A multi-aquatic vehicle combined reconfiguration docking method based on hierarchical predictive control
By modeling the wake as a conical hazardous potential field using a hierarchical predictive control method, and combining high-level planning and low-level robust control, a smooth trajectory that avoids the wake is generated. This solves the problems of wake interference and computational load in AUV self-reconfiguration, and achieves high-precision and stable multi-AUV self-reconfiguration docking.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NORTHWESTERN POLYTECHNICAL UNIV
- Filing Date
- 2026-04-10
- Publication Date
- 2026-06-05
Smart Images
Figure CN122151955A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of underwater robot cooperative control technology, specifically involving a multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control. It is used for trajectory planning and tracking control of follower AUVs dynamically docking with lead AUVs under complex flow field interference in multi-autonomous underwater vehicle (AUV) self-reconfiguration tasks. Background Technology
[0002] As the complexity of ocean exploration missions increases, the limitations of a single AUV in terms of endurance, payload capacity, and operational modes are becoming increasingly apparent. AUV self-reconfiguration technology, by physically connecting and combining multiple independent AUVs underwater to form a linear or other configuration multibody system, can effectively improve the overall propulsion efficiency of the system, achieve energy and data sharing, or accomplish large-scale tasks that a single vehicle cannot handle.
[0003] A review of existing passive sonar target detection and tracking methods reveals that they primarily focus on the target detection stage, while neglecting in-depth discussions of the nonlinear processing and observation delay issues inherent in the target tracking stage. Furthermore, although some track-before-detect techniques based on broadband passive sonar have improved target detection accuracy by processing raw data, these techniques also fail to effectively address the negative impact of observation delay on tracking accuracy. Therefore, current technologies are still insufficient for handling complex AUV self-reconfiguration tasks and cannot fully meet the needs of practical applications.
[0004] Specifically, current self-reconfiguration methods do not consider the wake interference zone, causing AUVs to suffer significant wake interference during the final docking stage, reducing control accuracy and even leading to reconfiguration failure. Meanwhile, end-of-life docking requires extremely high control accuracy and response speed, but traditional single-layer model predictive control, with limited onboard computing resources, struggles to simultaneously meet the demands of long-term time-domain safety planning and high-frequency real-time control. These issues make existing technologies insufficient for meeting the requirements of complex underwater environments and high-precision tracking in practical applications. Summary of the Invention
[0005] The purpose of this invention is to provide a multi-submarine vehicle combination reconfiguration and docking method based on hierarchical predictive control, which aims to solve the problems faced by the following AUV during the self-reconfiguration and docking process, such as large wake interference from the leading AUV, low dynamic tracking accuracy, and high computational load of traditional control methods.
[0006] To achieve the above objectives, the present invention employs the following technical solution: A multi-underwater vehicle combined reconfiguration and docking method based on hierarchical predictive control includes the following steps: Step 1: For the following autonomous underwater vehicle, establish its discrete-time nonlinear dynamic model; based on the motion state of the leading autonomous underwater vehicle, model the wake region generated by its tail thruster as a conical danger potential field that moves with the leading autonomous underwater vehicle, as a dynamic obstacle avoidance constraint. Step 2: Based on the error dynamics model of the follower autonomous underwater vehicle, the optimization problem is solved offline using linear matrix inequalities and semidefinite programming methods to obtain the robust feedback gain matrix and terminal weight matrix, and a robust terminal set with robust positive invariance under bounded perturbations is constructed. Step 3: In each control cycle, the following autonomous underwater vehicle (AUV) acquires its own status and relative pose information with the lead AUV in real time through its onboard sensor system, and updates the wake region at the current moment based on the real-time motion status of the lead AUV. Step 4: Construct a high-level reference trajectory planner; using a nonlinear model predictive control method, with the current state of the following autonomous underwater vehicle and the wake region as input, solve a finite-time domain optimization problem with an artificial potential field term to generate a smooth reference trajectory that avoids the wake and obstacles. Step 5: Construct a robust tracking controller at the bottom layer; adopt a pipe-based robust model predictive control method, take the reference trajectory generated by the high-level reference trajectory planner as the tracking target, combine the robust feedback gain matrix and robust terminal set designed offline, calculate the actual control quantity by solving the tracking control optimization problem and apply it to the following autonomous underwater vehicle. Step 6: Execute steps 3 to 5 in a periodic loop, so that the following autonomous underwater vehicle (AUV) smoothly approaches the leading AUV from the side and rear along the planned path, adjusts its attitude in the terminal phase, and finally completes physical locking with the tail docking mechanism of the leading AUV, realizing the self-reconfiguration docking of multiple AUVs.
[0007] Furthermore, the wake region generated by its tail thruster is modeled as a conical hazardous potential field moving with the leading autonomous underwater vehicle, including: Define the position of the leading AUV thruster as The reverse unit vector of the leading AUV's heading is The half-angle of the cone is The maximum influence length is ;time wake region as follows: ; in, Let be the coordinate vector of any point in space. For three-dimensional real space, For point The projection length onto the reverse heading vector of the leading AUV, superscript This represents the transpose of a vector. This represents the Euclidean norm of a vector.
[0008] Furthermore, based on the error dynamics model of the follower autonomous underwater vehicle, the optimization problem is solved offline using linear matrix inequalities and semidefinite programming methods to obtain the robust feedback gain matrix and terminal weight matrix. A robust terminal set with robust positive invariance under bounded perturbations is then constructed, specifically including: Error dynamics model based on follow-up autonomous underwater vehicle ,in For a moment The state error vector, For a moment The input error vector, Let be a bounded perturbation vector. and Given the linearized matrix of the nominal dynamics model, the following optimization problem is solved offline using linear matrix inequalities and semidefinite programming methods to minimize the terminal set volume: ; in, ; Let the inverse of the terminal Lyapunov matrix be a matrix that satisfies ; For affine feedback gain, satisfying ,in Let be the robust feedback gain matrix to be solved; and These are the weight matrices for the state and the input, respectively; It is the identity matrix; is a scalar variable representing the contraction caused by the disturbance; This is the orientation matrix for the state and input constraints; * denotes a symmetric block in the matrix. Represents the determinant of a matrix.
[0009] Solving the above optimization problem yields the optimal solution. , The robust feedback gain matrix is recovered by the following formula. and terminal weight matrix : ; Based on terminal weight matrix Construct a robust terminal set:
[0010] in, This is the error state vector following the AUV. For the real number space, Let be the dimension of the state vector. This is a preset scalar parameter, and the superscript T indicates transpose.
[0011] Furthermore, a nonlinear model predictive control method is employed, using the current state of the following autonomous underwater vehicle and the wake region as inputs, to solve a finite-time optimization problem with an artificial potential field term, generating a smooth reference trajectory that avoids the wake and obstacles. Specifically, this includes: In each planning cycle, based on the current status of the subsequent AUV and wake region Solve the following optimization problem: ; in, The reference state sequence is to be optimized; The reference input sequence to be optimized; Indicates at time Predicted future The reference state vector of the step; Indicates at time Predicted future The reference input vector for the step; Reference state Reference trajectory points in; The preset docking target state; The prediction time domain length of the planning layer; This is the weight matrix for the state tracking error; The weight matrix for controlling the input; Indicates Weighted quadratic form; Indicates Weighted quadratic form; Let be the artificial potential field function; For the artificial potential field term; The first element of the obtained optimal reference state sequence and optimal reference input sequence are used as the solution. , Reference trajectory for the current cycle Output to the underlying robust tracking control.
[0012] Furthermore, in the process of constructing the underlying robust tracking controller, a set of compressed states and input constraints is introduced. To counteract bounded disturbances Impact: ; in, and The underlying robust tracking controller at time 10:00 Predicted future The state vector and input vector of the step, This is the constraint set obtained by compressing the original state and input constraints.
[0013] Furthermore, the process of building the underlying robust tracking controller also includes: adding robust termination constraints to ensure that the system state converges to the pipeline near the reference trajectory. ; in, For the underlying robust tracking controller at time Predicted future The state vector of the step, For the optimal reference state sequence at time [time] Predicted future The reference state of the step, For robust terminal sets.
[0014] Furthermore, in each control cycle, the tracking control optimization problem with compressed state, input constraints, and robust terminal constraints is solved to obtain the optimal control sequence. The first control variable is then used as the actual thrust to act on the thruster of the following autonomous underwater vehicle to achieve high-frequency robust tracking.
[0015] Furthermore, steps 3 to 5 are executed periodically, in which the high-level reference trajectory planner continuously updates the reference trajectory and actively avoids dynamic wakes and obstacles; while the low-level robust tracking controller is used to continuously calculate control quantities to accurately track the reference trajectory; as the AUV smoothly approaches the leading AUV from the side and rear along the planned path, it adjusts its attitude in the end stage and finally engages with the tail docking mechanism of the leading AUV to complete the physical self-reconfiguration connection and realize the self-reconfiguration docking of multiple AUVs.
[0016] A terminal device includes a processor, a memory, and a computer program stored in the memory; when the processor executes the computer program, it implements the multi-underwater vehicle combined reconfiguration and docking method based on hierarchical predictive control.
[0017] A computer-readable storage medium storing a computer program; when executed by a processor, the computer program implements the multi-underwater vehicle combined reconfiguration and docking method based on hierarchical predictive control.
[0018] Compared with the prior art, the present invention has the following technical features: 1. Solving the wake interference problem in self-reconfiguration: By modeling the propeller wake of the leader AUV as a moving conical repulsive field and combining it with APF for high-level planning, the follower AUV can "predictively" plan a safe approach path that avoids strong turbulence regions, which significantly improves the success rate of self-reconfiguration docking.
[0019] 2. Strong robustness: The underlying layer adopts a pipe-based robust MPC design, which theoretically ensures that the control system still has closed-loop stability when facing uncertain hydrodynamic coefficients and external flow field disturbances, making it suitable for multi-vehicle collaborative operation in complex sea conditions.
[0020] 3. Computationally efficient: The layered architecture effectively separates long-term planning (low frequency) and short-term tracking (high frequency), significantly reducing the online computational burden and meeting the stringent real-time requirements of AUV self-reconfiguration tasks. Attached Figure Description
[0021] Figure 1 This is a schematic diagram of the coordinate system of the AUV model; Figure 2 This is the overall control flowchart of the AUV self-reconfiguration dynamic docking method; Figure 3 This is a diagram of the hierarchical MPC control framework structure; Figure 4 The docking trajectory of the AUV relative to the rover station is shown in (a) for t=10.2s, (b) for t=17.0s, (c) for t=22.2s, (d) for t=30.2s, and (e) and (f) for t=33.4s. Figure 5 It is the relative error of the AUV relative to the mobile station; Figure 6 It is the control input for the AUV's thrusters and servo motors. Detailed Implementation
[0022] This invention provides a multi-underwater vehicle (AUV) combined reconfiguration and docking method based on hierarchical predictive control. By constructing a hierarchical model predictive control (HMPC) framework, it utilizes high-level planning to actively avoid wake turbulence and obstacles, while the low-level controller employs robust tube technology to counteract disturbances, achieving safe and stable docking between the follower AUV and the lead AUV. The specific implementation steps of this invention are as follows: Step 1: For the multi-AUV self-reconfiguration docking scenario, construct the dynamic model of the following AUV; based on the motion state of the leading AUV, model the wake region generated by its tail thruster as a conical danger potential field that moves with the leading AUV, and this region is regarded as part of the dynamic obstacle avoidance constraint.
[0023] Step 1.1, Set up the self-reconfiguration docking task scenario: The leading AUV operates at a speed of Cruise, followed by an AUV approaching from behind; if Figure 1 As shown, a discrete-time nonlinear dynamic model of the AUV is established: (1) in, Indicates time The subsequent AUV state vector includes position, attitude, linear velocity, and angular velocity; For a moment The control input vector; For a moment The bounded perturbation vector; It is a nonlinear state transition function.
[0024] Step 1.2, as follows Figure 2 As shown, to simplify calculations and ensure safety, based on the motion state of the leading AUV, the wake region generated by its tail thruster is modeled as a conical hazardous potential field that moves with the leading AUV; the position of the leading AUV thruster is defined as... The reverse unit vector of the leading AUV's heading is The half-angle of the cone is The maximum influence length is ;time wake region as follows: (2) in, Let be the coordinate vector of any point in space. For three-dimensional real space, For point The projection length onto the reverse heading vector of the leading AUV, superscript This represents the transpose of a vector. This represents the Euclidean norm of the vector; this region is designated as a high-potential-energy repulsion zone in subsequent planning for proactive avoidance.
[0025] Step 2: For the error dynamics model of the follow-up AUV, design the robust feedback gain and terminal weight matrix offline, and construct a robust terminal set: Step 2.1, based on the nominal dynamics model of the follow-up AUV, establish the error dynamics model: (3) in, for The state error vector at time t. For the input error vector, It is a bounded perturbation vector; and This is the linearized matrix of the nominal dynamic model; the modeling of the dynamic model is a well-known technique and will not be elaborated here.
[0026] Step 2.2: Based on this error dynamics model, using linear matrix inequalities and semidefinite programming (SDP), solve the following SDP optimization problem offline to minimize the terminal set volume: (4) in, ; Let the inverse of the terminal Lyapunov matrix be a matrix that satisfies ; For affine feedback gain, satisfying ,in Let be the robust feedback gain matrix to be solved; and These are the weight matrices for the state and the input, respectively; It is the identity matrix; is a scalar variable representing the contraction caused by the disturbance; This is the orientation matrix for the state and input constraints; * denotes a symmetric block in the matrix. Represents the determinant of a matrix.
[0027] The objective function is minimized To shrink the terminal set while explicitly considering perturbations; solving the above optimization problem yields... , and optimal solution , and The robust feedback gain matrix is recovered by the following formula. and terminal weight matrix : (5) in, Represents matrix inversion; terminal weight matrix Used to subsequently define the robust terminal set, and the robust feedback gain matrix This is used for the compact constraint design of the underlying RTMPC to ensure that the system state can converge to the pipeline near the reference trajectory when there is a disturbance.
[0028] Step 2.2, based on the terminal weight matrix Construct a robust terminal set: (6) in, This is the error state vector following the AUV. For the real number space, Let be the dimension of the state vector. The size of the robust terminal set is determined by preset scalar parameters; the solution obtained This is used to quantify the impact of disturbances on system state contraction in robust terminal set design, ensuring that the terminal set has robust positive invariance under bounded disturbances.
[0029] The terminal set is robustly positive invariant under bounded perturbations, ensuring that the system state can converge to the pipe near the reference trajectory.
[0030] Step 3: Real-time status awareness and wake region update.
[0031] In each control cycle The AUV then uses its underwater acoustic positioning system (such as USBL), visual sensors, and DVL to calculate its own state and relative pose with the leading AUV in real time, including relative position, velocity, and attitude information. At the same time, based on the real-time motion state of the leading AUV, the wake region at the current moment is updated according to equation (2) in step 1.2. Spatial location.
[0032] Step 4, High-Level Reference Trajectory Planner (PMPC): Employs a nonlinear model predictive control method to solve a finite-time optimization problem with an artificial potential field term, generating a smooth reference trajectory that avoids wakes and obstacles.
[0033] This step employs nonlinear model predictive control for trajectory planning. An artificial potential field (APF) term is introduced into the cost function to construct a repulsive field targeting the wake cone region of the leader AUV and obstacles in the underwater environment. Unlike traditional hard constraints, the APF guides the AUV to generate a smooth reference trajectory. This trajectory naturally avoids the highly turbulent core region directly behind the leader AUV's thrusters during the approach phase, entering the docking point from a less disturbed location, thus ensuring the safety of the approach process. In each planning cycle, based on the current status of the subsequent AUV and wake region Solve the following optimization problem: (7) in, The reference state sequence is to be optimized; The reference input sequence to be optimized; Indicates at time Predicted future The reference state vector of the step; Indicates at time Predicted future The reference input vector for the step; Reference state Reference trajectory points in; The preset docking target state; The prediction time domain length of the planning layer; This is the weight matrix for the state tracking error; The weight matrix for controlling the input; Indicates Weighted quadratic form; Indicates Weighted quadratic form; Let be the artificial potential field function, when the reference trajectory point The wake region near the leading AUV At this point, the function value increases sharply, forcing the optimizer to generate a detour trajectory; This refers to the artificial potential field term.
[0034] The first element of the obtained optimal reference state sequence and optimal reference input sequence are used as the solution. , Reference trajectory for the current cycle Output to the underlying robust tracking control.
[0035] Step 5, Low-level Robust Tracking Controller (RTMPC): Employs a tube-based robust model predictive control method to calculate the actual thrust at high frequency, enabling the follower AUV to accurately track the upper-level reference trajectory.
[0036] The underlying robust tracking controller designed in this step is responsible for high-frequency tracking of the reference trajectory of the upper-level reference trajectory planner. To cope with the aftershock interference at the wake edge and the surrounding ocean currents, a tube-based robust MPC strategy is adopted. By solving the semidefinite programming (SDP) problem offline, the robust terminal penalty matrix and feedback gain are calculated to construct a robust terminal set. Through state compression and input constraints, it is ensured that as long as the initial error is within the "pipe," the system can maintain recursive feasibility and input state stability under bounded disturbances.
[0037] Step 5.1, the goal of the underlying control is to make the actual state... Tracking reference trajectory Solving the pipe model predictive control problem; therefore, a set of compressional states and input constraints is introduced. To counteract bounded disturbances Impact: (8) in, and The underlying robust tracking controller at time 10:00 Predicted future The state vector and input vector of the step; the set of input constraints It is obtained by compressing the original state and input constraints, and is used to reserve a safety margin for disturbances.
[0038] Step 5.2: Incorporate robust terminal constraints to ensure that the system state converges to the pipe near the reference trajectory, thus guaranteeing the closed-loop stability of the docking process. (9) in, For robust tracking control at the underlying time Predicted future The state vector of the step (predicting the end point in the time domain); In the optimal reference state sequence obtained in step 4, at time... Predicted future The reference state of the step; For robust terminal sets.
[0039] Step 5.3: In each control cycle, solve the tracking control optimization problem with constraints (8) and (9) to obtain the optimal control sequence, and set the first control variable... The actual thrust is applied to the thrusters of the following AUV to achieve high-frequency robust tracking.
[0040] Step 6: Collaborative control execution and self-reconfiguration completed.
[0041] Steps 3 to 5 are executed cyclically. The high-level reference trajectory planner continuously updates the reference trajectory and actively avoids dynamic wakes and obstacles. The low-level robust tracking controller is used to continuously calculate control variables to accurately track the reference trajectory. As the AUV smoothly approaches the leading AUV from the side and rear along the planned path, it adjusts its attitude in the end stage and finally engages with the tail docking mechanism of the leading AUV to complete the physical self-reconfiguration connection and realize the self-reconfiguration docking of multiple AUVs.
[0042] In this step, the high-level PMPC updates a collision-free, wake-avoiding reference trajectory at a low frequency; the low-level RTMPC calculates the actual thrust at a high frequency and controls the follow-up AUV to accurately track the trajectory until it achieves physical lock with the tail interface of the leading AUV.
[0043] To verify the effectiveness and robustness of the HMPC-based AUV dynamic docking method proposed in this invention, a comprehensive simulation scenario was constructed that included a moving lead AUV, unknown ocean current interference, and environmental obstacles for testing.
[0044] The lead AUV is set to cruise in a straight line at a speed of 1.5 m / s, while being disturbed by ocean currents of 0.3 m / s to the north and 0.2 m / s to the east. The environment includes static spherical obstacles and a conical wake interference zone (5° angle, 20 m long) generated behind the lead AUV's thrusters.
[0045] from Figure 4 As shown in the docking trajectory, the high-level PMPC planner of this invention can generate a smooth and safe reference trajectory. Subsequently, during the approach, the AUV actively steered at t=10.2s to avoid the static obstacle through the action of the APF.
[0046] from Figure 5 The results show that the underlying RTMPC controller maintained good tracking performance under bounded disturbances. The position error between the following AUV and the lead AUV eventually converged to 0.12m, the heading angle error was 0.009rad, and the pitch angle error was 0.022rad, achieving high-precision physical docking.
[0047] from Figure 6 The curves showing the changes in AUV control inputs after docking demonstrate the changes in the thruster and rudder during the docking process, reflecting the effective control performed by the underlying RTMPC controller while meeting input constraints.
[0048] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.
Claims
1. A method for reconfiguration and docking of multiple underwater vehicles based on hierarchical predictive control, characterized in that, Includes the following steps: Step 1: For the following autonomous underwater vehicle, establish its discrete-time nonlinear dynamic model; based on the motion state of the leading autonomous underwater vehicle, model the wake region generated by its tail thruster as a conical danger potential field that moves with the leading autonomous underwater vehicle, as a dynamic obstacle avoidance constraint. Step 2: Based on the error dynamics model of the follower autonomous underwater vehicle, the optimization problem is solved offline using linear matrix inequalities and semidefinite programming methods to obtain the robust feedback gain matrix and terminal weight matrix, and a robust terminal set with robust positive invariance under bounded perturbations is constructed. Step 3: In each control cycle, the following autonomous underwater vehicle (AUV) acquires its own status and relative pose information with the lead AUV in real time through its onboard sensor system, and updates the wake region at the current moment based on the real-time motion status of the lead AUV. Step 4: Construct a high-level reference trajectory planner; A nonlinear model predictive control method is adopted, taking the state of the current and following autonomous underwater vehicle and the wake region as input, to solve a finite time-domain optimization problem with artificial potential field terms, and generate a smooth reference trajectory that avoids the wake and obstacles. Step 5: Construct a robust tracking controller at the bottom layer; adopt a pipe-based robust model predictive control method, take the reference trajectory generated by the high-level reference trajectory planner as the tracking target, combine the robust feedback gain matrix and robust terminal set designed offline, calculate the actual control quantity by solving the tracking control optimization problem and apply it to the following autonomous underwater vehicle. Step 6: Execute steps 3 to 5 in a periodic loop, so that the following autonomous underwater vehicle (AUV) smoothly approaches the leading AUV from the side and rear along the planned path, adjusts its attitude in the terminal phase, and finally completes physical locking with the tail docking mechanism of the leading AUV, realizing the self-reconfiguration docking of multiple AUVs.
2. The multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control according to claim 1, characterized in that, The wake region generated by its tail thruster is modeled as a conical hazardous potential field that moves with the leading autonomous underwater vehicle, including: Define the position of the leading AUV thruster as The reverse unit vector of the leading AUV's heading is The half-angle of the cone is The maximum influence length is ;time wake region as follows: ； in, Let be the coordinate vector of any point in space. For three-dimensional real space, For point The projection length onto the reverse heading vector of the leading AUV, superscript This represents the transpose of a vector. This represents the Euclidean norm of a vector.
3. The multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control according to claim 1, characterized in that, Based on the error dynamics model of a follower autonomous underwater vehicle, the optimization problem is solved offline using linear matrix inequalities and semidefinite programming methods to obtain the robust feedback gain matrix and terminal weight matrix. A robust terminal set with robust positive invariance under bounded perturbations is then constructed, specifically including: Error dynamics model based on follow-up autonomous underwater vehicle ,in For a moment The state error vector, For a moment The input error vector, Let be a bounded perturbation vector. and Given the linearized matrix of the nominal dynamics model, the following optimization problem is solved offline using linear matrix inequalities and semidefinite programming methods to minimize the terminal set volume: ； in, ; Let the inverse of the terminal Lyapunov matrix be a matrix that satisfies ; For affine feedback gain, satisfying ,in Let be the robust feedback gain matrix to be solved; and These are the weight matrices for the state and the input, respectively; It is the identity matrix; is a scalar variable representing the contraction caused by the disturbance; This is the orientation matrix for the state and input constraints; * denotes a symmetric block in the matrix. Represents the determinant of a matrix. Solving the above optimization problem yields the optimal solution. , The robust feedback gain matrix is recovered by the following formula. and terminal weight matrix : ； Based on terminal weight matrix Construct a robust terminal set: in, This is the error state vector following the AUV. For the real number space, Let be the dimension of the state vector. This is a preset scalar parameter, and the superscript T indicates transpose.
4. The multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control according to claim 1, characterized in that, A nonlinear model predictive control method is employed, using the current and following state of the autonomous underwater vehicle (AUV) and the wake region as inputs, to solve a finite-time optimization problem with an artificial potential field term, generating a smooth reference trajectory that avoids the wake and obstacles. Specifically, this includes: In each planning cycle, based on the current status of the subsequent AUV and wake region Solve the following optimization problem: ； in, The reference state sequence is to be optimized; The reference input sequence to be optimized; Indicates at time Predicted future The reference state vector of the step; Indicates at time Predicted future The reference input vector for the step; For reference state Reference trajectory points in; The preset docking target state; The prediction time domain length of the planning layer; This is the weight matrix for the state tracking error; The weight matrix for controlling the input; Indicates Weighted quadratic form; Indicates Weighted quadratic form; Let be the artificial potential field function; For the artificial potential field term; The first element of the obtained optimal reference state sequence and optimal reference input sequence are used as the solution. , Reference trajectory for the current cycle Output to the underlying robust tracking control.
5. The multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control according to claim 1, characterized in that, In constructing the underlying robust tracking controller, a set of compressed states and input constraints is introduced. To counteract bounded disturbances Impact: ； in, and The underlying robust tracking controller at time 10:00 Predicted future The state vector and input vector of the step, This is the constraint set obtained by compressing the original state and input constraints.
6. The multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control according to claim 5, characterized in that, The process of building the underlying robust tracking controller also includes: adding robust termination constraints to ensure that the system state converges to the pipeline near the reference trajectory. ； in, For the underlying robust tracking controller at time Predicted future The state vector of the step, For the optimal reference state sequence at time [time] Predicted future The reference state of the step, For robust terminal sets.
7. The multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control according to claim 6, characterized in that, In each control cycle, the tracking control optimization problem with compressed state, input constraints, and robust terminal constraints is solved to obtain the optimal control sequence. The first control variable is then used as the actual thrust to act on the thruster of the following autonomous underwater vehicle to achieve high-frequency robust tracking.
8. The multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control according to claim 1, characterized in that, Steps 3 to 5 are executed periodically, in which the high-level reference trajectory planner continuously updates the reference trajectory and actively avoids dynamic wakes and obstacles; while the low-level robust tracking controller is used to continuously calculate control variables to accurately track the reference trajectory. As the AUV smoothly approaches the leading AUV from the side and rear along the planned path, it adjusts its attitude at the end stage and finally engages with the tail docking mechanism of the leading AUV, completing the physical self-reconfiguration connection and realizing the self-reconfiguration docking of multiple AUVs.
9. A terminal device, comprising a processor, a memory, and a computer program stored in the memory; characterized in that, When the processor executes the computer program, it implements the multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control as described in any one of claims 1-8.
10. A computer-readable storage medium storing a computer program; characterized in that, When the computer program is executed by the processor, it implements the multi-underwater vehicle combination reconfiguration and docking method based on hierarchical predictive control as described in any one of claims 1-8.