Unmanned ship attitude adjusting method, system, equipment and medium

Abstract

The invention discloses an unmanned ship attitude adjustment method, system, equipment and medium, and relates to the technical field of intelligent unmanned ship control and hydrodynamic environment perception, and the method comprises the steps: collecting hydrodynamic related parameters of a water area where an unmanned ship is located in the navigation process and real-time state data of the unmanned ship, and carrying out the feature extraction of multi-source hydrodynamic data; constructing a hydrodynamic change state recognition model based on an adaptive clustering neural network model, inputting the extracted multi-source hydrodynamic characteristic data into the model, combining model output with current state data of the unmanned ship, constructing a fine tuning control function, and generating a correction instruction through the fine tuning control function; the unmanned ship is adjusted according to the correction instruction, the navigation state of the adjusted unmanned ship is monitored, and the fine adjustment control function is dynamically optimized through a self-learning strategy. According to the method, multi-source hydrodynamic parameters and attitude state data are collected, input feature vectors are constructed, and complex disturbance types such as backflow, hidden surge and lateral vortex are accurately recognized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent unmanned boat control and hydrodynamic environment perception, and specifically provides an unmanned boat attitude adjustment method, system, device and medium. Background Technique

[0002] With the wide application of intelligent surface unmanned platforms in scenarios such as environmental monitoring, hydrographic surveying, water area security and emergency rescue, their autonomous navigation and attitude control capabilities have become core technical indicators. During the actual operation of an unmanned boat, it is often necessary to maintain a stable attitude in a complex and changeable hydrodynamic environment to ensure mission accuracy and structural safety. However, most of the existing attitude control systems are adjusted based on traditional proportional-integral-derivative control or linear control models, and rely on fixed parameters to adjust the speed, rudder angle or acceleration, making it difficult to adapt to non-linear environmental factors such as sudden water flow disturbances, swells, and vortices.

[0003] Currently, some studies have introduced intelligent algorithms such as machine learning or fuzzy control for hull attitude adjustment, but most of them focus on macro motion control such as path tracking and target avoidance, and still lack targeted strategies for fine attitude control under micro disturbances. Especially, there is a lack of a dynamic coupling mechanism between hydrodynamic disturbance identification and rudder surface response, resulting in a lag in rudder angle control response and insufficient attitude adjustment accuracy. In addition, traditional systems lack the ability to adaptively optimize control strategies and cannot continuously evolve and update the control model based on historical operation results. Summary of the Invention

[0004] In view of the above problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: the existing unmanned boat attitude control technology methods have insufficient ability to identify sudden hydrodynamic disturbances, low rudder surface adjustment accuracy, poor attitude response robustness, and how to achieve fine rudder angle adjustment and dynamic stable coupling control of attitude in a variable water area environment.

[0006] To solve the above technical problems, the present invention provides the following technical solution: an unmanned boat attitude adjustment method, which includes collecting hydrodynamic-related parameters of the water area where the unmanned boat is located during navigation and real-time state data of the unmanned boat, and extracting features from multi-source hydrodynamic data.

[0007] Based on an adaptive clustering neural network model, a hydrodynamic change state recognition model is constructed. The extracted multi-source hydrodynamic feature data is input into the model, and the model output is combined with the current state data of the unmanned boat to construct a fine-tuning control function, and a correction instruction is generated through the fine-tuning control function.

[0008] The unmanned boat is adjusted according to the correction instruction and the navigation state of the adjusted unmanned boat is monitored, and the fine-tuning control function is dynamically optimized through a self-learning strategy.

[0009] Constructing the fine-tuning control function includes calculating the similarity weight between the speed state and the reference speed mode, the non-linear interaction effect between the current pitch angle and the reference pressure gradient, the dynamic suppression mechanism of the acceleration difference, and the periodic memory offset of the disturbance response by using the reference disturbance parameter set corresponding to the obtained disturbance category label and the currently collected navigation state data of the unmanned boat as inputs.

[0010] According to the obtained nested coupling relationship of the functions, the fine-tuning control function is obtained and the rudder angle correction instruction is generated according to the fine-tuning control function.

[0011] As a preferred solution of the unmanned boat attitude adjustment method described in the present invention, wherein: the collection of the hydrodynamic related parameters of the water area where the unmanned boat is located during navigation includes collecting the flow velocity, water pressure, and water temperature data of the water area where the unmanned boat is located through a flow meter, a multi-point arranged water pressure sensor, and a temperature sensor.

[0012] Collecting the real-time state data of the unmanned boat includes collecting the rudder angle feedback, speed vector, acceleration, and attitude angle of the unmanned boat itself in the current state through a servo position encoder, GPS+IMU fusion, an IMU accelerometer, and an IMU gyroscope.

[0013] As a preferred solution of the unmanned boat attitude adjustment method described in the present invention, wherein: the feature extraction and pattern recognition of the multi-source hydrodynamic data include converting the hydrodynamic related parameters of the water area where the unmanned boat is located during navigation and the real-time state data of the unmanned boat into structured flow velocity change rate, water pressure gradient, lateral attitude change rate, acceleration modulus change rate, and rudder angle response error through a feature extraction formula, and forming an input vector with the flow velocity change rate, water pressure gradient, lateral attitude change rate, acceleration modulus change rate, and rudder angle response error.

[0014] As a preferred solution of the unmanned boat attitude adjustment method described in the present invention, wherein: constructing the hydrodynamic change state recognition model based on the adaptive clustering neural network model includes inputting the input vector into the adaptive clustering neural network model, and performing dynamic center expansion and weight adjustment in the input space by using the minimum distance matching function to obtain the reference disturbance parameter set corresponding to the disturbance category label.

[0015] When the distance between the current sample and all existing clustering centers exceeds the dynamic threshold, a new category node is automatically generated, and the current input is initialized as the new category center, and the disturbance parameter template structure of this category is recorded at the same time.

[0016] Obtaining the reference perturbation parameter set corresponding to the perturbation category label includes setting a set of predefined perturbation category template libraries, corresponding each template to a perturbation type, storing the training data generated by model training optimization in the template library, recording the statistical feature mean, coefficient of variation, and empirical model parameters of the perturbation category of the training data under historical samples, and indexing the corresponding reference perturbation parameter set in the template library when the perturbation category label is output.

[0017] Recording the training data includes updating the statistical parameters corresponding to each type of perturbation in the template library in real time by means of exponential moving average and weighted historical window.

[0018] As a preferred solution of the unmanned boat attitude adjustment method described in the present invention, wherein: constructing the fine-tuning control function includes using the reference perturbation parameter set corresponding to the obtained perturbation category label and the currently collected navigation state data of the unmanned boat as inputs, calculating the similarity weight between the current speed state and the reference speed mode through a Gaussian kernel function, modeling the nonlinear interaction between the current pitch angle and the reference pressure gradient through a hyperbolic tangent function, constructing a dynamic suppression mechanism for the acceleration difference through an exponential decay function, and introducing a sine function to simulate the periodic memory offset of the perturbation response.

[0019] According to the obtained function nesting and coupling relationship, the fine-tuning control function is obtained and the rudder angle correction instruction is generated according to the fine-tuning control function.

[0020] The rudder angle correction instruction includes restricting the rudder angle correction instruction according to the actual maximum rudder angle deflection range of the unmanned boat, trimming and compressing the generated rudder angle correction instruction according to the actual maximum rudder angle deflection range until the boundary value of the actual maximum rudder angle deflection range is reached.

[0021] As a preferred solution of the unmanned boat attitude adjustment method described in the present invention, wherein: adjusting the unmanned boat according to the correction instruction includes superimposing the rudder angle correction instruction and the original basic control instruction to form the final executed rudder angle instruction, sending the final executed rudder angle instruction to the rudder surface servo mechanism of the unmanned boat through the execution controller, and driving the rudder machine to adjust to the specified angle according to the final executed rudder angle instruction to complete the dynamic response control of the current attitude.

[0022] Monitoring the navigation state of the unmanned boat after adjustment includes collecting the real-time state feedback data after the adjustment action is executed, comparing the collected feedback data with the system expected attitude or stable state target, and calculating the attitude error function.

[0023] The real-time state feedback data includes attitude angle parameters, acceleration vectors, actual feedback values of the rudder angle, input of perturbation characteristics and correction values at corresponding moments before and after adjustment.

[0024] As a preferred solution of the unmanned boat attitude adjustment method described in the present invention, wherein: the dynamic optimization of the fine-tuning control function includes calculating the attitude error function of the current control process according to the adjusted navigation state of the unmanned boat, determining the change trend of the error value within a set time window, and when the error function continuously exceeds the preset control error threshold, marking the current control result as a low-response sample and updating the control parameters.

[0025] Construct an optimization data set from the input feature vectors, disturbance class labels, output correction amounts, and feedback attitude response values in the low-response samples.

[0026] Adjust the exponential decay factor based on the relationship between the feedback response amplitude and acceleration difference according to the optimization data set, adjust the memory offset factor based on the historical offset trend of the rudder angle correction, and perform batch sliding window optimization on the coupling function term used to model the disturbance effect in the function structure.

[0027] Another object of the present invention is to provide an unmanned boat attitude adjustment system, which can solve the problem that the current unmanned boat attitude adjustment technology lacks a fine-grained modeling and dynamic control mechanism for the response to hydrodynamic disturbances by constructing a rudder angle fine-tuning control module based on disturbance recognition and multi-function chimerism.

[0028] As a preferred solution of the unmanned boat attitude adjustment system described in the present invention, wherein: it includes a collection and extraction module, a fine-tuning and correction module, and a monitoring and optimization module. The collection and extraction module is used to collect the hydrodynamic-related parameters of the water area where the unmanned boat is located during navigation and the real-time state data of the unmanned boat, and extract the features of the multi-source hydrodynamic data. The fine-tuning and correction module is used to construct a hydrodynamic change state recognition model based on the adaptive clustering neural network model, input the extracted multi-source hydrodynamic feature data into the model, combine the model output with the current state data of the unmanned boat to construct a fine-tuning control function, and generate a correction instruction through the fine-tuning control function. The monitoring and optimization module is used to adjust the unmanned boat according to the correction instruction and monitor the navigation state of the adjusted unmanned boat, and dynamically optimize the fine-tuning control function through a self-learning strategy.

[0029] A computer device includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the unmanned boat attitude adjustment method.

[0030] A computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, it implements the steps of the unmanned boat attitude adjustment method.

[0031] Advantages of the present invention: The unmanned boat attitude adjustment method provided by the present invention can accurately identify complex disturbance types such as backflows, surges, and lateral vortices by collecting multi-source hydrodynamic parameters and attitude state data, constructing input feature vectors, and introducing an adaptive clustering neural network for disturbance recognition, thereby enhancing the system's ability to recognize environmental changes and robustness. By constructing a non-linear fine-tuning control function that includes a coupling structure of a Gaussian kernel function, a hyperbolic tangent function, an exponential decay function, and a sine function, the current state is deeply combined with the disturbance reference parameters, improving the response sensitivity and adjustment accuracy of the rudder angle control to non-linear disturbances. By executing the correction instruction and real-time monitoring the attitude response of the unmanned boat after adjustment, and combining the attitude error function and the time threshold determination mechanism, the control effect can be quantitatively evaluated, providing an objective basis for subsequent function optimization. The present invention achieves better results in terms of the accuracy of hydrodynamic disturbance recognition, the fineness of rudder surface control, and the system's adaptability and self-learning ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0033] Figure 1 It is the overall flowchart of an unmanned boat attitude adjustment method provided by the first embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the drawings of the specification. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0035] Example 1, referring to Figure 1 , which is an embodiment of the present invention, provides an unmanned boat attitude adjustment method, including:

[0036] S1: Collect the hydrodynamic parameters of the water area where the unmanned boat is located during navigation and the real-time state data of the unmanned boat, and extract the features of the multi-source hydrodynamic data.

[0037] Collect the flow velocity, water pressure, and water temperature data of the water area where the unmanned boat is located through a flow meter, a multi-point arranged water pressure sensor, and a temperature sensor.

[0038] Collecting the real-time status data of the unmanned boat includes collecting the rudder angle feedback, velocity vector, acceleration, and attitude angle of the unmanned boat itself in the current state through the rudder position encoder, GPS+IMU fusion, IMU accelerometer, and IMU gyroscope.

[0039] Convert the hydrodynamic-related parameters of the water area where the unmanned boat is located during navigation and the real-time status data of the unmanned boat into structured flow rate change rate, water pressure gradient, lateral attitude change rate, acceleration modulus change rate, and rudder angle response error through the feature extraction formula, and form an input vector with the flow rate change rate, water pressure gradient, lateral attitude change rate, acceleration modulus change rate, and rudder angle response error.

[0040] A preferred scheme for forming the input vector is:

[0041]

[0042] Among them, Δv t represents the velocity change rate, represents the water pressure gradient, Δψ t represents the lateral attitude change rate, represents the acceleration change rate, ε θ represents the rudder angle response error, X t represents the input vector.

[0043] S2: Construct a hydrodynamic change state recognition model based on the adaptive clustering neural network model, input the extracted multi-source hydrodynamic feature data into the model, combine the model output with the current state data of the unmanned boat, construct a fine-tuning control function, and generate a correction instruction through the fine-tuning control function.

[0044] Input the input vector into the adaptive clustering neural network model, and perform dynamic center expansion and weight adjustment in the input space using the minimum distance matching function to obtain the reference perturbation parameter set corresponding to the perturbation category label.

[0045] A preferred scheme for judging the category label using the minimum distance matching function is:

[0046]

[0047] Among them, C k represents the currently recognized perturbation category label, arg min represents taking the index value that makes the subsequent function value the smallest, j represents the candidate category index, X t represents the feature vector input at the current moment, W j represents the clustering center vector corresponding to the perturbation category, ||·|| represents the Euclidean norm of the vector, ||X t -W j || 2 represents the squared distance between the input vector and the clustering center.

[0048] When the distances between the current sample and all existing clustering centers exceed the dynamic threshold, a new category node is automatically generated, and the current input is initialized as the new category center. Meanwhile, the perturbation parameter template structure of this category is recorded.

[0049] Obtaining the reference perturbation parameter set corresponding to the perturbation category label includes setting a group of predefined perturbation category template libraries, corresponding each template to the perturbation type, storing the training data generated by model training optimization in the template library, and recording the statistical feature means, coefficient of variation, and empirical model parameters of the perturbation category of the training data under historical samples. When the perturbation category label is output, the corresponding reference perturbation parameter set in the template library is indexed.

[0050] A preferred scheme for obtaining the reference perturbation parameter set corresponding to the perturbation category label is:

[0051]

[0052] Where represents the reference velocity vector of perturbation category k at the (t + 1)-th update, represents the reference velocity vector of perturbation category k at the t-th update, represents the actual velocity vector collected by the current unmanned boat, ρ represents the exponential moving average weight factor, and (1 - ρ) represents the retention ratio of the historical velocity to the new template value.

[0053] When Create a new category C k+1 And initialize the center weight W k+1 = X t , and establish the perturbation category template T k+1 .

[0054] Recording the training data includes updating the statistical parameters corresponding to each type of perturbation in the template library in real time through the exponential moving average and weighted historical window methods.

[0055] Taking the reference perturbation parameter set corresponding to the obtained perturbation category label and the current navigation state data of the collected unmanned boat as inputs, calculating the similarity weight between the current velocity state and the reference velocity pattern through the Gaussian kernel function, modeling the non-linear interaction between the current pitch angle and the reference pressure gradient through the hyperbolic tangent function, constructing a dynamic suppression mechanism for the acceleration difference through the exponential decay function, and introducing the sine function to simulate the periodic memory offset of the perturbation response.

[0056] A preferred scheme for constructing the fine-tuning control function is:

[0057]

[0058] Among them, A represents the speed similarity weight coefficient, e represents the base of the natural logarithm, v m represents the current unmanned boat speed vector, v k represents the reference speed vector of disturbance category k, a k represents the speed diffusion coefficient of disturbance category k, B represents the attitude-pressure nonlinear interaction expression term, Δp k represents the reference pressure gradient of disturbance category k, γ m represents the pitch angle of the current unmanned boat, η m represents the acceleration modulus value of the current unmanned boat, ψ k represents the reference roll angle of disturbance category k, δ θ represents the output rudder angle correction amount, β k represents the rudder angle sensitivity factor of disturbance category k, tanh(·) represents the hyperbolic tangent function, λ k represents the exponential decay factor of disturbance category k, represents the acceleration vector of the current unmanned boat, represents the reference acceleration vector of disturbance category k, ξ k represents the periodic disturbance memory factor of disturbance category k, sin(·) represents the sine function, and θ0 represents the current basic rudder angle control instruction.

[0059] According to the obtained nested coupling relationship of functions, a fine-tuning control function is obtained and a rudder angle correction instruction is generated according to the fine-tuning control function.

[0060] A preferred scheme for generating the correction instruction is:

[0061] θ c = θ0 + δ θ

[0062] Among them, θ c represents the final executed rudder angle instruction value.

[0063] [[ID=4&]]The rudder angle correction instruction includes that the rudder angle correction instruction is restricted according to the actual maximum rudder angle deflection range of the unmanned boat, and the generated rudder angle correction instruction is trimmed and compressed according to the actual maximum rudder angle deflection range until the boundary value of the actual maximum rudder angle deflection range is reached.

[0064] A preferred scheme for restricting according to the actual maximum rudder angle deflection range of the unmanned boat is:

[0065] θ c = min(max(θ c , -θ max ), θ max )

[0066] Among them, θ max represents the upper limit of the maximum rudder angle deflection range of the unmanned boat, θmin Represents the lower limit of the minimum rudder angle deflection range of the unmanned boat.

[0067] S3: Adjust the unmanned boat according to the correction instruction and monitor the navigation state of the unmanned boat after adjustment, and dynamically optimize the fine-tuning control function through the self-learning strategy.

[0068] Superimpose the rudder angle correction instruction and the original basic control instruction to form the final executed rudder angle instruction, and send the final executed rudder angle instruction to the rudder surface servo mechanism of the unmanned boat through the execution controller, and drive the rudder machine to adjust to the specified angle according to the final executed rudder angle instruction to complete the dynamic response control of the current attitude.

[0069] Monitoring the navigation state of the unmanned boat after adjustment includes collecting real-time state feedback data after the adjustment action is executed, comparing the collected feedback data with the system expected attitude or stable state target, and calculating the attitude error function.

[0070] A preferred scheme for calculating the attitude error function is:

[0071]

[0072] Among them, ε(t) represents the attitude error function, φ t represents the roll angle at the current moment, φ ref represents the target roll angle, ψ t represents the current roll angle, ψ ref represents the reference roll angle, γ t represents the reference roll angle, γ ref represents the target pitch angle.

[0073] The real-time state feedback data includes attitude angle parameters, acceleration vectors, actual feedback values of rudder angles, disturbance characteristic inputs and correction values at corresponding moments before and after adjustment.

[0074] According to the navigation state of the unmanned boat after adjustment, calculate the attitude error function of the current control process, judge the change trend of the error value within the set time window, and when the error function continuously exceeds the preset control error threshold, mark the current control result as a low-response sample and update the control parameters.

[0075] Construct the input feature vector, disturbance category label, output correction amount, and feedback attitude response value in the low-response sample into an optimization data set.

[0076] Adjust the exponential decay factor according to the relationship between the feedback response amplitude and the acceleration difference based on the optimization data set, adjust the memory offset factor based on the historical offset trend of the rudder angle correction, and perform batch sliding window optimization on the coupling function term used to model the disturbance effect in the function structure.

[0077] It should be noted that the control function after parameter update will replace the original function when the next disturbance category is triggered, so as to achieve dynamic control optimization with long-term adaptability.

[0078] A preferred scheme for self-learning optimization of the control function is as follows:

[0079]

[0080] Among them, represents the rudder angle sensitivity factor of disturbance category k at the t-th iteration, η represents the learning rate, represents the partial derivative of the error function with respect to the sensitivity factor.

[0081] The optimization process is carried out in an online update manner, and the control effect after each round of optimization is recorded as the evaluation index of the new round of samples, so as to realize the continuous evolution ability and adaptive learning ability of the unmanned boat attitude adjustment control logic.

[0082] Embodiment 2 is an embodiment of the present invention, which provides an unmanned boat attitude adjustment method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0083] First of all, in order to verify the practicability and control advantages of the unmanned boat attitude adjustment method under complex hydrodynamic disturbance conditions, a set of simulation experiments are constructed to simulate the medium-speed navigation task of the unmanned boat in the typical nearshore shallow water area. The test scenario is set with disturbance sources including typical unstable factors such as flow velocity change, vortex disturbance, water pressure unevenness, wind and wave interference, aiming to verify the attitude maintenance, correction ability and self-learning response performance of the system under multi-source disturbances.

[0084] The test comparison object is the traditional PID controller system and the self-learning fine-tuning control deployed by the scheme of the present invention. The test platform is a general modular unmanned boat simulation architecture, and the control layer supports plug-and-play control logic switching. The collected data includes rudder angle feedback, attitude angle, flow velocity, water pressure, acceleration modulus, etc., which are recorded in real time through IMU+GPS fusion and three-axis hydrodynamic sensors, and are transmitted to the control decision at a frequency of 50Hz per second. For the present invention, an adaptive clustering neural network is used to classify the multi-source disturbance characteristics collected, automatically identify the corresponding disturbance parameter templates, and calculate key elements such as velocity similarity, Gaussian kernel distance, pitch angle-pressure relationship, etc. to construct a rudder angle fine-tuning control function. The control function outputs correction instructions in real time, which are superimposed on the original control instructions of the steering gear system and then sent out. After the rudder surface is executed, the adjustment feedback is continuously collected and the attitude error function is calculated. If the error is higher than the threshold for a long time, the control factor learning optimization mechanism is started to automatically correct the disturbance sensitivity coefficient and the weight of the periodic response function.

[0085] To evaluate the performance of the method, 6 groups of sample tests using the present invention were recorded and compared with 1 group of traditional PID control. The whole process of the unmanned boat from the occurrence of disturbance to attitude stabilization was continuously monitored for each group of data, and key performance indicators including the maximum attitude error, rudder angle correction amplitude, response delay, stabilization time, error self-learning optimization amplitude, etc. were statistically analyzed. The experimental data is shown in Table 1.

[0086] Table 1 Experimental data table

[0087]

[0088] As can be seen from the tabular data, when using traditional PID, the maximum attitude error reaches 3.8°. Among the six groups of samples using the fine-tuning control system of the present invention, the maximum attitude error is controlled between 1.8° and 2.2°, and the average reduction amplitude exceeds 41%. This advantage stems from the disturbance recognition mechanism introduced in the present invention, which can dynamically load the most matching response template according to the disturbance type and perform fine correction in combination with the current state, avoiding the "under-regulation" or "over-regulation" problems caused by traditional PID relying on fixed parameters.

[0089] In terms of the average rudder angle correction amplitude, the control object is 12.5°, and the control systems of each group of the present invention fluctuate between 7.6° and 8.2°, indicating that the present invention not only improves the attitude response efficiency, but also reduces the consumption of rudder angle resources, improves the energy control efficiency, and is suitable for long endurance scenarios. In terms of the disturbance recognition accuracy rate, each sample of the present invention exceeds 91%, while traditional control does not have this ability, further verifying the effectiveness of the adaptive clustering recognition model in modeling navigation disturbances. In terms of control delay, the response time of the present invention is controlled between 202 ms and 215 ms, which has a significant advantage compared with 350 ms of PID control, and has a higher course recovery ability especially when facing rapid change disturbances.

[0090] Particularly noteworthy is that the attitude error self-learning optimization rate of the present invention remains between 17.8% and 20.1% under continuous operation, indicating that it has obvious parameter self-evolution ability and can adapt to environmental changes for a long time without manual intervention. Due to the lack of an online learning mechanism, the control effect of traditional PID gradually decreases after multiple disturbances, and the risk of instability increases.

[0091] The experimental data proves that the proposed method is comprehensively superior to the traditional control method in terms of disturbance recognition accuracy, attitude correction amplitude, rudder angle energy consumption control, response delay, and system self-learning performance, and has high innovation and engineering feasibility, especially suitable for high-precision autonomous control tasks in dynamically complex waters.

[0092] Embodiment 3, which is an embodiment of the present invention, provides an unmanned boat attitude adjustment system, including a collection and extraction module 100, a fine-tuning correction module 200, and a monitoring and optimization module 300.

[0093] Among them, S4: The acquisition and extraction module 100 is used to acquire the hydrodynamic related parameters of the water area where the unmanned boat is located during navigation and the real-time state data of the unmanned boat, and extract the features of the multi-source hydrodynamic data.

[0094] It should also be noted that the acquisition and extraction module 100, as the input layer of the system, is responsible for acquiring the multi-source hydrodynamic parameters of the water area where the unmanned boat is located and the state data of the boat body itself in real time during the navigation of the unmanned boat. This module not only completes the data acquisition, but also undertakes the task of feature extraction of the original multi-dimensional data, including the structured generation of key indicators such as the flow rate change rate, the water pressure gradient, the roll angle change rate, and the rudder angle response error. These feature data are used as input vectors and directly called by the fine-tuning and correction module 200.

[0095] S5: The fine-tuning and correction module 200 is used to construct a hydrodynamic change state recognition model based on the adaptive clustering neural network model, input the extracted multi-source hydrodynamic feature data into the model, combine the model output with the current state data of the unmanned boat, construct a fine-tuning control function, and generate a correction instruction through the fine-tuning control function.

[0096] It should also be noted that the fine-tuning and correction module 200 is the central decision-making unit of the system. After receiving the input vector from the acquisition and extraction module 100, it inputs it into the disturbance state recognition model constructed based on the adaptive clustering neural network, automatically judges the current hydrodynamic disturbance type and matches the corresponding disturbance parameter template. Subsequently, this module combines the disturbance category output with the current state of the unmanned boat, constructs a fine-tuning control function, and generates a rudder angle correction instruction for fine-tuning the attitude. The generated instruction is then sent to the servo system of the rudder through the system control bus for execution.

[0097] S6: The monitoring and optimization module 300 is used to adjust the unmanned boat according to the correction instruction and monitor the navigation state of the unmanned boat after adjustment, and dynamically optimize the fine-tuning control function through the self-learning strategy.

[0098] It should also be noted that the monitoring and optimization module 300 is located in the output-feedback layer of the system. On the one hand, it acquires the actual attitude state of the unmanned boat after executing the correction instruction and the sensor feedback data, and calculates the current attitude error function; on the other hand, it judges the effectiveness of the control strategy according to the feedback data. When the error continuously exceeds the set threshold, this module triggers the self-learning optimization process, constructs an optimization data set based on historical samples, updates the parameter items in the fine-tuning control function online, and returns the updated control factor to the fine-tuning and correction module 200 to realize the dynamic evolution of the control model.

[0099] If a function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0100] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0101] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROM). Additionally, a computer-readable medium can even be paper or other suitable media on which a program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or processing it in other suitable ways as necessary, and then storing it in a computer memory.

[0102] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following techniques well known in the art or a combination thereof can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for adjusting the attitude of an unmanned boat, characterized in that, Including: Collect the hydrodynamic related parameters of the water area where the unmanned boat is located during navigation and the real-time state data of the unmanned boat, and extract the features of multi-source hydrodynamic data; Based on the adaptive clustering neural network model, construct a hydrodynamic change state recognition model, input the extracted multi-source hydrodynamic feature data into the model, combine the model output with the current state data of the unmanned boat, construct a fine-tuning control function, and generate a correction instruction through the fine-tuning control function; Adjust the unmanned boat according to the correction instruction and monitor the navigation state of the unmanned boat after adjustment, and dynamically optimize the fine-tuning control function through the self-learning strategy; Constructing the fine-tuning control function includes calculating the similarity weight between the speed state and the reference speed mode, the non-linear interaction effect between the current pitch angle and the reference pressure gradient, the dynamic suppression mechanism of the acceleration difference, and the periodic memory offset of the disturbance response by taking the reference disturbance parameter set corresponding to the obtained disturbance category label and the currently collected navigation state data of the unmanned boat as inputs; According to the obtained function nesting and coupling relationship, obtain the fine-tuning control function and generate a rudder angle correction instruction according to the fine-tuning control function.

2. The unmanned boat attitude adjustment method according to claim 1, wherein: The collection of the hydrodynamic related parameters of the water area where the unmanned boat is located during navigation includes Collecting the flow velocity, water pressure, and water temperature data of the water area where the unmanned boat is located through a flow velocity meter, a water pressure sensor arranged at multiple points, and a temperature sensor; Collecting the real-time state data of the unmanned boat includes collecting the rudder angle feedback, velocity vector, acceleration, and attitude angle of the unmanned boat itself in the current state through a rudder position encoder, GPS+IMU fusion, an IMU accelerometer, and an IMU gyroscope.

3. The unmanned boat attitude adjustment method according to claim 1 or 2, characterized in that: The feature extraction and pattern recognition of the multi-source hydrodynamic data include Converting the hydrodynamic related parameters of the water area where the unmanned boat is located during navigation and the real-time state data of the unmanned boat into structured flow velocity change rate, water pressure gradient, lateral attitude change rate, acceleration modulus change rate, and rudder angle response error through a feature extraction formula, and forming an input vector with the flow velocity change rate, water pressure gradient, lateral attitude change rate, acceleration modulus change rate, and rudder angle response error.

4. The unmanned boat attitude adjustment method according to claim 3, characterized in that: The construction of the hydrodynamic change state recognition model based on the adaptive clustering neural network model includes Input the input vector into the adaptive clustering neural network model, and use the minimum distance matching function to perform dynamic center expansion and weight adjustment in the input space to obtain the reference disturbance parameter set corresponding to the disturbance category label; When the distance between the current sample and all existing cluster centers exceeds the dynamic threshold, automatically generate a new category node, initialize the current input as the new category center, and record the disturbance parameter template structure of this category at the same time; Obtaining the reference disturbance parameter set corresponding to the disturbance category label includes setting a set of predefined disturbance category template libraries, corresponding each template to the disturbance type, storing the training data generated by model training and optimization into the template library, recording the statistical feature mean, coefficient of variation, and empirical model parameters of the disturbance category of the training data under historical samples, and indexing the corresponding reference disturbance parameter set in the template library when the disturbance category label is output. The recorded training data includes updating in real time the statistical parameters corresponding to each type of perturbation in the template library by means of exponential moving average and weighted historical window.

5. The unmanned boat attitude adjustment method according to any one of claims 1, 2 or 4, characterized in that: The construction of the fine-tuning control function includes taking the reference perturbation parameter set corresponding to the obtained perturbation category label and the currently collected navigation state data of the unmanned boat as inputs, calculating the similarity weight between the current speed state and the reference speed mode through a Gaussian kernel function, modeling the non-linear interaction effect between the current pitch angle and the reference pressure gradient through a hyperbolic tangent function, constructing a dynamic suppression mechanism for the acceleration difference through an exponential decay function, and introducing a sine function to simulate the periodic memory offset of the perturbation response; obtaining the fine-tuning control function according to the obtained function nested coupling relationship and generating a rudder angle correction instruction according to the fine-tuning control function; The rudder angle correction instruction includes that the rudder angle correction instruction is limited according to the actual maximum rudder angle deflection range of the unmanned boat, and the generated rudder angle correction instruction is trimmed and compressed according to the actual maximum rudder angle deflection range until the boundary value of the actual maximum rudder angle deflection range is reached.

6. The unmanned boat attitude adjustment method according to claim 5, characterized in that: The adjustment of the unmanned boat according to the correction instruction includes superimposing the rudder angle correction instruction and the original basic control instruction to form the final executed rudder angle instruction, sending the final executed rudder angle instruction to the rudder surface servo mechanism of the unmanned boat through the execution controller, and driving the rudder machine to adjust to the specified angle according to the final executed rudder angle instruction to complete the dynamic response control of the current attitude; Monitoring the navigation state of the unmanned boat after adjustment includes collecting real-time state feedback data after the adjustment action is executed, comparing the collected feedback data with the expected attitude or stable state target of the system, and calculating the attitude error function; The real-time state feedback data includes attitude angle parameters, acceleration vectors, actual feedback values of the rudder angle, perturbation feature inputs and correction values at corresponding times before and after adjustment.

7. The unmanned boat attitude adjustment method according to any one of claims 1, 2, 4 or 6, characterized in that: The dynamic optimization of the fine-tuning control function includes calculating the attitude error function of the current control process according to the navigation state of the unmanned boat after adjustment, determining the change trend of the error value within the set time window, and when the error function continuously exceeds the preset control error threshold, marking the current control result as a low-response sample and updating the control parameters; constructing an optimization data set from the input feature vectors, perturbation category labels, output correction amounts, and feedback attitude response values in the low-response samples; adjusting the exponential decay factor according to the relationship between the feedback response amplitude and the acceleration difference based on the optimization data set, adjusting the memory offset factor according to the historical offset trend of the rudder angle correction, and performing batch sliding window optimization on the coupling function items used to model the perturbation effect in the function structure.

8. A system adopting the unmanned boat attitude adjustment method according to any one of claims 1 to 7, characterized in that: including an acquisition and extraction module (100), a fine-tuning and correction module (200), and a monitoring and optimization module (300); The acquisition and extraction module (100) is used to acquire the hydrodynamic related parameters of the water area where the unmanned boat is located during navigation and the real-time state data of the unmanned boat, and extract the features of the multi-source hydrodynamic data The fine-tuning and correction module (200) is used to construct a hydrodynamic change state recognition model based on the adaptive clustering neural network model, input the extracted multi-source hydrodynamic feature data into the model, combine the model output with the current state data of the unmanned boat, construct a fine-tuning control function, and generate a correction instruction through the fine-tuning control function; The monitoring and optimization module (300) is used to adjust the unmanned boat according to the correction instruction and monitor the navigation state of the unmanned boat after adjustment, and dynamically optimize the fine-tuning control function through a self-learning strategy.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the unmanned boat attitude adjustment method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the unmanned boat attitude adjustment method according to any one of claims 1 to 7 are implemented.

Images