A shipboard roll damping device for a launch vehicle sea launch vessel and method of use

Abstract

The application discloses a kind of launch vehicle sea launching shipborne roll reduction devices and its use method, belong to the technical field of rocket launching, including dynamic platform, dynamic platform is placed on the hull, dynamic platform is sequentially provided with roll reduction module and static platform, roll reduction module is placed between dynamic platform and static platform, two ends of roll reduction module are respectively hinged with static platform, dynamic platform is mutually separated, and the hull is provided with roll prediction module, and the roll prediction module includes sea state acquisition sensor and prediction hull roll angle system, and the rocket is placed on the static platform.The application uses large-tonnage oil cylinder or electric cylinder as linear actuator, combined with multi-stage damping and high-strength fixed structure, with the characteristics of fast response, large control torque and high reliability, can realize high-precision real-time suppression of platform roll motion during launch vehicle sea launching preparation stage under complex sea conditions.

Description

Technical Field

[0001] This invention relates to the technical field of rocket launch, and in particular to a shipborne anti-roll device for sea-launched carrier rockets and its usage method. Background Technology

[0002] In recent years, with the diversification of my country's space launch missions, sea-based launch technology for carrier rockets has gradually become an important direction in the aerospace field. Compared with land-based rocket launches, sea-based rocket launches have advantages such as higher launch safety, flexible orbital inclination, and controllable recovery area. However, because the launch platform is located in the marine environment, environmental factors such as wind and waves can cause six degrees of freedom motion of the platform. Specifically, sea-based launch vessels typically adopt a slender design, which results in a larger longitudinal moment of inertia, thus naturally possessing strong pitch suppression capabilities, and its interference with the launch mission is relatively weak. However, due to the relatively small lateral restoring moment of the hull, the roll motion is often large in amplitude and fast in response, becoming the core disturbance source restricting the verticality maintenance and launch safety of sea-based rockets.

[0003] The relevant technologies can be referenced from the existing shipborne roll reduction systems, which mainly fall into three categories: First, passive roll reduction devices (such as U-shaped roll reduction tanks), which rely on fluid inertia or rudder surface control to generate counter torque. They have a simple structure but slow response and are difficult to eliminate the effects of roll. Second, active fin plates or hydraulic rudder surface systems, which achieve compensation through servo control. However, these systems are large, the hydraulic mechanisms are complex to maintain, and they consume a lot of energy.

[0004] Currently, domestic and international marine rocket launch platforms still mainly rely on the passive stability and attitude adjustment of the hull structure, lacking a roll reduction scheme that can be controlled in real time and respond quickly.

[0005] Regarding the aforementioned technologies, the applicant has identified the following shortcomings in existing launch platforms: Six-degree-of-freedom or three-degree-of-freedom platforms have complex structures, making it difficult to maintain balance under complex sea conditions; furthermore, the actuators of multi-degree-of-freedom platforms have low load-bearing capacity, making it difficult to support large instruments such as rockets. Traditional hydraulic rudder systems exhibit significant control lag under complex sea conditions, failing to meet the high stability requirements of the platform during launch preparation, thus presenting the following technical challenges:

[0006] 1. Passive roll stabilization systems have slow response and limited adjustment range;

[0007] 2. Hydraulic or mechanical active devices have complex structures and high maintenance costs;

[0008] 3. It cannot achieve high-precision, low-energy-consumption real-time control for roll.

[0009] In conclusion, it is necessary to develop a roll reduction device that is simple in structure, fast in response, and can independently control the roll of the launch platform in order to improve the attitude stability and safety of the sea launch platform under complex sea conditions. Summary of the Invention

[0010] This invention addresses the shortcomings of existing technologies by providing a shipborne anti-roll device for sea-launched carrier rockets and its usage method.

[0011] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0012] A shipborne roll reduction device for sea-launched carrier rockets includes a moving platform mounted on the hull. A roll reduction module and a stationary platform are sequentially mounted on the moving platform. The roll reduction module is positioned between the moving platform and the stationary platform. Both ends of the roll reduction module are hinged to the stationary platform and the moving platform, respectively. The stationary platform and the moving platform are arranged separately. A roll prediction module is mounted on the hull. The roll prediction module includes a sea state acquisition sensor and a system for predicting the hull roll angle. The rocket is mounted on the stationary platform.

[0013] Furthermore, the anti-roll module includes a linear actuator and a support column. The linear actuator and the support column are respectively placed on both sides of the moving platform. The two ends of the linear actuator are respectively hinged to the stationary platform and the moving platform. One end of the support column is fixedly connected to the moving platform, and the end of the support column away from the moving platform is hinged to the stationary platform.

[0014] Furthermore, the roll reduction module includes an actuator control system. The linear actuator is equipped with an actuator thrust detection device and an actuator extension length detection device. The actuator control system is electrically connected to the actuator thrust detection device and the actuator extension length detection device, respectively. The actuator control system is also electrically connected to the sea state acquisition sensor and the hull roll angle prediction system, respectively.

[0015] Furthermore, the static platform is equipped with an angle detection sensor, which is electrically connected to the actuator control system and the hull roll angle prediction system, respectively.

[0016] Furthermore, the moving platform is provided with several heat insulation baffles, which are placed at the inner edge of the linear actuator and the support column.

[0017] A method for using a shipborne roll reduction device for sea-launched carrier rockets includes the following steps:

[0018] S1: The sea state acquisition sensor continuously collects and acquires the current sea state environment, performs preliminary processing and output of the collected data, and stably transmits the processed sea state data to the ship roll angle prediction system.

[0019] S2: The system for predicting the ship's roll angle performs modeling analysis and dynamic calculation on the received sea state data. Combined with the current motion state of the ship, it predicts the attitude change trend of the ship in the future, obtains the roll angle and change trend data of the ship, and outputs the prediction results to the actuator control system in advance.

[0020] S3: The actuator control system calculates the control strategy based on the roll angle and its trend data, generates corresponding control command signals, drives each linear actuator to perform pre-adjustment actions, and makes the linear actuators retract or extend according to the predetermined stroke, so as to actively compensate and adjust the static platform before the hull attitude changes.

[0021] S4: After the linear actuator completes the pre-adjustment action, the actuator thrust detection device and the actuator extension length detection device are used to detect whether the linear actuator has completed the pre-adjustment action. The angle detection sensor is used to detect the current attitude of the static platform in real time, obtain the actual roll angle and attitude deviation data of the static platform, and feed the detection results back to the actuator control system for subsequent control judgment.

[0022] S5: When the detection result shows that the static platform has not reached the preset balance state or still has attitude deviation, the actuator control system recalculates the control command based on the feedback data, continues to drive the linear actuator to perform dynamic adjustment, and cyclically executes the detection and control process until the static platform returns to the target balance state and remains stable.

[0023] Furthermore, in the sensor data acquisition and synchronization processing of S1, the sea state acquisition sensor utilizes an inertial navigation system and a high-precision attitude reference unit built into the ship's center of gravity to synchronously acquire the ship's original roll angle under the action of waves at a sampling frequency of no less than 100Hz. angular velocity and vertical heave displacement The sensor coordinate system data is uniformly transformed to the ship's fixed coordinate system through a coordinate transformation matrix, and a low-pass Butterworth filter is used to eliminate high-frequency noise caused by engine vibration, thereby obtaining a smooth attitude sequence reflecting the characteristics of the sea waves. :

[0024]

[0025] in Indicates the current sampling time. Used to index past raw input data, Used to index previously filtered output data. Represents the feedforward coefficient. Indicates the feedback coefficient;

[0026] The nonlinear dynamic modeling and characteristic analysis of ship roll is based on fluid dynamics theory, establishing the ship's roll under random wave moments. The second-order nonlinear differential equations of motion under these conditions describe the physical relationship between the ship's displacement, restoring moment, and roll angle, providing mechanical constraints for the prediction model:

[0027]

[0028] in, For rotational inertia, The moment of inertia caused by the added mass, The linear damping coefficient is... For drainage volume, It is a statically stable lever arm function. This indicates the active roll reduction control torque. This represents the roll acceleration. Indicates the roll rate;

[0029] An attention-based LSTM-KAN deep learning prediction architecture is constructed, and a hybrid prediction structure is used to process temporally correlated wave sequence data. A forgetting gate is implemented. Input gate and output gate Extracting the periodic and trend characteristics of wave motion, and introducing attention mechanism weights. The core formula for processing the surge characteristics in historical data is as follows:

[0030]

[0031]

[0032] in Used for storing and transmitting long-term temporal characteristics of ship roll motion across time domains This represents the implicit state output at the current moment. and This represents the weight matrix and bias terms generated during network training.

[0033] Furthermore, in S1, the multi-step advance prediction of sea state involves an inherent physical lag time from the linear actuator receiving the signal to completing the extension and retraction action. Using a trained prediction model, the future is recursively calculated. Predicted roll angle at time :

[0034]

[0035] in This represents a deep learning prediction function based on the LSTM-KAN architecture. Represents historical attitude sequence data. This represents the weight matrix of the prediction model. This represents the bias term vector of the prediction model;

[0036] The predicted output acts as a feedforward compensation signal, enabling the anti-roll module to act before the wave impact, thus minimizing the dynamic response lag from the order of magnitude.

[0037] Online real-time correction and adaptive updating of prediction errors are implemented, a sliding window error monitoring mechanism is established, and the residuals between observed and predicted values ​​are calculated in real time.

[0038] in express The predicted residual at time, express The actual observed value at time [time]. Indicates based on Information about time The prior predictions made at each time point are used to smooth the prediction curves in the subsequent time domain through proportional-integral gain, and newly collected sea state data are periodically fed back to the network model for online fine-tuning.

[0039] Furthermore, in S2, the state-space discretization modeling of the controlled object is based on the geometric and topological relationship between the linear actuator and the static platform, transforming complex mechanical motion into a mathematical state-space representation and defining state vectors. Control input For the actuator system instructions, the continuous system is discretized as follows:

[0040]

[0041] in, This incorporates future wave disturbance terms, enabling the control model to predict disturbances. Represents the system state matrix. Represents the control input matrix;

[0042] Construct a multi-objective model predictive control optimization cost function. In each sampling period, the actuator control system in the prediction time domain... Finding the optimal control sequence Its objective function comprehensively considers the leveling accuracy of the static platform, the motion smoothness of the actuator, and the system energy consumption:

[0043]

[0044] Where Q is the attitude deviation weight, R is the control increment weight, and J is the objective function. To predict the time domain, To control the time domain, In order to be in The future predicted at every moment The yaw angle of the stationary platform. To control the increment, representing the change in control command between two adjacent time points, the optimal leveling path is obtained by solving this quadratic programming problem in real time.

[0045] Real-time injection of hard constraints on the physical boundaries of the actuator: During the process of solving for the optimal control quantity, the physical limits of the actuator are explicitly introduced as boundary constraints to ensure the executability of the control commands. Combined with the real-time feedback from the stroke detection device and the thrust detection device, the following constraints are set:

[0046]

[0047] in These represent the physical minimum and maximum values ​​of the extension / retraction stroke of the linear actuator. For the first The current actual length of a linear actuator Preparing the stretching increment for the linear actuator to be executed. To control the absolute value of the change in the command, This is the maximum allowable control increment step size of the system. For the first The real-time thrust output by a linear actuator. Let L be the length of the lever arm of the linear actuator relative to the rotation axis of the stationary platform, where the force application point is located. The maximum ultimate bending moment that the structure can withstand. This is the sum of the total compensating torques generated by all linear actuators on the static platform.

[0048] Furthermore, in S2, the multi-linear actuator cross-coupling leveling and synchronization control is designed for a multi-cylinder fully synchronized drive layout. To eliminate displacement deviations caused by rocket center of gravity shift or internal leakage of linear actuators, a cross-coupling controller is established to calculate the... Position error between one linear actuator and the rest of the linear actuators And superimposed synchronous compensation gain :

[0049]

[0050] For the first The final control command quantity for a linear actuator The original optimal control quantity calculated by the model predictive control algorithm. For the first The cumulative synchronization error of a linear actuator relative to all other linear actuators in the system;

[0051] For robust adaptive correction of rocket launch impact loads, the system utilizes actuator thrust detection devices to record sudden load changes at the moment of launch. When the load slope exceeds a preset threshold, the actuator control system automatically switches from leveling mode to stiffness mode. It calculates the counteracting torque using a disturbance observer and, in conjunction with the thermal protection of the heat insulation baffle, suppresses the static platform vibration caused by the emission reaction force.

[0052]

[0053] This represents the final control command output to the linear actuator. This represents the basic leveling control quantity calculated by the model predictive control algorithm. This indicates the disturbance compensation gain or control gain.

[0054] In summary, compared with the prior art, the beneficial effects of the above technical solution are:

[0055] 1. This application uses a large-tonnage hydraulic cylinder or electric cylinder as a linear actuator, combined with multi-stage vibration reduction and a high-strength fixed structure, which has the characteristics of fast response, large control torque and high reliability. It can achieve high-precision real-time suppression of platform roll motion during the preparation stage of launch vehicle at sea under complex sea conditions.

[0056] 2. This application has the effect of significantly reducing system cost and complexity. By using a single-axis compensation structure with a support column on one side and a multi-cylinder synchronous hydraulic drive on the opposite side, i.e., the support column and the linear actuator are set on the opposite side, the total number of hydraulic cylinders is reduced, thereby significantly reducing the complexity, weight, energy consumption and maintenance cost of the hydraulic system.

[0057] 3. This application significantly improves control accuracy and response speed by adopting linear actuator technology with separate dynamic and static platforms and fully synchronous drive of four cylinders on one side. All linear actuators only perform unidirectional extension and retraction movements without differential reverse actions, which greatly simplifies the control method, improves the synchronization of the system, and significantly shortens the response time of the device, thereby enabling faster and more accurate compensation for hull roll.

[0058] 4. This application significantly improves the system's resistance to sea state disturbances and its roll advance suppression performance. Through a sea state prediction + feedforward compensation control strategy, the device can complete the linear actuator action in advance before the actual disturbance of waves, enabling the static platform to achieve advanced steady-state control and avoiding the lag error that is unavoidable in traditional feedback control. This technology significantly reduces the transient attitude deviation of the static platform, improves the stability margin during the rocket launch preparation stage, and enables the static platform to maintain a near-horizontal attitude under medium and strong sea states, providing higher reliability assurance for the verticality control and launch safety of the launch vehicle.

[0059] 5. This application improves the stability and reliability of the system attitude control by introducing an angle detection sensor on the static platform to achieve real-time monitoring and feedback correction of the actual attitude of the static platform. This avoids the cumulative error that may be caused by relying solely on the prediction model, and enables the static platform to always maintain within the allowable attitude range under complex sea conditions, thereby improving the overall control accuracy and system operation safety.

[0060] 6. This application enhances the safety and durability of the device under rocket launch conditions. The heat insulation baffle effectively blocks the high-temperature radiation generated by the rocket exhaust flame, reduces the risk of performance degradation or structural damage to the support column and hydraulic actuator due to heat, ensures the reliable operation of key actuators, and thus improves the service life and engineering reliability of the anti-roll module in actual launch missions.

[0061] 7. This application improves the reliability and safety of the system. By collecting and analyzing the real-time data and historical records of the thrust and extension stroke of the linear actuator, the accuracy of the control model and the reliability of predictive control can be effectively improved. This enables the system to continuously optimize the control strategy according to different sea conditions, reduce execution errors and improve the attitude stability of the static platform. At the same time, it has the ability to identify and warn of abnormal working conditions, thereby enhancing the overall system's safety, adaptability and reliability in engineering applications. Attached Figure Description

[0062] Figure 1 This is a schematic diagram of the overall structure of an embodiment of the present invention;

[0063] Figure 2 This is a front view of the anti-roll module in an embodiment of the present invention;

[0064] Figure 3 This is a left view of the roll reduction module when the ship rolls to the left in an embodiment of the present invention;

[0065] Figure 4 This is a right view of the roll reduction module when the ship rolls to the right in an embodiment of the present invention.

[0066] Explanation of reference numerals in the attached figures:

[0067] 1. Roll prediction module; 11. Sea state acquisition sensor; 12. Ship roll angle prediction system; 2. Roll reduction module; 21. Actuator control system; 22. Linear actuator; 23. Actuator thrust detection device; 24. Actuator extension length detection device; 25. Support column; 3. Static platform; 31. Angle detection sensor; 4. Moving platform; 5. Rocket; 6. Heat insulation baffle; 7. Hull. Detailed Implementation

[0068] The principles and features of the present invention are described below with reference to all the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0069] This invention discloses a shipborne roll reduction device for sea-launched carrier rockets and its usage method.

[0070] Reference Figures 1 to 4 As shown, a shipborne roll reduction device for sea-launched carrier rockets includes a moving platform 4, which is placed on the hull 7. A roll reduction module 2 and a stationary platform 3 are sequentially arranged on the moving platform 4. The roll reduction module 2 is placed between the moving platform 4 and the stationary platform 3. The two ends of the roll reduction module 2 are respectively hinged to the stationary platform 3 and the moving platform 4. The stationary platform 3 and the moving platform 4 are arranged separately. A roll prediction module 1 is provided on the hull 7. The roll prediction module 1 includes a sea state acquisition sensor 11 and a system 12 for predicting the ship's roll angle. The rocket 5 is placed on the stationary platform 3.

[0071] The roll reduction module 2 includes a linear actuator 22 and a support column 25. The linear actuator 22 and the support column 25 are respectively placed on both sides of the moving platform 4. The two ends of the linear actuator 22 are hinged to the stationary platform 3 and the moving platform 4, respectively. One end of the support column 25 is fixedly connected to the moving platform 4, and the end of the support column 25 away from the moving platform 4 is hinged to the stationary platform 3. The roll reduction module 2 includes an actuator control system 21. The linear actuator 22 is equipped with an actuator thrust detection device 23 and an actuator extension length detection device 24. The actuator control system 21 is electrically connected to the actuator thrust detection device 23 and the actuator extension length detection device 24, respectively. The actuator control system 21 is also electrically connected to the sea state acquisition sensor 11 and the hull roll angle prediction system 12, respectively.

[0072] An angle detection sensor 31 is installed on the static platform 3. The angle detection sensor 31 is electrically connected to the actuator control system 21 and the hull roll angle prediction system 12. Several heat insulation baffles 6 are installed on the moving platform 4. The heat insulation baffles 6 are placed at the inner edge of the linear actuator 22 and the support column 25.

[0073] This invention mainly consists of a ship roll prediction module 1 based on sea state perception and deep learning (sea state acquisition sensor 11, ship roll angle prediction system 12), a ship roll reduction module 2 based on a control system (actuator control system 21, linear actuator 22, actuator thrust detection device 23, actuator extension length detection device 24, support column 25), a static platform 3, an angle detection sensor 31 for the static platform 3, a moving platform 4, and a heat insulation baffle 6. The moving platform 4 is connected to the main body of the ship 7 via the support column 25, while the static platform 3 is separately arranged from the moving platform 4 via the linear actuator 22, achieving active compensation of the platform's attitude.

[0074] Reference Figures 1 to 4 As shown, a method for using a shipborne anti-roll device for sea-launched carrier rockets includes the following steps:

[0075] S1: The sea state acquisition sensor 11 continuously collects and acquires the current sea state environment, performs preliminary processing and output of the collected data, and stably transmits the processed sea state data to the ship roll angle prediction system 12.

[0076] S11: Sea state acquisition sensor 11 data acquisition and synchronization processing;

[0077] Using an inertial navigation system (INS) and a high-precision attitude reference unit (VRU) built into the ship's center of mass, the original roll angle of the ship 7 under the action of waves is synchronously acquired at a sampling frequency of not less than 100Hz. angular velocity and vertical heave displacement The sensor coordinate system data was uniformly transformed to the fixed coordinate system of the ship's hull using a coordinate transformation matrix, and a low-pass Butterworth filter was used to eliminate high-frequency noise caused by engine vibration, thereby obtaining a smooth attitude sequence reflecting the characteristics of the sea waves. :

[0078]

[0079] in Indicates the current sampling time. Used to index past raw input data, Used to index previously filtered output data. Represents the feedforward coefficient. This represents the feedback coefficient.

[0080] S12: Nonlinear dynamic modeling and characteristic analysis of hull roll 7;

[0081] Based on fluid dynamics theory, the system is established for hull 7 under random wave moments. The second-order nonlinear equations of motion under the influence of the hull's displacement, restoring moment, and roll angle are used to describe the physical relationship between these factors, providing mechanical constraints for the prediction model.

[0082]

[0083] in, For rotational inertia, The moment of inertia caused by the added mass, The linear damping coefficient is... For drainage volume, It is a statically stable lever arm function. This indicates the active roll reduction control torque. This represents the roll acceleration. This represents the roll angular velocity. This equation reveals the self-stabilizing characteristics of the system in the open-loop state.

[0084] S13: Construct an LSTM-KAN deep learning prediction architecture based on attention mechanism;

[0085] We utilize a hybrid prediction structure (LSTM-KAN) to process wave sequence data with strong temporal correlation. This is achieved by setting a forget gate. Input gate and output gate This method extracts the periodicity and trend characteristics of wave motion. An attention mechanism weighting is introduced. The core formula for processing the surge characteristics in historical data is as follows:

[0086]

[0087]

[0088] in Used for storing and transmitting long-term temporal characteristics of the ship's hull roll motion across time domains. This represents the implicit state output at the current moment. and This represents the weight matrix and bias terms generated during network training.

[0089] This model can identify the inherent nonlinear laws governing wave evolution, thus maintaining high prediction accuracy under complex sea conditions.

[0090] S14: Enable multi-step advance prediction of sea conditions;

[0091] Considering the inherent physical lag time between the linear actuator 22 receiving the signal and completing the extension / retraction action. Using a trained prediction model, the future is recursively calculated. Time (of which) Predicted roll angle :

[0092]

[0093] in This represents a deep learning prediction function based on the LSTM-KAN architecture. Represents historical attitude sequence data. This represents the weight matrix of the prediction model. This represents the bias term vector of the prediction model.

[0094] The predicted output acts as a feedforward compensation signal, enabling the anti-roll module 2 to act before the wave impact, thereby minimizing the dynamic response lag from the order of magnitude.

[0095] S15: Online real-time correction and adaptive updating of prediction errors;

[0096] Establish a sliding window error monitoring mechanism to calculate the residual between observed and predicted values ​​in real time. ,(in express The predicted residual at time, express The actual observed value at time [time]. Indicates based on Information about time (Prior predictions made at each moment). The prediction curve in the subsequent time domain is smoothed and corrected by proportional-integral gain, and newly acquired sea state data is periodically fed back to the network model for online fine-tuning to ensure that the prediction model is not distorted or drifted when the sea state changes abruptly.

[0097] S2: The hull roll angle prediction system 12 performs modeling analysis and dynamic calculation on the received sea state data, and predicts the attitude change trend of the hull 7 in the future period of time by combining the current motion state of the hull 7, obtains the roll angle and change trend data of the hull 7, and outputs the prediction results to the actuator control system 21 in advance.

[0098] S21: Discretization modeling of the state space of a controlled object;

[0099] Based on the geometric topological relationship between the linear actuator 22 and the static platform 3, the complex mechanical motion is transformed into a mathematical state-space representation. A state vector is defined. Control input For the actuator system instructions, the continuous system is discretized as follows:

[0100]

[0101] in, The future wave disturbance term provided by step S14 enables the control model to predict disturbances. Represents the system state matrix. This represents the control input matrix.

[0102] S22: Construct a multi-objective model predictive control optimization cost function;

[0103] Within each sampling period, the actuator control system 21 in the prediction time domain Finding the optimal control sequence Its objective function comprehensively considers the leveling accuracy of the static platform 3, the motion smoothness of the actuator, and the system energy consumption:

[0104]

[0105] Where Q is the attitude deviation weight, R is the control increment weight, and J is the objective function. To predict the time domain, To control the time domain, In order to be in The future predicted at every moment The yaw angle of the stationary platform. The control increment represents the change in control command between two adjacent time points. The optimal leveling path is obtained by solving this quadratic programming problem in real time.

[0106] S23: Real-time injection of hard constraints on the physical boundaries of the actuator;

[0107] In solving for the optimal control quantity, the physical limits of the actuator are explicitly introduced as boundary constraints to ensure the executability of the control commands. Combining the real-time feedback from the stroke detection device and the thrust detection device, the following constraints are set:

[0108]

[0109] in The physical minimum and maximum values ​​of the extension stroke of the linear actuator 22 are given. For the first The current actual length of the linear actuator 22, The telescoping increment is prepared for execution by the linear actuator 22. To control the absolute value of the change in the command, This is the maximum allowable control increment step size of the system. For the first The real-time thrust output by the linear actuator 22 The lever arm length of the linear actuator 22 relative to the rotation axis of the stationary platform 3 is given by the point of force application of the linear actuator 22. The maximum ultimate bending moment that the structure can withstand. This is the sum of the total compensating torques generated by all linear actuators 22 on the static platform 3.

[0110] S24: Cross-coupling leveling and synchronization control of multiple linear actuators 22;

[0111] For a multi-cylinder fully synchronized drive layout, a cross-coupled controller is established to eliminate displacement deviations caused by the offset of the rocket's center of gravity or internal leakage of the linear actuator 22. The calculation of the... Position error between one linear actuator 22 and the other linear actuators 22 And superimposed synchronous compensation gain :

[0112]

[0113] For the first The final control command quantity of each linear actuator 22 The original optimal control quantity calculated by the Model Predictive Control (MPC) algorithm. For the first The cumulative synchronization error of one linear actuator 22 relative to all other linear actuators 22 in the system.

[0114] The algorithm ensures that the piston rod extension and retraction heights of multiple linear actuators 22 maintain synchronous accuracy when supporting the load of rocket 5, thus avoiding local torsional stress on the static platform 3.

[0115] S25: Robust adaptive correction for rocket 5 launch impact load;

[0116] The actuator thrust detection device 23 records the sudden change in load at the moment of launch. When the load slope exceeds a preset threshold, the actuator control system 21 automatically switches from leveling mode to stiffness mode. The reverse canceling torque is calculated by the disturbance observer, and combined with the thermal protection of the heat insulation baffle 6, the vibration of the static platform 3 caused by the emission reaction force is suppressed.

[0117]

[0118] This indicates the final control command output to the linear actuator 22. This represents the basic leveling control quantity calculated by the model predictive control algorithm. This indicates the disturbance compensation gain or control gain. This strategy ensures that the static platform 3 can maintain horizontal stability at the moment of deorbiting, meeting the launch requirements for precise orbit insertion.

[0119] S3: The actuator control system 21 calculates the control strategy based on the roll angle and the trend of change, generates the corresponding control command signal, drives each linear actuator 22 to perform pre-adjustment action, so that the linear actuator 22 retracts or extends according to the predetermined stroke, and actively compensates and adjusts the static platform 3 before the attitude change of the hull 7 occurs.

[0120] S4: After the linear actuator 22 completes the pre-adjustment action, the actuator thrust detection device 23 and the actuator extension length detection device 24 detect whether the linear actuator 22 has completed the pre-adjustment action, and the angle detection sensor 31 detects the current attitude of the static platform 3 in real time to obtain the actual roll angle and attitude deviation data of the static platform 3, and feeds the detection results back to the actuator control system 21 for subsequent control judgment.

[0121] S5: When the detection result shows that the static platform 3 has not reached the preset balance state or still has attitude deviation, the actuator control system 21 recalculates the control command according to the feedback data, continues to drive the linear actuator 22 to perform dynamic adjustment, and cyclically executes the detection and control process until the static platform 3 returns to the target balance state and remains stable.

[0122] The implementation principle of a shipborne anti-roll device for sea-launched carrier rockets and its usage method according to an embodiment of the present invention is as follows:

[0123] During the sea launch preparation phase, taking a launch vessel with a total length of approximately 150m, a beam of approximately 38m, and a static platform with three sides of approximately 14m as an example, when in sea state 4 and with persistent swells on the sea surface, the device achieves high-precision steady-state control through the following process:

[0124] First, the sea state acquisition sensor 11 synchronously acquires the raw roll angle, angular velocity, and vertical heave displacement at the center of mass of the hull 7, output by the inertial navigation system, at a sampling frequency of no less than 100Hz, and uses a low-pass filter to remove high-frequency mechanical vibration interference above 10Hz. Next, the roll prediction module 1 performs real-time calculations using an attention-based LSTM-KAN deep learning architecture. Due to a physical execution lag of approximately 0.5s to 1.0s, the roll prediction module 1 anticipates the motion trend within the next 1.5 seconds. For example, when it detects that a periodic wave is about to cause the hull 7 to roll 3° to the left, the system outputs this angle parameter to the control terminal in advance.

[0125] Subsequently, the actuator control system 21 generates a compensation command, driving the linear actuators 22 arranged on one side to perform a pre-adjustment action. Based on the side length of the 14m stationary platform 3, the linear actuators 22 need to accurately achieve a vertical elongation displacement of approximately 94.4cm to offset the 3° tilt angle before the wave impact occurs. During the action execution, the system monitors the status of the linear actuators 22 in real time through the actuator thrust detection device 23 and the actuator extension length detection device 24, and works with the angle detection sensor 31 on the stationary platform 3 to capture residual attitude deviations.

[0126] Finally, the anti-roll module 2 adaptively corrects itself based on the feedback data, driving the linear actuator 22 to perform dynamic compensation, ensuring that the static platform 3 always maintains a near-horizontal state under complex sea state disturbances, providing reliable protection for the verticality maintenance and safe deorbiting of the launch vehicle 5.

[0127] This application proposes and implements for the first time a single-axis active roll compensation structure (i.e., linear actuator 22) with a fixed support column 25 on one side and a multi-cylinder synchronous hydraulic drive on the opposite side. Compared with the traditional dual-side multi-actuator differential drive, it reduces the number of actuators, significantly reduces system complexity, weight, energy consumption and maintenance costs, while retaining the advantages of high thrust and high rigidity.

[0128] The launch platform of the sea-based carrier rocket adopts a technology scheme of separation of dynamic platform 4 and static platform 3 + single-side four-cylinder fully synchronous drive. All linear actuators 22 only perform unidirectional telescopic movement (without differential reverse action), which greatly simplifies the control method, improves the system synchronization, and significantly shortens the response time.

[0129] This application provides a complete predictive control compensation system based on active feedforward control using sea state perception and attitude prediction. The device integrates a predictive hull roll angle system 12 and an actuator control system 21, changing the traditional roll reduction device's lag mode of relying solely on real-time attitude sensor feedback for post-accident correction. This system can predict the future roll trend of the hull 7 in advance based on current sea state data, and instruct the linear actuator 22 to perform extension and retraction actions before the wave impact causes significant roll of the hull 7, achieving a leap from passive following suppression to active predictive compensation.

[0130] By arranging angle detection sensors 31 at key positions on the static platform 3, the attitude angle of the static platform 3 relative to the horizontal plane is monitored in real time, and the detection results are fed back to the actuator control system 21, which complements the ship roll angle prediction system.

[0131] On the one hand, the system 12 for predicting the hull roll angle is used to achieve feedforward compensation of the roll trend. On the other hand, the angle detection sensor 31 of the static platform 3 is used to correct the compensation effect in real time, thereby avoiding the cumulative attitude error caused by model error, sudden sea state change or hydraulic execution deviation, and significantly improving the stability and reliability of the attitude control of the static platform 3.

[0132] Heat insulation baffles 6 are installed on the outside of the support column 25 and the retractable linear actuator 22 to form an active protection structure against the high temperature of the rocket's exhaust plume. Targeted thermal protection of key load-bearing and actuating components effectively blocks the direct impact of high-temperature airflow and radiant heat on the hydraulic linear actuator 22 and the support column 25, avoiding the risk of performance degradation, material aging, or failure caused by high temperatures, and ensuring that the anti-roll module 2 can operate safely and stably during and before and after launch.

[0133] An actuator thrust detection device 23 and an actuator extension length detection device 24 are added to the actuator control system 21 to monitor and record the output thrust and extension stroke of each linear actuator 22 in real time during attitude adjustment and rocket 5 launch. The detection data is transmitted to the control module for storage and analysis, thereby establishing the correlation between the operating parameters of the actuator and the attitude response of the platform. This provides data support for subsequent control parameter optimization, model correction and system performance improvement, and realizes the upgrade of the actuator control system 21.

[0134] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A shipborne roll reduction device for sea-launched carrier rockets, characterized in that: The system includes a moving platform (4), which is placed on the hull (7). The moving platform (4) is equipped with a roll reduction module (2) and a static platform (3) in sequence. The roll reduction module (2) is placed between the moving platform (4) and the static platform (3). The two ends of the roll reduction module (2) are respectively hinged to the static platform (3) and the moving platform (4). The static platform (3) and the moving platform (4) are arranged separately. The hull (7) is equipped with a roll prediction module (1). The roll prediction module (1) includes a sea state acquisition sensor (11) and a system (12) for predicting the hull roll angle. The rocket (5) is placed on the static platform (3). The anti-roll module (2) includes a linear actuator (22) and a support column (25). The linear actuator (22) and the support column (25) are respectively placed on both sides of the moving platform (4). The two ends of the linear actuator (22) are respectively hinged to the stationary platform (3) and the moving platform (4). One end of the support column (25) is fixedly connected to the moving platform (4), and the end of the support column (25) away from the moving platform (4) is hinged to the stationary platform (3). The roll reduction module (2) includes an actuator control system (21). The linear actuator (22) is equipped with an actuator thrust detection device (23) and an actuator extension length detection device (24). The actuator control system (21) is electrically connected to the actuator thrust detection device (23) and the actuator extension length detection device (24) respectively. The actuator control system (21) is also electrically connected to the sea state acquisition sensor (11) and the ship roll angle prediction system (12) respectively. An angle detection sensor (31) is provided on the static platform (3). The angle detection sensor (31) is electrically connected to the actuator control system (21) and the hull roll angle prediction system (12). Includes the following steps: S1: The sea state acquisition sensor (11) continuously collects and acquires the current sea state environment, and performs preliminary processing and output of the collected data, and stably transmits the processed sea state data to the prediction ship roll angle system (12). S2: The hull roll angle prediction system (12) performs modeling analysis and dynamic calculation on the received sea state data, and predicts the attitude change trend of the hull (7) in the future period of time in combination with the current motion state of the hull (7), obtains the roll angle and change trend data of the hull (7), and outputs the prediction results to the actuator control system (21) in advance. The state-space discretization modeling of the controlled object is based on the geometric topological relationship between the linear actuator (22) and the static platform (3), which transforms the complex mechanical motion into a mathematical state-space expression and defines the state vector. Control input For the actuator system instructions, the continuous system is discretized as follows: in, This incorporates future wave disturbance terms, enabling the control model to predict disturbances. Represents the system state matrix. Represents the control input matrix. Indicates the current sampling time; Construct a multi-objective model predictive control optimization cost function. In each sampling period, the actuator control system (21) in the prediction time domain... Finding the optimal control sequence within Its objective function comprehensively considers the leveling accuracy of the static platform (3), the motion smoothness of the actuator, and the system energy consumption: Where Q is the attitude deviation weight, R is the control increment weight, and J is the objective function. To predict the time domain, To control the time domain, In order to be in The future predicted at every moment The yaw angle of the stationary platform (3) of the step. To control the increment, representing the change in control command between two adjacent time points, the optimal leveling path is obtained through real-time calculation; Real-time injection of hard constraints on the physical boundaries of the actuator: During the process of solving for the optimal control quantity, the physical limits of the actuator are explicitly introduced as boundary constraints to ensure the executability of the control commands. Combined with the real-time feedback from the stroke detection device and the thrust detection device, the following constraints are set: in The physical minimum and maximum values ​​of the extension stroke of the linear actuator (22) are given. For the first The current actual length of the linear actuator (22), The stretching increment is prepared for the linear actuator (22) to execute. To control the absolute value of the change in the command, This is the maximum allowable control increment step size of the system. For the first The real-time thrust output by the linear actuator (22) The lever arm length of the linear actuator (22) relative to the rotation axis of the stationary platform (3) is the point of force application. The maximum ultimate bending moment that the structure can withstand. The sum of the total compensating torques generated by all linear actuators (22) on the static platform (3); S3: The actuator control system (21) calculates the control strategy based on the roll angle and change trend data, generates the corresponding control command signal, drives each linear actuator (22) to perform pre-adjustment action, so that the linear actuator (22) retracts or extends according to the predetermined stroke, and performs active compensation adjustment on the static platform (3) before the attitude change of the hull (7) occurs. S4: After the linear actuator (22) completes the pre-adjustment action, the actuator thrust detection device (23) and actuator extension length detection device (24) are used to detect whether the linear actuator (22) has completed the pre-adjustment action. The current attitude of the static platform (3) is detected in real time by the angle detection sensor (31) to obtain the actual roll angle and attitude deviation data of the static platform (3). The detection results are fed back to the actuator control system (21) for subsequent control judgment. S5: When the detection result shows that the static platform (3) has not reached the preset balance state or still has attitude deviation, the actuator control system (21) recalculates the control command according to the feedback data, continues to drive the linear actuator (22) to perform dynamic adjustment, and cyclically executes the detection and control process until the static platform (3) returns to the target balance state and remains stable.

2. The shipborne roll reduction device for sea-launched carrier rockets according to claim 1, characterized in that: The moving platform (4) is provided with several heat insulation baffles (6), which are placed at the inner edge of the linear actuator (22) and the support column (25).

3. The method of using a shipborne anti-roll device for sea-launched carrier rockets according to claim 2, characterized in that, In the sensor data acquisition and synchronization processing of S1, the sea state acquisition sensor (11) uses the inertial navigation system and high-precision attitude reference unit built into the center of mass of the hull (7) to synchronously acquire the original roll angle of the hull (7) under the action of the waves at a sampling frequency of not less than 100Hz. angular velocity and vertical heave displacement The sensor coordinate system data is uniformly transformed to the fixed coordinate system of the ship (7) through a coordinate transformation matrix, and a low-pass Butterworth filter is used to eliminate high-frequency noise caused by engine vibration, so as to obtain a smooth attitude sequence reflecting the characteristics of the waves. : in Indicates the current sampling time. Used to index past raw input data, Used to index previously filtered output data. Represents the feedforward coefficient. Indicates the feedback coefficient; The nonlinear dynamic modeling and characteristic analysis of the hull (7) roll is based on fluid dynamics theory, establishing the hull (7) roll nonlinear dynamic modeling and characteristic analysis under random wave moment. The second-order nonlinear motion differential equations under action are used to describe the physical relationship between the ship's (7) displacement, restoring moment and roll angle, providing mechanical constraints for the prediction model: in, For rotational inertia, The moment of inertia caused by the added mass, The linear damping coefficient is... For drainage volume, It is a statically stable lever arm function. This indicates the active roll reduction control torque. This represents the roll acceleration. Indicates the roll rate; An attention-based LSTM-KAN deep learning prediction architecture is constructed, and a hybrid prediction structure is used to process temporally correlated wave sequence data. A forgetting gate is implemented. Input gate and output gate Extracting the periodic and trend characteristics of wave motion, and introducing attention mechanism weights. The core formula for processing the surge characteristics in historical data is as follows: in Used for storing and transmitting long-term temporal characteristics of the hull (7) roll motion across time domains. This represents the implicit state output at the current moment. and This represents the weight matrix and bias terms generated during network training.

4. The method of using a shipborne anti-roll device for sea-launched carrier rockets according to claim 3, characterized in that, In S1, the multi-step advance prediction of sea state involves an inherent physical lag time from the linear actuator (22) receiving the signal to completing the extension and retraction action. Using a trained prediction model, the future is recursively calculated. Predicted roll angle at time : in This represents a deep learning prediction function based on the LSTM-KAN architecture. Represents historical attitude sequence data. This represents the weight matrix of the prediction model. This represents the bias term vector of the prediction model; The predicted output acts as a feedforward compensation signal, enabling the anti-roll module (2) to act before the wave impact, thereby minimizing the dynamic response lag from the order of magnitude. Online real-time correction and adaptive updating of prediction errors are implemented, a sliding window error monitoring mechanism is established, and the residuals between observed and predicted values ​​are calculated in real time. in express The predicted residual at time, express The actual observed value at time [time]. Indicates based on Information about time The prior predictions made at each time point are used to smooth the prediction curves in the subsequent time domain through proportional-integral gain, and newly collected sea state data are periodically fed back to the network model for online fine-tuning.

5. The method of using a shipborne anti-roll device for sea-launched carrier rockets according to claim 4, characterized in that, In S2, the cross-coupled leveling and synchronization control of multiple linear actuators (22) is designed for a layout with fully synchronized multi-cylinder drive. To eliminate displacement deviations caused by rocket (5) center of gravity shift or internal leakage of linear actuators (22), a cross-coupled controller is established to calculate the first... Position error between one linear actuator (22) and the other linear actuators (22) And superimposed synchronous compensation gain : For the first The final control command quantity of each linear actuator (22), The original optimal control quantity calculated by the model predictive control algorithm. For the first The cumulative synchronization error of the linear actuator (22) relative to all other linear actuators (22) in the system; For robust adaptive correction of the launch impact load of the rocket (5), the system uses the actuator thrust detection device (23) to record the load mutation at the moment of launch. When the load slope is detected to exceed the preset threshold, the actuator control system (21) automatically switches from leveling mode to stiffness mode, calculates the reverse offset torque through the disturbance observer, and, in conjunction with the thermal protection effect of the heat insulation baffle (6), suppresses the vibration of the static platform (3) caused by the emission reaction force. This indicates the final control command output to the linear actuator (22). This represents the basic leveling control quantity calculated by the model predictive control algorithm. This indicates the disturbance compensation gain or control gain.

Images