Ship multi-target cooperative control method and system, computer equipment and medium

Abstract

The invention provides a ship multi-target cooperative control method and system, computer equipment and a medium, and belongs to the crossing field of intelligent ship control and compressed sensing technology, and the method comprises the steps: firstly building a ship state space model, and representing a control sequence under a sparse base; then, constructing a multi-target optimization problem which comprehensively considers path tracking errors, control energy, sequence sparsity and control smoothness; quickly solving by using an accelerated near-end gradient method to obtain a sparse control sequence; and finally, dynamically generating a control instruction through rolling optimization and a self-adaptive sparseness mechanism. The compressed sensing theory is systematically applied to control sequence generation for the first time, high-precision path tracking, effective rolling suppression, energy efficiency improvement and actuator protection can be achieved at the same time, and the problems that a traditional method is frequent in control instruction, high in calculation complexity and reduced in performance under severe sea conditions are solved.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of intelligent ship control and compressed sensing technology, and specifically relates to a method, system, computer equipment and medium for multi-objective cooperative control of ships. Background Technology

[0002] As the global shipping industry transforms towards intelligent and green technologies, the performance requirements for ship automatic control systems, as a core technology for ensuring navigation safety and improving operational efficiency, are becoming increasingly stringent. During navigation, ships must simultaneously meet multiple objectives, including path tracking accuracy, roll stability control, energy consumption optimization, and actuator life protection. This poses extremely high challenges to the synergy and robustness of control technologies.

[0003] The classic control technology in the current field of ship control is represented by Proportional-Integral-Derivative (PID) control. It generates control commands by performing proportional, integral, and derivative operations on the deviation signal. Due to its simple structure and ease of engineering implementation, it is widely used in traditional ship heading and speed control scenarios. Furthermore, improved algorithms derived from PID dynamically adjust parameters to adapt to different navigation conditions. Modern control technologies include Model Predictive Control (MPC), Adaptive Control, and Sliding Mode Control. Among these, MPC, a modern control technology, establishes a ship dynamics model to predict the system state over a future period, solving optimization problems under constraints such as actuator amplitude and speed, and has the potential to handle multi-objective optimization. Adaptive Control improves its adaptability to time-varying ship parameters by estimating system parameters or disturbances online. Sliding Mode Control utilizes discontinuous control laws to make the system state move along a preset sliding surface, exhibiting strong robustness against disturbances.

[0004] However, the parameter tuning of traditional PID control and its derivative algorithms in classical control techniques relies on experience or offline optimization, making it difficult to adapt to dynamic changes under complex sea conditions (such as sudden changes in wind and waves, and course disturbances). Their control logic focuses only on a single objective (such as minimizing course deviation), failing to balance energy consumption optimization and actuator protection. This leads to frequent fluctuations in control commands, severe wear on actuators such as steering gear and propellers, high maintenance costs, and significant energy waste. While modern control technology, MPC, can handle multi-constraint, multi-objective problems, the computational complexity of optimization problems increases exponentially with the extension of the prediction time domain and the increase in state variables, making it difficult to meet real-time control requirements under the limited onboard computing resources of ships. Adaptive control, while possessing some robustness, does not consider the sparsity of control sequences, still exhibiting problems such as control action redundancy and low energy utilization. The discontinuous control law of sliding mode control easily induces "chattering," exacerbating actuator wear, and has limited adaptive adjustment capabilities to disturbances.

[0005] In summary, existing ship control technologies suffer from high computational complexity and severe actuator wear, necessitating a new theoretical framework for a multi-objective collaborative control scheme that can systematically address these issues. Summary of the Invention

[0006] To address the aforementioned problems, this invention provides a method, system, computer equipment, and medium for multi-objective cooperative control of ships.

[0007] To achieve the above objectives, the present invention provides a method for multi-objective cooperative control of ships, comprising: Collect target navigation data, real-time navigation status data, and real-time environmental data of the target vessel.

[0008] Based on the inherent physical parameters of the target ship and the preset sparse basis matrix, a state-space model of ship navigation is constructed. Under the constraints of ship dynamics, actuator amplitude inequality, and actuator rate inequality, the target navigation data, real-time navigation state data, and real-time environmental data are substituted into the state-space model to obtain the predictive control sequence for the next N time moments.

[0009] The predicted control sequence for the next N time steps is projected onto a preset sparse basis matrix to obtain a sparse coefficient vector. A multi-objective optimization function is constructed, comprising a sparse regularization term and control smoothness, with the objectives of minimizing the error between the predicted control sequence and the target navigation data, minimizing the energy of the predicted control sequence, minimizing the number of non-zero elements in the sparse coefficient vector, and minimizing the rate of change of the predicted control sequence between adjacent time steps. Elements in the sparse coefficient vector with amplitudes less than a set threshold are set to zero, generating the optimal sparse solution of the multi-objective optimization function.

[0010] The optimal sparse decoder is decoded into a real control sequence, and the first control command in the real control sequence is extracted to obtain an instantaneous control command; based on the instantaneous control command, multi-target cooperative control of the target ship is realized.

[0011] Preferably, the step of setting the elements in the sparse coefficient vector with magnitudes less than a set threshold to zero to generate the optimal sparse solution of the multi-objective optimization function specifically includes: decomposing the multi-objective optimization function into smooth and non-smooth parts using the accelerated proximal gradient method; performing gradient descent on the smooth part and performing sparsification processing on the non-smooth part in combination with a soft threshold function, so that the elements in the sparse coefficient vector with magnitudes less than the set threshold are set to zero, thereby generating the optimal sparse solution; the smooth part refers to the error term between the predictive control sequence and the target navigation data and the energy term of the predictive control sequence; the non-smooth part refers to the number of non-zero elements in the sparse coefficient vector and the rate of change term of the predictive control sequence at adjacent time points.

[0012] Preferably, the preset sparse basis matrix is ​​a time-domain basis matrix or a frequency-domain basis matrix; the time-domain basis matrix is ​​an identity matrix, and the frequency-domain basis matrix is ​​a discrete Fourier transform basis matrix or a discrete cosine transform basis matrix.

[0013] Preferably, the real-time navigation status data includes: ship position, the angle between the ship's heading and due north, the ship's roll angle around its longitudinal axis, the ship's forward or reverse speed along the heading, the ship's lateral movement speed perpendicular to the heading, the rate of change of the heading angle, and the rate of change of the roll angle; the real-time environmental data includes: wind direction and wind speed, wave height, and wave direction.

[0014] Preferably, the method further includes: collecting new navigation status data of the target vessel after executing the real-time control command; using the new navigation status data as a new initial state, calculating the tracking error between the initial state and the target navigation data, dynamically adjusting the weights of the sparse regularization term and control smoothness of the next control cycle based on the tracking error, and constructing a new multi-objective optimization function with the adjusted weights and solving it to obtain the real-time control command for the next control cycle; thereby realizing multi-objective cyclical collaborative control of the target vessel.

[0015] Preferably, the tracking error of the initial state and target navigation data is calculated, and the weights of the sparse regularization term and control smoothness of the next control cycle are dynamically adjusted based on the tracking error. Specifically, this includes: calculating a sparsity adjustment factor based on the tracking error of the initial state and target navigation data, and using the adjustment factor to dynamically adjust the sparse weights and control smoothness weights of the multi-objective optimization function in the next cycle.

[0016] Preferably, the optimal sparse solution is decoded into a real control sequence by multiplying the optimal sparse solution by a preset sparse basis matrix to obtain the real control sequence.

[0017] The present invention also provides a ship multi-objective cooperative control system, comprising: The data acquisition module is used to collect target navigation data, real-time navigation status data, and real-time environmental data of the target vessel.

[0018] The computation module is used to construct a state-space model of ship navigation based on the inherent physical parameters of the target ship and a preset sparse basis matrix. Under the constraints of ship dynamics, actuator amplitude inequality, and actuator rate inequality, the target navigation data, real-time navigation state data, and real-time environmental data are substituted into the state-space model to obtain a predictive control sequence for the next N time moments. The predictive control sequence for the next N time moments is projected onto the preset sparse basis matrix to obtain a sparse coefficient vector. With the objectives of minimizing the error between the predictive control sequence and the target navigation data, minimizing the energy of the predictive control sequence, minimizing the number of non-zero elements in the sparse coefficient vector, and minimizing the rate of change of the predictive control sequence between adjacent time moments, a multi-objective optimization function containing sparse regularization terms and control smoothness is constructed. Elements in the sparse coefficient vector with amplitudes less than a set threshold are set to zero to generate the optimal sparse solution of the multi-objective optimization function.

[0019] The control module is used to decode the optimal sparse decoder into a real control sequence, extract the first control command from the real control sequence to obtain an instantaneous control command, and realize multi-target cooperative control of the target ship based on the instantaneous control command.

[0020] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement any of the steps in the ship multi-objective cooperative control method.

[0021] The present invention also provides a computer-readable storage medium storing a computer program that, when loaded by a processor, can execute any of the steps in the ship multi-objective cooperative control method.

[0022] The multi-objective cooperative control method for ships provided by this invention has the following beneficial effects: The introduction of ship dynamics constraints and actuator amplitude and rate inequality constraints delineates the control boundary from both physical laws and equipment limits, effectively preventing actuator overload operation, reducing equipment damage risks, and ensuring navigation safety. Based on this, a multi-objective optimization function simultaneously considers four objectives: minimizing route tracking error, minimizing control energy, sparse control actions, and smooth rate of change of actions. This successfully resolves the contradiction between precision and energy consumption in traditional control, enabling the ship to stably follow the target route while eliminating redundant control actions through sparsification, reducing ineffective losses of equipment such as steering gear and propellers, and lowering fuel consumption and maintenance costs. Finally, by extracting the first control command from the actual control sequence as the immediate control command, the rigid full-sequence execution mode is abandoned, ensuring that the control command can quickly adapt to the current navigation state, reducing redundant calculation processes, and achieving timely response to dynamic environmental disturbances such as waves and wind speed. Ultimately, this achieves efficient, stable, and reliable operation of the multi-objective cooperative control of ships. Attached Figure Description

[0023] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart of a multi-objective cooperative control method for ships according to an embodiment of the present invention; Figure 2 This is a diagram illustrating the overall architecture of a ship control system based on compressed sensing, according to an embodiment of the present invention. Figure 3 This is a flowchart of the sparse control sequence optimization process according to an embodiment of the present invention; Figure 4 This is a flowchart illustrating the performance verification process according to an embodiment of the present invention. Figure 5 This is a flowchart illustrating the performance monitoring process according to an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0026] Figure 2 This paper illustrates the overall architecture of a ship control system based on an embodiment. The system comprises multiple cooperating modules, including an environmental perception system, a disturbance estimation module, a path planning module, a reference trajectory generation module, a multi-objective optimizer, a sparse control sequence generation module, a control reconstruction module, and ship actuators. The environmental perception system collects real-time external environmental data (such as wind, waves, and obstacles), and the disturbance estimation module estimates the impact of external disturbances on the ship based on this data. The path planning module generates a global navigation path, while the reference trajectory generation module outputs a smooth reference trajectory. The multi-objective optimizer comprehensively considers control performance, energy consumption, and sparsity objectives, generating the optimal solution through an optimization algorithm. The sparse control sequence generation module utilizes compressed sensing and sparse optimization techniques (such as sparse representation and sparse optimization) to compress the control sequence into a sparse form, reducing computational burden. The control reconstruction module converts the sparse sequence into actual control commands, which are ultimately executed by the ship actuators. This architecture, by integrating multi-objective optimization and sparse control, achieves efficient and energy-saving ship motion control, highlighting the innovation of this invention in reducing computational complexity and improving energy efficiency.

[0027] Based on this, the present invention provides a method for multi-objective cooperative control of ships, specifically as follows: Figure 1 As shown, it includes: S1. Collect target navigation data, real-time navigation status data, and real-time environmental data of the target vessel.

[0028] The target navigation data is pre-defined. Real-time navigation status data includes the ship's position, the angle between the ship's heading and true north, the ship's roll angle about its longitudinal axis, the ship's forward or reverse speed along the heading, the ship's lateral speed perpendicular to the heading, the rate of change of the heading angle, and the rate of change of the roll angle. Real-time environmental data refers to the surrounding environment during navigation.

[0029] S2. Based on the inherent physical parameters of the target ship and a preset sparse basis matrix, construct a state-space model of the ship's navigation. Under the constraints of ship dynamics, actuator amplitude inequality, and actuator rate inequality, substitute the target navigation data, real-time navigation state data, and real-time environmental data into the state-space model to obtain a predictive control sequence for the next N time moments. Project the predictive control sequence for the next N time moments onto the preset sparse basis matrix to obtain a sparse coefficient vector. With the objectives of minimizing the error between the predictive control sequence and the target navigation data, minimizing the energy of the predictive control sequence, minimizing the number of non-zero elements in the sparse coefficient vector, and minimizing the rate of change of the predictive control sequence between adjacent time moments, construct a multi-objective optimization function containing sparse regularization terms and control smoothness. Set the elements in the sparse coefficient vector with amplitudes less than a set threshold to zero to generate the optimal sparse solution of the multi-objective optimization function.

[0030] Ship system modeling and sparse representation within the compressed sensing framework. A ship control system model based on compressed sensing theory is established, with the sparsity of the control sequence as the core optimization objective. The inherent physical parameters of the target ship (length, hydrodynamic coefficients, actuator parameters, etc.) are obtained to construct a state-space model of the ship's motion.

[0031] (1) Where, x( k )∈R n This is the system state vector, updated in real time during navigation, including ship position, heading angle (the angle between the ship's bow and true north), roll angle (the ship's roll angle about its longitudinal axis), pitch speed (the ship's roll angle about its longitudinal axis), sway speed (the ship's lateral movement speed perpendicular to its bow), bow angular velocity, and roll angular velocity, etc.; u( k )∈R m The control input vector at time k includes the main thruster thrust and rudder angle; y( k )∈R p Let w( be the system output vector at time k, and let w( be a subset or linear combination of the state vectors); k ) and v( kThese are process noise and measurement noise, which represent the influence of environmental condition data such as wind direction and speed, wave height, and wave direction. k Indicates different times. , It is the system input matrix, used to characterize the natural evolution of the ship's state when there is no control input and no noise interference (its value is determined by the ship's inherent physical parameters, such as moment of inertia and fluid damping coefficient). This is the system output matrix. Where x( k ), u( k ) and y( k (and the inherent physical parameters of the target ship)

[0032] By substituting target navigation data, real-time navigation status data, and real-time environmental data into a state-space model, predictive control sequences for the next N time steps are obtained. Based on compressed sensing theory, the control sequences for the next N time steps are... U =[ ]∈ℝ mN In a certain sparse basis matrix Ψ∈ℝ mN×mN Projecting below:

[0033] (2) Where Ψ∈R mN×mN Let s be a sparse basis matrix, s∈R mN This is a sparse coefficient vector. The key lies in the choice of the sparse basis:

[0034] (1) Time-domain basis: The control sequence is directly optimized using the identity matrix (Ψ=I). U This inherent property makes it zero or close to zero most of the time.

[0035] (2) Frequency domain basis: The Discrete Fourier Transform (DFT) basis or Discrete Cosine Transform (DCT) basis is used to concentrate the control energy on a few frequency components, thereby suppressing high-frequency, energy-consuming small adjustments.

[0036] Define the measurement matrix Φ∈R M×mN Compressed sensing is applied to the control sequence to construct a compressed measurement model: (3) Where z is the virtual control energy packet, which contains the reconstructed control sequence. U The required sufficient information, according to compressed sensing theory, dimension M Much lower than the original sequence dimension mN This lays the foundation for reducing computational complexity. Simultaneously, it determines the minimum number of measurements required to guarantee accurate reconstruction.

[0037] (4) in, C A constant factor, K This represents the sparsity level (the sparse coefficient vector s has at most K elements that are 0). m To control the dimension of the input vector, N This refers to the number of moments in the future that are predicted.

[0038] Based on the compressed sensing framework, a multi-objective optimization problem is constructed that comprehensively considers control performance, energy efficiency, and actuator protection. The objective function is designed as follows:

[0039] (5) in, The first item represents the expected reference status output (target navigation data). Output tracking error penalty to ensure path tracking accuracy, component matrix This is the state diagonal weight matrix, where each component is used to weigh the penalty for the deviation between the actual state and the expected output; the second term... Controlling energy penalties, optimizing energy consumption, and component matrices. To control the diagonal weight matrix, each component is used to weigh the magnitude penalty of the control quantity; the third term The sparse regularization term induces sparsity in the control sequence, forcing most elements in the optimized sparse vector s to be zero, thereby achieving sparsification of the control sequence; the fourth term To control the smoothness term, drastic changes in the control variable are penalized, protecting the actuator from damage caused by frequent starts and stops and large movements. Γ is the system response matrix, and D is the difference operator matrix. λ and μ This is the regularization scalar parameter.

[0040] The ship's state-space model is introduced as an equality constraint into the optimization problem. Output sequence. Y Through system matrix and control sequence U and initial state x 0 (Real-time navigation status data) are correlated to form dynamic-level equality constraints, which are expressed as formula (6):

[0041] (6) in, The initial state response matrix is: (7) ℋ∈ℝ PN×mN It is the control response matrix: (8) Further considering the physical constraints of the ship's actuators, these are also converted into linear constraints of s. The formulas for the actuator amplitude inequality constraint and the actuator rate inequality constraint are shown in equation (9):

[0042] (9) Among them, u min and u max Δumin and Δumax are the amplitude limits of the actuator, such as the maximum deflection angle of the rudder and the maximum thrust of the propeller; Δumin and Δumax are the rate limits of the actuator, such as the maximum rate of change of the rudder angle and the maximum rate of adjustment of the thrust; D is the differential operator matrix, which describes the rate of change of the control quantity.

[0043] Figure 3 The execution flow of the sparse optimization algorithm, the core of which generates sparse control sequences, is described in detail. After starting sparse optimization, sparse coefficients are initialized and a sparsity target is set. Then, the optimization problem is constructed, including using accelerated gradient descent for fast convergence, handling non-smooth terms with a proximal operator, applying L1 regularization to promote sparsity, applying a soft threshold function to further compress coefficients, updating sparse coefficients, performing constraint projection operations to ensure solution feasibility, and verifying magnitude and rate of change constraints to prevent abrupt changes in control commands. Afterwards, feasibility adjustments are made to handle constraint conflicts. The process then proceeds to convergence judgment: if convergence fails, the iteration point is updated and the optimization step is returned; if convergence has occurred, sparsity is evaluated. If the sparsity target is met, the optimization result is output; otherwise, the regularization parameters are adjusted and the optimization is repeated. This process, through iterative optimization and constraint handling, ensures the sparsity and optimal performance of the control sequence, highlighting the innovation of this invention in real-time computation and sparse control, and improving system response speed and stability.

[0044] Design of a Sparse Optimization Algorithm Based on Compressed Sensing. For the constructed sparse optimization problem, a fast solution algorithm based on the proximal gradient method is designed. The objective function is decomposed into smooth and non-smooth parts:

[0045] (10) An iterative solution is performed using the accelerated proximal gradient method: (11) Among them, s (k) v is a sparse coefficient vector. (k) To accelerate the variable, α For step size parameters, k Indicates different times, This is a proximal operator used to control sparsity and smoothness (non-smooth parts), where k+1 represents the updated next time-space variable. Let represent the gradient of the smooth portion. The optimal sparse vector s is obtained by iterating through equation (11) until convergence. * The optimal sparse vector is used as the true control sequence, and then through... U*= Ψs * The optimal sparse control sequence is obtained. Ψ For sparse basis matrices, only the first control variable is executed. (Immediate control instructions) are given, and this rolling optimization process is repeated at the next moment. Proximal operator The calculation formula is as follows:

[0046] (12) in, For the gradient descent iterative operator, is the input to the proximal operator; s is the sparse coefficient vector. λ The weights are for sparse regularization terms; D is the difference operator matrix, which calculates the adjacent differences in the sequence and measures the smoothness of the control sequence. μ By controlling the weights of the smoothing term, drastic changes in the control sequence can be suppressed. μ The larger the value, the smaller the adjacent changes in the control command. The aforementioned proximal operator... Each element has an analytical solution, which can directly generate sparse solutions, greatly improving computational efficiency.

[0047] Design an adaptive sparsity control mechanism to dynamically adjust the sparsity level based on navigation conditions and performance requirements. Define the sparsity adjustment factor. :

[0048] (13) Where γ0 is the baseline sparsity, η is the adjusted intensity parameter, and e( k Let be the tracking error vector, and ϵ be the error tolerance. The principle is that as the tracking error e( k As the tracking error e(e) increases, the sparsity adjustment factor decreases. The system needs to avoid sparsifying the control sequence to provide a non-zero control sequence, thereby reducing tracking error and improving tracking accuracy. k As the sparsity adjustment factor decreases, the system tends towards sparsity of the control sequence to avoid frequent rudder inputs and reduce energy consumption. Furthermore, the regularization parameter is dynamically and flexibly adjusted based on the sparsity adjustment factor.

[0049] (14) in, λ 0 , μ 0 This is the nominal value. Indicates the first kIn each iteration, the real-time weight of the control smoothing term corresponds to the adjustment weight of the control sequence smoothness, which is dynamically adjusted with each iteration. In this way, the system automatically reduces sparsity when fine control is required (such as berthing and departure, or rough sea conditions) to provide more continuous control; and increases sparsity when navigating smoothly in open waters to achieve pulsed energy-saving control.

[0050] S3. Decode the optimal sparse decoder into a real control sequence, and extract the first control command from the real control sequence to obtain an instantaneous control command; realize multi-target cooperative control of the target ship based on the instantaneous control command.

[0051] U*= Ψs * In this context, Ψ is the sparse basis matrix, and s* is the optimal sparse solution. From... U The first control command extracted from * is obtained u 1 *, u 1 * represents the actual control command. The subsequent repeated rolling optimization process achieves dynamic multi-objective collaborative control.

[0052] A performance evaluation system for a ship multi-objective cooperative sparse optimization method based on compressed sensing is established, and the main evaluation indicators include: 11. Controlling reconstruction error: The control sequence obtained by this invention... U* With a computationally expensive optimal control sequence that does not consider sparsity U opt Compare the results. This error should be small enough to demonstrate the effectiveness of the sparsity control:

[0053] (15) 12. Sparsity efficiency index: measures the sparsity of the control sequence.

[0054] (16) in, K actual The number of non-zero control actions, K max This represents the total number of control actions.

[0055] 3. Multi-objective comprehensive performance indicators: (17) in, w 1 , w 2 , w 3 Let be the weighting coefficient, satisfying w 1+w 2 +w 3 =1; Let these represent the path tracking objective function, energy loss objective function, and sparse control objective function, respectively, and let the overall index be... The effectiveness of this invention is demonstrated by comparing it with traditional methods such as MPC and PID under the same simulation scenario. If the ship deviates too much from the target route... sparse It allows for more intensive (but not too sparse) control, ensuring navigation accuracy; E recon Controlling the reconstruction error is used to verify whether the sparsed control sequence deviates from the optimal effect, so as to avoid the ship losing control in order to save energy.

[0056] Figure 4 This paper demonstrates the performance verification process, validating the effectiveness and superiority of the method of this invention. The process begins with "Start Performance Verification," setting up test scenarios (such as different sea states) and comparing algorithms, including running PID control, traditional MPC, and the method of this invention. Subsequently, operational data is collected, and performance indicators are calculated, including roll suppression effect, path tracking accuracy, control energy consumption, and control motion sparsity. After generating a performance comparison report, a statistical significance test is performed to determine if the performance advantage is significant. If the advantage is significant, the effectiveness of the method is verified; otherwise, parameter tuning is performed. Next, robustness testing is conducted, including verification under different sea states and sensitivity analysis, to evaluate the system's performance under extreme conditions. Finally, a verification report is generated, and the results are verified. This process, through comprehensive comparison and testing, demonstrates the significant advantages of the method of this invention in terms of control accuracy, energy consumption, and sparsity, highlighting its innovation in practicality and reliability.

[0057] Figure 5 The project showcases a performance evaluation and adaptive adjustment module for monitoring and optimizing system operation. This module includes sparsity evaluation, control performance evaluation, energy efficiency evaluation, adaptive parameter adjustment, and real-time optimization. The sparsity evaluation module analyzes the sparsity of the control sequence to ensure compliance with objectives; the control performance evaluation module assesses path tracking accuracy and roll suppression effectiveness; and the energy efficiency evaluation module calculates control energy consumption. Based on these evaluation results, the adaptive parameter adjustment module dynamically adjusts optimization algorithm parameters (such as regularization coefficients) to adapt to different operating conditions. The real-time optimization module then uses the adjusted parameters for online optimization to ensure continuous and efficient system operation. Through continuous monitoring and feedback adjustments, this module achieves system adaptability and robustness, highlighting the innovation of this invention in optimization capabilities and energy efficiency management under dynamic environments.

[0058] The purpose of this invention is to overcome the shortcomings of the prior art and provide a multi-objective sparse cooperative optimization method for ships based on compressed sensing. It innovatively transforms compressed sensing theory from a signal reconstruction paradigm to a control generation paradigm. By actively designing and solving a sparse control sequence optimization problem, it simultaneously achieves high-precision path tracking, effective roll suppression, energy efficiency improvement, and actuator protection. It is particularly suitable for intelligent control of ship path tracking, roll stabilization, and energy consumption optimization under complex sea conditions.

[0059] This invention is the first to systematically transfer compressed sensing theory from the field of signal analysis to the field of control generation, providing a novel sparse optimization paradigm for ship control. Within a unified convex optimization framework, it achieves synergistic optimization and effective trade-offs among four objectives: path tracking, roll suppression, energy saving and consumption reduction, and actuator protection, rather than a simple weighted average. Through sparse control, it can effectively reduce minute control actions by up to 30%-50%, directly reducing energy consumption and actuator mechanical wear, and extending equipment lifespan. Benefiting from the fast convergence of the projected near-end gradient algorithm and the analytical solution properties of the near-end operator, this invention reduces online computation time by 50%-70% compared to traditional algorithms, meeting the real-time requirements of ship control. Compressed sensing theory itself has inherent robustness to measurement noise and incomplete data; combined with an adaptive sparsity mechanism, this results in stronger stability of the control system under sensor noise and unknown ocean disturbances.

[0060] This invention is the first to introduce compressed sensing theory into the field of ship motion control. Through the optimized design of sparse control sequences, it achieves synergistic optimization of control performance and energy efficiency, filling a technological gap in this field. This invention is applicable to intelligent control systems for various types of ships, especially suitable for long-haul marine engineering vessels and commercial transport vessels, possessing significant engineering application value and substantial economic benefits.

[0061] Based on the same inventive concept, the present invention also provides a ship multi-objective cooperative control system, comprising: The data acquisition module is used to collect target navigation data, real-time navigation status data, and real-time environmental data of the target vessel.

[0062] The computation module is used to construct a state-space model of ship navigation based on the inherent physical parameters of the target ship and a preset sparse basis matrix. Under the constraints of ship dynamics, actuator amplitude inequality, and actuator rate inequality, the target navigation data, real-time navigation state data, and real-time environmental data are substituted into the state-space model to obtain a predictive control sequence for the next N time moments. The predictive control sequence for the next N time moments is projected onto the preset sparse basis matrix to obtain a sparse coefficient vector. With the objectives of minimizing the error between the predictive control sequence and the target navigation data, minimizing the energy of the predictive control sequence, minimizing the number of non-zero elements in the sparse coefficient vector, and minimizing the rate of change of the predictive control sequence between adjacent time moments, a multi-objective optimization function containing sparse regularization terms and control smoothness is constructed. Elements in the sparse coefficient vector with amplitudes less than a set threshold are set to zero to generate the optimal sparse solution of the multi-objective optimization function.

[0063] The control module is used to decode the optimal sparse decoder into a real control sequence, extract the first control command from the real control sequence to obtain an instantaneous control command, and realize multi-target cooperative control of the target ship based on the instantaneous control command.

[0064] This invention also provides a computer device. At the hardware level, this computer device includes a processor, an internal bus, a network interface, memory, and non-volatile storage, and may also include other hardware required for business operations. The processor reads the corresponding computer program from the non-volatile storage into the memory and then runs it to implement the aforementioned ship multi-objective cooperative control method.

[0065] The present invention also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described ship multi-objective cooperative control method.

[0066] Specific limitations regarding the computational system for the ship multi-objective cooperative control method can be found in the limitations of the ship multi-objective cooperative control method described above, and will not be repeated here. Each module in the aforementioned ship multi-objective cooperative control system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0067] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. Furthermore, the above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for multi-objective cooperative control of ships, characterized in that, include: Collect target navigation data, real-time navigation status data, and real-time environmental data of the target vessel; Based on the inherent physical parameters of the target ship and the preset sparse basis matrix, a state-space model of ship navigation is constructed. Under the constraints of ship dynamics, actuator amplitude inequality, and actuator rate inequality, the target navigation data, real-time navigation state data, and real-time environmental data are substituted into the state-space model to obtain the predictive control sequence for the next N time moments. The predicted control sequence for the next N time steps is projected onto a preset sparse basis matrix to obtain a sparse coefficient vector. A multi-objective optimization function, including a sparse regularization term and control smoothness, is constructed with the objectives of minimizing the error between the predicted control sequence and the target navigation data, minimizing the energy of the predicted control sequence, minimizing the number of non-zero elements in the sparse coefficient vector, and minimizing the rate of change of the predicted control sequence between adjacent time steps. Elements in the sparse coefficient vector with amplitudes less than a set threshold are set to zero to generate the optimal sparse solution of the multi-objective optimization function. The optimal sparse decoder is decoded into a real control sequence, and the first control command in the real control sequence is extracted to obtain an instantaneous control command; based on the instantaneous control command, multi-target cooperative control of the target ship is realized.

2. The ship multi-objective cooperative control method according to claim 1, characterized in that, The step of setting elements in the sparse coefficient vector with magnitudes less than a set threshold to zero to generate the optimal sparse solution of the multi-objective optimization function specifically includes: decomposing the multi-objective optimization function into smooth and non-smooth parts using the accelerated proximal gradient method; performing gradient descent on the smooth part and performing sparsification processing on the non-smooth part in combination with a soft threshold function, so that elements in the sparse coefficient vector with magnitudes less than the set threshold are set to zero, thereby generating the optimal sparse solution; the smooth part refers to the error term between the predictive control sequence and the target navigation data, and the energy term of the predictive control sequence; the non-smooth part refers to the number of non-zero elements in the sparse coefficient vector and the rate of change term of the predictive control sequence at adjacent time points.

3. The ship multi-objective cooperative control method according to claim 1, characterized in that, The preset sparse basis matrix is ​​a time-domain basis matrix or a frequency-domain basis matrix; the time-domain basis matrix is ​​an identity matrix, and the frequency-domain basis matrix is ​​a discrete Fourier transform basis matrix or a discrete cosine transform basis matrix.

4. The ship multi-objective cooperative control method according to claim 1, characterized in that, The real-time navigation status data includes: ship position, the angle between the ship's heading and due north, the ship's roll angle around its longitudinal axis, the ship's forward or reverse speed along the heading, the ship's lateral speed perpendicular to the heading, the rate of change of the heading angle, and the rate of change of the roll angle; the real-time environmental data includes: wind direction and wind speed, wave height, and wave direction.

5. The ship multi-objective cooperative control method according to claim 1, characterized in that, Also includes: Collect new navigation status data of the target vessel after executing the real-time control command; Using the new navigation state data as the new initial state, the tracking error between the initial state and the target navigation data is calculated. Based on the tracking error, the weights of the sparse regularization term and control smoothness in the next control cycle are dynamically adjusted. A new multi-objective optimization function is constructed using the adjusted weights and solved to obtain the real-time control command for the next control cycle. This achieves multi-objective cyclical collaborative control of the target vessel.

6. The ship multi-objective cooperative control method according to claim 5, characterized in that, The tracking error of the initial state and target navigation data is calculated, and the weights of the sparse regularization term and control smoothness of the next control cycle are dynamically adjusted based on the tracking error. Specifically, this includes: calculating the sparsity adjustment factor based on the tracking error of the initial state and target navigation data, and using the adjustment factor to dynamically adjust the sparse weights and control smoothness weights of the multi-objective optimization function in the next cycle.

7. The ship multi-objective cooperative control method according to claim 1, characterized in that, The optimal sparse solution is decoded into a real control sequence by multiplying the optimal sparse solution with a preset sparse basis matrix to obtain the real control sequence.

8. A ship multi-objective cooperative control system, characterized in that, include: The data acquisition module is used to collect target navigation data, real-time navigation status data, and real-time environmental data of the target vessel. The computation module is used to construct a state-space model of ship navigation based on the inherent physical parameters of the target ship and a preset sparse basis matrix. Under the constraints of ship dynamics, actuator amplitude inequality, and actuator rate inequality, the target navigation data, real-time navigation state data, and real-time environmental data are substituted into the state-space model to obtain a predictive control sequence for the next N time moments. The predictive control sequence for the next N time moments is projected onto the preset sparse basis matrix to obtain a sparse coefficient vector. With the objectives of minimizing the error between the predictive control sequence and the target navigation data, minimizing the energy of the predictive control sequence, minimizing the number of non-zero elements in the sparse coefficient vector, and minimizing the rate of change of the predictive control sequence between adjacent time moments, a multi-objective optimization function containing sparse regularization terms and control smoothness is constructed. Elements in the sparse coefficient vector with amplitudes less than a set threshold are set to zero to generate the optimal sparse solution of the multi-objective optimization function. The control module is used to decode the optimal sparse decoder into a real control sequence, extract the first control command from the real control sequence to obtain an instantaneous control command, and realize multi-target cooperative control of the target ship based on the instantaneous control command.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is loaded by the processor, it is able to perform the steps of the method according to any one of claims 1 to 7.

Images