ROV trajectory tracking optimization control method for complex environment

Abstract

The invention discloses an ROV trajectory tracking optimization control method for a complex environment, and relates to the technical field of underwater robot control, and the method comprises the steps: building a north-east coordinate system as a fixed coordinate system, building an underwater robot body coordinate system as a motion coordinate system, defining six-degree-of-freedom motion parameters, and building a complete dynamic model; according to the actual operation condition and the structural characteristics of the underwater robot, the complete dynamic model is simplified into a four-degree-of-freedom control model, and thrust distribution is carried out according to a direct logic distribution method; designing a double-closed-loop sliding mode surface containing an integral term and an improved index reaching law, and using a hyperbolic tangent function as a switching term to complete the improved design of an outer-loop position controller and an inner-loop speed controller; and designing a multi-objective fitness function, and carrying out iterative optimization on parameters of the sliding mode controller based on an improved crow search algorithm to obtain an optimal parameter combination adaptive to thrust limitation and complex trajectories. According to the invention, the tracking precision of the ROV in a complex environment is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of underwater robot control technology, and in particular to an optimized control method for ROV trajectory tracking in complex environments. Specifically, it is a comprehensive control strategy method for ROV's precise trajectory tracking and thrust-constrained adaptation requirements in complex fishing environments. Background Technology

[0002] In recent years, remotely operated underwater vehicles (ROVs) have become important equipment for seafood harvesting operations, especially in the harvesting of benthic organisms such as sea cucumbers, where they can replace manual labor, reduce safety risks, and improve operational efficiency. However, harvesting ROVs face challenges from complex marine environments and operating conditions in actual operations, and the accuracy of their trajectory tracking control and thrust adaptability directly affect the harvesting results.

[0003] Existing control technologies for fishing ROVs have many limitations: traditional PID-based control methods struggle to cope with complex constraints such as dynamic load changes and thrust limitations, resulting in insufficient parameter adaptability; while traditional sliding mode control possesses a certain degree of robustness, it exhibits significant chattering, which can easily damage the thruster and affect control accuracy; key controller parameters often rely on manual experience for setting, making it difficult to achieve the expected control effect under conditions such as thrust limitations and complex trajectories, resulting in poor adaptability; and existing parameter optimization techniques are mostly single-objective optimizations, lacking a synergistic consideration of tracking accuracy, system stability, and thrust constraints.

[0004] In relevant research both domestically and internationally, some improved control schemes have attempted to address the aforementioned issues, but they still have shortcomings: some studies only verify single-degree-of-freedom or simple trajectories without considering the complex actual working conditions of fishing operations; some parameter optimization schemes do not incorporate thrust-constrained characteristics, resulting in insufficient adaptability to changes in fishing load and thruster saturation constraints; and there is a lack of integrated schemes that deeply integrate intelligent optimization algorithms with sliding mode control to achieve parameter self-tuning to adapt to complex trajectories and thrust constraints.

[0005] In summary, existing technologies cannot simultaneously meet the comprehensive requirements of fishing ROVs for trajectory tracking accuracy, thrust-limited adaptability, parameter self-optimization, and system stability in complex environments. There is an urgent need for an integrated trajectory tracking optimization control scheme that can adapt to complex fishing conditions. Summary of the Invention

[0006] To overcome the aforementioned problems in the existing technology, this invention proposes an optimized control method for ROV trajectory tracking in complex environments.

[0007] The technical solution adopted by this invention to solve its technical problem is: a method for optimizing ROV trajectory tracking control in complex environments, comprising the following steps: Step 1, Mathematical Modeling and Simplification of Underwater Robot: Establish the northeast-east coordinate system as the fixed coordinate system and the underwater robot body coordinate system as the motion coordinate system. Define six degrees of freedom motion parameters and establish a complete dynamic model. Based on the actual operation and the structural characteristics of the underwater robot, simplify the complete dynamic model into a four-degree-of-freedom control model and distribute the thrust according to the direct logic allocation method. Step 2, Design a dual-loop sliding diaphragm controller: Design a dual-loop sliding diaphragm surface with integral terms and an improved exponential reaching law. Use the hyperbolic tangent function as the switching term to complete the improved design of the outer loop position controller and the inner loop velocity controller, and verify the asymptotic stability of the improved system. Step 3, optimize the slurry controller parameters: design a multi-objective fitness function, and iteratively optimize the slurry controller parameters obtained in Step 2 based on the improved crow search algorithm to obtain the optimal parameter combination that adapts to thrust constraints and complex trajectories; Step 4, System Integration and Simulation Verification: Integrate the four-degree-of-freedom control model obtained in Step 1, the sliding membrane controller obtained in Step 2, and the improved crow search parameter optimization module obtained in Step 3 to construct a complete underwater robot trajectory tracking optimization control system. Verify the system's control performance under undisturbed conditions through planar trajectory and three-dimensional complex trajectory simulation tests.

[0008] The above-mentioned ROV trajectory tracking optimization control method for complex environments, in step 1, the four-degree-of-freedom control model is as follows: ; ; ; ; in, The inertial mass matrix of a four-degree-of-freedom ROV system is... The Coriolis centripetal force matrix of a four-degree-of-freedom ROV dynamic system. The damping force matrix of a four-degree-of-freedom ROV dynamic system. To control the input vector, Let m be the external disturbance force and torque vector, and m be the rigid body mass of the ROV. Add a mass coefficient to the oscillation. Add a mass coefficient to the sway. Add a mass coefficient to the heave. Add an additional moment of inertia coefficient to the yaw. Let be the moment of inertia of the ROV about the Z-axis. The relative lateral velocity of the water flow. The relative longitudinal velocity of the water flow. is the cross-damping coefficient, representing the effect of yaw rate on lateral force. The linear damping coefficient for sway is... The second-order damping coefficient for sway is... For the cross-coupling damping derivative, This is the eccentricity damping coefficient. This is the yaw secondary damping coefficient. The cross-damping coefficient represents the effect of lateral velocity on yaw moment. For the cross-coupling damping derivative, The vertical linear damping coefficient is... The second-order damping coefficient of the heave is given. The longitudinal linear damping coefficient is... is the second-order damping coefficient for oscillation.

[0009] The above-mentioned ROV trajectory tracking optimization control method for complex environments, wherein the expression for the double closed-loop sliding diaphragm surface containing the integral term in step 2 is: Position sliding surface: ; Velocity sliding surface: ; in, Indicates position tracking error. Indicates the parameters of the sliding surface. Indicates speed tracking error. This represents the parameters of the sliding surface.

[0010] The improved exponential reaching law in the above-mentioned ROV trajectory tracking optimization control method for complex environments is specifically as follows: outer loop control law for: Inner loop control law for: ; Where J represents the transformation matrix, This represents the switching gain coefficient of the outer loop controller. The scale parameter represents the hyperbolic tangent function. This represents the sliding surface gain matrix of the outer loop controller. The derivative representing the desired position. This represents the restoring force and torque vectors of a four-degree-of-freedom ROV system. This represents the sliding surface gain matrix of the inner loop controller. This represents the switching gain coefficient of the inner loop controller. This represents the scaling parameter of the inner loop hyperbolic tangent function.

[0011] The above-mentioned ROV trajectory tracking optimization control method for complex environments, in step 3, specifically includes the multi-objective fitness function as follows: ; ; ; ; in, Represents the tracking error function; This represents the function that controls input restrictions. Represents the system stability function. To control input limits; t represents time, e(t) represents error. Indicates control input, Represents the sliding mode approach rate. , , This represents the weighting coefficient.

[0012] The aforementioned ROV trajectory tracking optimization control method for complex environments, specifically includes the improved crow search algorithm comprising: replacing random initialization with tent mapping. ,in Represents the nth mapping, Represents the mapping parameters. This represents the (n+1)th mapping; When updating the crow's position, a weight matrix is ​​introduced. : ; in It is a linear decay weight matrix. A random number that is uniformly distributed in the range [0,1]. For the probability of flight, For the first The place where crows hide their food is essentially the first A crow passed by The optimal position after the next iteration. For the first Only crows are there Perceived probability in the next iteration. This is the step size scaling factor. For the random step size of Levi's flight, This represents the globally optimal position within the entire crow flock. Let k be the maximum number of iterations. This represents the initial maximum value of the weights. This represents the final minimum value of the weight.

[0013] The beneficial effects of this invention are: 1. Improved modeling accuracy and adaptability to working conditions: A six-degree-of-freedom model is constructed based on the Fossen method and simplified into a four-degree-of-freedom control model. The thrust distribution scheme is designed in conjunction with the actual arrangement of the thruster. After maneuverability simulation verification, the maneuverability of the model meets the actual operation expectations, providing reliable theoretical support for the control strategy and solving the problem of poor adaptability between the existing model and the actual fishing working conditions.

[0014] 2. Chattering Suppression and Tracking Accuracy Optimization: The improved sliding mode controller effectively suppresses chattering in traditional sliding mode control through the design of an integral sliding surface and a hyperbolic tangent function approach law, while simultaneously improving trajectory tracking accuracy. Simulation results show that in multi-heading angle trajectory tracking, the sway error is optimized by 45%, the sway error by 89.2%, and the chattering amplitude is reduced by more than 85%.

[0015] 3. Enhanced parameter self-optimization and thrust adaptability: The improved crow search algorithm enables multi-objective optimization of control parameters, avoiding the limitations of manual experience settings. Under conditions of limited thrust and complex three-dimensional trajectories (spirals, irregular fixed depths), the trajectory tracking convergence speed is increased by more than 80%, the maximum peak error is reduced by 50% to 90%, and the thrust output is always within the limited range, significantly improving adaptability.

[0016] 4. Excellent system integrity and engineering practicality: The system constructs a progressive control system of "modeling-control-optimization", with each module working together to balance tracking accuracy, thrust constraints, stability and adaptability. Through simulation verification in actual fishing scenarios, the average trajectory tracking error of the ROV is ≤0.0015m, which fully meets the engineering requirements of complex fishing operations and is easy to apply in practical engineering. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the coordinate system of the underwater robot of the present invention; Figure 2 This is a flowchart of the improved crow search algorithm of this invention; Figure 3 This is a simulation comparison diagram of the SMC of this invention and the improved SMC for tracking the sway direction; Figure 4 This is a comparison of simulation results for the SMC of this invention and the improved SMC in the sway direction tracking. Figure 5 This is a simulation comparison diagram of the sway direction error of the SMC of this invention and the improved SMC; Figure 6 This is a simulation comparison diagram of the sway direction error of the SMC of this invention and the improved SMC; Figure 7 This is a simulation comparison diagram of the yaw direction error of the SMC of this invention and the improved SMC; Figure 8 This is a simulation comparison diagram of the sway direction error of the SMC of this invention and the improved SMC; Figure 9 This is a simulation comparison diagram of the sway direction error of the SMC of this invention and the improved SMC; Figure 10 This is a simulation comparison diagram of the heading angle error of the SMC of this invention and the improved SMC; Figure 11 These are simulation comparison diagrams of the sway direction error of the SMC of this invention and the improved SMC, where (a) is the sway direction error of the improved SMC and (b) is the sway direction error of the traditional SMC. Figure 12 This is a simulation comparison diagram of the sway direction error of the SMC of this invention and the improved SMC; Figure 13 This is a simulation comparison diagram of the heading angle error of the SMC of this invention and the improved SMC; Figure 14 This is a simulation diagram of the irregular fixed-depth trajectory tracking control of the present invention, where (a) is the trajectory before parameter optimization and (b) is the trajectory after parameter optimization; Figure 15 This is a comparison chart of the tracking error in the sway direction of the irregular fixed-depth trajectory tracking of the present invention (0-500s); Figure 16 This is a comparison chart of the tracking error in the sway direction of the irregular fixed-depth trajectory tracking of the present invention (0-30s); Figure 17 This is a comparison chart of the tracking error of the heave direction in the irregular fixed-depth trajectory tracking of the present invention, wherein (A) is the overall chart, (a) is a magnified view of the position from 50 to 80 seconds in the overall chart; (b) is a magnified view of the position from 160 to 175 seconds in the overall chart; and (c) is a magnified view of the position from 250 to 280 seconds in the overall chart. Figure 18 This is a simulation diagram of the initial 10 seconds of spiral trajectory tracking according to the present invention; Figure 19 This is a comparison chart of the tracking errors in the oscillation direction of the spiral trajectory tracking of this invention; Figure 20 This is a comparison chart of the tracking error in the sway direction of the spiral trajectory tracking of the present invention; Figure 21 This is a comparison chart of the tracking errors in the helical trajectory tracking and sway direction of this invention; Figure 22 This is a comparison chart of the tracking errors in the yaw direction of the spiral trajectory tracking of this invention; Figure 23This is a simulation diagram of the thrust curve of the present invention, wherein (a) is a schematic diagram of the thrust T1 changing with time; (b) is a schematic diagram of the thrust T2 changing with time; (c) is a schematic diagram of the thrust T3 changing with time; (d) is a schematic diagram of the thrust T4 changing with time; (e) is a schematic diagram of the thrust T5 changing with time; and (f) is a schematic diagram of the thrust T6 changing with time. Detailed Implementation

[0018] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] This embodiment discloses an optimized control scheme for ROV trajectory tracking in complex environments. It falls under the category of trajectory tracking control technology for fishing ROVs performing autonomous fishing tasks in complex seabed environments, specifically addressing conditions such as suppressing flutter and thrust limitation. This technology is applicable to the implementation of intelligent control systems for ROVs in scenarios such as fisheries fishing and marine benthic organism collection. Therefore, the technical solution claimed in this invention belongs to the category of trajectory tracking and parameter optimization control methods in motion control systems for fishing ROVs. It is directly applied to the trajectory tracking and parameter self-optimization execution module of the ROV autonomous fishing control system. It does not belong to the general control technology for underwater robots or the category of simple algorithm improvement, nor does it fall under the category of ROV hardware structure design. Instead, it is a comprehensive control strategy method specifically designed for the precise trajectory tracking and thrust-limited adaptation requirements of ROVs in complex fishing environments. Through a progressive design of "modeling-control-optimization," a complete intelligent control system is constructed to achieve precise trajectory tracking of the ROV under complex fishing conditions and thrust-limited conditions. The specific technical solution is as follows: Step 1: Mathematical Modeling and Simplification of ROV Fishing Coordinate system and parameter definition: such as Figure 1 As shown, the Northeast Elevation (NED) coordinate system is established as the fixed coordinate system, and the ROV body coordinate system is established as the motion coordinate system. The six degrees of freedom motion parameters (sway, roll, heave, roll, pitch, and yaw) are defined in accordance with the SNAME specification.

[0020] Kinematic equations: Define the position vector in a fixed coordinate system Velocity vector in body coordinate system The relationship between the two is as follows: In the formula, The transformation matrix is ​​defined as follows: In the formula, represent , represent , represent .

[0021] The combined calculations yield the complete kinematic equations of the ROV: Dynamic equations: Define the vector of the resultant external force (moment) acting on an underwater robot: The rigid body motion dynamics equations of an underwater robot can be expressed as: in Other forces acting on an underwater robot mainly include gravity, buoyancy, propulsion force, umbilical cable, wind, waves, currents, and interference from external objects.

[0022] For the rigid body inertia matrix: Coriolis centripetal force matrix: If the axes of the body coordinate system are chosen on the principal axes of inertia of the underwater robot, and the origin of the coordinate system is set at the center of gravity of the underwater robot, then , At this point, the rigid body inertial mass matrix Coriolis centripetal matrix They can be simplified as follows: Since rigid body dynamics are independent of ocean currents, a commonly used ROV dynamics simplification method is adopted: assuming that the ocean current is... If the body is non-rotational and constant, then the rigid body dynamics satisfy: In the formula: The velocity of the ROV relative to the ocean current; for The speed of the ROV; for The speed of the ocean current .

[0023] Therefore, based on Fossen's proposed dynamics model for marine vehicles, and combined with the above analysis of the kinematics, dynamics, and thruster configuration matrix of underwater robots: .

[0024] The inertial mass matrix of the ROV system can be obtained. , is represented as: Coriolis centripetal force matrix of ROV dynamic system It can be expressed by the following formula: Damping force matrix of fishing ROV dynamic system It can be expressed by the following formula: Model simplification and thrust distribution: The kinematic equations of a four-degree-of-freedom ROV are: Considering the actual operational requirements and the structural characteristics of ROVs, their sway and roll motions possess inherent stability and do not require active control. Therefore, this embodiment selects the four key degrees of freedom (longitudinal, lateral, vertical, and yaw) that have the most significant impact on operational performance as the main control objects. Based on this simplification principle, the complete dynamic model can be simplified to the following four-degree-of-freedom control model: The inertial mass matrix of the four-degree-of-freedom ROV system can be obtained. , is represented as: Coriolis force matrix of a four-DOF ROV dynamic system It can be expressed by the following formula: Damping force matrix of a four-DOF ROV dynamic system It can be expressed by the following formula: Direct logic allocation method has become the preferred thrust allocation scheme for underwater robots with fixed thruster angles due to its high computational efficiency, strong engineering applicability, and ease of programming implementation.

[0025] When positive longitudinal thrust is required, the horizontal thrusters are allocated according to the following rules: When reverse longitudinal thrust is required, the horizontal thrusters are allocated according to the following rules: When positive lateral thrust is required, the horizontal thrusters are allocated according to the following rules: When reverse lateral thrust is required, the horizontal thrusters are allocated according to the following rules: When a positive yaw moment is required, the horizontal thrusters are distributed according to the following rules: When a reverse yaw moment is required, the horizontal thrusters are distributed according to the following rules: When positive vertical thrust is required, thrusters 5 and 6 are allocated according to the following rules: Where T1, T2, T3, T4, T5, and T6 represent six thrusters, with T1, T2, T3, and T4 being horizontal thrusters and T5 and T6 being vertical thrusters; N represents the desired yaw control torque. This indicates the installation distance of the horizontal thruster relative to the x-plane (the plane of sway symmetry). The value represents the installation distance of the horizontal thruster relative to the y-plane (the plane of sway symmetry), α represents the installation vector angle of the thruster, and β represents the installation vector angle of the thruster.

[0026] Step 2: Design of a Low-Jitter Dual Closed-Loop Sliding Mode Controller Four-degree-of-freedom model adaptation: Based on the simplified four-degree-of-freedom model, a dual-closed-loop sliding mode control architecture with outer-loop position control and inner-loop speed control is designed to achieve precise control of position tracking and speed regulation, respectively.

[0027] Traditional sliding mode controller basic design: Define the sliding surface of the outer loop position controller as follows: In the formula: For position tracking error, These are the parameters of the sliding surface.

[0028] Differentiating the expression, we get: set up The desired speed for the inner loop speed controller. The error between the expected speed and the actual speed is then... .

[0029] Substituting into the four-degree-of-freedom kinematic equations, we can obtain Then the equation yields: .

[0030] Since the control objective requires the inner loop speed controller to achieve it. Therefore, the formula can be simplified to: .

[0031] Choose a commonly used convergence law: Combining the above formula, we can obtain the expected speed input. That is, the outer loop control law is: .

[0032] Define the sliding surface of the inner loop speed controller as: In the formula: For speed tracking error, These are the parameters of the sliding surface.

[0033] Differentiating the expression, we get: Substituting into a four-degree-of-freedom dynamic model: Then the equation yields: .

[0034] Choose a commonly used convergence law: Combining the above formula, we can obtain the inner loop control input. That is, the inner loop control law is: .

[0035] Improved sliding mode controller optimization: Define the sliding surface of the outer loop position controller as follows: In the formula: It is a 4×4 positive definite diagonal matrix.

[0036] Differentiating the expression, we get: .

[0037] set up The desired speed for the inner loop speed controller. The error between the expected speed and the actual speed is then... .

[0038] Substituting into the four-degree-of-freedom kinematic equations, we can obtain Then the equation yields: .

[0039] Since the control objective requires the inner loop speed controller to achieve it. Therefore, the formula can be simplified to: For the outer-loop sliding mode controller, an improved special reaching law is designed. Furthermore, to eliminate the chattering problem inherent in the sliding mode controller, a hyperbolic tangent function is used as the switching term instead of the sign function. The improved special reaching law is as follows: ; Combining the above formula, we can obtain the expected speed input. That is, the outer loop control law is: .

[0040] Define the sliding surface of the inner loop speed controller as: Differentiating the expression, we get: Substituting into a four-degree-of-freedom dynamic model: Then the equation yields: .

[0041] For the inner-loop sliding mode controller, an improved special reaching law is designed. Furthermore, to eliminate the chattering problem inherent in the sliding mode controller, a hyperbolic tangent function is used as the switching term instead of the sign function. The improved special reaching law is as follows: ; Combining the above formula, we can obtain the inner loop control input. That is, the inner loop control law is: ; Improved designs were completed for the outer loop position controller and the inner loop speed controller, and the asymptotic stability of the improved system was verified.

[0042] Step 3: Parameter optimization based on the improved crow search algorithm 1. Multi-objective fitness function design: Sliding mode controllers not only need to optimize control error, but may also involve multiple objectives (such as stability, response speed, control input constraints, robustness, etc.). Therefore, the fitness function is improved into a multi-objective optimization problem, optimizing different objectives separately: minimizing control error to ensure the most basic tracking accuracy of the trajectory tracking control system; controlling input constraints to ensure that the control input is within a certain range, preventing system divergence caused by limited thruster thrust in practical applications; and system stability to ensure that the robot remains stable under disturbances and uncertainties.

[0043] Considering the actual mission objectives of underwater robots for fishing, the priorities of these objectives are ranked. Given that the actual operation involves repeatedly traversing the entire marine ranch or aquaculture farm, the expected value of system error is reduced, with greater emphasis placed on system stability and minimizing the impact of control input constraints. Based on this, a multi-objective fitness function is designed: First, select the evaluation index that minimizes error (ITSE), namely: When considering control inputs, a target needs to be added to prevent the control signal from becoming too large or exceeding the control input range. This can be achieved by penalizing excessively large control inputs. This paper defines the limit of the control input as follows: The control input is required to meet the following constraints: Finally, the stability of the system is considered. For sliding mode control systems, the Lyapunov method is typically used to analyze system stability. The stability of sliding mode control requires that the system state always remains on the sliding surface, attracting all trajectories of the system on this surface. In Lyapunov stability theory, if the Lyapunov function V(x) of the system satisfies that its derivative ≤ 0, then the system is stable. However, the stability of sliding mode control not only requires the system to remain on the sliding surface, but also that the system state can quickly and smoothly enter the sliding surface. Therefore, this paper quantifies the degree of system stability by calculating the approach rate of the sliding surface: In summary, the improved fitness function is expressed as: ; in, Represents the tracking error function; This represents the function that controls input restrictions. Represents the system stability function. To control input limits; t represents time, e(t) represents error. Indicates control input, Represents the sliding mode approach rate. , , This represents the weighting coefficient.

[0044] Improvements to the Crow Search Algorithm: To address the shortcomings of the traditional Crow Search algorithm, such as slow convergence speed and susceptibility to local optima, two optimizations are implemented: First, the Tent chaotic map is used instead of random initialization to improve population diversity and global search capability. The Tent map, also known as the tent map, has a graphical distribution that resembles a tent, as its name suggests. It is also a common piecewise linear chaotic map, as shown in the following equation: The behavior of Tent mappings varies with parameters. Different characteristics arise from variations in these properties, including stability, periodicity, and chaos. When When the value is 0.5, Tent has the most classic mapping method.

[0045] Secondly, the traditional crow position update process relies on fixed parameters to control the crow's movement step size, lacking a dynamic balance between global and local searches. This method is prone to getting stuck in local optima in the early stages, and often suffers from insufficient accuracy in later stages due to a lack of precise search. To address these shortcomings, a weight matrix is ​​introduced into the position update process. , in The weight matrix is ​​designed to decay linearly, shifting the focus of the update process from global search in the early stages to local search in the later stages, thus improving the overall balance of the process. This is illustrated by the following formula: Furthermore, when crow tracking fails, Lévy flight is used to generate long-step perturbations, improving the crow's ability to escape local optima. The crow search algorithm flow is as follows: Figure 2 As shown.

[0046] 3. Parameter optimization execution: Improve the key parameters of the sliding mode controller (sliding surface parameters) , , , , , and switching gain , Using the improved crow search algorithm as the optimization variable, the optimal parameter combination that adapts to the limited thrust and complex trajectory is obtained through iterative optimization, thus solving the problem of insufficient adaptability of human experience parameters.

[0047] Step 4: System Integration and Simulation Verification A complete ROV trajectory tracking optimization control system was constructed by integrating a simplified four-degree-of-freedom model, an improved sliding mode controller, and an improved crow search parameter optimization module. Simulation tests were conducted on planar trajectories with multiple heading angles, rectangles, and circles, as well as complex three-dimensional trajectories such as irregular fixed-depth and spiral lines, to verify the system's control performance under undisturbed conditions.

[0048] The technical solution of the present invention will be described in detail below with reference to specific embodiments. The embodiments take sea cucumber harvesting ROV as the application object to verify the actual effect of the control scheme of the present invention.

[0049] System initialization and parameter settings Hardware parameter configuration: The actual ROV prototype for sea cucumber harvesting was selected, and the key parameters are as follows: Body weight: 80kg, dimensions: 1.2m × 0.8m × 0.6m; Propulsion system: Equipped with 6 thrusters, the horizontal thrusters (T1-T4) have a maximum thrust of 282N, and the vertical thrusters (T5-T6) have a maximum thrust of 250N. The output capabilities of each degree of freedom that this ROV system can achieve are shown in Table 1.

[0050] Table 1 Model parameters were determined: hydrodynamic parameters were obtained based on experimental measurements and simulation calibration, and key parameters are shown in Table 2. Add a mass coefficient to the oscillation. Add a mass coefficient to the sway. Add a mass coefficient to the heave. Add an additional moment of inertia coefficient to the yaw. The linear damping coefficient for sway is... The second-order damping coefficient for sway is... This is the cross-damping coefficient. For the cross-coupling damping derivative, This is the eccentricity damping coefficient. This is the yaw secondary damping coefficient. The cross-damping coefficient represents the effect of lateral velocity on yaw moment. For the cross-coupling damping derivative, The vertical linear damping coefficient is... The second-order damping coefficient of the heave is given. This is the cross-damping coefficient. For the cross-coupling damping derivative, The longitudinal linear damping coefficient is... The second damping coefficient for oscillation is . This is the cross-damping coefficient. This is the derivative of the cross-coupling damping.

[0051] Table 2 2. Implementation of the overall controller structure The controller in this embodiment adopts an architecture of "dual closed-loop sliding mode control + improved crow search algorithm optimization", and the specific implementation logic is as follows: The outer loop is the position control loop: it receives the desired trajectory command and calculates the desired speed by improving the sliding mode controller; the inner loop is the speed control loop: it outputs the thruster control signal based on the error between the desired speed and the actual speed.

[0052] Optimization layer: By improving the crow search algorithm, the core parameters of the sliding mode controller are optimized in real time to adapt to environmental changes and load fluctuations.

[0053] 3. Specific implementation of improved sliding mode controller Sliding surface design: To address the requirement of tracking the position and velocity of fishing ROVs, a double closed-loop sliding mode surface with an integral term is designed, the specific expression of which is as follows: Position sliding surface: ; Velocity sliding surface: .

[0054] Improved reaching law design: To suppress chattering problems in traditional sliding mode control, a hyperbolic tangent function is used instead of the sign function, implementing an improved exponential reaching law: Location ring reaching law: ; Velocity cycle reaching law: .

[0055] Derivation and Implementation of Control Law: Based on Lyapunov stability theory, the final control law is derived as follows: Speed ​​Expected Input That is, the outer loop control law is: ; Inner loop control input That is, the inner loop control law is: .

[0056] 4. Algorithm parameter settings: Improved crow search algorithm parameters: The improved fitness function is expressed as: Population size (N=50), maximum number of iterations (T=200), perception probability (p=0.8), flight probability (l=0.5); weights of the multi-objective fitness function ( ), ( ), ( ).

[0057] Mathematical model verification Model verification simulation: Perform four-degree-of-freedom maneuverability simulation: 1. Sway Simulation: Controlling the ROV to move along the X-axis, analyzing its trajectory, velocity response, and stability under longitudinal propulsion force to ensure the model accurately reflects the dynamic characteristics of longitudinal motion. Simulation analysis is performed under two conditions: Condition 1: Apply a longitudinal thrust of 100N only along the X-axis, and set the initial pose to... The simulation time is 50 seconds, with a step size of 0.1 seconds. Condition 2: Apply a longitudinal thrust with varying thrust along the X-axis. The initial pose is set to... The simulation time is 50s with a step size of 0.1s.

[0058] Under condition one, it can be concluded that even when subjected to only longitudinal thrust, the ROV still experiences a displacement along the positive Z-axis in the vertical direction. The reason for this is that, in the body coordinate system, the center of gravity is located 0.04m below the center of buoyancy. When the ROV experiences longitudinal thrust, a pitching moment is simultaneously generated, causing the ROV to submerge. As the ROV's longitudinal and vertical velocities increase to 0.73m / s and 0.12m / s respectively, they balance the generated drag, and the velocity stops changing.

[0059] 2. Sway Simulation: Control the ROV to move along the Y-axis and analyze its trajectory under lateral thrust, especially lateral offset and attitude changes, to ensure the model accurately reflects the dynamic characteristics of lateral motion. Simulation analysis is performed under two conditions. Condition 1: A lateral thrust of 100N is applied only along the Y-axis, and the initial pose is set to... The simulation time is 50s with a step size of 0.1s.

[0060] Condition 2: Apply a lateral thrust with varying thrust along the Y-axis. The initial pose is set to... The simulation time is 50s with a step size of 0.1s.

[0061] Condition 3: Apply a lateral thrust of 282 N along the Y-axis. Initial pose is set to... The simulation time is 500s with a step size of 0.1s.

[0062] Analyzing the same scenario, the ROV only experiences a lateral thrust of 100N, yet it still exhibits a 0.2m displacement along the positive Z-axis in the vertical direction. This is also due to the ROV's center of gravity being slightly lower than its center of buoyancy. However, the difference lies in the fishing ROV's yaw angle and longitudinal displacement. This is because the fishing ROV's forward-mounted fishing hose has a certain degree of structural asymmetry, affecting the stability of these two degrees of freedom. The analysis shows a lateral displacement of approximately 31m, while the longitudinal displacement is only about 0.2m, which is within an acceptable range. The bow angle also fluctuates only within 0.2rad, a negligible angular deviation in a specific fishing scenario. Ultimately, the ROV's lateral velocity increases to 0.63m / s, balancing with the generated drag, and the velocity stops changing.

[0063] If the lateral thrust is gradually increased, as in condition two, the ROV will have a displacement of 0.2m in the positive Z-axis direction in the vertical direction. This is because the ROV's center of gravity is slightly lower than its center of buoyancy. The results show that the lateral displacement is about 41m and the longitudinal displacement is about 1.7m, which is a significant increase compared to the 0.2m displacement in condition one. The ROV's lateral velocity eventually stabilizes at 1.13m / s, the longitudinal velocity fluctuates within an error range of 0.3m / s, and the vertical velocity fluctuates within an error range of 0.02m / s. However, the ROV's heading angle changes to 0.34rad, which has a slight impact on the ROV's trajectory.

[0064] Under condition three, the ROV's lateral velocity eventually stabilized at 1.17 m / s. The longitudinal velocity fluctuated within an error range of 0.31 m / s, but the ROV's heading angle changed by 3.3 rad. The analysis of the results showed a longitudinal displacement of approximately -310 m, while the lateral position remained almost unchanged, severely affecting the expected lateral motion. Therefore, it can be concluded that for this type of fishing ROV, a short-term, slight lateral thrust is generally in line with the expected motion, but continuous maximum lateral thrust should be avoided as much as possible.

[0065] 3. Heave Simulation: Control the ROV to move along the Z-axis and analyze its trajectory under vertical thrust to ensure the model accurately reflects the dynamic characteristics of vertical motion. Simulation analysis is performed under three conditions: Condition 1: Apply a vertical thrust of 282N only along the Z-axis, and set the initial pose to... The simulation time is 50 seconds, with a step size of 0.1 seconds. Condition 2: Apply a vertical thrust of -282N only along the Z-axis, and set the initial pose to... The simulation time is 50 seconds, with a step size of 0.1 seconds. Condition 3: Apply a vertical thrust with varying thrust along the Z-axis. The initial pose is set to... The simulation time is 50s with a step size of 0.1s.

[0066] Analyzing the simulation results under condition one, when a continuous maximum thrust is applied vertically to the ROV, the ROV experiences a displacement of approximately 50m along the positive Z-axis. Simultaneously, a longitudinal offset of approximately 1.257m is observed, due to the influence of the damping matrix cross terms. The applied vertical thrust causes longitudinal displacement, but the offset is very small and negligible. Ultimately, the ROV's vertical velocity increases to 0.956m / s, balancing with the generated drag, and the velocity stops changing.

[0067] Analyzing the simulation results under condition two, when a continuous maximum reverse thrust is applied vertically to the ROV, the ROV experiences a displacement of approximately 50m in the reverse direction along the Z-axis. Simultaneously, a longitudinal offset of approximately -1.25m is observed, similar to the effect of applying the maximum forward thrust under condition one. The lateral position and bow angle do not change significantly. Ultimately, the ROV's vertical velocity increases to -0.96m / s, balancing with the resulting drag, and the velocity stops changing.

[0068] Analyzing the simulation results under condition three, varying thrusts were applied to the ROV vertically: a positive thrust of 100N for the first 10 seconds, followed by a negative thrust of 50N, then a positive thrust of 150N, and finally a negative thrust of 200N. The vertical displacement of the ROV conformed to the expected motion. Simultaneously, a longitudinal offset of approximately 0.15m was observed, which was negligible compared to the vertical offset. The vertical velocity of the ROV remained relatively stable after stabilization. In summary, the simulation results under all three conditions demonstrate that the vertical motion of the ROV met expectations.

[0069] 4. Yaw Simulation: Control the ROV to simulate helical motion and analyze its trajectory under the applied bow torque to ensure the model accurately reflects the dynamic characteristics of yaw motion. The following simulation analysis is performed: A longitudinal thrust of 100N is continuously applied along the X-axis, and a vertical thrust of 100N is applied along the Z-axis. The bow torque is set to 20Nm, and the initial pose is set to... The simulation time is 50s with a step size of 0.1s.

[0070] During yaw motion, the ROV's trajectory resembles a spiral, initially exhibiting rapid changes in longitudinal position before stabilizing. The ROV's longitudinal velocity stabilizes at 0.08 m / s, its sway velocity stabilizes at -0.3 m / s without affecting its overall motion, its vertical velocity stabilizes at 0.56 m / s, and its yaw angular velocity eventually stabilizes at 0.98 rad / s. Analysis shows that the ROV's yaw motion conforms to the expected state.

[0071] Disturbance-free trajectory tracking simulation of sliding mode controller Multi-heading angle trajectory tracking: Initial pose set to The controller parameters are set to , , , , , , , The simulation time is 100s, the step size is 0.01s, and the expected trajectory is set as follows: 30° straight flight from 0 to 25s, 50° straight flight from 25 to 50s, and 90° straight flight from 50 to 200s.

[0072] Figure 3 Simulation results show that, under multi-heading angle tracking, both the improved SMC and the traditional SMC can achieve the expected displacement target in the sway direction. However, the traditional SMC exhibits significant chattering, specifically an error band of about 0.4m near the target value during the steady-state phase. In contrast, the improved SMC shows good tracking performance. Further analysis is needed to examine the error fluctuations at the target value switching point and the steady-state phase, and to comprehensively compare the performance differences between the two.

[0073] Figure 4 Simulation results show that, under multi-heading angle tracking, both the improved SMC and the traditional SMC can achieve the expected displacement target in the sway direction, but the traditional SMC will have obvious chattering, while the improved SMC has a good tracking effect.

[0074] Figure 5 This is a comparison of the sway direction errors of the improved SMC controller and the traditional SMC controller in multi-heading angle tracking. Simulation results show that the former has a lower peak error and a mean error of 0.0922m, while the latter has a more volatile error, with a peak error close to 0.5m and a mean error of 0.1679m. Analysis indicates that the improved SMC controller improves the mean absolute error accuracy in the sway direction by approximately 45% compared to the traditional SMC controller.

[0075] Figure 6 This is a comparison of the sway direction error between the improved SMC controller and the traditional SMC controller in multi-heading angle tracking. Simulation results show that the former has a mean error of 0.0169m, while the latter exhibits significant error fluctuations, with a peak value approaching 0.5m and a mean error of 0.1563m. Based on the mean error, the improved SMC controller achieves approximately 89.2% optimization in sway direction error compared to the traditional SMC controller.

[0076] Figure 7This chart compares the yaw direction errors of the improved SMC controller and the traditional SMC controller in multi-heading angle tracking. Simulation results show that the average error of the former is 0.1904m, while the average error of the latter is 0.3333m. Based on the average error, the improved SMC controller optimizes the yaw direction error by approximately 42.87% compared to the traditional SMC controller. However, further analysis is needed to determine the specific details at each target switching value.

[0077] Based on the simulation results, it was found that in ROV target trajectory tracking with fixed heading and multiple heading angles, the improved SMC (Signal Control Mechanism) exhibited better stability, reduced chattering, and significantly improved tracking accuracy. The traditional SMC showed noticeable chattering fluctuations across multiple time periods, particularly during path switching and rapid changes, where the control system failed to quickly and effectively eliminate these chattering fluctuations. In contrast, the improved SMC, by using a hyperbolic tangent function instead of the traditional sign function, was able to suppress chattering more quickly, ensuring stable system operation.

[0078] Rectangular trajectory tracking: Initial pose set to The controller parameters are set to , , , , , , , The simulation time is 200s, the step size is 0.01s, and the expected trajectory is set as follows: 90° straight flight for 0~25s, 0° straight flight for 25~50s, -90° straight flight for 50~75s, and 0° straight flight for 75~100s.

[0079] Error comparison results Figure 8 , Figure 9 , Figure 10 Simulation results show that the improved SMC controller optimizes the sway direction error by about 67.7% compared to the traditional SMC controller, reducing the error by more than two-thirds; optimizes the sway direction error by about 66.7%; and optimizes the heading angle error by about 9.2%, reducing the error by 10%.

[0080] Based on the comprehensive simulation results, it was found that in ROV rectangular target trajectory tracking, the improved sliding mode control has better stability than the traditional sliding mode control, reduces chattering, and significantly improves tracking accuracy. The tracking errors in both the longitudinal and transverse directions were reduced by two-thirds, demonstrating a significant optimization effect.

[0081] Circular trajectory tracking: Initial pose set to The controller parameters are set to , , , , , , , The simulation time is 75s with a step size of 0.01s.

[0082] comprehensive Figure 11 , Figure 12 , Figure 13 Simulation results show that in ROV rectangular target trajectory tracking, the improved sliding mode control (SMC) offers better stability compared to the traditional SMC, reduces chattering, and significantly improves tracking accuracy. Specifically, the improved SMC controller optimizes the sway direction error by approximately 15.1% compared to the traditional SMC controller, reducing the error by more than one-seventh. In the yaw direction, it optimizes by approximately 29.8%, reducing the error by about one-third. Regarding heading angle tracking, the improved SMC controller optimizes the heading angle error by approximately 98% compared to the traditional SMC controller, significantly reducing error fluctuations. This indicates that the improved SMC controller demonstrates its performance advantage more effectively when target value switching speeds are faster.

[0083] Improved Raven Search Algorithm Parameter Optimization Simulation Irregular fixed-depth trajectory tracking: Initial pose set to The simulation time is 500s, with a step size of 0.01s. From Figure 14 Simulation results show that, in a simple vertical + longitudinal combined motion trajectory, both the optimized and unoptimized parameters can perform the trajectory tracking task well from a macroscopic perspective, further proving the effectiveness of the optimized SMC controller in tracking. However, further quantitative analysis is needed to consider the microscopic accuracy and response time.

[0084] analyze Figure 15 , Figure 16 The simulation comparison results shown quantitatively analyze the longitudinal tracking performance before and after parameter optimization. In the initial stage, the peak tracking error of the controller before parameter optimization reached 0.015m, and the error converged to within 0.01m after about 20 seconds. After parameter optimization, the peak tracking error of the controller reached 0.0015m, only one-tenth of the former. The overall error fluctuation converged rapidly within 1.5s, and the convergence speed was optimized by 92.5%. In the later stage, both tended to a stable state. Obviously, different controller parameters will affect the initial peak error and the convergence speed. The tracking performance after optimization using the Raven Search algorithm is better than the tracking performance of parameters selected by manual trial and error.

[0085] analyze Figure 17 The simulation comparison results shown can quantitatively analyze the heave direction tracking effect before and after parameter optimization. Figure 17 (a) In stage (a), before parameter optimization, the peak tracking error of the controller reached 0.0013m, and after about 10 seconds, the error converged to within 0.01m. After parameter optimization, the peak tracking error of the controller reached 0.0011m, and the overall error fluctuation converged rapidly within 1.5s, with the convergence speed optimized by 85%. Figure 17 In stage (b), the controller before parameter optimization took 11 seconds to converge the error to within 0.01m, while the tracking effect optimized using the crow search algorithm converged in approximately 1.3 seconds, a speedup of nearly 85.8%. Figure 17 (c) In stage 2, the difference between the two convergence speeds remains above 80%.

[0086] Spiral trajectory tracking: Initial pose set to The simulation time is 100. s, step size 0.01 s.

[0087] In actual trajectory tracking tasks, by Figure 18 It can be seen that using manually trial-and-error sliding mode controller parameters instead of the Raven Search algorithm easily leads to poor tracking performance in the initial stage. Excessive sliding mode gain is affected by thrust limitations, resulting in insufficient output thrust and impacting the initial tracking performance. If the sliding mode gain is too small, there will be a significant deviation between the actual trajectory and the desired spiral trajectory. Especially in the initial stage, unoptimized parameters will lead to frequent sliding surface switching, causing a noticeable chattering effect. The optimized tracking performance can track the target trajectory more quickly in the initial stage.

[0088] The tracking accuracy before and after optimization can be visually observed from the error curve graph. Figure 19 Before parameter optimization, the maximum peak error in the sway direction reached -2m, which is similar to the maximum peak error after parameter optimization. However, the former took 3s to track the target trajectory, while the latter only took about 0.6s to track it, which is nearly 80% faster in terms of convergence speed.

[0089] Regarding sway, this can be observed from the error curve. Figure 20 The difference in tracking accuracy before and after optimization is readily apparent. Before parameter optimization, the maximum peak error in the sway direction reached 0.6m, which is significantly different from the maximum peak error of 0.26m after parameter optimization. This shows that in the sway direction, after parameter optimization, the maximum peak error of trajectory tracking was reduced by approximately 56.7%. In terms of convergence time, the former took 8.3s to track the target trajectory, while the latter only took about 1.2s to track it, resulting in a convergence speed increase of nearly 85.5%.

[0090] Regarding sag, it can be observed from the error curve. Figure 21The difference in tracking accuracy before and after optimization is readily apparent. Before parameter optimization, the maximum peak error in the heave direction reached 0.04m, which is significantly different from the maximum peak error of 0.01m after parameter optimization. This shows that in the heave direction, the maximum peak error of trajectory tracking was reduced by about 75% after parameter optimization. In terms of convergence time, the former took 6.3s to track the target trajectory, while the latter only took about 0.7s to track it, resulting in a convergence speed increase of nearly 88.9%.

[0091] Regarding yaw, this can be observed from the error curve. Figure 22 The difference in tracking accuracy before and after optimization is readily apparent. Before parameter optimization, the maximum peak error in the yaw direction reached 0.05 rad, which is significantly different from the maximum peak error of 0.015 rad after parameter optimization. It can be seen that in the heave direction, after parameter optimization, the maximum peak error of trajectory tracking was reduced by about 70%. In terms of convergence time, the former took 7 seconds to track the target trajectory, while the latter only took about 0.7 seconds to track it, which is nearly 90% faster in terms of convergence speed.

[0092] analyze Figure 23 The results show that the ROV needs to adjust its attitude or overcome initial inertia in the initial stage, which causes significant fluctuations in the range of 0-50 seconds. After 50 seconds, the thrust curve remains basically stable, and the overall thrust is within the entire limit range, which is in line with expectations.

[0093] This embodiment implements the design of an ROV trajectory tracking controller in complex environments. First, a detailed mathematical model of the ROV is studied, constructing a six-degree-of-freedom dynamic and kinematic model, laying the theoretical foundation for subsequent control strategy design. Regarding the control method, a trajectory tracking controller based on SMC is proposed, and the parameters of the SMC are optimized by introducing an improved reaching law and a crow search algorithm. The optimized controller effectively improves the tracking accuracy of the ROV in complex dynamic environments. A series of simulations demonstrate the superiority of the improved SMC in ROV trajectory tracking. Compared with the traditional SMC, the improved SMC controller can improve the tracking error accuracy by more than 50% under different trajectory tracking conditions.

[0094] This embodiment optimizes the design of an ROV trajectory tracking controller under thrust-constrained conditions. For the ROV trajectory tracking problem in underwater operations under thrust-constrained conditions, an improved sliding mode controller parameter optimization strategy based on the Raven Search algorithm is proposed. By introducing the Tent chaotic mapping to optimize the population initialization process, and combining a dynamic weight matrix with the Lévy flight mechanism, the balance between global exploration and local exploitation in traditional algorithms is effectively solved. A multi-objective fitness function is designed to integrate tracking error, thrust constraints, and stability indices, achieving coordinated optimization of control performance. Simulation results for spiral trajectory tracking and irregular constant-depth trajectory tracking show that the overall convergence speed of the optimized controller is increased by more than 80%, and the maximum peak error is reduced by more than 50%, achieving high-precision trajectory tracking.

[0095] The above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art can make various modifications or equivalent substitutions to the present invention within its scope and spirit, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of the present invention.

Claims

1. A method for optimized control of ROV trajectory tracking in complex environments, characterized in that, Includes the following steps: Step 1, Mathematical Modeling and Simplification of Underwater Robot: Establish the northeast-east coordinate system as the fixed coordinate system and the underwater robot body coordinate system as the motion coordinate system. Define six degrees of freedom motion parameters and establish a complete dynamic model. Based on the actual operation and the structural characteristics of the underwater robot, simplify the complete dynamic model into a four-degree-of-freedom control model and distribute the thrust according to the direct logic allocation method. Step 2, Design a dual-loop sliding diaphragm controller: Design a dual-loop sliding diaphragm surface with integral terms and an improved exponential reaching law. Use the hyperbolic tangent function as the switching term to complete the improved design of the outer loop position controller and the inner loop velocity controller, and verify the asymptotic stability of the improved system. Step 3, optimize the slurry controller parameters: design a multi-objective fitness function, and iteratively optimize the slurry controller parameters obtained in Step 2 based on the improved crow search algorithm to obtain the optimal parameter combination that adapts to thrust constraints and complex trajectories; Step 4, System Integration and Simulation Verification: Integrate the four-degree-of-freedom control model obtained in Step 1, the sliding membrane controller obtained in Step 2, and the improved crow search parameter optimization module obtained in Step 3 to construct a complete underwater robot trajectory tracking optimization control system. Verify the system's control performance under undisturbed conditions through planar trajectory and three-dimensional complex trajectory simulation tests.

2. The ROV trajectory tracking optimization control method for complex environments according to claim 1, characterized in that, The four-degree-of-freedom control model in step 1 is as follows: ； ； ； ； in, The inertial mass matrix of a four-degree-of-freedom ROV system is... The Coriolis centripetal force matrix of a four-degree-of-freedom ROV dynamic system. The damping force matrix of a four-degree-of-freedom ROV dynamic system. To control the input vector, Let m be the external disturbance force and torque vector, and m be the rigid body mass of the ROV. Add a mass coefficient to the oscillation. Add a mass coefficient to the sway. Add a mass coefficient to the heave. Add an additional moment of inertia coefficient to the yaw. Let be the moment of inertia of the ROV about the Z-axis. The relative lateral velocity of the water flow. The relative longitudinal velocity of the water flow. is the cross-damping coefficient, representing the effect of yaw rate on lateral force. The linear damping coefficient for sway is... The second-order damping coefficient for sway is... For the cross-coupling damping derivative, This is the eccentricity damping coefficient. This is the yaw secondary damping coefficient. The cross-damping coefficient represents the effect of lateral velocity on yaw moment. For the cross-coupling damping derivative, The vertical linear damping coefficient is... The second-order damping coefficient of the heave is given. The longitudinal linear damping coefficient is... is the second-order damping coefficient for oscillation.

3. The ROV trajectory tracking optimization control method for complex environments according to claim 2, characterized in that, The expression for the double-closed-loop sliding surface containing the integral term in step 2 is as follows: Position sliding surface: ; Velocity sliding surface: ; in, Indicates position tracking error. Indicates the parameters of the sliding surface. Indicates speed tracking error. This represents the parameters of the sliding surface.

4. The ROV trajectory tracking optimization control method for complex environments according to claim 3, characterized in that, The improved exponential convergence law is specifically as follows: outer loop control law for: Inner loop control law for: ； Where J represents the transformation matrix, This represents the switching gain coefficient of the outer loop controller. The scale parameter represents the hyperbolic tangent function. This represents the sliding surface gain matrix of the outer loop controller. The derivative representing the desired position. This represents the restoring force and torque vectors of a four-degree-of-freedom ROV system. This represents the sliding surface gain matrix of the inner loop controller. This represents the switching gain coefficient of the inner loop controller. This represents the scaling parameter of the inner loop hyperbolic tangent function.

5. The ROV trajectory tracking optimization control method for complex environments according to claim 4, characterized in that, The multi-objective fitness function in step 3 is specifically as follows: ； ； ； ； in, Represents the tracking error function; This represents the function that controls input restrictions. Represents the system stability function. To control input limits; t represents time, e(t) represents error. Indicates control input, Represents the sliding mode reaching rate. , , This represents the weighting coefficient.

6. The ROV trajectory tracking optimization control method for complex environments according to claim 5, characterized in that, The improved crow search algorithm specifically includes: replacing random initialization with tent mapping. ,in Represents the nth mapping, Represents the mapping parameters. This represents the (n+1)th mapping; When updating the crow's position, a weight matrix is ​​introduced. : ； in It is a linear decay weight matrix. A random number that is uniformly distributed in the range [0,1]. For the probability of flight, For the first The place where crows hide their food is essentially the first A crow passed by The optimal position after the next iteration. For the first Only crows are there Perceived probability in the next iteration. This is the step size scaling factor. For the random step size of Levi's flight, This represents the globally optimal position within the entire crow flock. Let k be the maximum number of iterations. This represents the initial maximum value of the weights. This represents the final minimum value of the weight.

Images