Compensation method and system for marine embarkation gangway based on model predictive control and decoupling

Abstract

The application discloses a marine boarding bridge compensation method and system based on model prediction control and decoupling, and relates to the technical field of offshore wind power operation and boarding, wherein the method comprises the following steps: collecting six-degree-of-freedom motion parameters of the boarding bridge, completing motion decoupling through a dynamic model, and splitting the motion into multiple single-degree-of-freedom subsystems; building a state space prediction model for each subsystem, constructing a rolling optimization problem with the minimum tracking error and control quantity as the target, converting the rolling optimization problem into a quadratic programming problem, and solving the quadratic programming problem by using an active set algorithm to obtain an optimal control sequence; and simultaneously, online correcting the model by means of a radial basis function neural network, compensating for model errors and external disturbances, and guaranteeing the control effect. The application can realize dynamic decoupling of six-degree-of-freedom motion of the marine boarding bridge, improve control accuracy and system robustness, reduce calculation complexity, and meet real-time control requirements.

Description

Technical Field

[0001] This invention relates to the field of offshore wind power operation and maintenance boarding technology, and in particular to a marine boarding bridge compensation method and system based on model predictive control and decoupling. Background Technology

[0002] As a key piece of equipment connecting ships and offshore platforms, marine boarding bridges face complex marine environmental challenges during operation. Environmental loads such as waves and winds can cause six degrees of freedom motion of the ship, resulting in relative displacement and attitude changes between the boarding bridge and the target platform, which seriously affects the safe transfer of personnel and equipment.

[0003] Existing motion compensation methods for boarding bridges directly employ PID control with fixed parameters or fuzzy control based on empirical rules, without fully considering the strong coupling characteristics between the six degrees of freedom motion and the time-varying uncertainty of the marine environment. This may lead to mutual interference between the control commands of each degree of freedom, or model mismatch causing a decrease in compensation accuracy, thereby affecting the safety and efficiency of personnel boarding. Summary of the Invention

[0004] The main objective of this invention is to provide a compensation method for marine boarding bridges based on model predictive control and decoupling.

[0005] Another objective of this invention is to propose a marine boarding bridge compensation device based on model predictive control and decoupling.

[0006] The third objective of this invention is to provide an electronic device.

[0007] The fourth objective of this invention is to provide a non-transitory computer-readable storage medium.

[0008] To achieve the above objectives, a first aspect of the present invention proposes a marine boarding bridge compensation method based on model predictive control and decoupling, comprising:

[0009] S1, obtain the six-degree-of-freedom motion parameters of the boarding bridge, and dynamically decouple the six-degree-of-freedom motion based on the dynamic model, transforming the coupled multi-degree-of-freedom system into multiple independent single-degree-of-freedom subsystems; S2. For each independent single-degree-of-freedom subsystem, a state-space prediction model is established, and a rolling optimization problem is constructed based on the prediction model within the preset prediction time domain and control time domain. S3 transforms the rolling optimization problem into a quadratic programming problem, and uses the activity set algorithm for fast solution to obtain the optimal control sequence that satisfies the control input constraints, state constraints, and output constraints; S4 utilizes a radial basis function neural network to learn model uncertainties online, and corrects the output of the prediction model in real time based on the learning results, compensating for the impact of model errors and external disturbances on control performance.

[0010] Optionally, dynamic decoupling of the six-degree-of-freedom motion based on a dynamic model includes: Obtain the inertia matrix of the boarding bridge and the inertia matrix Perform singular value decomposition:

[0011] in, The inertia matrix, For generalized coordinates, It is a left singular vector matrix. It is a singular value diagonal matrix. for The transpose of the matrix, It is a right singular vector matrix; Decoupling control input is defined based on the decomposition results:

[0012] in, To control the input vector, The matrix of Coriolis force and centrifugal force. The gravity vector For Jacobian matrices, This represents the external disturbance force vector. Transform the system dynamics equations into To achieve dynamic decoupling of six degrees of freedom motion, where for The second derivative with respect to time.

[0013] Optionally, for each independent single-degree-of-freedom subsystem, a state-space prediction model is established. Based on the prediction model, a rolling optimization problem is constructed within a preset prediction time domain and control time domain, including: The dynamic characteristics of the system are described using a state-space model:

[0014]

[0015] in, For state vectors, for The system state vector at any given time. for Predict the state vector at each time step. Here is the state transition matrix. To control the input matrix, To control the input, for Control the input vector at all times. The perturbation input matrix, For interference, for External disturbance vector at time. For output, for The actual output vector of the system at time step, Output matrix for observation; Define the prediction time domain and control time domain Set the optimization objective function:

[0016] in, and This is the weight matrix. and For weighted L2 norm, Use the reference trajectory and set constraints. and .

[0017] Optionally, the rolling optimization problem can be transformed into a quadratic programming problem, including: Transform the optimization problem into and set constraints. ,in For Hessian matrix, The gradient vector, and For constrained matrices and vectors.

[0018] Optionally, model uncertainties can be learned online using radial basis function neural networks, including: The model prediction error is calculated using a radial basis function neural network.

[0019] in, For radial basis functions, As the center vector, As weight, The number of neurons; The weights and covariance matrix are updated using the recursive least squares method:

[0020]

[0021] in, for The vector of parameters to be estimated in the time-matter model. For the updated parameter vector, Let k be the covariance matrix at time k. The updated covariance matrix, For the regression vector, for transpose, for The actual output value of the system at any given time. for The time-based model predicts the output value.

[0022] To achieve the above objectives, a second aspect of the present invention provides a marine boarding bridge compensation device based on model predictive control and decoupling, comprising: The decoupling module is used to obtain the six-degree-of-freedom motion parameters of the boarding bridge, and dynamically decouples the six-degree-of-freedom motion based on the dynamic model, transforming the coupled multi-degree-of-freedom system into multiple independent single-degree-of-freedom subsystems. The optimization module is used to establish a state-space prediction model for each independent single-degree-of-freedom subsystem, and to construct a rolling optimization problem based on the prediction model within a preset prediction time domain and control time domain. The solution module is used to transform the rolling optimization problem into a quadratic programming problem, and uses the activity set algorithm to solve it quickly, so as to obtain the optimal control sequence that satisfies the control input constraints, state constraints and output constraints; The compensation module is used to learn model uncertainties online using radial basis function neural networks, and correct the output of the prediction model in real time based on the learning results, thereby compensating for the impact of model errors and external disturbances on control performance.

[0023] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0024] To achieve the above objectives, a third aspect of this application provides an electronic device, including a processor and a memory; wherein the processor runs a program corresponding to the executable program code stored in the memory, for implementing the marine boarding bridge compensation method based on model predictive control and decoupling as described in the first aspect embodiment.

[0025] To achieve the above objectives, the fourth aspect of this application proposes a non-transitory computer-readable storage medium storing a computer program that, when executed by a processor, implements the marine boarding bridge compensation method based on model predictive control and decoupling as described in the first aspect embodiment.

[0026] The embodiments of the present invention have the following beneficial effects: By transforming the inertia matrix through singular value decomposition, dynamic decoupling of the six-degree-of-freedom motion is achieved, reducing the coupling influence between degrees of freedom and improving control accuracy. A state-space model is used to describe the system's dynamic characteristics, and a fast solution algorithm is designed to meet real-time requirements while ensuring control performance. RBFNN is used to learn model uncertainties online, improving the system's robustness to model errors and external disturbances. Attached Figure Description

[0027] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart of a marine boarding bridge compensation method based on model predictive control and decoupling, provided for an embodiment of the present invention; Figure 2 This is a structural diagram of a marine boarding bridge compensation device based on model predictive control and decoupling, provided in an embodiment of the present invention. Detailed Implementation

[0028] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0029] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0030] The following description, with reference to the accompanying drawings, describes a marine boarding bridge compensation method and apparatus based on model predictive control and decoupling, according to embodiments of the present invention.

[0031] Example 1 This invention provides a compensation method for marine boarding bridges based on model predictive control and decoupling, such as... Figure 1 As shown, the method includes the following steps: S1: Obtain the six-degree-of-freedom motion parameters of the boarding bridge, and dynamically decouple the six-degree-of-freedom motion based on the dynamic model, transforming the coupled multi-degree-of-freedom system into multiple independent single-degree-of-freedom subsystems.

[0032] To overcome the problems of strong coupling and strong interference in the six-degree-of-freedom motion of the marine boarding bridge, this application decomposes the coupled system into independent subsystems through multi-sensor collaborative perception modeling and SVD dynamic decoupling algorithm, providing a reliable model basis for subsequent high-precision motion compensation control.

[0033] In this application embodiment, the application completes the all-round and high-precision collection of the bridge operation status information and marine environment information entirely based on the self-configured sensor module. Through the collaborative work of multiple types of sensing units, it comprehensively obtains complete basic data to support dynamic modeling and dynamic decoupling calculation, ensuring the authenticity and accuracy of subsequent model construction and decoupling calculation.

[0034] In the embodiments of this application, the sensor module configured in this application specifically includes four types of sensing units: multi-axis IMU, LIDAR system, force sensor, and environmental sensors such as anemometer and wave height meter. Each sensing unit has a clear division of labor and works together to achieve full-dimensional monitoring of the motion state, stress state, and external marine conditions of the boarding bridge.

[0035] In this embodiment, a multi-axis IMU is specifically installed at key stress and motion positions of the ship body and the boarding bridge body. It can detect the complete six-degree-of-freedom motion parameters of the boarding bridge in real time and with high precision, accurately capture dynamic information such as random swaying of the ship and changes in the attitude of the boarding bridge itself, and provide raw motion data for subsequent attitude calculation.

[0036] In this embodiment, the LIDAR system continuously monitors the relative spatial position and relative attitude deviation between the walkway and the target docking platform, providing real-time feedback on docking gap and attitude offset, thus offering precise ranging and positioning data for walkway position compensation and attitude correction. Force sensors collect overall load parameters and structural connection stress parameters of the walkway in real time during operation, enabling continuous monitoring of structural stress changes and effectively preventing structural deformation and operational failures caused by overload.

[0037] In the embodiments of this application, environmental sensors such as anemometers and wave height meters collect wind speed and wave height parameters in real time under marine operation scenarios, accurately obtain external interference conditions of complex marine environments, realize all-round perception of marine operation environment, and provide environmental data support for subsequent interference compensation modeling.

[0038] In this embodiment, based on the acquisition of full-dimensional perception data, and combined with the structural characteristics and motion laws of the boarding bridge, a complete system dynamics model adapted to marine dynamic conditions is constructed. This dynamics model can accurately characterize the intrinsic relationship between six-degree-of-freedom coupled motion, structural forces, gravity, and external disturbances. The system dynamics equations are specifically expressed as follows:

[0039] in, For generalized coordinates, Here is the system inertia matrix. The matrix of Coriolis force and centrifugal force. The gravity vector To control the input vector, For Jacobian matrices, This represents the external disturbance force vector.

[0040] Used to uniformly represent the overall position and attitude state of the boarding bridge in six degrees of freedom space, its specific expression is as follows: , , , These correspond to the translational displacement parameters of the bridge's center of mass along the X, Y, and Z axes in the inertial coordinate system, respectively. , , These correspond to the roll, pitch, and yaw angles of the covered bridge around the x, y, and z axes of the body coordinate system, respectively, fully covering all motion forms of the covered bridge in three-way translation and three-way rotation.

[0041] This is the system inertia matrix, used to accurately characterize the mass distribution and motion inertia characteristics of the overall structure of the boarding bridge. The diagonal elements of the matrix... The equivalent mass parameters corresponding to the three-axis translational motion of the covered bridge, and the diagonal elements of the matrix. The rotational inertia parameters corresponding to the three-axis rotational motion of the covered bridge, and the off-diagonal elements of the matrix. Inertial matrix is ​​used to characterize the inertial coupling relationships between different degrees of freedom of motion. The specific values ​​and matrix composition are determined by the overall structural form of the boarding bridge, the mass distribution law, and the selection method of the body coordinate system, which can truly reflect the inherent coupling characteristics of the system.

[0042] The matrix represents the Coriolis force and centrifugal force. This matrix uses the system's generalized coordinates and generalized velocity as variables to characterize the influence of the Coriolis force and centrifugal force generated during the composite rotational motion of the bridge on the system's dynamic characteristics. It can accurately depict the virtual force disturbances caused by multi-degree-of-freedom composite motion and improve the dynamic response characteristics of the dynamic model.

[0043] The gravity vector contains the gravity component in the translational direction and the gravity torque component in the rotational direction. It can comprehensively reflect the influence of gravity on the static equilibrium state and dynamic motion response of the bridge and is a key parameter to ensure the accuracy of static modeling.

[0044] The control input vector is used to characterize the driving force and driving torque of each degree of freedom output by the actuator module of this application, and is the input source for the active control of the system.

[0045] The Jacobian matrix enables velocity mapping and force conversion between generalized coordinate space and Cartesian coordinate space. It allows for unified conversion of motion parameters and force parameters in different coordinate systems, providing support for the unified solution of dynamic equations.

[0046] It is an external disturbance force vector, mainly including external disturbance loads such as wave force and wind force under marine conditions, and accurately covers various disturbance factors of complex marine environment on the movement of the boardwalk.

[0047] In the embodiments of this application, considering the strong coupling, strong disturbance, and nonlinear characteristics of the six-degree-of-freedom motion of the marine boarding bridge, traditional control methods are difficult to achieve high-precision motion compensation. Therefore, this application innovatively adopts a dynamic decoupling method based on SVD singular value decomposition to effectively eliminate the coupling interference between the degrees of freedom and realize the decoupling simplification of the multi-coupled system.

[0048] In this embodiment, the system inertia matrix M(q) is first subjected to singular value decomposition, and the corresponding decomposition formula is:

[0049] in, The left singular vector matrix obtained by decomposition, The singular value diagonal matrix obtained by decomposition, The right singular vector matrix obtained by decomposition, It is the transpose of the right singular vector matrix.

[0050] In this embodiment, based on the singular value decomposition results of the inertia matrix, this application redefines a novel decoupling control input for adaptive decoupling control, and the specific calculation formula is as follows:

[0051] in, To control the input vector, The matrix of Coriolis force and centrifugal force. The gravity vector For Jacobian matrices, This represents the external disturbance force vector.

[0052] In this embodiment, through the singular value decomposition and decoupling control input transformation processes described above, this application can equivalently simplify the originally highly coupled and strongly nonlinear complex system dynamic equations to: This transformation method can completely eliminate the inertial and motion coupling relationships between the six degrees of freedom motions of the boarding bridge, decomposing the originally heavily coupled complex six-degree-of-freedom system into six independent and non-interfering single-degree-of-freedom subsystems. This greatly simplifies the design difficulty and computational complexity of the subsequent controller, effectively avoiding problems such as control lag, attitude overshoot, and insufficient compensation accuracy caused by multi-degree-of-freedom coupling, and laying a solid, accurate, and concise model foundation for subsequent high-precision model prediction and compensation control.

[0053] In this embodiment, a multi-sensor module is used to collect data on the movement of the covered bridge and the environment, and a dynamic model is constructed. The SVD singular value decomposition dynamic decoupling algorithm is then used to eliminate the coupling interference of degrees of freedom, laying the foundation for the subsequent construction of the rolling optimization problem.

[0054] S2. For each independent single-degree-of-freedom subsystem, a state-space prediction model is established. Based on the prediction model, a rolling optimization problem is constructed within the preset prediction time domain and control time domain.

[0055] To achieve high-precision tracking of the corridor bridge trajectory and safe and stable system operation, this application establishes a state-space prediction model and constructs a constrained rolling optimization objective function to provide a basis for solving the control variables.

[0056] In this embodiment, this step is a key step in the predictive modeling and optimization target construction of the motion compensation control system for the marine boarding bridge. Building upon the aforementioned six-degree-of-freedom dynamic decoupling processing results, this application decomposes the strongly coupled six-degree-of-freedom system into multiple independent single-degree-of-freedom subsystems. For each independent subsystem, a high-precision predictive model adapted to its own dynamic characteristics is built, providing a solid model support and computational foundation for the subsequent rolling optimization solution of model predictive control and the calculation of the optimal control sequence.

[0057] In this application embodiment, a precise discrete state-space model is used to fully characterize the dynamic operation law, state change characteristics, and disturbance rejection characteristics of each independent single-degree-of-freedom subsystem. This model can accurately adapt to the time-varying characteristics and dynamic response characteristics of the bridge motion under marine conditions. The state-space model used in this application specifically includes state iteration equations and output observation equations, with the corresponding formulas as follows:

[0058] in, For state vectors, for The system state vector at any given time. for Predict the state vector at each time step. Here is the state transition matrix. To control the input matrix, To control the input, for Control the input vector at all times. The perturbation input matrix is... For interference, for External disturbance vector at time. For output, for The actual output vector of the system at time step, This is the output matrix for observation.

[0059] In this embodiment, to achieve high-precision tracking of the boarding bridge's trajectory while ensuring stable control system operation and suppressing high-frequency jitter and ineffective energy consumption of the actuators, this application innovatively adopts a finite-time domain rolling optimization control strategy, pre-setting a prediction time domain that matches the system's control characteristics. With control time domain Based on the system's preset standard reference trajectory, a rolling optimization objective function adapted to the decoupled subsystem of this application is constructed. The corresponding optimization objective function formula is as follows:

[0060] in, and This is the weight matrix. and For weighted L2 norm, Use the reference trajectory and set constraints. and .

[0061] In this application embodiment, in order to further improve the operational safety, stability and reliability of the control system and avoid abnormal operating conditions such as control overshoot, output overload and mechanism over-limit, this application specifically constructs a comprehensive optimization constraint system to perform all-round limit control on control input and system output.

[0062] In this embodiment, control input constraints are set. The control output is constrained to upper and lower limits to prevent overload of control commands from damaging the actuator; at the same time, system output constraints are set. The system output parameters, such as location, are limited to ensure that the covered bridge always operates within a safe operating range. By setting a multi-dimensional constraint system, the rolling optimization solution process can always be carried out within the safe and feasible domain, taking into account both optimal control effect and system operation safety, and providing standardized constraint conditions for solving the subsequent optimal control sequence.

[0063] In this embodiment, a control model is constructed by building a state-space model and setting rolling optimization objectives and constraints, laying the foundation for obtaining the optimal control sequence.

[0064] S3 transforms the rolling optimization problem into a quadratic programming problem and uses the activity set algorithm for fast solution to obtain the optimal control sequence that satisfies the control input constraints, state constraints, and output constraints.

[0065] To quickly solve for compliant control commands and ensure the safe operation of equipment, this application transforms the optimization problem into a quadratic programming form, uses the activity set algorithm to solve for the optimal control sequence, and implements state monitoring and safety protection based on the controller.

[0066] In this embodiment, this step is the core link in the motion compensation control system of the marine boarding bridge, responsible for solving the control variables. Building upon the optimization model and constraints established in the previous step, it focuses on transforming the optimization problem into its formal form and performing rapid numerical solutions. Ultimately, it outputs the optimal control sequence that meets the requirements, thereby driving the entire actuator to operate stably and complete the motion compensation action of the boarding bridge. In this embodiment, this application combines the computational characteristics of quadratic programming algorithms, performing an equivalent transformation on the constrained rolling optimization objective function established above, uniformly converting it into a standard quadratic programming (QP) solution form. The corresponding optimization calculation formula after the transformation is:

[0067] Simultaneously, corresponding constraints are set: .

[0068] in, For Hessian matrix, The gradient vector and the two work together to construct the standard form of the quadratic programming objective function, ensuring that the mathematical expression of the optimization operation is standardized and consistent. For the constraint matrix, As constraint vectors, this application uses these two sets of parameters to fully encompass all control input constraints, operating state constraints, and output constraints of the system, allowing various constraints to participate in the optimization solution process simultaneously.

[0069] In this embodiment, to improve the efficiency and stability of the solution, the active set algorithm is used to perform iterative calculations on the transformed quadratic programming problem. At the same time, slack variables are introduced to flexibly adapt the hard and soft constraints of the system, which can effectively resolve the conflict between different constraints, avoid problems such as solution interruption and result failure caused by constraint contradictions, and ensure the continuous and reliable solution process.

[0070] In this embodiment of the application, after multiple rounds of iterative calculations, the system can quickly calculate the optimal control increment sequence within the prediction time domain, and then select the optimal control sequence that fully meets all the constraints by combining preset rules.

[0071] In this embodiment, by issuing control commands according to the optimal control sequence, the motion posture and position changes of each degree of freedom of the boarding bridge can accurately follow the preset reference trajectory, successfully complete the basic motion compensation control, and effectively counteract the motion disturbances caused by the marine environment.

[0072] In this embodiment, all computational processing in this stage is independently completed by the controller module. This module can receive and process the system's control input signals in real time, collect operating status data, and verify various output constraints. It also continuously evaluates the real-time operating condition of the entire machine. Once an abnormal operating condition is detected, the controller module will immediately and automatically activate the safety protection mechanism, promptly take emergency measures, and comprehensively ensure the operational safety of the entire boarding bridge equipment during operation.

[0073] In this embodiment, the optimization problem is transformed into a quadratic programming problem and solved using the activity set algorithm to obtain the optimal control sequence to drive the mechanism, laying the foundation for subsequent correction of the prediction model output.

[0074] S4 utilizes a radial basis function neural network to learn model uncertainties online, and corrects the output of the prediction model in real time based on the learning results, compensating for the impact of model errors and external disturbances on control performance.

[0075] To mitigate the impact of model errors and ocean disturbances, this application employs a radial basis function neural network to correct the prediction model online, in conjunction with the actuator and data module, thereby improving the system's control accuracy and environmental adaptability.

[0076] In this embodiment, this step belongs to the error compensation and robust optimization link of the entire marine boarding bridge motion compensation control system. Under the premise that the basic control commands are output normally, it is specifically designed to optimize the dynamic model parameter deviation, system unmodeled error and time-varying external interference such as ocean wind and waves that are common in actual operation.

[0077] In the embodiments of this application, the above-mentioned problems will directly lead to a decrease in the control accuracy of the system. To address this, this application introduces the RBFNN radial basis function neural network, which is used to complete the online adaptive correction of model uncertainty, thereby comprehensively improving the control accuracy of the entire control system and enhancing the system's environmental adaptability and operational stability under complex marine conditions.

[0078] In this embodiment, the system continuously monitors its operating status using a radial basis function neural network during operation and calculates the prediction error generated by the model in real time. The corresponding error calculation formula is as follows:

[0079] in, Represents radial basis functions. This is the center vector corresponding to the neural network. These are the network weight coefficients. The total number of neurons configured for the neural network. This application utilizes the collaborative fitting operation of multiple groups of neurons to comprehensively and accurately identify various model errors in the system, providing a reliable calculation basis for subsequent error compensation work.

[0080] In this embodiment, in order to enable the radial basis function neural network to have continuous online learning capabilities and dynamically adjust its own parameters according to changes in working conditions, this application uses the recursive least squares method to iteratively update the network weights and covariance matrix in real time.

[0081] The formula for updating network weights is:

[0082] The formula for updating the covariance matrix is:

[0083] in, express The weight vector of the neural network at time step. This represents the weight vector at the next time step after iterative updates; for The covariance matrix at time t, This is the updated covariance matrix; For the regression vector, It is the transpose of the regression vector; Is the system in The actual output value at any given time. The predicted output value calculated from the original model.

[0084] In this embodiment, after the system completes error identification and iterative update of neural network parameters, it will perform real-time correction on the output of the original prediction model based on the calculated error amount. The model output correction formula is as follows:

[0085] In this embodiment of the application, relying on this neural network online correction mechanism, the inherent errors of the model itself, parameter drift during operation, and control deviations caused by external ocean disturbances such as waves and wind can be offset in real time, continuously optimizing the calculation accuracy of the prediction model and ensuring that the control output always meets the actual working conditions.

[0086] In this embodiment, the actuator module is responsible for executing various motion compensation commands. The six-degree-of-freedom parallel Stewart platform is specifically used to perform high-precision fine-tuning of the position and attitude of the end of the corridor bridge. The servo hydraulic cylinder system is responsible for driving the main structure of the corridor bridge to achieve large-scale movements such as lifting, pitching, and rotating. The two types of actuators cooperate with each other to accurately respond to the optimal control commands issued by the controller and complete the full-dimensional motion compensation actions.

[0087] In this embodiment, the communication and data processing module provides comprehensive data support for the entire system's operation. The high-speed data acquisition system within the module can simultaneously aggregate and fuse raw data collected by multiple sensors, ensuring real-time data transmission and processing. The wireless communication link can smoothly complete data interaction between the ship and the docking platform, enabling information exchange between the two devices. The data storage unit will completely archive various historical data generated by the system operation, allowing staff to conduct control effect reviews, algorithm performance analysis, and subsequent algorithm iteration and optimization based on the historical data.

[0088] In this embodiment, the model error is corrected online by a radial basis function neural network, and combined with the actuator and data processing module, the high-precision and robust motion compensation of the corridor bridge is achieved.

[0089] Example 2 This invention relates to a marine boarding bridge compensation system based on model predictive control and decoupling, comprising the following components: Sensor modules: Multi-axis IMU (installed at key locations on the ship and boarding bridge) for measuring six-degree-of-freedom motion parameters; LiDAR system for real-time monitoring of the relative position and attitude of the boarding bridge and the target platform; force sensor for monitoring the load and connection force of the boarding bridge; environmental sensors (anemometer, wave height meter, etc.) for acquiring external environmental parameters.

[0090] Controller module: decomposes the coupled six-degree-of-freedom motion into independent subsystems; establishes a system prediction model based on physical models and data-driven methods; uses the MPC algorithm to solve for the optimal control sequence through rolling optimization; simultaneously handles control input, state, and output constraints; evaluates the system state in real time and triggers protection mechanisms.

[0091] Actuator module: A six-degree-of-freedom parallel robot (Stewart platform) enables precise attitude adjustment at the end of the walkway; a servo hydraulic cylinder system drives the heave, pitch, and rotation of the main structure of the walkway.

[0092] Communication and data processing module: High-speed data acquisition system synchronously processes data from multiple sensors; wireless communication link enables data transmission between the ship and the platform; data storage unit records historical operational data for analysis and optimization.

[0093] Example 3 This invention provides a marine boarding bridge compensation device based on model predictive control and decoupling, such as... Figure 2 As shown, the device includes: The decoupling module 100 is used to obtain the six-degree-of-freedom motion parameters of the boarding bridge, and to dynamically decouple the six-degree-of-freedom motion based on the dynamic model, transforming the coupled multi-degree-of-freedom system into multiple independent single-degree-of-freedom subsystems. The optimization module 200 is used to establish a state-space prediction model for each independent single-degree-of-freedom subsystem, and to construct a rolling optimization problem based on the prediction model within a preset prediction time domain and control time domain. The solver module 300 is used to transform the rolling optimization problem into a quadratic programming problem, and uses the activity set algorithm to solve it quickly, so as to obtain the optimal control sequence that satisfies the control input constraints, state constraints and output constraints; The compensation module 400 is used to learn the model uncertainty online using a radial basis function neural network, and correct the output of the prediction model in real time based on the learning results, thereby compensating for the impact of model error and external disturbances on control performance.

[0094] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0095] Example 4 To implement the methods of the above embodiments, the present invention also provides an electronic device, which includes a memory and a processor; wherein the processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the various steps of the methods described above.

[0096] Example 5 To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described in the foregoing embodiments.

[0097] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0098] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0099] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A compensation method for marine boarding bridges based on model predictive control and decoupling, characterized in that, include: S1, obtain the six-degree-of-freedom motion parameters of the boarding bridge, and dynamically decouple the six-degree-of-freedom motion based on the dynamic model, transforming the coupled multi-degree-of-freedom system into multiple independent single-degree-of-freedom subsystems; S2. For each independent single-degree-of-freedom subsystem, a state-space prediction model is established, and a rolling optimization problem is constructed based on the prediction model within the preset prediction time domain and control time domain. S3 transforms the rolling optimization problem into a quadratic programming problem and uses the activity set algorithm for fast solution to obtain the optimal control sequence that satisfies the control input constraints, state constraints, and output constraints. S4 utilizes a radial basis function neural network to learn model uncertainties online, and corrects the output of the prediction model in real time based on the learning results, compensating for the impact of model errors and external disturbances on control performance.

2. The method according to claim 1, characterized in that, The dynamic decoupling of six-degree-of-freedom motion based on a dynamic model includes: Obtain the inertia matrix of the boarding bridge and the inertia matrix Perform singular value decomposition: in, The inertia matrix, For generalized coordinates, It is a left singular vector matrix. It is a singular value diagonal matrix. for The transpose of the matrix, It is a right singular vector matrix; Decoupling control inputs are defined based on the decomposition results: in, To control the input vector, The matrix of Coriolis force and centrifugal force. The gravity vector For Jacobian matrices, This represents the external disturbance force vector. Transform the system dynamics equations into To achieve dynamic decoupling of six degrees of freedom motion, where for The second derivative with respect to time.

3. The method according to claim 1, characterized in that, For each independent single-degree-of-freedom subsystem, a state-space prediction model is established. Based on the prediction model, a rolling optimization problem is constructed within a preset prediction time domain and control time domain, including: The dynamic characteristics of the system are described using a state-space model: in, For state vectors, for The system state vector at any given time. for Predict the state vector at each time step. Here is the state transition matrix. To control the input matrix, To control the input, for Control the input vector at all times. The perturbation input matrix, For interference, for External disturbance vector at time. For output, for The actual output vector of the system at time step, Output matrix for observation; Define the prediction time domain and control time domain Set the optimization objective function: in, and This is the weight matrix. and For weighted L2 norm, Use the reference trajectory and set constraints. and .

4. The method according to claim 1, characterized in that, The transformation of the rolling optimization problem into a quadratic programming problem includes: Transform the optimization problem into and set constraints. ,in For Hessian matrix, The gradient vector, and For constrained matrices and vectors.

5. The method according to claim 1, characterized in that, The uncertainty of the model learned online using radial basis function neural networks includes: The model prediction error is calculated using a radial basis function neural network. in, For radial basis functions, As the center vector, As weight, The number of neurons; The weights and covariance matrix are updated using the recursive least squares method: in, for The vector of parameters to be estimated in the time-matter model. For the updated parameter vector, Let k be the covariance matrix at time k. The updated covariance matrix, For the regression vector, for transpose, for The actual output value of the system at any given time. for The time-based model predicts the output value.

6. A marine boarding bridge compensation device based on model predictive control and decoupling, characterized in that, include: The decoupling module is used to obtain the six-degree-of-freedom motion parameters of the boarding bridge, and dynamically decouples the six-degree-of-freedom motion based on the dynamic model, transforming the coupled multi-degree-of-freedom system into multiple independent single-degree-of-freedom subsystems. The optimization module is used to establish a state-space prediction model for each independent single-degree-of-freedom subsystem, and to construct a rolling optimization problem based on the prediction model within a preset prediction time domain and control time domain. The solution module is used to transform the rolling optimization problem into a quadratic programming problem, and uses the activity set algorithm to solve it quickly, so as to obtain the optimal control sequence that satisfies the control input constraints, state constraints and output constraints; The compensation module is used to learn model uncertainties online using radial basis function neural networks, and correct the output of the prediction model in real time based on the learning results, thereby compensating for the impact of model errors and external disturbances on control performance.

7. An electronic device, characterized in that, Including processor and memory; The processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the method as described in any one of claims 1-5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-5.

Images