Ship actuator motion control method based on fusion optimization algorithm

Abstract

The present invention proposes a ship actuator motion control method based on a fusion optimization algorithm, which relates to the field of ship actuator control technology, including: S1 presetting multiple ship control modes, each control mode corresponding to a specific optimization target; S2 using multi-source sensors to collect ship motion state and environmental parameter data in real time, fusing the collected multi-source sensor data, and estimating the current motion state of the ship; S3 selecting the ship control mode according to the estimated motion state, current task requirements and environmental conditions; S4 using a fusion optimization algorithm, based on the optimization target of the selected ship control mode and the estimated motion state, dynamically optimizing the parameters of the hybrid controller; S5 based on the optimized parameters, generating control instructions of the hybrid controller, passing the control instructions to the corresponding actuator, and the actuator operates according to the control instructions. The present invention solves the problem that the existing control method is insufficiently adaptable under complex sea conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of ship actuator control, and in particular to a ship actuator motion control method based on a fusion optimization algorithm. Background Art

[0002] Ship actuator control is an important research direction in the field of ship engineering. Its purpose is to achieve precise control of ship motion by controlling various ship actuators, such as main propellers, rudders, thrusters, and fin stabilizers. As modern ships develop towards large-scale, intelligent, and multifunctional, the accuracy and reliability of actuator control are increasingly required. Ships face various external interferences such as wind, waves, and currents in complex sea conditions, which brings great challenges to actuator control.

[0003] In the field of ship actuator control, PID (proportional-integral-differential) controller is one of the most commonly used control methods. PID controller is widely used in ship control systems due to its simple structure and good robustness. It controls the system by calculating the proportional term, integral term and differential term of the error, and taking the weighted sum of these three terms as the control output. The advantage of PID controller is that it is easy to implement and adjust, and has a certain adaptability to changes in system parameters. However, when dealing with strong nonlinearity, large time lag and time-varying systems, the performance of traditional PID controllers may be limited. Therefore, how to optimize the parameters of PID controllers to adapt to complex ship dynamics and marine environment has become one of the research focuses.

[0004] For the optimization of PID controllers, researchers have proposed a variety of algorithms, such as genetic algorithms, particle swarm optimization algorithms, ant colony algorithms, etc. These algorithms simulate the evolution or group behavior of nature to search for the optimal solution in the solution space to achieve automatic adjustment of PID controller parameters. However, the existing optimization algorithms still have some shortcomings when dealing with high-dimensional complex problems. First, a single optimization algorithm is prone to fall into local optimality and it is difficult to obtain a global optimal solution. Second, most algorithms do not fully consider the influence of ship dynamic characteristics and environmental factors, resulting in the optimization results may not be ideal in practical applications. Summary of the invention

[0005] In view of this, the present invention proposes a ship actuator motion control method based on a fusion optimization algorithm, aiming to solve the problem of insufficient adaptability of existing control methods in complex sea conditions. The method fuses multi-source sensor data, applies an adaptive Kalman filter algorithm for state estimation, and uses a two-stage fusion optimization algorithm to optimize the parameters of the hybrid controller. This method can adapt to different control modes, consider the influence of environmental factors, improve the control performance of ships in various sea conditions, and achieve more accurate and reliable actuator control.

[0006] The technical solution of the present invention is implemented as follows: The present invention provides a ship actuator motion control method based on a fusion optimization algorithm, comprising:

[0007] S1 presets multiple ship control modes, each of which corresponds to a specific optimization target;

[0008] S2 uses multi-source sensors to collect ship motion status and environmental parameter data in real time, performs fusion processing on the collected multi-source sensor data, and estimates the current motion status of the ship;

[0009] S3 selects the ship control mode based on the estimated motion state, current mission requirements and environmental conditions;

[0010] S4 uses a fusion optimization algorithm to dynamically optimize the hybrid controller parameters based on the optimization objective of the selected ship control mode and the estimated motion state;

[0011] S5 generates control instructions for the hybrid controller based on the optimized parameters, transmits the control instructions to the corresponding actuators, and the actuators operate according to the control instructions.

[0012] Based on the above solution, preferably, the ship control mode includes:

[0013] Heading hold mode, whose optimization goal is to minimize heading deviation;

[0014] Track tracking mode, whose optimization goal is to minimize track deviation;

[0015] Dynamic positioning mode, whose optimization goal is to minimize position and heading deviation in complex sea conditions;

[0016] Energy-saving cruise mode, whose optimization goal is to minimize fuel consumption;

[0017] Collision avoidance mode, whose optimization goal is to maximize the safe distance.

[0018] Based on the above solution, preferably, step S2 includes:

[0019] S21 uses multi-source sensors to collect raw data in real time, including IMU data, GNSS data, magnetic compass data, water velocity data, meteorological data and wave data;

[0020] S22 preprocesses the collected raw data, including removing outliers, unifying timestamps, and converting coordinate systems;

[0021] S23 uses the Kalman filter algorithm to fuse the pre-processed multi-source sensor data to obtain the ship's state estimation, including position, speed, attitude and angular velocity;

[0022] S24 For water velocity data, meteorological data, and wave data, short-term forecasts are made using time series analysis methods to compensate for sensor delays.

[0023] Based on the above solution, preferably, step S23 includes:

[0024] Define the state vector of the ship y as follows: ,

[0025] in, , , are longitude, latitude and altitude respectively, , , are the north velocity, east velocity and vertical velocity respectively, is the roll angle, is the pitch angle, is the yaw angle, , , are the roll angular velocity, pitch angular velocity and yaw angular velocity respectively;

[0026] Build a system model, including:

[0027] ,

[0028] ,

[0029] in, is the state vector, is the observation vector, is the state transition matrix, is the observation matrix, is the control input, is the control input matrix, and are process noise and observation noise respectively;

[0030] Make predictions for the states and covariances:

[0031] ,

[0032] ,

[0033] In the formula, is the prior state estimate at time k, is the posterior state estimate at time k-1, is the prior error covariance matrix at time k, is the posterior error covariance matrix at time k-1, is the tracking factor, is the process noise covariance matrix;

[0034] Calculate the Kalman gain and update the state and covariance:

[0035] ,

[0036] ,

[0037] ,

[0038] In the formula, is the Kalman gain, is the measurement noise covariance matrix, is the actual observed value, is the posterior state estimate at time k, is the identity matrix;

[0039] Adaptive Noise Covariance Adjustment:

[0040] ,

[0041] ,

[0042] ,

[0043] In the formula, For the innovation sequence, is the estimated actual measurement noise covariance, is the sliding window size, For the forgetting factor;

[0044] Tracking filter adjustment:

[0045] ,

[0046] ,

[0047] In the formula, is the trace operation of the matrix, is the adjustment step size of the tracking factor;

[0048] For different sensor data, adjust the observation matrix accordingly ;

[0049] Output state estimation:

[0050] ,

[0051] In the formula, , , are the estimated longitude, latitude, and altitude, , , are the estimated north, east and vertical velocities, , , are the estimated roll, pitch and yaw angles, , , are the estimated roll angular velocity, pitch angular velocity, and yaw angular velocity.

[0052] Based on the above solution, preferably, for different sensor data, the observation matrix is ​​adjusted accordingly. ,include:

[0053] For IMU data, The matrix is ​​set to associate only the pose in the state vector , , and angular velocity , , quantity;

[0054] For GNSS data, The matrix is ​​set up to associate only the positions in the state vector , , and speed , , Quantity;

[0055] For magnetic compass data, The matrix is ​​set to associate only the yaw angle in the state vector Quantity.

[0056] Based on the above solution, preferably, step S4 includes:

[0057] S41 build hybrid controllers, including PID controllers, fuzzy controllers, and neural network controllers;

[0058] S42 defines an optimization objective function J, and sets a weight coefficient according to the optimization objective of the selected control mode and the estimated motion state;

[0059] S43 uses a fusion optimization algorithm to iteratively optimize the parameters of the hybrid controller. During the optimization, a solution is a set of parameters of the hybrid controller. When the iteration is completed, an optimal set of parameters of the hybrid controller is output;

[0060] S44 applies the optimized optimal hybrid controller parameters to the hybrid controller.

[0061] Based on the above solution, preferably, the fusion optimization algorithm includes a first optimization algorithm and a second optimization algorithm, and step S43 includes:

[0062] S431 Initialize the parameters of the first optimization algorithm and the second optimization algorithm, including the first particle number N, the maximum number of iterations Max iter , inertia weight, acceleration coefficient, momentum coefficient, convergence coefficient, number of second particles M, enclosure coefficient and spiral coefficient; and set the parameter range of the hybrid controller, including K of the PID controller p , K i , K d , fuzzy rule parameters of fuzzy controller, hidden layer weights and biases of neural network controller;

[0063] S432 performs a preliminary search using the first optimization algorithm:

[0064] a. Initialize the position and velocity of the first particle, each first particle represents a set of hybrid controller parameters;

[0065] b. Evaluate the fitness of each first particle, that is, calculate the value of the objective function J according to the current control mode;

[0066] c. Update the individual optimal position p of each first particle i and the global optimal position p g ;

[0067] d. Update the first particle according to the speed and position update formula;

[0068] e. Repeat steps bd until the preset number of iterations or convergence condition is reached;

[0069] S433 using the candidate solution obtained by the first optimization algorithm as an initial second particle of the second optimization algorithm;

[0070] S434 uses the second optimization algorithm to perform a global search:

[0071] a. Use the enclosing coefficient to update the position of each second particle and explore the parameter space;

[0072] b. Use the spiral coefficient to update the position and refine the local optimal solution;

[0073] c. Randomly select the second particle for global search;

[0074] S435 dynamically adjusts and optimizes the objective function according to the current control mode, and calculates the value of the objective function J;

[0075] S436 Iterate steps S434-S435 until one of the termination conditions is met: the maximum number of iterations Max is reachediter , The improvement of the optimal solution after n consecutive iterations is less than the preset threshold , the objective function J reaches the preset optimization target value;

[0076] S437 outputs the optimal solution as the parameters of the hybrid controller, including the K of the PID controller p , K i , K d , fuzzy rule parameters of the fuzzy controller, hidden layer weights and biases of the neural network controller.

[0077] Based on the above solution, preferably, the expression of the objective function J is as follows:

[0078] Heading hold mode:

[0079] ,

[0080] In the formula, represents the objective function in heading hold mode, is the estimated yaw angle, represents the expected value of the yaw angle, represents the estimated yaw rate, represents the integrated square error in heading hold mode, , , is the weight coefficient;

[0081] Tracking mode:

[0082] ,

[0083] In the formula, represents the objective function in track tracking mode, , , is the estimated longitude, latitude, and altitude, and the subscript d represents the corresponding expected value. represents the integrated square error in track tracking mode, , , is the weight coefficient;

[0084] Dynamic positioning mode:

[0085] ,

[0086] In the formula, represents the objective function in dynamic positioning mode, , , are the estimated north, east and vertical velocities, Indicates the integrated square error in dynamic positioning mode, , , , is the weight coefficient;

[0087] Energy-saving cruise mode:

[0088] ,

[0089] In the formula, represents the objective function in the energy-saving cruise mode, FC represents the fuel consumption rate, represents the integrated square error in the energy-saving cruise mode, , , , is the weight coefficient;

[0090] Collision avoidance mode:

[0091] ,

[0092] In the formula, represents the objective function in collision avoidance mode, Indicates the distance to the obstacle. Indicates the distance from the original route. represents the integrated square error in collision avoidance mode, , , is the weight coefficient.

[0093] Based on the above solution, preferably, in step S432, the speed update formula is:

[0094] ,

[0095] In the formula, and are the velocity and position of the i-th first particle at the t-th iteration, represents the optimal position of an individual, represents the global optimal position, represents the global worst position, is the inertia weight, , , , is the acceleration factor, , , , is a random number between [0,1], f(E) is the environmental factor function, which takes into account the ship dynamics constraints;

[0096] The position update formula is:

[0097] ,

[0098] In the formula, is the momentum coefficient, is the convergence coefficient.

[0099] Based on the above solution, preferably, step S5 includes:

[0100] S51 obtains current state information, including estimated motion state and environment information;

[0101] S52 calculates the error between the actual state and the expected state according to the current control mode, where:

[0102] In heading hold mode, calculate heading deviation;

[0103] In track tracking mode, calculate position deviation and heading deviation;

[0104] In dynamic positioning mode, calculate position deviation, heading deviation and speed deviation;

[0105] In energy-saving cruise mode, calculate speed deviation and heading deviation;

[0106] In collision avoidance mode, the distance to obstacles and the distance from the original route are calculated;

[0107] S53 uses a hybrid controller to calculate the control amount based on the error:

[0108] The PID controller calculates the proportional, integral, and differential control quantities;

[0109] The fuzzy controller performs fuzzy reasoning to obtain the fuzzy control quantity;

[0110] The neural network controller calculates the neural network control quantity through forward propagation;

[0111] The outputs of each controller are integrated to obtain the final control quantity of the hybrid controller;

[0112] S54 converts the control quantity into specific actuator control instructions, including the main propeller speed and direction, rudder angle, thrust and direction of the side thrusters, and fin stabilizer angle;

[0113] S55 transmits the generated control instructions to the corresponding actuators through the control system;

[0114] S56 After receiving the control command, each actuator performs the corresponding action, including adjusting the speed and direction of the main thruster, rotating the steering gear to a specified angle, generating a specified thrust and direction for the thruster, and adjusting the fin stabilizer to a specified angle;

[0115] S57 continuously monitors the operating status of the actuators and the response of the ship, and feeds the monitoring data back to the control system.

[0116] Compared with the prior art, the present invention has the following beneficial effects:

[0117] (1) By fusing multi-source sensor data, adaptive Kalman filtering and a two-stage fusion optimization algorithm, precise control of the ship's actuators is achieved. This method can adapt to different control modes, consider the influence of environmental factors, and dynamically optimize controller parameters, thereby improving the ship's control performance and adaptability in complex sea conditions and enhancing the accuracy and reliability of ship motion control;

[0118] (2) Adaptive Kalman filtering algorithm is used to fuse multi-source sensor data, and tracking factors and adaptive noise covariance adjustment are introduced to improve the accuracy and stability of state estimation. By adjusting the observation matrix of different sensor data, the advantages of various sensors are fully utilized, effectively improving the accuracy and robustness of ship state estimation;

[0119] (3) Multiple ship control modes are preset, and specific optimization objective functions are defined for each mode, so that the control system can flexibly switch control strategies according to current mission requirements and environmental conditions. This multi-mode control method improves the adaptability and flexibility of the ship control system and can meet the needs of different navigation missions;

[0120] (4) The two-stage fusion optimization algorithm is adopted, combining the preliminary search of the first optimization algorithm and the global search of the second optimization algorithm, which effectively avoids the problem that a single algorithm is prone to falling into the local optimum. By dynamically adjusting the optimization objective function and taking into account the influence of environmental factors, the optimization results are more in line with actual navigation needs and the effect of controller parameter optimization is improved;

[0121] (5) A hybrid controller consisting of a PID controller, a fuzzy controller, and a neural network controller was constructed, and the parameters of each controller were dynamically optimized through a fusion optimization algorithm. This hybrid control strategy fully utilizes the advantages of various controllers and can better cope with nonlinear, time-varying, and uncertain problems in ship motion control. BRIEF DESCRIPTION OF THE DRAWINGS

[0122] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0123] Figure 1is a flow chart of the method of the present invention;

[0124] Figure 2 This is a flow chart of the fusion optimization algorithm of the present invention. DETAILED DESCRIPTION

[0125] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0126] like Figure 1 As shown, the present invention provides a ship actuator motion control method based on a fusion optimization algorithm, comprising:

[0127] S1 presets multiple ship control modes, each of which corresponds to a specific optimization target;

[0128] S2 uses multi-source sensors to collect ship motion status and environmental parameter data in real time, performs fusion processing on the collected multi-source sensor data, and estimates the current motion status of the ship;

[0129] S3 selects the ship control mode based on the estimated motion state, current mission requirements and environmental conditions;

[0130] S4 uses a fusion optimization algorithm to dynamically optimize the hybrid controller parameters based on the optimization objective of the selected ship control mode and the estimated motion state;

[0131] S5 generates control instructions for the hybrid controller based on the optimized parameters, transmits the control instructions to the corresponding actuators, and the actuators operate according to the control instructions.

[0132] Specifically, in one embodiment of the present invention, multiple control modes are pre-set according to the requirements of the ship under different navigation tasks and environmental conditions. These modes include: heading keeping mode, track tracking mode, dynamic positioning mode, energy-saving cruise mode, and collision avoidance mode. Specific optimization goals are defined for each control mode, which reflect the core requirements of the ship motion control in this mode:

[0133] Heading hold mode, whose optimization goal is to minimize heading deviation;

[0134] Track tracking mode, whose optimization goal is to minimize track deviation;

[0135] Dynamic positioning mode, whose optimization goal is to minimize position and heading deviation in complex sea conditions;

[0136] Energy-saving cruise mode, whose optimization goal is to minimize fuel consumption;

[0137] Collision avoidance mode, whose optimization goal is to maximize the safe distance.

[0138] In order to achieve smooth switching between different control modes, the triggering conditions and switching rules for mode switching need to be defined in advance, such as changes in current mission requirements, changes in environmental conditions (such as sea conditions, wind force, etc.), manual switching instructions from operators, and changes in ship status (such as speed, position, etc.).

[0139] Specifically, in one embodiment of the present invention, step S2 includes:

[0140] S21 uses multi-source sensors to collect raw data in real time, including IMU data, GNSS data, magnetic compass data, water flow velocity data, meteorological data and wave data; among them, IMU data includes acceleration and angular velocity; GNSS data includes latitude and longitude, altitude and speed; magnetic compass data provides heading information; water flow velocity data is obtained by Rongguo Acoustic Doppler Current Meter; meteorological data includes wind speed, wind direction, air pressure, etc.; wave data includes wave height, wave direction, wave period, etc.

[0141] S22 preprocesses the collected raw data, including removing outliers, unifying timestamps and coordinate system conversion; specifically, using statistical methods (such as the 3σ criterion) to detect and eliminate abnormal data; synchronizing data from different sensors to a unified time base; and converting data from different sensors to a unified coordinate system, such as the hull coordinate system.

[0142] S23 uses the Kalman filter algorithm to fuse the pre-processed multi-source sensor data to obtain the ship's state estimation, including position, speed, attitude and angular velocity;

[0143] Specifically, step S23 includes:

[0144] Define the state vector of the ship y as follows:

[0145] ,

[0146] in, , , are longitude, latitude and altitude respectively, , , are the north velocity, east velocity and vertical velocity respectively, is the roll angle, is the pitch angle, is the yaw angle, , , They are the roll angular velocity, pitch angular velocity and yaw angular velocity respectively.

[0147] Build a system model, including:

[0148] ,

[0149] ,

[0150] in, is the state vector, is the observation vector, is the state transition matrix, is the observation matrix, is the control input, is the control input matrix, and are process noise and observation noise respectively.

[0151] The prediction step is to use the state equation to predict the current state, calculate the prior error covariance matrix, and introduce the tracking factor π to improve the adaptability to mutations. The prediction formula is as follows:

[0152] ,

[0153] ,

[0154] In the formula, is the prior state estimate at time k, is the posterior state estimate at time k-1, is the prior error covariance matrix at time k, is the posterior error covariance matrix at time k-1, is the process noise covariance matrix; is the tracking factor.

[0155] The update step is to calculate the Kalman gain, update the state estimate using the observed value, and update the posterior error covariance matrix. The update formula is as follows:

[0156] ,

[0157] ,

[0158] ,

[0159] In the formula, is the Kalman gain, is the measurement noise covariance matrix, is the actual observed value, is the posterior state estimate at time k, is the identity matrix.

[0160] Adaptive noise covariance adjustment, specifically calculating the innovation sequence, using a sliding window to estimate the actual measurement noise covariance, updating the measurement noise covariance matrix, and introducing a forgetting factor To balance the impact of historical data and new data. The formula is as follows:

[0161] ,

[0162] ,

[0163] ,

[0164] In the formula, For the innovation sequence, is the estimated actual measurement noise covariance, is the sliding window size, For the forgetting factor.

[0165] Tracking filter adjustment, specifically dynamically adjusting the tracking factor according to the size of the innovation sequence , to adapt to the speed of state change. The formula is as follows:

[0166] ,

[0167] ,

[0168] In the formula, is the trace operation of the matrix, is the adjustment step size of the tracking factor.

[0169] For different sensor data, adjust the observation matrix accordingly ; The specific adjustment methods are as follows:

[0170] For IMU data, The matrix is ​​set to associate only the pose in the state vector , , and angular velocity , , Quantity;

[0171] For GNSS data, The matrix is ​​set up to associate only the positions in the state vector , , and speed , , Quantity;

[0172] For magnetic compass data, The matrix is ​​set to associate only the yaw angle in the state vector Quantity.

[0173] Output state estimation:

[0174] ,

[0175] In the formula, , , are the estimated longitude, latitude, and altitude, , , are the estimated north, east and vertical velocities, , , are the estimated roll, pitch and yaw angles, , , are the estimated roll angular velocity, pitch angular velocity, and yaw angular velocity.

[0176] S24 For water velocity data, meteorological data, and wave data, short-term forecasts are made using time series analysis methods to compensate for sensor delays.

[0177] Specifically, the time series analysis method adopts the ARIMA model, and the prediction results are used to compensate for sensor delays and improve the real-time performance of state estimation.

[0178] It should be noted that the data fusion and state estimation process of this application realizes the accurate estimation of the current motion state of the ship. This process fully utilizes the advantages of multi-source sensors and effectively handles sensor noise and system uncertainty through the adaptive Kalman filter algorithm. At the same time, by introducing tracking factors and adaptive noise covariance adjustment, the filter's adaptability to sudden changes and nonlinear situations is improved. The short-term prediction of environmental data further enhances the real-time and predictive capabilities of the system. These estimation results provide a reliable basis for subsequent control decisions.

[0179] Specifically, in one embodiment of the present invention, step S3 includes:

[0180] Integrate input data, including the estimated motion state, which is the output of step S2, the current mission requirements, from the predetermined sailing plan, environmental conditions, including data such as water current, wind speed, waves, and predicted short-term changes.

[0181] Define clear selection conditions for each control mode, for example: Heading hold mode: when you need to maintain a fixed heading and the environment has little interference. Track tracking mode: when you need to navigate along a predetermined path. Dynamic positioning mode: when you need to remain stationary or move slightly at a specific location. Energy-saving cruise mode: when you are sailing long distances and the time requirement is not strict. Collision avoidance mode: when a potential collision risk is detected.

[0182] Prioritize different modes to handle situations where multiple conditions are met at the same time. Typically, safety-related modes (such as collision avoidance mode) have the highest priority.

[0183] A built-in decision algorithm is used to implement mode selection. This embodiment uses a decision tree to implement the process as follows:

[0184] If (collision risk detected): select collision avoidance mode

[0185] elif (need to maintain a specific position): select dynamic positioning mode

[0186] elif (there is a scheduled route): Select the track tracking mode

[0187] elif (need to fix the heading): select heading hold mode

[0188] elif (long distance sailing and time requirements are not strict): select energy-saving cruise mode

[0189] else: keep the current mode

[0190] To avoid sudden control changes, achieve a smooth transition when switching modes: set a transition time during which the control output from one mode to another is gradually switched.

[0191] Specifically, in one embodiment of the present invention, step S4 includes:

[0192] S41 builds hybrid controllers including PID controller, fuzzy controller and neural network controller.

[0193] Specifically, the process of building a hybrid controller is:

[0194] PID controller, defines three parameters: proportional gain K p , integral gain K i , differential gain K d The control law is defined as: .

[0195] Fuzzy controllers, defining fuzzy sets of input and output variables, designing fuzzy rule base, selecting fuzzy reasoning methods, Mamdani method, and defuzzification methods, centroid method.

[0196] Neural network controller, design the network structure, specifically a three-layer feedforward network, define the activation function, specifically the sigmoid function, and initialize the weights and biases.

[0197] S42 defines the optimization objective function J, and sets the weight coefficient according to the optimization objective of the selected control mode and the estimated motion state.

[0198] Specifically, the expression of the objective function J is as follows:

[0199] Heading hold mode:

[0200] ,

[0201] In the formula, represents the objective function in heading hold mode, is the estimated yaw angle, represents the expected value of the yaw angle, represents the estimated yaw rate, represents the integrated square error in heading hold mode, , , is the weight coefficient;

[0202] Tracking mode:

[0203] ,

[0204] In the formula, represents the objective function in track tracking mode, , , is the estimated longitude, latitude, and altitude, and the subscript d represents the corresponding expected value. represents the integrated square error in track tracking mode, , , is the weight coefficient;

[0205] Dynamic positioning mode:

[0206] ,

[0207] In the formula, represents the objective function in dynamic positioning mode, , , are the estimated north, east and vertical velocities, Indicates the integrated square error in dynamic positioning mode, , , , is the weight coefficient;

[0208] Energy-saving cruise mode:

[0209] ,

[0210] In the formula, represents the objective function in the energy-saving cruise mode, FC represents the fuel consumption rate, represents the integrated square error in the energy-saving cruise mode, , , , is the weight coefficient;

[0211] Collision avoidance mode:

[0212] ,

[0213] In the formula, represents the objective function in collision avoidance mode, Indicates the distance to the obstacle. Indicates the distance from the original route. represents the integrated square error in collision avoidance mode, , , is the weight coefficient.

[0214] In this embodiment, the weight coefficient is set according to the optimization target of the selected control mode and the estimated motion state, that is, the weight coefficient is set in each control mode. i The value of .

[0215] It should be noted that the weight coefficient is not fixed, but can be adjusted dynamically according to the current state. For example, when approaching the target point, the weight of position control is increased. When the sea conditions are bad, the weight related to stability is increased. Taking the heading hold mode as an example, if the current heading deviation is large, set w1=0.6, w2=0.3, w3=0.1; if the heading is basically correct but fluctuates greatly, set w1=0.3, w2=0.5, w3=0.2.

[0216] S43 uses a fusion optimization algorithm to iteratively optimize the parameters of the hybrid controller. During the optimization, a solution is a set of parameters of the hybrid controller. When the iteration is completed, an optimal set of parameters of the hybrid controller is output;

[0217] Specifically, Figure 2 As shown, the fusion optimization algorithm includes a first optimization algorithm and a second optimization algorithm, and step S43 includes:

[0218] S431 Initialize the parameters of the first optimization algorithm and the second optimization algorithm, including the first particle number N, the maximum number of iterations Maxiter , inertia weight, acceleration coefficient, momentum coefficient, convergence coefficient, number of second particles M, enclosure coefficient and spiral coefficient; and set the parameter range of the hybrid controller, including K of the PID controller p , K i , K d , fuzzy rule parameters of fuzzy controller, hidden layer weights and biases of neural network controller;

[0219] S432 performs a preliminary search using the first optimization algorithm:

[0220] a. Initialize the position and velocity of the first particle, each first particle represents a set of hybrid controller parameters;

[0221] b. Evaluate the fitness of each first particle, that is, calculate the value of the objective function J according to the current control mode;

[0222] c. Update the individual optimal position p of each first particle i and the global optimal position p g ;

[0223] d. Update the first particle according to the speed and position update formula;

[0224] The speed update formula is:

[0225] ,

[0226] In the formula, and are the velocity and position of the i-th first particle at the t-th iteration, represents the optimal position of an individual, represents the global optimal position, represents the global worst position, is the inertia weight, , , , is the acceleration factor, , , , is a random number between [0,1], f(E) is the environmental factor function, which takes into account the ship dynamics constraints;

[0227] The position update formula is:

[0228] ,

[0229] In the formula, is the momentum coefficient, is the convergence coefficient.

[0230] e. Repeat steps bd until the preset number of iterations or convergence condition is reached;

[0231] S433 using the candidate solution obtained by the first optimization algorithm as an initial second particle of the second optimization algorithm;

[0232] S434 uses the second optimization algorithm to perform a global search:

[0233] a. Use the enclosing coefficient to update the position of each second particle and explore the parameter space:

[0234] ,

[0235] In the formula, , , . represents the current position of the second particle, is the updated position, is the current optimal position, A and C are coefficient vectors, a is a linearly decreasing enclosing coefficient, and r is a random vector between [0,1].

[0236] b. Use the spiral coefficient to update the position and refine the local optimal solution:

[0237] ,

[0238] In the formula, is the updated position, represents the distance to the optimal solution, is the spiral coefficient, Represents a random number between [-1,1].

[0239] c. Randomly select the second particle for global search:

[0240] ,

[0241] In the formula, represents the position of a randomly selected second particle.

[0242] S435 dynamically adjusts and optimizes the objective function according to the current control mode, and calculates the value of the objective function J;

[0243] S436 Iterate steps S434-S435 until one of the termination conditions is met: the maximum number of iterations Max is reached iter , The improvement of the optimal solution after n consecutive iterations is less than the preset threshold , the objective function J reaches the preset optimization target value; specifically, the improvement of the optimal solution for n consecutive iterations is less than the preset threshold It is expressed as: .

[0244] S437 outputs the optimal solution as the parameters of the hybrid controller, including the K of the PID controller p , K i , K d , fuzzy rule parameters of the fuzzy controller, hidden layer weights and biases of the neural network controller.

[0245] S44 applies the optimized optimal hybrid controller parameters to the hybrid controller.

[0246] Specifically, the PID controller is updated as: .in, is the error signal.

[0247] The fuzzy controller is updated as: .in, represents the membership function, c, a, b are the parameters of the Gaussian membership function.

[0248] The neural network controller is updated to: .in, is the neuron weight, is the neuron input, Bias for neurons.

[0249] Specifically, in one embodiment of the present invention, step S5 includes:

[0250] S51 obtains current state information, including estimated motion state and environment information.

[0251] The estimated motion state is obtained according to step S2, and specifically includes estimated longitude, latitude, altitude, northing, easting, vertical velocity, roll angle, pitch angle, yaw angle, roll angular velocity, pitch angular velocity, and yaw angular velocity.

[0252] Environmental information includes information sent by sonar or radar, which includes the location of obstacles, as well as environmental conditions such as wind speed and flow rate monitored by anemometers and current meters.

[0253] It can also include feedback information from various actuators, such as propeller speed, rudder angle, etc.

[0254] S52 calculates the error between the actual state and the expected state according to the current control mode, where:

[0255] In heading hold mode, the heading deviation is calculated; specifically, As heading deviation.

[0256] In the track tracking mode, the position deviation and heading deviation are calculated; the position deviation is: , the heading deviation is .

[0257] In dynamic positioning mode, the position deviation, heading deviation and speed deviation are calculated; among them, the position deviation , the heading deviation is , the speed deviation is: .

[0258] In the energy-saving cruise mode, the speed deviation and heading deviation are calculated; the speed deviation is: , the heading deviation is .

[0259] In collision avoidance mode, the distance to the obstacle and the distance from the original route are calculated; the distance to the obstacle: , the distance from the original route .

[0260] S53 uses a hybrid controller to calculate the control amount based on the error:

[0261] The PID controller calculates the proportional, integral and differential control quantities; specifically, the control quantities are calculated according to the control law formula.

[0262] The fuzzy controller performs fuzzy reasoning to obtain fuzzy control quantities; specifically, the error is converted into a fuzzy set, reasoning is performed using a fuzzy rule base, and the fuzzy output is converted into an exact value.

[0263] The neural network controller calculates the neural network control quantity through forward propagation; specifically, the output y of the neural network is used as the control quantity.

[0264] The outputs of each controller are integrated to obtain the final control quantity of the hybrid controller; the weighted summation method is used to calculate the final control quantity.

[0265] S54 converts the control quantity into specific actuator control instructions, including main propeller speed and direction, rudder angle, thruster thrust and direction, and fin stabilizer angle.

[0266] S55 transmits the generated control instruction to the corresponding actuator through the control system; for example, the control instruction is transmitted through the CAN bus.

[0267] S56 After receiving the control command, each actuator performs the corresponding action, including adjusting the speed and direction of the main thruster, rotating the steering gear to the specified angle, generating the specified thrust and direction of the thruster, and adjusting the fin stabilizer to the specified angle.

[0268] S57 continuously monitors the operating status of the actuators and the response of the ship, and feeds the monitoring data back to the control system.

[0269] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A ship actuator motion control method based on a fusion optimization algorithm, characterized in that: include: S1 presets multiple ship control modes, each of which corresponds to a specific optimization goal; Ship control modes include: Heading hold mode, whose optimization goal is to minimize heading deviation; Track tracking mode, whose optimization goal is to minimize track deviation; Dynamic positioning mode, whose optimization goal is to minimize position and heading deviation in complex sea conditions; Energy-saving cruise mode, whose optimization goal is to minimize fuel consumption; Collision avoidance mode, whose optimization goal is to maximize the safe distance; S2 uses multi-source sensors to collect ship motion status and environmental parameter data in real time, fuses the collected multi-source sensor data, and estimates the current motion status of the ship; S3 selects the ship control mode based on the estimated motion state, current mission requirements and environmental conditions; S4 uses a fusion optimization algorithm to dynamically optimize the hybrid controller parameters based on the optimization objectives and estimated motion states of the selected ship control mode; S5 generates control instructions for the hybrid controller based on the optimized parameters, and transmits the control instructions to the corresponding actuators, which operate according to the control instructions.

2. The ship actuator motion control method based on the fusion optimization algorithm according to claim 1 is characterized in that: Step S2 includes: The S21 uses multi-source sensors to collect raw data in real time, including IMU data, GNSS data, magnetic compass data, water velocity data, meteorological data, and wave data; S22 preprocesses the collected raw data, including removing outliers, unifying timestamps, and converting coordinate systems; S23 uses the Kalman filter algorithm to fuse the pre-processed multi-source sensor data to obtain the ship's state estimation, including position, speed, attitude and angular velocity; S24 uses time series analysis methods to make short-term predictions for water velocity data, meteorological data, and wave data to compensate for sensor delays.

3. The ship actuator motion control method based on fusion optimization algorithm as claimed in claim 2 is characterized in that: Step S23 includes: The state vector y of the ship is defined as follows: Among them, λ, μ, h are longitude, latitude and altitude respectively, v N ,v E ,v D are the north velocity, east velocity and vertical velocity respectively, θ is the roll angle, is the pitch angle, ψ is the yaw angle, ω x ,ω y ,ω z are the roll angular velocity, pitch angular velocity and yaw angular velocity respectively; Build a system model, including: y(k)=F(k)y(k-1)+G(k)u(k)+q(k) z(k)=H(k)y(k)+a(k) Where k represents the kth moment, y(k) is the state vector, z(k) is the observation vector, F(k) is the state transfer matrix, H(k) is the observation matrix, u(k) is the control input, G(k) is the control input matrix, q(k) and a(k) are the process noise and observation noise respectively; Make predictions for the states and covariances: P(k|k-1)=π·F(k)P(k-1|k-1)F(k) T +Q(k) In the formula, is the prior state estimate at time k, is the a posteriori state estimate at time k-1, P(k|k-1) is the a priori error covariance matrix at time k, P(k-1|k-1) is the a posteriori error covariance matrix at time k-1, π is the tracking factor, and Q(k) is the process noise covariance matrix; Adaptive Noise Covariance Adjustment: In the formula, ∈(k) is the innovation sequence, is the estimated actual measurement noise covariance, B is the sliding window size, τ is the forgetting factor, ∈(m) is the innovation sequence of the mth time step, R(k) is the measurement noise covariance matrix at time k, and R(k-1) represents the measurement noise covariance matrix at time k-1; Tracking filter adjustment: if tr[∈(k)∈(k) T ]>tr[H(k)P(k|k-1)H(k) T +R(k)]:π=π+Δπ else:π=π-Δπ Where tr is the trace operation of the matrix, and Δπ is the adjustment step of the tracking factor; For different sensor data, adjust the observation matrix H(k) accordingly; Output state estimation: In the formula, is the posterior state estimate at time k, are the estimated longitude, latitude, and altitude, are the estimated north, east and vertical velocities, are the estimated roll, pitch and yaw angles, are the estimated roll angular velocity, pitch angular velocity, and yaw angular velocity.

4. The ship actuator motion control method based on the fusion optimization algorithm as claimed in claim 3 is characterized in that: For different sensor data, the observation matrix H(k) is adjusted accordingly, including: For IMU data, the H(k) matrix is ​​set to associate only the posture θ in the state vector, ψ and angular velocity ω x ,ω y ,ω z Quantity; For GNSS data, the H(k) matrix is ​​set to associate only the position λ,μ,h and velocity v in the state vector N ,v E ,v D Quantity; For magnetic compass data, the H(k) matrix is ​​set to associate only the yaw angle ψ component in the state vector.

5. The ship actuator motion control method based on fusion optimization algorithm as claimed in claim 3 is characterized in that: Step S4 includes: S41 builds hybrid controllers, including PID controllers, fuzzy controllers, and neural network controllers; S42 defines an optimization objective function J, and sets a weight coefficient according to the optimization objective of the selected control mode and the estimated motion state; S43 uses a fusion optimization algorithm to iteratively optimize the parameters of the hybrid controller. During the optimization, a solution is a set of parameters of the hybrid controller. When the iteration is completed, an optimal set of parameters of the hybrid controller is output; S44 applies the optimized parameters of the optimal hybrid controller to the hybrid controller.

6. The ship actuator motion control method based on fusion optimization algorithm as claimed in claim 5 is characterized in that: The fusion optimization algorithm includes a first optimization algorithm and a second optimization algorithm. Step S43 includes: S431 Initialize the parameters of the first optimization algorithm and the second optimization algorithm, including the first particle number N, the maximum number of iterations Max iter , inertia weight, acceleration coefficient, momentum coefficient, convergence coefficient, second particle number M, enclosure coefficient and spiral coefficient; and set the parameter range of the hybrid controller, including the proportional gain K of the PID controller p , integral gain K i , differential gain K d , fuzzy rule parameters of fuzzy controller, hidden layer weights and biases of neural network controller; S432 performs a preliminary search using the first optimization algorithm: a. Initialize the position and velocity of the first particle, each first particle represents a set of hybrid controller parameters; b. Evaluate the fitness of each first particle, that is, calculate the value of the objective function J according to the current control mode; c. Update the individual optimal position p of each first particle i and the global optimal position p g ; d. Update the first particle according to the speed and position update formula; e. Repeat steps bd until the preset number of iterations or convergence condition is reached; S433 uses the candidate solution obtained by the first optimization algorithm as the initial second particle of the second optimization algorithm; S434 uses the second optimization algorithm to perform a global search: a. Use the enclosing coefficient to update the position of each second particle and explore the parameter space; b. Use the spiral coefficient to update the position and refine the local optimal solution; c. Randomly select the second particle for global search; S435 dynamically adjusts the optimization objective function according to the current control mode and calculates the value of the objective function J; S436 Iterate steps S434-S435 until one of the termination conditions is met: the maximum number of iterations Max is reached iter , the improvement of the optimal solution after n consecutive iterations is less than the preset threshold ε, and the objective function J reaches the preset optimization target value; S437 outputs the optimal solution as the parameters of the hybrid controller, including the K of the PID controller p , K i , K d , fuzzy rule parameters of the fuzzy controller, hidden layer weights and biases of the neural network controller.

7. The ship actuator motion control method based on fusion optimization algorithm according to claim 6 is characterized in that: The expression of the objective function J is as follows: Heading hold mode: Where J1 represents the objective function in heading keeping mode, is the estimated yaw angle, ψ d represents the expected value of the yaw angle, represents the estimated yaw rate, ISE1 represents the integrated square error in heading hold mode, w1, w2, w3 are weight coefficients; Tracking mode: Where, J2 represents the objective function in track tracking mode, are the estimated longitude, latitude, and altitude, the subscript d represents the corresponding expected value, ISE2 represents the integrated square error in the track tracking mode, and w4, w5, and w6 are weight coefficients; Dynamic positioning mode: Where, J3 represents the objective function in dynamic positioning mode, are the estimated north, east, and vertical velocities, ISE3 represents the integrated square error in dynamic positioning mode, w7, w8, w9, w 10 is the weight coefficient; Energy-saving cruise mode: Where J4 represents the objective function in the energy-saving cruise mode, FC represents the fuel consumption rate, ISE4 represents the integrated square error in the energy-saving cruise mode, and w 11 ,w 12 ,w 13 ,w 14 is the weight coefficient; Collision avoidance mode: J5=-w 15 ·min(D obs )+in 16 ·D dev +in 17 ·ISE5 Where, J5 represents the objective function in collision avoidance mode, D obs Indicates the distance to the obstacle, D dev represents the distance from the original route, ISE5 represents the integrated square error in the collision avoidance mode, and w 15 ,w 16 ,w 17 is the weight coefficient.

8. The ship actuator motion control method based on fusion optimization algorithm as claimed in claim 6 is characterized in that: In step S432, the speed update formula is: In the formula, and are the velocity and position of the i-th first particle at the t-th iteration, represents the optimal position of an individual, represents the global optimal position, represents the global worst position, w is the inertia weight, c1, c2, c3, c4 are acceleration coefficients, r1, r2, r3, r4 are random numbers between [0, 1], and f(E) is the environmental factor function, which takes into account the ship dynamics constraints; The position update formula is: Where α is the momentum coefficient and β is the convergence coefficient.

9. The ship actuator motion control method based on fusion optimization algorithm as claimed in claim 6, characterized in that: Step S5 includes: S51 obtains current state information, including estimated motion state and environment information; S52 calculates the error between the actual state and the desired state according to the current control mode, where: In heading hold mode, calculate heading deviation; In track tracking mode, calculate position deviation and heading deviation; In dynamic positioning mode, calculate position deviation, heading deviation and speed deviation; In energy-saving cruise mode, calculate speed deviation and heading deviation; In collision avoidance mode, the distance to obstacles and the distance from the original route are calculated; S53 uses a hybrid controller to calculate the control amount based on the error: The PID controller calculates the proportional, integral, and differential control quantities; The fuzzy controller performs fuzzy reasoning to obtain the fuzzy control quantity; The neural network controller calculates the neural network control quantity through forward propagation; The outputs of each controller are integrated to obtain the final control quantity of the hybrid controller; S54 converts the control quantity into specific actuator control instructions, including the main propeller speed and direction, rudder angle, thrust and direction of the side thrusters, and fin stabilizer angle; S55 transmits the generated control instructions to the corresponding actuators through the control system; After receiving the control command, each actuator in S56 executes the corresponding action, including adjusting the speed and direction of the main thruster, rotating the steering gear to a specified angle, generating a specified thrust and direction for the thruster, and adjusting the fin stabilizer to a specified angle; S57 continuously monitors the operating status of the actuators and the response of the ship, and feeds the monitoring data back to the control system.

Images