean currents, and strong high-altitude winds affecting cargo ships were accurately captured. Through differentiated granularity of key and non-key areas, fine-grained sampling was implemented in key areas to accurately capture deviation fluctuations, while coarse-grained sampling was used in non-key areas to save on-board resources. This overcomes the shortcomings of traditional systems where fixed-granularity partitioning cannot adapt to differences in deviation fluctuations, easily leading to insufficient compensation in key areas and waste of resources in non-key areas. Through spatial angle domain grid partitioning and physical space mapping, each spatial sampling unit can correspond to the physical area where the cargo ship is actually sailing, ensuring the reliability of energy supply for ocean-going cargo ships during transoceanic voyages.

[0077] In a preferred embodiment of the present invention, based on the statistical characteristics of the deviation data within multiple spatial sampling units, an adaptive filtering algorithm is used for dynamic evaluation and compensation to generate a primary adjustment amount for beam pointing, which may include:

[0078] In this embodiment of the invention, the acquisition of feature deviation values ​​recorded by multiple spatial sampling units at the current moment and the construction of a pointing deviation dataset specifically includes: First, determining the current time node; for each previously divided spatial sampling unit, starting the on-board data acquisition module to collect the feature deviation values ​​within the unit in real time. The feature deviation values ​​include the azimuth angle deviation value and pitch angle deviation value corresponding to the unit. The acquisition frequency is once every 0.5 seconds, and the acquisition is carried out 10 times in a row to obtain 10 sets of real-time deviation data for each unit; for spatial sampling units located in the equatorial magnetic disturbance region, in addition to the 10 sets of data collected in real time, 20 sets of historical deviation data within the past 5 minutes of the unit are retrieved from the on-board historical database to avoid distortion of instantaneous deviation data caused by magnetic disturbance; for sampling units corresponding to ship navigation areas affected by sudden strong ocean currents, 5 sets of deviation data from each of the 3 adjacent units around the unit are collected to verify the rationality of the current unit data.

[0079] Subsequently, a pointing deviation dataset is constructed. Using the independent identifier of the spatial sampling unit as an index, the real-time deviation data, supplementary historical deviation data, and adjacent unit verification data of each unit are correlated to form a structured dataset of unit identifier - azimuth deviation value set - pitch deviation value set. For example, the key area unit identified as 708 contains 10 sets of real-time azimuth deviation values, 20 sets of historical azimuth deviation values, and 15 sets of azimuth deviation verification data of surrounding units. The construction logic of the pitch deviation value set is consistent with that of the azimuth deviation set.

[0080] Statistical analysis is performed on the pointing deviation dataset, calculating the statistical mean and variance of pointing deviation within multiple spatial sampling units to form corresponding statistical feature vectors. Specifically, this includes: performing statistical analysis on the deviation value set of each spatial sampling unit in the pointing deviation dataset, first calculating the statistical mean of pointing deviation, and then calculating the variance. The specific process is as follows: Calculating the statistical mean: Taking the azimuth deviation value set of a certain spatial sampling unit as an example, first, all azimuth deviation values ​​in the set are added sequentially to obtain the sum of the azimuth deviation values ​​of that unit; then, this sum is divided by the total number of deviation values ​​to obtain the statistical mean of the azimuth pointing deviation of that unit. The calculation logic for the pitch statistical mean is exactly the same, that is, all data in the pitch deviation value set of that unit are added together, and then divided by the total number of pitch deviation values ​​to obtain the statistical mean of the pitch pointing deviation.

[0081] Variance calculation, taking the set of azimuth deviation values ​​of the unit as an example, first subtracts the statistical mean of the azimuth of the unit from each azimuth deviation value in the set to obtain the difference between each deviation value and the mean; then squares each difference to obtain the squared difference of each deviation value; next, sums all the squared differences to obtain the sum of the squared differences; finally, divides this sum by the total number of deviation values ​​to obtain the variance of the azimuth pointing deviation of the unit. The calculation logic of the pitch variance is the same as that of the azimuth, that is, by subtracting the mean from the deviation value, squaring, summing, and dividing by the number, the variance of the pitch pointing deviation is obtained.

[0082] After calculating the mean and variance for each unit, the azimuth mean, azimuth variance, pitch mean, and pitch variance for each unit are arranged sequentially according to the identification order of the spatial sampling units, forming the corresponding statistical feature vectors. For example, the azimuth mean of unit 708 is 0.35 degrees, the azimuth variance is 0.02 degrees², the pitch mean is 0.28 degrees, and the pitch variance is 0.015 degrees². The corresponding mean and variance for unit 709 are arranged sequentially, ultimately forming a feature vector sequence containing statistical data from all units.

[0083] Based on the statistical feature vector, the process noise covariance and measurement noise covariance parameters of the adaptive Kalman filter algorithm are dynamically configured. Specifically, this includes dynamically configuring the process noise covariance and measurement noise covariance parameters of the adaptive Kalman filter algorithm based on the mean and variance of each spatial sampling unit in the statistical feature vector. The specific configuration logic is as follows: Process noise covariance configuration. The process noise covariance reflects the uncertainty of the system. When the deviation variance of a spatial sampling unit is large, it indicates that the deviation fluctuation is severe, and the uncertainty of the system increases, requiring an increase in the process noise covariance. The specific calculation method is that the process noise covariance value is equal to the deviation variance of that unit. Multiply by a fluctuation coefficient, which is determined based on the fluctuation frequency of the deviation. If the fluctuation frequency of the unit deviation is 3 times every 10 seconds (magnetic disturbance scenario), the fluctuation coefficient is 1.2; if the fluctuation frequency is 1 time every 20 seconds (non-disturbance scenario), the fluctuation coefficient is 0.8. For example, the azimuth process noise covariance of the unit in the magnetic disturbance region is 0.02 degrees² × 1.2 = 0.024 degrees²; the azimuth process noise covariance of the unit in the non-disturbance region is 0.005 degrees² × 0.8 = 0.004 degrees². The configuration logic of the pitch process noise covariance is consistent with that of the azimuth, that is, it is calculated by multiplying the unit deviation variance by the fluctuation coefficient.

[0084] Measurement noise covariance configuration is crucial because it reflects the uncertainty of measurement data. A large difference between the statistical mean of the deviation of the spatial sampling unit and the initial expected value of the beam pointing indicates a significant deviation from the ideal state, increasing measurement uncertainty and necessitating an increase in the measurement noise covariance. Specifically, the measurement noise covariance is calculated as the absolute value of (statistical mean of deviation - initial expected value) multiplied by a deviation coefficient. This coefficient is determined based on the range of the difference between the mean and the expected value: a coefficient of 1.1 for a difference greater than 0.3 degrees and a coefficient of 0.9 for a difference less than 0.1 degrees. For example, for a unit where the difference between the mean and the expected value is 0.35 degrees, the azimuth measurement noise covariance is 0.35 degrees × 1.1 = 0.385 degrees²; for a unit where the difference is 0.08 degrees, the azimuth measurement noise covariance is 0.08 degrees × 0.9 = 0.072 degrees². The configuration logic for the pitch measurement noise covariance is consistent with that for the azimuth, and it is calculated by multiplying the absolute value of the difference between the mean and the expected value by the deviation coefficient. Finally, the process noise covariance and measurement noise covariance parameters of each spatial sampling unit are stored according to the unit identifier and used as input parameters for the adaptive Kalman filter algorithm.

[0085] An adaptive Kalman filter algorithm with configured parameters is used to filter the statistical mean of the pointing deviations of multiple spatial sampling units at the current time to obtain a real-time estimate. Specifically, the algorithm first initializes the state variables of the adaptive Kalman filter algorithm, setting the initial state variable value to the real-time estimate of the deviation of the spatial sampling unit at the previous time. Then, it enters the prediction stage of the filtering process. Based on the state estimate of the previous time and the system state transition matrix, the predicted state value at the current time is calculated, i.e., the current predicted state value = the state estimate of the previous time × 1.0 + process noise. For example, if the azimuth state estimate of the previous time is 0.3 degrees, the process noise covariance is 0.024 degrees², and the randomly generated process noise is 0.02 degrees, then the current predicted azimuth state value = 0.3 degrees × 1.0 + 0.02 degrees = 0.32 degrees.

[0086] The process then enters the update phase of the filtering process. The calculated mean deviation of the unit at the current moment is used as the measured value. Combined with the configured measurement noise covariance, the Kalman gain is calculated (Kalman gain = predicted state covariance × (predicted state covariance + measurement noise covariance) reciprocal, where the predicted state covariance is the previous state covariance × 1.0 + process noise covariance). For example, if the azimuth state covariance at the previous moment is 0.02 degrees² and the process noise covariance is 0.024 degrees², then the predicted state covariance = 0.02 degrees² × 1.0 + 0.024 degrees² = 0.044 degrees²; and the measurement noise covariance is 0.385 degrees², then the Kalman gain = 0.044 degrees² ÷ (0.044 degrees² + 0.385 degrees²) ≈ 0.102.

[0087] Finally, the predicted state value is updated using the Kalman gain to obtain the real-time estimate for the current moment. The real-time estimate = predicted state value + Kalman gain × (measured value - predicted state value). For example, if the predicted state value is 0.32 degrees, the measured value is 0.35 degrees, and the Kalman gain is 0.102, then the real-time azimuth estimate = 0.32 degrees + 0.102 × (0.35 degrees - 0.32 degrees) ≈ 0.323 degrees. For each spatial sampling unit, filtering is performed according to the above initialization, prediction, and update process to obtain the real-time estimate of the azimuth and pitch deviations for each unit. Especially in equatorial magnetic disturbances or sudden strong ocean currents, the filtering algorithm can smooth out the drastic fluctuations in the deviation data through dynamically configured noise covariance, avoiding large jumps in the estimated value.

[0088] Based on real-time estimates, the initial adjustments for the azimuth and elevation dimensions of the beam pointing are calculated comprehensively. Specifically, this includes: First, determining the target reference values ​​for beam pointing. The azimuth target reference value is the initially expected azimuth angle of the beam pointing, and the elevation target reference value is the initially expected elevation angle of the beam pointing. Then, assigning weight coefficients to each spatial sampling unit. Spatial sampling units in critical areas have a greater impact on energy transmission due to deviations, so their weight coefficient is set to 1.2; units in non-critical areas have a weight coefficient of 0.8. The weight coefficients are set based on the degree of deviation influence in the area where the unit is located; critical areas have a higher degree of deviation influence, so their weight coefficients are larger. The initial adjustment for the azimuth dimension is calculated by first weighting and averaging the real-time azimuth deviation estimates of all spatial sampling units. The method involves multiplying the real-time azimuth estimate of each cell by its weighting coefficient to obtain a weighted estimate for each cell. Then, the weighted estimates of all cells are summed sequentially to obtain a total weighted estimate. Simultaneously, the weighting coefficients of all cells are summed sequentially to obtain a total weight. Finally, the total weighted estimate is divided by the total weight to obtain a weighted average estimate of the azimuth deviation. The initial azimuth adjustment is calculated as: azimuth target reference value - weighted average estimate of the azimuth deviation. For example, if the azimuth target reference value is 120.0 degrees and the weighted average estimate is 0.32 degrees, then the initial azimuth adjustment is 120.0 degrees - 0.32 degrees = 119.68 degrees. This means the beam azimuth needs to be adjusted to 119.68 degrees to compensate for the 0.32-degree deviation.

[0089] The logic for calculating the primary adjustment in the pitch dimension is the same as that in the azimuth dimension. First, calculate the weighted average estimate of the real-time pitch deviation estimates of all units (each unit's pitch deviation estimate × weight coefficient, summed and divided by the total weight). Then, subtract this weighted average estimate from the target reference value in the pitch direction to obtain the primary adjustment in the pitch dimension. For example, if the target reference value in the pitch direction is 30.0 degrees and the weighted average estimate is 0.28 degrees, then the primary adjustment in the pitch dimension = 30.0 degrees - 0.28 degrees = 29.72 degrees. Finally, integrate the primary adjustment in the azimuth dimension and the primary adjustment in the pitch dimension to form a complete primary beam pointing adjustment, which is used for verification and secondary correction.

[0090] By collecting and supplementing deviation data under magnetic disturbance and strong ocean current scenarios in real time to construct a dataset, the problem of incomplete deviation analysis caused by single data collection is reduced. By calculating the mean and variance of each unit in detail, a precise basis for the configuration of filter parameters is provided, avoiding the defect of fixed parameter filtering being out of touch with actual deviation characteristics. Based on the deviation statistical characteristics, the noise covariance of Kalman filtering is dynamically configured, enabling the filtering algorithm to adaptively cope with deviation fluctuations caused by magnetic disturbance and strong ocean current, improving the stability of deviation estimation and ensuring the stability of energy transmission.

[0091] In a preferred embodiment of the present invention, the primary adjustment amount is verified, the compensation residual is calculated, and the primary adjustment amount is further corrected based on the compensation residual to finally obtain the adjusted beam pointing parameters, which may include:

[0092] In this embodiment of the invention, the initial adjustment values ​​of azimuth and pitch are input into the beam pointing control parameters to achieve the initial correction of the beam pointing. Specifically, this includes: First, extracting the initial adjustment values ​​of azimuth and pitch from the adjustment value generation, clarifying their specific values, and specifying that the beam pointing control parameters include azimuth control threshold, pitch control threshold, parameter activation sequence, adjustment rate limit, etc. The initial adjustment value of azimuth is input into the azimuth target control value field of the control parameters, and the initial adjustment value of pitch is input into the pitch target control value field. In the magnetic disturbance region, to avoid mutual interference between azimuth and pitch axis adjustments, the activation order of the control parameters is set to azimuth first, then pitch. That is, the initial adjustment value of azimuth is written into the satellite attitude control circuit first, and after the azimuth adjustment is stable, the initial adjustment value of pitch is written into the satellite attitude control circuit. If a sudden strong ocean current is detected, the adjustment rate limit is temporarily increased from 0.5 degrees / second to 1 degree / second to ensure that the initial correction can quickly respond to the dynamic drift of the ship.

[0093] Simultaneously, a real-time monitoring mechanism for control parameters is activated, reading the actual effective value of the control parameters every 0.1 seconds and comparing it with the input primary adjustment amount. If the difference between the two exceeds 0.03 degrees, a calibration command is immediately sent to correct the effective value to the primary adjustment amount, avoiding correction deviations caused by circuit transmission delays, and finally completing the initial correction of beam pointing.

[0094] After the initial correction, the satellite's actual attitude data is reacquired using onboard sensors, and the new actual beam pointing state is calculated based on the updated attitude data. Specifically, this involves: immediately after the initial correction, initiating a multi-sensor collaborative acquisition process, with an acquisition duration of 5 seconds and a acquisition frequency of once every 0.2 seconds, acquiring a total of 25 sets of actual satellite attitude data. The sensors used include fiber optic gyroscopes, star sensors, and magnetometers. The specific acquisition and processing process is as follows: Fiber optic gyroscope data acquisition: angular velocity data for the roll axis, pitch axis, and yaw axis are acquired separately. Each set of data contains angular velocity values ​​for three axes. After acquisition, abnormal data exceeding the normal range is first removed, and then the remaining valid data are summed along each axis to obtain the sum of the angular velocities for each axis. The sum is then divided by the... The number of effective data points is used to obtain the average angular velocity for each axis. The average angular velocity is then multiplied by the acquisition time to obtain the angular change of each axis relative to the attitude after the initial correction, which serves as the base value for the attitude data. For star sensor data acquisition, one image of a star in space is captured every 0.2 seconds, for a total of 25 images. The pixel coordinates of the stars in each image are compared with the built-in star database to identify at least 5 matching stars. The satellite zenith angle and azimuth angle corresponding to each image are calculated. The average zenith angle and azimuth angle of the 25 images are taken as the attitude data measured by the star sensor and used to calibrate the angular change of the gyroscope. Specifically, the calibrated zenith angle is obtained by subtracting the zenith angle change calculated by the gyroscope from the average zenith angle of the star sensor. Similarly, the calibrated azimuth angle is obtained.

[0095] Magnetometer data acquisition and correction: In the equatorial magnetic disturbance region, real-time magnetic field strength data is collected from the magnetometer, 25 sets every 0.2 seconds, and the average value is calculated as the real-time magnetic field data. Historical average magnetic field data for the satellite's orbital region is retrieved from the onboard historical database. The real-time magnetic field data is subtracted from the historical average magnetic field data to obtain the magnetic field deviation. The magnetic field deviation is then multiplied by a correction factor of 0.0001 degrees / nano to obtain the attitude error correction value caused by the magnetic disturbance. This correction value is superimposed on the attitude data after calibration by the gyroscope and star sensor to obtain the final actual attitude data. Finally, a new actual pointing state is calculated based on the actual attitude data. Combined with the satellite's current orbital parameters, the actual attitude data is converted into the satellite's attitude direction in the inertial coordinate system. The real-time relative geometric relationship between the satellite and the target ship is then calculated. Finally, the relative geometric relationship is combined with the satellite's attitude direction to calculate the new azimuth and pitch actual pointing angles of the beam, forming the new actual pointing state.

[0096] The new actual pointing state is compared with the initial expected beam pointing value to calculate the residual pointing error. Based on the residual pointing error, the distribution characteristics and variation law of the residual error are analyzed to obtain the error analysis results. Specifically, this includes: First, extracting the initial expected beam pointing value and the new actual pointing state, and calculating the residual pointing errors in the azimuth and elevation directions respectively. The residual pointing error is calculated by subtracting the initial expected azimuth angle from the new actual pointing angle to obtain the azimuth residual pointing error; and by subtracting the initial expected elevation angle from the new actual pointing angle to obtain the elevation residual pointing error. Subsequently, the distribution characteristics and variation law of the residual pointing error are analyzed. The variation patterns were analyzed, and the residual errors in the azimuth and elevation directions were divided into intervals of 0.05 degrees. The frequency of residual errors in each interval was counted. If the frequency of a certain interval exceeded 30%, the residual errors were determined to be concentrated in that interval, which is a systematic deviation. The variation trend of residual errors was plotted in chronological order to observe whether the error showed a specific trend over time. If the absolute value of the error gradually increased over time, it was determined that there was a cumulative deviation. If the error fluctuated within a certain range, it was determined that there was a fluctuating deviation. The analysis results of the distribution characteristics and variation patterns were compiled to form the error analysis results, clarifying the type, concentration interval and variation trend of residual errors.

[0097] Based on the error analysis results, the correction amount for the primary adjustment is calculated, and the primary adjustment amount and the correction amount are integrated to generate the final beam pointing parameters. Specifically, if the residual error is a systematic deviation, the median value of the interval is taken as the correction amount. The azimuth correction amount is the absolute value of -0.275 degrees, because the residual error is negative, indicating that the actual pointing is too small, and the primary adjustment amount needs to be increased to offset the deviation. Similarly, for the pitch direction, if the residual error is concentrated in the range of -0.27 degrees to -0.22 degrees, with a median value of -0.245 degrees, the pitch correction amount is 0.245 degrees. If the residual error is a cumulative deviation, the cumulative rate of the error is first calculated, and then the correction amount is calculated as the cumulative rate × the correction validity period, based on the expected correction validity period, to offset the cumulative deviation within the next 10 seconds. If the residual error is a fluctuating deviation, the maximum value of the fluctuation range is taken as the correction amount to ensure that the correction amount can cover the maximum fluctuation deviation.

[0098] The initial adjustment and correction values ​​are added together to obtain the preliminary integrated parameters. Then, it is verified whether the integrated parameters are within the maximum allowable range of satellite attitude adjustment. If they are within the range, they are determined as the final beam pointing parameters. If they are outside the range, the parameters are adjusted to the maximum allowable value, and the overshoot reason is recorded for optimizing the error analysis logic. Finally, the final beam pointing parameters are stored in the on-board parameter database and sent to the control unit of the beam transmitter for energy transmission control.

[0099] By inputting the initial adjustment amount into the control parameters and monitoring its effectiveness in real time, the initial correction deviation caused by parameter transmission delay is avoided. It can respond quickly, especially in the case of sudden strong ocean currents, ensuring that the initial correction is in line with the ship's dynamics. By collecting and calibrating attitude data in collaboration with multiple sensors, the influence of magnetic disturbances on the sensors is effectively corrected, reducing the pointing calculation error caused by the distortion of data from a single sensor.

[0100] In a preferred embodiment of the present invention, based on the adjusted beam pointing parameters, the on-board beam transmitting device is driven, and the beam transmission performance index is fed back in real time, forming a closed-loop on-board autonomous decision-making and control process to achieve stable energy transmission to moving ships, which may include:

[0101] In this embodiment of the invention, beam control commands are generated based on the final beam pointing parameters, and the on-board beam transmitter is driven to perform pointing operations to form a transmission beam pointing towards the moving ship. Specifically, this includes: first, extracting the final determined beam pointing parameters, including the azimuth target angle and the pitch target angle, and converting these two angle parameters into drive signal parameters for the beam transmitter. The drive signal parameter corresponding to the azimuth target angle is the angle value multiplied by the conversion coefficient of the azimuth axis of the beam transmitter to obtain the number of azimuth drive pulses; similarly, the pitch target angle is multiplied by the pitch axis conversion coefficient to obtain the number of pitch drive pulses.

[0102] Dynamic response parameters are incorporated when generating beam control commands. If the ship's real-time speed exceeds 15 knots, a 5% compensation pulse is added to the number of drive pulses to ensure the beam can quickly track the ship's movement. If equatorial magnetic disturbance is detected, the output voltage of the drive signal is increased from 12 volts to 15 volts to enhance the motor drive torque and avoid drive jamming caused by magnetic disturbance. The generated control commands are sent to the drive circuit of the on-board beam transmitter. The drive circuit controls the azimuth and pitch axis motors to rotate according to the commands, driving the beam transmitting antenna to adjust its pointing. At the same time, the angle sensor built into the device provides real-time feedback of the actual rotation angle. Every 0.1 seconds, the feedback angle is compared with the target angle. If the difference exceeds 0.02 degrees, a compensation command is immediately sent until the difference between the actual pointing angle and the target angle is less than 0.02 degrees, ultimately forming a transmission beam that accurately points to the moving ship.

[0103] Based on the transmission beam, downlink signal parameters are monitored in real time, and transmission performance indicators reflecting the receiving status of the ship terminal are collected to obtain the energy reception efficiency data at the current moment. Specifically, after the transmission beam is formed, the on-board signal monitoring and data backhaul link with the ship terminal are activated. Downlink signal parameters are monitored in real time, including received signal strength indication, signal-to-noise ratio, signal transmission delay, and signal jitter amplitude. When collecting transmission performance indicators reflecting the receiving status of the ship terminal, the dynamic characteristics of the moving ship are considered. When the ship is sailing at a constant speed, the above signal parameters are collected once every 1 second; when the ship is affected by strong ocean currents and accelerates or turns, the collection frequency is increased to once every 0.2 seconds. Next, ensure the capture of rapidly changing reception status; calculate the energy reception efficiency data at the current moment. The energy reception efficiency is equal to the actual signal power received by the ship terminal divided by the satellite beam's transmission power, and then multiplied by 100 to obtain a percentage value. For example, if the satellite transmission power is 1000 milliwatts and the ship's receiving power is 850 milliwatts, then the energy reception efficiency is 850 milliwatts divided by 1000 milliwatts multiplied by 100, which equals 85%. At the same time, correlate the signal-to-noise ratio, transmission delay, jitter amplitude, and energy reception efficiency to form a performance index dataset containing multi-dimensional information. When the signal-to-noise ratio is below 10 dB or the jitter amplitude is greater than 5 milliseconds, it is marked as an performance anomaly and used for parameter correction.

[0104] Energy reception efficiency data is used as real-time observations and input into a neural network model and an adaptive Kalman filter algorithm for joint state identification and parameter correction, generating parameter correction values. Specifically, the following steps are taken: First, a neural network model for state identification is constructed. The input layer contains 5 neurons, corresponding to energy reception efficiency, signal-to-noise ratio, transmission delay, jitter amplitude, and real-time ship speed, respectively. Two hidden layers are set: the first layer has 12 neurons, and the second layer has 8 neurons. The output of each neuron is calculated using a sigmoid activation function, mapping the value to between 0 and 1. The output layer contains 2 neurons, outputting the prediction deviation correction values ​​for azimuth and pitch, respectively. The model training data consists of historical data from the past 6 months. The performance index dataset includes scenarios such as magnetic disturbances, strong ocean currents, and ship constant / variable speed. During training, the connection weights between neurons in each layer are adjusted by comparing the difference between the predicted correction and the actual correction. In each iteration, the weights are adjusted to the original weights minus the difference multiplied by the learning rate, which is set to 0.01, until the prediction error is less than 0.05 degrees. The energy reception efficiency data at the current moment (e.g., 85%) and the associated signal parameters (signal-to-noise ratio of 15 dB, latency of 20 ms, jitter amplitude of 2 ms, and speed of 12 knots) are used as real-time observations and input into the trained neural network model to obtain the azimuth prediction correction (e.g., 0.03 degrees) and pitch prediction correction (e.g., 0.02 degrees) output by the model.

[0105] Simultaneously, the same set of real-time observations are input into the adaptive Kalman filter algorithm. The difference between the energy receiving efficiency and the ideal efficiency (95%) is used as the measurement residual (85% minus 95% equals -10%). Combined with dynamically configured noise covariance parameters, the correction amount of the filter output (0.02 degrees in azimuth and 0.01 degrees in pitch) is calculated. The output results of the neural network and the Kalman filter are weighted and integrated. Because the neural network is better at identifying nonlinear interference, it is given a weight of 60%. Kalman filtering provides more stable tracking of linear changes, so it is given a 40% weight. The parameter correction is calculated as follows: the azimuth parameter correction is equal to the neural network azimuth prediction correction multiplied by 60% plus the Kalman filter azimuth correction multiplied by 40% (0.03 degrees × 60% + 0.02 degrees × 40% = 0.026 degrees); the pitch parameter correction is calculated similarly (0.02 degrees × 60% + 0.01 degrees × 40% = 0.016 degrees), and the final integrated parameter correction is obtained.

[0106] Based on the parameter correction, the beam pointing parameters are adjusted in real time in a closed loop, and the adjusted parameters are then reapplied to the beam transmitting device to form a spaceborne autonomous decision-making and control loop, thereby achieving continuous and stable energy transmission to moving ships. Specifically, this includes: based on the parameter correction, a real-time closed-loop adjustment mechanism is initiated, and the adjustment cycle is dynamically set according to the energy receiving efficiency. When the efficiency is higher than 90%, the adjustment cycle is 1 second; when the efficiency is lower than 80% (unstable state, such as when the ship is impacted by strong ocean currents), the adjustment cycle is shortened to 0.2 seconds.

[0107] Each time an adjustment is made, the parameter correction is added to the current beam pointing parameter. The new azimuth pointing parameter is equal to the current azimuth parameter plus the azimuth correction; the new pitch pointing parameter is equal to the current pitch parameter plus the pitch correction. The adjusted parameters are then re-input into the beam control command generation process to generate new drive pulse counts and voltage parameters, driving the beam transmitter to adjust its pointing again. Simultaneously, the energy receiving efficiency after adjustment is continuously monitored. If the efficiency improves, the current correction logic is maintained; if the efficiency does not change significantly or decreases, the weights of the neural network are reduced by 10%, the weights of the Kalman filter are increased by 10%, the parameter correction is recalculated, and the adjustment is repeated, forming an onboard autonomous decision-making and control loop. This ensures that the beam always points accurately at the ship under conditions such as ship movement, magnetic disturbances, and strong ocean currents, achieving continuous and stable energy transmission.

[0108] By dynamically generating beam control commands and calibrating the drive unit in real time, the beam can quickly respond to ship displacement. By monitoring transmission performance indicators in multiple dimensions and dynamically adjusting the acquisition frequency, the performance changes caused by magnetic disturbances and strong ocean currents are accurately captured, avoiding the one-sidedness of monitoring a single parameter. By forming a control loop through real-time closed-loop adjustment, the beam pointing is dynamically optimized, ensuring continuous power supply to key equipment such as cargo ship navigation systems and improving the reliability of ocean voyages.

[0109] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0110] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0111] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An orbit control and management system for shortwave satellites, characterized in that, include: The data acquisition module is used to collect the position data, motion status data, and environmental disturbance data of the moving vessel to form a dynamic tracking input set; The trajectory prediction module is used to input the dynamic tracking input set into a pre-trained neural network model for processing, in order to obtain a prediction result dataset including the ship's future trajectory and position offset. Specifically, it includes: based on the dynamic tracking input set, using the pre-trained neural network model to perform multi-step prediction of the ship's motion trajectory, generating a trajectory sequence of the ship within a preset time period in the future, including the ship's position coordinates at multiple prediction times; based on the trajectory sequence, combined with real-time acquired environmental disturbance data, calculating the estimated position offset of the ship relative to the current satellite orbit at each prediction time; associating and integrating the trajectory sequence with the corresponding position offset, adding timestamp information, and constructing a structured prediction result dataset. The attitude pointing module is used to autonomously generate attitude adjustment commands from the satellite based on the prediction result dataset and calculate the initial expected pointing value of the beam. The onboard actuators adjust the satellite attitude in real time according to the attitude adjustment commands to obtain the initial pointing state of the beam and the corresponding actual pointing deviation data. Specifically, the onboard actuators receive attitude adjustment commands and drive the satellite's attitude control mechanism to perform corresponding roll, pitch, and yaw three-axis adjustments. After the three-axis adjustments are completed, the actual attitude data representing the satellite's position in space is acquired through onboard sensors. Based on the actual attitude data and combined with the geometric relationship between the satellite and the target ship, the actual pointing state of the communication beam is calculated. The actual pointing state is compared with the initial expected pointing value of the beam to calculate the azimuth angle deviation and pitch angle deviation, which constitute the actual pointing deviation data. The deviation analysis module is used to divide the beam coverage area corresponding to the actual pointing deviation data into multiple spatial sampling units. Based on the statistical characteristics of the deviation data in multiple spatial sampling units, an adaptive filtering algorithm is used for dynamic evaluation and compensation to generate the primary adjustment amount of beam pointing. The adjustment generation module is used to verify the primary adjustment, calculate the compensation residual, and perform a secondary correction on the primary adjustment based on the compensation residual, ultimately obtaining the adjusted beam pointing parameters. The control execution module, based on the adjusted beam pointing parameters, drives the onboard beam transmitter and provides real-time feedback on beam transmission efficiency indicators, forming a closed-loop onboard autonomous decision-making and control process to achieve stable energy transmission to moving vessels. Specifically, it includes: generating beam control commands based on the adjusted beam pointing parameters and driving the onboard beam transmitter to perform pointing operations, forming a transmission beam pointing towards the moving vessel; monitoring downlink signal parameters in real-time based on the transmission beam and collecting transmission efficiency indicators reflecting the receiving status of the vessel terminal to obtain current energy reception efficiency data; inputting the energy reception efficiency data as real-time observations into a neural network model and an adaptive Kalman filter algorithm for joint state identification and parameter correction, generating parameter correction values; and adjusting the beam pointing parameters in real-time using the parameter correction values, and reapplying the adjusted parameters to the beam transmitter to form an onboard autonomous decision-making and control loop, achieving continuous and stable energy transmission to moving vessels.

2. The orbit control and management system for shortwave satellites according to claim 1, characterized in that, Based on the prediction dataset, the satellite autonomously generates attitude adjustment commands and calculates the initial expected beam pointing value, including: Based on the ship's future trajectory information and satellite orbit parameters in the prediction result dataset, the three-axis attitude adjustment required by the satellite to establish and maintain energy transmission beam coverage over the target ship is calculated. The three-axis attitude adjustment values ​​are converted into a standard command format to drive the satellite to perform attitude maneuvers, and specific attitude adjustment commands are generated. Based on the expected attitude of the satellite determined by the attitude adjustment command, and combined with the real-time relative geometric relationship between the satellite and the target ship, the initial expected value of the beam pointing that will align the center of the communication beam with the target ship is calculated.

3. The orbit control and management system for shortwave satellites according to claim 2, characterized in that, The beam coverage area corresponding to the actual pointing deviation data is divided into regions to form multiple spatial sampling units, including: Based on the statistical distribution characteristics of the actual pointing deviation data, the key areas within the beam coverage area that require fine-grained compensation are determined. Based on the boundary range of the key area, the granularity of spatial division is defined, and the azimuth span and elevation span of each sampling unit are determined. Based on the azimuth and elevation spans, the beam coverage area is divided into grids in the spatial angular domain; Each grid is assigned an independent identifier and a mapping relationship with the physical space is established, thereby forming multiple spatial sampling units.

4. The orbit control and management system for shortwave satellites according to claim 3, characterized in that, Based on the statistical characteristics of deviation data within multiple spatial sampling units, an adaptive filtering algorithm is used for dynamic evaluation and compensation to generate a primary adjustment for beam pointing, including: Obtain the feature deviation values ​​recorded by multiple spatial sampling units at the current time, and construct a bias dataset; Statistical analysis is performed on the pointing deviation dataset, and the statistical mean and variance of pointing deviation within multiple spatial sampling units are calculated to form the corresponding statistical feature vectors. Based on the statistical feature vector, the process noise covariance and measurement noise covariance parameters of the adaptive Kalman filter algorithm are dynamically configured. An adaptive Kalman filter algorithm with configured parameters is used to filter the statistical mean of the pointing deviation of multiple spatial sampling units at the current time to obtain a real-time estimate. Based on real-time estimates, the primary adjustment amounts for the azimuth and elevation dimensions of the beam pointing are calculated.

5. The orbit control and management system for shortwave satellites according to claim 4, characterized in that, The primary adjustment is verified, the compensation residual is calculated, and the primary adjustment is then corrected a second time based on the compensation residual to obtain the adjusted beam pointing parameters, including: The initial adjustment values ​​for azimuth and elevation are input into the beam pointing control parameters to achieve the initial correction of the beam pointing. After the initial correction, the satellite's actual attitude data is reacquired through onboard sensors, and the new actual beam pointing state is calculated based on the updated attitude data. The new actual pointing state is compared with the initial expected value of beam pointing, the residual pointing error is calculated, and the distribution characteristics and variation law of the residual pointing error are analyzed based on the residual pointing error to obtain the error analysis results. Based on the error analysis results, the correction amount for the primary adjustment is calculated, and the primary adjustment amount and the correction amount are integrated to generate the adjusted beam pointing parameters.

6. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the system as described in any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the system as described in any one of claims 1 to 5.

Images