Intelligent control method and system for underwater magnetic coupling propeller

Abstract

The invention relates to the technical field of underwater propulsion, and discloses an intelligent control method and system for an underwater magnetic coupling propeller. The method comprises the following steps: detecting a magnetic field of a magnetic coupling unit through a Hall sensor array to obtain magnetic flux density distribution data; permanent magnet array hysteresis characteristics are modeled according to the magnetic flux density distribution data, and hysteresis compensation parameters are obtained; inputting the hysteresis compensation parameter into a magnetic field genetic algorithm to optimize magnetic pole configuration to obtain an optimal magnetic pole phase difference; distributing a torque control instruction according to the optimal magnetic pole phase difference to obtain a target excitation current value; the magnetic field is adjusted through the hysteresis compensation PID controller, and magnetic coupling torque output is obtained. By establishing an intelligent control method of hysteresis characteristic accurate modeling, magnetic field genetic optimization and hysteresis compensation PID control, the control precision and transmission efficiency of the magnetic coupling propeller in a complex marine environment are improved.

Description

Technical Field

[0001] The present application relates to the field of underwater propulsion technology, and in particular to an intelligent control method and system for an underwater magnetic coupling thruster. Background Art

[0002] Traditional underwater thrusters primarily utilize mechanical seals for power transmission, with a rotating shaft passing through a sealed housing to drive a propeller to generate propulsion. These thrusters are widely used in underwater robots, submersibles, and marine engineering equipment. Their control methods typically rely on motor speed regulation and propeller pitch control to precisely adjust the propulsion force. As marine engineering expands into harsh environments such as the deep sea and polar regions, higher requirements are placed on the reliability and control accuracy of thrusters.

[0003] However, existing mechanical seal thrusters present key challenges, including a high risk of seal leakage, expensive maintenance costs, and insufficient long-term reliability. Mechanical seals are prone to failure in high-pressure deep-sea environments, leading to seawater intrusion and equipment damage. Furthermore, traditional speed control methods are unable to effectively cope with dynamic load changes in complex marine environments. While magnetically coupled thrusters address seal leakage, existing magnetic coupling control methods primarily employ simple open-loop control or traditional PID control, lacking effective compensation for hysteresis nonlinearity and the ability to coordinate control between multiple magnetic coupling units.

[0004] Further analysis revealed that magnetically coupled thrusters face progressive technical challenges in practical applications, including the hysteresis effect affecting control accuracy, magnetic field interference between multiple magnetic coupling units leading to reduced transmission efficiency, and a lack of adaptive optimization algorithms to address different operating conditions. The root cause of these issues is the failure of existing control methods to establish an accurate mathematical model of hysteresis characteristics, the lack of intelligent optimization algorithms that consider magnetic field interactions, and the lack of a comprehensive control strategy that integrates hysteresis compensation and real-time feedback. This limits the control performance and operational reliability of magnetically coupled thrusters in complex marine environments. Summary of the Invention

[0005] This application provides an intelligent control method and system for underwater magnetic coupling thrusters, addressing the issues of hysteresis nonlinearity affecting control accuracy and magnetic field interference between multiple magnetic coupling units, which reduces transmission efficiency. By establishing an intelligent control method with precise hysteresis characteristic modeling, magnetic field genetic optimization, and hysteresis-compensated PID control, the control accuracy and transmission efficiency of magnetic coupling thrusters in complex marine environments are improved.

[0006] In the first aspect, the present application provides an intelligent control method for an underwater magnetic coupling thruster, which includes: performing magnetic field detection processing on the magnetic coupling unit through a Hall sensor array to obtain magnetic flux density distribution data; performing hysteresis characteristic modeling processing on the permanent magnet array according to the magnetic flux density distribution data to obtain hysteresis compensation parameters; inputting the hysteresis compensation parameters into a magnetic field genetic algorithm to optimize the magnetic pole configuration to obtain an optimal magnetic pole phase difference; performing magnetic field perception Lagrangian distribution processing on the torque control instruction according to the optimal magnetic pole phase difference to obtain a target excitation current value; performing magnetic field regulation processing on the target excitation current value through a hysteresis compensation PID controller to obtain a magnetic coupling torque output.

[0007] In a second aspect, the present application provides an underwater magnetic coupling thruster intelligent control system, the underwater magnetic coupling thruster intelligent control system comprising: A detection module is used to detect and process the magnetic field of the magnetic coupling unit through a Hall sensor array to obtain magnetic flux density distribution data; A modeling module, configured to perform hysteresis characteristic modeling on the permanent magnet array according to the magnetic flux density distribution data to obtain hysteresis compensation parameters; A processing module, configured to input the hysteresis compensation parameters into a magnetic field genetic algorithm to optimize the magnetic pole configuration and obtain an optimal magnetic pole phase difference; a distribution module, configured to perform magnetic field-aware Lagrangian distribution processing on the torque control command according to the optimal magnetic pole phase difference to obtain a target excitation current value; The regulating module is used to perform magnetic field regulation processing on the target excitation current value through a hysteresis compensation PID controller to obtain a magnetic coupling torque output.

[0008] In the third aspect, an underwater magnetic coupling thruster intelligent control device is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the underwater magnetic coupling thruster intelligent control device executes the above-mentioned underwater magnetic coupling thruster intelligent control method.

[0009] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, which, when executed on a computer, enables the computer to execute the above-mentioned underwater magnetic coupling thruster intelligent control method.

[0010] In the technical solution provided by this application, the magnetic flux density distribution data obtained by the Hall sensor array provides an accurate basic data source for the subsequent hysteresis characteristic modeling, ensuring the accuracy and reliability of the hysteresis compensation parameters. The hysteresis characteristic modeling process is targeted at the special working environment of the underwater magnetic coupling thruster, and a hysteresis loop mathematical model that takes into account the influence of seawater pressure and temperature is established, which effectively solves the problem of insufficient control accuracy caused by the traditional control method ignoring the nonlinear characteristics of hysteresis. The introduction of the magnetic field genetic algorithm realizes the intelligent optimization of the magnetic pole configuration, automatically finds the optimal magnetic pole phase difference through evolutionary calculation, overcomes the limitations of manual experience setting, and significantly improves the magnetic coupling transmission efficiency. The magnetic field sensing Lagrangian allocation process converts the optimized magnetic pole configuration into a specific control allocation strategy, and reasonably allocates the load of each magnetic coupling unit through the constrained optimization method, avoiding the overload operation of a single unit and extending the service life of the equipment. The design of the hysteresis compensation PID controller combines the advantages of feedforward hysteresis compensation and feedback PID control. It can not only actively compensate for the influence of the hysteresis effect, but also respond to external interference and load changes in real time, achieving high-precision magnetic field regulation and stable torque output.

[0011] The swarm intelligence search characteristics of the magnetic field genetic algorithm are particularly suitable for solving the complex problem of multi-parameter coupling optimization in magnetic coupling systems. The global search capability of the algorithm ensures that the global optimal solution that maximizes the magnetic coupling efficiency can be found, avoiding falling into the local optimum. The magnetic field-aware Lagrangian method shows good numerical stability and convergence when dealing with magnetic saturation constraints and adjacent magnetic field interference constraints. Its enhanced Lagrangian function can effectively handle inequality constraints, ensuring that the control allocation results both meet physical constraints and achieve optimal performance. The hysteresis compensation lookup table technology in the hysteresis-compensated PID controller fully utilizes the repeatability of the hysteresis phenomenon and achieves fast and accurate feedforward compensation through pre-stored compensation data, significantly reducing the negative impact of the hysteresis effect on control performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 This is a schematic diagram of an embodiment of an intelligent control method for an underwater magnetic coupling thruster in an embodiment of the present application; Figure 2 This is a schematic diagram of an embodiment of an intelligent control system for an underwater magnetic coupling thruster in an embodiment of the present application; Figure 3 It is a schematic block diagram of the structure of the underwater magnetic coupling thruster intelligent control device in an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The embodiments of the present application provide an intelligent control method and system for an underwater magnetically coupled thruster. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or are inherent to these processes, methods, products or devices.

[0015] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In one embodiment of the present application, an intelligent control method for an underwater magnetic coupling thruster includes: Step S101: Perform magnetic field detection processing on the magnetic coupling unit through the Hall sensor array to obtain magnetic flux density distribution data; Step S102: performing hysteresis characteristic modeling on the permanent magnet array according to the magnetic flux density distribution data to obtain hysteresis compensation parameters; Step S103: inputting the hysteresis compensation parameters into the magnetic field genetic algorithm to optimize the magnetic pole configuration and obtain the optimal magnetic pole phase difference; Step S104: performing magnetic field sensing Lagrangian distribution processing on the torque control command according to the optimal magnetic pole phase difference to obtain a target excitation current value; Step S105 : The target excitation current value is subjected to magnetic field regulation processing by a hysteresis compensation PID controller to obtain a magnetic coupling torque output.

[0016] It is understandable that the execution subject of this application can be an underwater magnetic coupling thruster intelligent control system, or a terminal or a server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0017] Specifically, the Hall sensor array performs magnetic field detection processing on the magnetic coupling unit. Based on the Hall effect principle, the Hall sensor generates a voltage signal when a magnetic field passes through a semiconductor material. The magnetic field strength is determined by measuring the strength of this voltage signal. The controller performs a full-scale scan of each of the three magnetic coupling units. The magnetic field strength data collected by each sensor is converted into a digital signal via an analog-to-digital converter, forming an initial magnetic field strength data matrix. Subsequently, a three-dimensional coordinate mapping algorithm maps each magnetic field strength data point to its corresponding spatial coordinate, generating a table of correspondences between magnetic field strength and position coordinates. Vector decomposition calculates the magnetic flux density components in the three orthogonal directions of X, Y, and Z. The magnetic flux density component in each direction reflects the magnetic field strength distribution characteristics in that direction. The pressure sensor built into the oil-filled pressure compensation structure simultaneously detects the seawater pressure corresponding to the current water depth. The pressure data and the magnetic flux density components are fused using a weighted fusion algorithm to generate magnetic flux density distribution data that accounts for depth.

[0018] Hysteresis modeling is a core technical step. The hysteresis loop describes the nonlinear relationship between the magnetic flux density and magnetic field strength of a magnetic material under an alternating magnetic field. The controller extracts three key parameters from the magnetic flux density distribution data: coercivity, remanence, and saturation magnetic flux density. Coercivity represents the reverse magnetic field strength required to completely demagnetize the magnetic material, remanence represents the magnetic flux density retained after the external magnetic field is removed, and saturation magnetic flux density represents the maximum magnetic flux density at which the material reaches magnetic saturation. An excitation current sequence is applied to the permanent magnet array according to a preset step pattern. Each current value generates a specific magnetic field response. The controller collects the magnetic field response data in real time, forming a current-magnetic field response dataset. A least-squares fitting algorithm determines the mathematical model parameters of the hysteresis loop by minimizing the squared error between the measured data and the theoretical curve, thereby establishing a mathematical equation describing the hysteresis characteristics. A temperature compensation algorithm corrects the hysteresis parameters based on the magnet temperature detected by the temperature sensor. Since the coercivity and remanence of magnetic materials vary at different temperatures, the temperature compensation algorithm adjusts the hysteresis compensation parameters by looking up a temperature-magnetic parameter mapping table.

[0019] Optimizing magnetic pole configuration using a magnetic field genetic algorithm is a key innovation in intelligent control. The magnetic pole phase difference refers to the angular difference between adjacent magnetic poles. The excitation current amplitude determines the magnetic field strength, and the magnetic field direction angle determines the spatial orientation of the magnetic field. Ternary genetic encoding converts these three parameters into a binary code string, with each individual representing a magnetic pole configuration. The fitness function uses magnetic coupling transfer efficiency as an evaluation criterion, calculating the transfer efficiency value for each configuration through magnetic field simulation. A probabilistic selection operator calculates the probability of selection for each individual based on the fitness value, with individuals with higher fitness values having a greater probability of selection. Magnetic field physics-constrained crossover ensures that the new individuals generated by the crossover operation still meet the laws of magnetic field physics, avoiding unrealizable magnetic field configurations. Dynamic mutation probability automatically adjusts the probability of mutation based on the number of evolutionary generations, initially adopting a higher mutation probability to increase population diversity and later reducing the mutation probability to retain high-performing individuals. After 50 generations of iterative evolution, the algorithm converged to the optimal magnetic pole phase difference configuration that maximizes magnetic coupling efficiency.

[0020] The magnetic field-aware Lagrangian allocation process intelligently distributes torque control requirements to each magnetic coupling unit. The torque control command includes the desired propulsion force magnitude and steering torque direction. The three-magnetic coupling unit decomposition process geometrically decomposes the overall control requirement into specific propulsion force and steering torque components for each unit based on the optimal magnetic pole phase difference configuration. The constrained optimization objective function aims to minimize the energy consumption difference between each unit while satisfying magnetic saturation constraints and adjacent magnetic field interference threshold limits. The enhanced Lagrangian multiplier method transforms the constrained optimization problem into an unconstrained optimization problem by introducing Lagrangian multipliers. The magnetic field-aware factor modifies the Lagrangian multiplier value based on the deviation between the actual magnetic field strength reported by the Hall sensor and the desired magnetic field strength, ensuring real-time and accurate control allocation. The current distribution calculation calculates the target excitation current required for each magnetic coupling unit based on the modified Lagrangian multiplier value.

[0021] The hysteresis-compensated PID controller combines traditional PID control with hysteresis compensation techniques. The magnetic field control error signal is calculated from the deviation between the target excitation current and the actual magnetic field strength. A hysteresis compensation lookup table pre-stores hysteresis compensation values for different magnetic field strengths and rates of change. The controller searches for the corresponding hysteresis pre-compensation current value based on the current magnetic field state. The PID controller's proportional phase responds to the current error, the integral phase eliminates steady-state error, and the differential phase predicts the error trend. The weighted summation of the outputs of these three phases forms the basic PID control variable. The hysteresis pre-compensation current value is then added to the basic PID control variable for correction, compensating for the hysteresis effect on control accuracy. The final hysteresis-compensated PID control output drives the electromagnet to adjust the magnetic field strength, achieving precise magnetic coupling torque transmission. This control method overcomes the technical challenges of conventional thruster seal leakage, hysteresis nonlinearity, and multi-unit coordinated control. Through intelligent algorithm optimization and real-time compensation control, it ensures stable and reliable operation of underwater magnetic coupling thrusters in complex marine environments.

[0022] In a specific embodiment, the process of executing step S101 may specifically include the following steps: Performing a full-scale scanning detection process on the magnetic field strength of the three magnetic coupling units to obtain initial magnetic field strength data; The initial magnetic field intensity data is input into a three-dimensional coordinate mapping algorithm for spatial positioning processing to obtain the corresponding relationship between the magnetic field intensity and the position coordinates; Based on the corresponding relationship, the magnetic flux density is vector-decomposed to obtain the magnetic flux density components in the X, Y, and Z directions; Perform pressure detection on the oil-filled pressure compensation structure according to the magnetic flux density component to obtain the current water depth pressure parameter; The magnetic flux density component and the water depth pressure parameter are fused to obtain the magnetic flux density distribution data.

[0023] Specifically, the magnetic field detection processing of the underwater magnetic coupling thruster uses a Hall sensor array to achieve accurate magnetic field strength measurement. The Hall sensor works based on the Hall effect principle. When carriers move in a magnetic field, a potential difference perpendicular to the magnetic field and current direction is generated. The magnitude of the potential difference is proportional to the magnetic field strength. The full-range scanning detection processing refers to the Hall sensor array measuring the magnetic field around the three magnetic coupling units point by point according to a preset spatial path. Each magnetic coupling unit is equipped with multiple Hall sensors, and the sensors are evenly distributed along the circumference. During the scanning process, the sensors collect magnetic field strength values at different angular positions in sequence. The controller receives the analog voltage signal output by the sensor and converts the voltage signal into a digital quantity through an analog-to-digital converter to form an initial magnetic field strength data matrix containing the magnetic field strength value and the corresponding measurement position information. The rows of the data matrix represent different measurement points, and the columns represent the magnetic field strength value and spatial coordinate information.

[0024] The spatial positioning process of the three-dimensional coordinate mapping algorithm establishes a correspondence between each magnetic field strength data point and its precise position in three-dimensional space. The algorithm first establishes a three-dimensional rectangular coordinate system with the center of the thruster as the origin. Then, the three-dimensional coordinate value of each measurement point is calculated based on the physical installation position and scanning path of the Hall effect sensor. The spatial positioning process converts the local coordinates of the sensor into coordinate values in the global coordinate system through a coordinate transformation matrix. The coordinate transformation includes two parts: rotation transformation and translation transformation. The rotation transformation eliminates the influence of the sensor installation angle, and the translation transformation determines the position of the sensor relative to the coordinate origin. The mapping algorithm associates each magnetic field strength value in the initial magnetic field strength data with its corresponding three-dimensional coordinate, generating a correspondence table between magnetic field strength and position coordinates. The relationship table is indexed by the three-dimensional coordinate and the magnetic field strength as the numerical content, forming a complete description of the spatial magnetic field distribution.

[0025] The vector decomposition processing of magnetic flux density is based on the vector characteristics of the magnetic field. The magnetic field strength data in the correspondence table is calculated according to the components of three orthogonal directions. The vector decomposition processing first determines the directions of the three mutually perpendicular coordinate axes X, Y, and Z. The X axis points to the forward direction of the propeller, the Y axis points to the side of the propeller, and the Z axis points to the vertical direction of the propeller. The decomposition algorithm calculates the projection components of the magnetic field vector at each measurement point in the directions of the three coordinate axes based on the direction angle of the magnetic field vector. The X-direction magnetic flux density component reflects the magnetic field strength in the propulsion direction, the Y-direction magnetic flux density component reflects the magnetic field strength for lateral steering, and the Z-direction magnetic flux density component reflects the magnetic field strength for vertical lifting. The decomposition calculation converts the modulus and direction angle of the magnetic field vector into three component values through trigonometric function operations. Each component value contains size and positive and negative sign information. The positive and negative signs indicate the directional relationship of the magnetic field direction relative to the coordinate axis.

[0026] The pressure detection processing of the oil-filled pressure compensation structure determines the degree of influence of the current water depth environment on the magnetic field distribution based on the changes in the magnetic flux density component. The pressure detection processing infers the compression effect of seawater pressure on the magnetic coupling gap by analyzing the numerical change amplitude of the magnetic flux density component. When the water depth increases, the seawater pressure increases. The pressure acting on the oil-filled cavity causes the cavity to deform slightly. The deformation causes small changes in the magnetic coupling gap. The gap change directly affects the distribution characteristics of the magnetic flux density. The pressure sensor is installed inside the oil-filled pressure compensation structure to monitor the oil pressure value in the cavity in real time. The pressure detection algorithm compares the oil pressure value with the standard atmospheric pressure to obtain the seawater pressure parameter corresponding to the current water depth. The seawater pressure parameter represents the pressure environment characteristics at the current working depth.

[0027] The data fusion process comprehensively calculates the magnetic flux density components and the water depth pressure parameters to generate magnetic flux density distribution data that takes into account the influence of depth. The fusion algorithm first establishes a correlation model between the magnetic flux density components and the water depth pressure parameters. The model describes the correction relationship between pressure changes and the magnetic flux density distribution. The fusion calculation numerically synthesizes the magnetic flux density components and pressure parameters in three directions through a weighted average method. The weighting coefficient is determined according to the sensitivity of the magnetic field in different directions to pressure changes. The magnetic field in the forward direction is most sensitive to pressure changes, and the sensitivity in the lateral and vertical directions decreases in turn. The magnetic flux density distribution data generated by the fusion process contains two parts: the original magnetic flux density information and the pressure correction information. The original information reflects the basic magnetic field characteristics of the magnetic coupling unit, and the correction information reflects the influence of the depth environment on the magnetic field distribution. The combination of the two parts of information forms a complete description of the magnetic field environment.

[0028] In a specific embodiment, the process of executing step S102 may specifically include the following steps: The magnetic flux density distribution data is processed by hysteresis loop feature extraction to obtain the coercive force, remanence and saturation magnetic induction parameters; Different excitation current sequences are input into the permanent magnet array to collect and process the magnetic field response, thereby obtaining a magnetic field response data set; The hysteresis loop is fitted by the least square method based on the magnetic field response data set to obtain the hysteresis loop mathematical model; The hysteresis compensation coefficient is calculated for different working conditions according to the hysteresis loop mathematical model to obtain a hysteresis compensation lookup table; The magnet temperature detected by the temperature sensor is input into the temperature compensation algorithm for hysteresis correction processing to obtain the hysteresis compensation parameter.

[0029] Specifically, the hysteresis loop feature extraction process identifies key characteristic parameters of magnetic materials by analyzing the nonlinear variations in magnetic flux density distribution data. A hysteresis loop is a closed-loop curve formed between the magnetic flux density of a magnetic material under an alternating magnetic field and the applied magnetic field strength. This curve reflects the characteristics of the magnetization and demagnetization processes of the magnetic material. The feature extraction algorithm first performs numerical differentiation on the magnetic flux density distribution data and identifies the key characteristic points of the hysteresis loop by analyzing the extreme points of the data's rate of change. Coercivity is the reverse magnetic field strength required to completely demagnetize a magnetic material. The extraction algorithm determines the coercivity parameter by finding the magnetic field strength corresponding to zero magnetic flux density. Remanence is the magnetic flux density retained by a magnetic material after the external magnetic field is removed. The algorithm determines the remanence parameter by finding the magnetic flux density value at zero applied magnetic field strength. Saturation magnetic flux density is the maximum magnetic flux density when a magnetic material reaches magnetic saturation. The algorithm determines the saturation magnetic flux density parameter by finding the asymptotic value of the magnetic flux density curve. The feature extraction process separates these three parameters from the magnetic flux density distribution data to form a basic parameter set that describes the hysteresis characteristics of the permanent magnet array.

[0030] The magnetic field response acquisition and processing of the excitation current sequence applies excitation currents of varying intensities to the permanent magnet array according to a preset current variation pattern and simultaneously acquires the corresponding magnetic field response data. The excitation current sequence includes three states: forward current, reverse current, and zero current. The current variation follows a step function pattern, with each current value maintained at a fixed interval to ensure the magnetic field reaches a steady state. The magnetic field response acquisition uses Hall sensors to monitor the changes in magnetic field intensity around the permanent magnet array in real time, with the acquisition frequency synchronized with the frequency of the excitation current variation. The response dataset is constructed by pairing and storing each excitation current value with its corresponding magnetic field response value, forming a data set of current-magnetic field responses. The dataset also records the timestamp of each response measurement for analysis of the dynamic characteristics of the magnetic field response. During the acquisition and processing, the controller continuously monitors the actual output value of the excitation current and uses current sensor feedback to correct any deviation between the set and actual current values to ensure the accuracy of the excitation current sequence. The magnetic field response dataset contains complete response data for the three processes of forward excitation, reverse excitation, and demagnetization, forming a comprehensive database describing the dynamic magnetization characteristics of the permanent magnet array.

[0031] The least-squares fitting process constructs a mathematical model of the hysteresis loop based on the magnetic field response dataset. The least-squares method is a numerical optimization algorithm that determines model parameters by minimizing the squared error between measured data and the theoretical model. The fitting algorithm first establishes a theoretical mathematical expression for the hysteresis loop, which includes the unknown parameters describing the hysteresis characteristics. The fitting process uses the current-magnetic field data pairs in the magnetic field response dataset as fitting samples and constructs a sum-of-squares error function to evaluate the degree of agreement between the theoretical model and the measured data. The algorithm determines the optimal values of the unknown parameters in the theoretical model by minimizing the sum-of-squares error function. This solution utilizes numerical optimization methods such as gradient descent or Newton's method. The resulting mathematical model of the hysteresis loop accurately describes the magnetization response characteristics of the permanent magnet array under different excitation conditions. The model parameters include key parameters such as the hysteresis loop's shape coefficient, saturation coefficient, and loss coefficient. This mathematical model addresses the problem of quantitatively describing hysteresis characteristics and lays the theoretical foundation for subsequent hysteresis compensation calculations.

[0032] The hysteresis compensation coefficient calculation process uses the hysteresis loop mathematical model to calculate the corresponding compensation coefficient values for different operating conditions. These operating conditions include different combinations of parameters such as excitation current intensity, magnetic field frequency, and load impedance. Each operating condition corresponds to specific hysteresis loss characteristics. The calculation algorithm inputs the operating parameters into the hysteresis loop mathematical model and calculates the theoretical magnetic field response value for that condition. This value is then compared with the expected ideal response value to calculate the hysteresis deviation. The hysteresis compensation coefficient is calculated using a proportional relationship algorithm. The compensation coefficient is equal to the ratio of the hysteresis deviation to the excitation current intensity, which reflects the degree of hysteresis loss generated per unit excitation current. The calculation process calculates the compensation coefficient for all preset operating conditions one by one, generating a hysteresis compensation lookup table covering the entire operating range. The lookup table uses the operating parameters as the index and the hysteresis compensation coefficient as the numerical content, forming a data structure that facilitates real-time search. The establishment of the lookup table enables the controller to quickly determine the required hysteresis compensation amount based on the current operating conditions, avoiding the time delay associated with real-time calculation.

[0033] The hysteresis correction process in the temperature compensation algorithm addresses the temperature-dependent drift of the hysteresis characteristics of magnetic materials. Temperature sensors are installed at key locations within the permanent magnet array to monitor the magnet's operating temperature in real time. The temperature compensation algorithm establishes a correlation model between hysteresis characteristic parameters and temperature. This model describes how coercivity, remanence, and saturation magnetic induction vary with temperature. The hysteresis correction process first reads the real-time temperature data from the temperature sensor. Then, based on the temperature-hysteresis correlation model, it calculates the hysteresis correction at the current temperature. The correction algorithm numerically combines the temperature correction with the baseline compensation coefficient in the hysteresis compensation lookup table to generate the actual hysteresis compensation parameters that account for temperature effects. This calculation utilizes a weighted superposition method, with the temperature correction weight determined based on the degree of temperature deviation, with a higher weight associated with a greater temperature deviation. The generation of hysteresis compensation parameters comprehensively considers both operating conditions and the temperature environment, resulting in accurate compensation data tailored to the actual operating environment.

[0034] In a specific embodiment, the process of executing step S103 may specifically include the following steps: The hysteresis compensation parameters are used as constraints to perform ternary gene encoding on the magnetic pole phase difference, excitation current amplitude and magnetic field direction angle, and an initial population of 100 individuals is obtained. Perform magnetic coupling transfer efficiency simulation calculation on the magnetic pole configuration scheme of each individual in the initial population to obtain the corresponding fitness function value; Based on the fitness function value, the population individuals are probabilistically selected, cross-constrained by magnetic field physical constraints, and subjected to dynamic mutation probability mutation processing to obtain a new generation of population that meets the magnetic saturation constraint. The new generation population was fed into the iterative evolutionary algorithm for 50 generations of cyclic evolution to obtain the convergent individuals with the maximum magnetic coupling efficiency. The convergent individuals are subjected to ternary gene decoding processing to obtain the optimal magnetic pole phase difference.

[0035] Specifically, the ternary genetic encoding process converts three continuous variables—magnetic pole phase difference, excitation current amplitude, and magnetic field angle—into a discrete encoding format that can be processed by the genetic algorithm. The hysteresis compensation parameter serves as a constraint on the encoding range. The magnetic pole phase difference represents the angular difference between adjacent magnetic poles. Its value range is limited by the coercivity, a parameter used in the hysteresis compensation parameter. Magnetic materials with higher coercivity require a larger phase difference to overcome hysteresis resistance. The excitation current amplitude determines the magnetic field strength, with its upper limit constrained by the saturation magnetic induction, a parameter used in the hysteresis compensation parameter. Exciting currents exceeding the saturation point do not produce a stronger magnetic field but instead increase energy consumption. The magnetic field angle determines the spatial orientation of the magnetic field. Its adjustable range is affected by the remanence parameter. Materials with higher remanence require a larger reverse magnetic field to eliminate residual magnetization when the magnetic field direction changes. The encoding algorithm first calculates the valid value range for each variable based on the hysteresis compensation parameter. This range is then divided into several discrete encoding segments, each corresponding to a binary code value. Ternary genetic encoding concatenates the binary codes of the three variables to form the complete genetic sequence of a single individual. The length of the genetic sequence depends on the required encoding accuracy; higher accuracy requires more encoding bits. The initial population generation algorithm randomly generates one hundred different gene sequence combinations, each of which represents a magnetic pole configuration scheme. The population size is set to one hundred individuals in order to strike a balance between computational complexity and optimization effect.

[0036] The magnetic coupling transfer efficiency simulation process evaluates the performance of each magnetic pole configuration by establishing a magnetic field physical model. The simulation first decodes the individual's genetic sequence into specific values for the magnetic pole phase difference, excitation current amplitude, and magnetic field angular orientation. These parameters are then input into the magnetic field simulation model for calculation. Magnetic coupling transfer efficiency is defined as the ratio of output torque to input power, reflecting the efficiency of the magnetic coupling system in converting electrical energy into mechanical torque. The simulation model uses finite element analysis to calculate the magnetic field distribution and flux orientation for different magnetic pole configurations. The total magnetic flux in the magnetic coupling region is then integrated to determine the magnitude of the magnetic flux, which directly affects the efficiency of torque transfer. The calculation also considers the effects of hysteresis and eddy current losses on transfer efficiency. Hysteresis losses are quantitatively calculated using hysteresis compensation parameters, while eddy current losses are estimated based on the frequency of magnetic field fluctuations and conductor material properties. The simulation output is an efficiency value between zero and one, which serves as the fitness function of the genetic algorithm to assess the individual's performance within the population. The higher the fitness function value, the better the magnetic coupling transmission efficiency of the magnetic pole configuration scheme, and the greater the probability of being selected in subsequent genetic operations.

[0037] Probabilistic selection, crossover with physical constraints, and dynamic probabilistic mutation are the core evolutionary operations of the genetic algorithm. These operations screen and recombine individuals in the population based on their fitness function values. The probabilistic selection algorithm calculates the probability of selection based on each individual's fitness function value. Individuals with higher fitness values have a greater probability of selection. The selection process uses proportional selection, meaning that the probability of an individual being selected is proportional to its fitness value. The crossover with physical constraints recombines the genetic sequences of two parent individuals to produce an offspring individual. The crossover process must satisfy the constraints of magnetic field physics laws, such as the law of flux continuity and Ampere's circuit law. The constraint verification algorithm verifies the physical feasibility of the offspring individuals after the crossover operation. Individuals that do not meet the physical constraints are regenerated or modified. The dynamic probabilistic mutation operation randomly changes certain sites in the individual's genetic sequence. The mutation probability is dynamically adjusted based on the number of evolutionary generations. Initially, a higher mutation probability is used to increase population diversity, while later, a lower probability is used to maintain the stability of the best individuals. The magnetic saturation constraint acts as a boundary constraint in the mutation operation, ensuring that the mutated individuals remain within the physically feasible parameter range. A new generation of population is generated from the current population through selection, crossover and mutation operations. The new population maintains a size of one hundred individuals to prepare for the next round of evolutionary iteration.

[0038] The iterative evolutionary algorithm repeats genetic operations for fifty generations, each generating a new population through selection, crossover, and mutation. An iteration counter increments from the first generation and increases by one after each evolutionary generation. The algorithm stops iterating when the counter reaches fifty. A convergence determination mechanism monitors the fitness of the population during the iteration process. The algorithm is considered to have converged when the change in the optimal fitness value for several consecutive generations is less than a preset threshold. The converged individual with the highest fitness function value over fifty generations represents the magnetic pole configuration that achieves the optimal magnetic coupling transmission efficiency under the current constraints. The algorithm records the optimal individual and average fitness of each generation during the evolution process for analysis of the convergence characteristics and stability of the optimization process. The number of iterations set to fifty is based on a balance between the complexity of the magnetic field optimization problem and computational resources. This ensures that the algorithm has sufficient evolutionary space to find the global optimal solution while maintaining an acceptable computational time.

[0039] The ternary gene decoding process converts the binary gene sequence of the convergent individual back into specific numerical values for the magnetic pole phase difference, excitation current amplitude, and magnetic field angular orientation. The decoding algorithm first divides the gene sequence into three independent binary segments based on the number of bits encoded during the encoding process, with each segment corresponding to an optimization variable. A binary-to-decimal conversion calculation converts each binary segment into a corresponding integer value. These integer values are then converted to actual physical parameter values based on the numerical mapping relationship determined during encoding. The optimal magnetic pole phase difference is the first of the three parameters obtained through decoding, and it determines the optimal angular configuration relationship between the three magnetic coupling units. The decoding process also includes parameter validity verification to ensure that the decoded parameter values are within the physically achievable range and meet the constraints of the hysteresis compensation parameters. The optimal magnetic pole phase difference serves as an input parameter for the subsequent magnetic field-aware Lagrangian allocation algorithm, guiding the specific magnetic field control allocation calculations.

[0040] In a specific embodiment, the process of performing ternary gene encoding processing on the magnetic pole phase difference, the excitation current amplitude, and the magnetic field direction angle using the hysteresis compensation parameter as a constraint condition may specifically include the following steps: The hysteresis compensation coefficient in the hysteresis compensation parameter is converted into a magnetic field constraint boundary value, and the range of the magnetic pole phase difference value is limited to obtain the phase difference encoding interval; The excitation current amplitude and magnetic field direction angle are binary-coded based on the phase difference coding interval to obtain the gene string length of each individual; The gene string length is randomly initialized according to the ternary combination method to obtain the encoding matrix of magnetic pole phase difference, excitation current amplitude and magnetic field direction angle; Perform magnetic saturation constraint test on each gene position in the coding matrix to obtain valid individuals that meet physical constraints; The effective individuals are supplemented and generated according to the population size requirements to obtain an initial population of 100 individuals.

[0041] Specifically, the conversion of the hysteresis compensation coefficient into magnetic field constraint boundary values uses a numerical mapping algorithm to convert the compensation coefficient in the hysteresis compensation parameters into the constraint range required for genetic algorithm encoding. The hysteresis compensation coefficient reflects the degree of nonlinear response of the magnetic material under different excitation conditions. A larger coefficient value indicates a more pronounced hysteresis effect and a correspondingly more restricted magnetic field adjustment range. The conversion algorithm first establishes a mapping relationship between the compensation coefficient and the magnetic field intensity threshold. The upper limit constraint for the magnetic pole phase difference is determined by finding the maximum allowable magnetic field intensity corresponding to the compensation coefficient. The limiting process for the magnetic pole phase difference value range is based on two constraints: the physical constraint requires that the phase difference cannot exceed a 360-degree periodic range, while the hysteresis constraint requires that changes in the phase difference cannot cause magnetic field reversals that exceed the material coercivity threshold. The limiting algorithm performs a logical AND operation on these two constraints, taking the more stringent constraint as the actual value range. The phase difference encoding range is achieved by discretizing a continuous angular range into a finite number of encoding values. The encoding precision determines the level of discretization. Higher precision results in finer divisions in the encoding range, resulting in a greater number of corresponding encoding bits.

[0042] Binary encoding uses the required number of bits in the phase difference encoding interval to design a unified encoding length for the excitation current amplitude and magnetic field angular direction. The encoding range for the excitation current amplitude is determined by the saturation current and minimum effective current of the permanent magnet array. The saturation current is the minimum current value required to reach magnetic saturation in the magnetic material; currents exceeding this value do not produce a stronger magnetic field effect. The minimum effective current is the minimum current threshold that produces a detectable magnetic field change; currents below this value are of no practical significance for magnetic coupling control. The encoding range for the magnetic field angular direction covers a full circular angle from 0 to 360 degrees, with angular accuracy determined by the precision requirements of magnetic coupling control. The binary encoding algorithm divides the value range of each variable into powers of two, with the number of divisions equal to the power of two in the encoding bit number. The selection of the encoding bit number requires a balance between encoding accuracy and computational complexity. The gene string length calculation sums the encoding bit number of the three variables to obtain the total encoding length of a single individual. This length determines the gene sequence size and computational complexity of the genetic algorithm operation.

[0043] The ternary random initialization process concatenates the coding segments for the magnetic pole phase difference, excitation current amplitude, and magnetic field angle in a fixed order to form a complete gene sequence. The initialization algorithm uses a pseudo-random number generator to produce a random binary sequence with a uniform distribution. This randomness ensures that the initial population has a high diversity and covers the entire search space. The ternary structure divides the gene sequence into three consecutive coding segments: the first segment encodes the magnetic pole phase difference, the second segment encodes the excitation current amplitude, and the third segment encodes the magnetic field angle. The boundaries between the segments are strictly determined by the number of coding bits. The encoding matrix is generated by arranging multiple gene sequences in rows to form a two-dimensional matrix structure. The number of rows in the matrix is equal to the population size, and the number of columns is equal to the length of the gene sequence. Each element in the matrix is either zero or one, corresponding to the two states of the binary code. The row vector of the matrix represents a complete magnetic pole configuration. Random initialization ensures that each position in the encoding matrix has an equal probability of being assigned a value of zero or one, preventing the initial population from being biased towards a specific configuration pattern.

[0044] The magnetic saturation constraint check verifies the physical feasibility of each gene position in the encoding matrix, ensuring that the decoded parameter values will not cause the magnetic material to enter an uncontrollable magnetic saturation state. The check algorithm first decodes the gene sequence into corresponding values for the magnetic pole phase difference, excitation current amplitude, and magnetic field direction angle. These values are then input into a magnetic field calculation model to estimate the resulting magnetic field strength. The magnetic saturation constraint requires that the calculated magnetic field strength cannot exceed the saturation magnetic induction of the magnetic material. Configurations exceeding this threshold will cause a sharp decline in magnetic coupling performance. The constraint check also includes magnetic field gradient verification to ensure that the variation in magnetic field strength between adjacent magnetic poles does not exceed the material's hysteresis loop range. The check algorithm scans each gene position one by one, marking and correcting those that do not meet the constraints. Correction methods include random regeneration and boundary value replacement strategies. Valid individual screening retains gene positions that pass the constraint check, while individuals that do not meet the constraints are excluded or regenerated, ensuring that all individuals in the population represent physically feasible magnetic pole configurations.

[0045] The supplementary generation process adjusts the number of valid individuals based on the population size requirements. If the number of valid individuals is insufficient, additional individuals are generated to meet the population size of 100 individuals. The supplementary algorithm uses an improved random generation strategy to generate new gene sequences while satisfying the magnetic saturation constraint. Newly generated individuals must pass a similarity test with existing individuals to avoid excessive duplication in the population. The similarity test assesses the degree of difference between individuals by calculating the Hamming distance between gene sequences. The Hamming distance represents the number of different characters at corresponding positions in two equal-length binary sequences. A distance threshold is set to ensure that the newly generated individuals remain sufficiently different from existing individuals, maintaining the diversity characteristics of the population. The final construction of the initial population organizes the 100 valid and differentiated individuals into a complete population structure, with each individual having a gene sequence of the same length and satisfying the magnetic saturation constraint. Population statistics are recorded, including parameters such as the number of individuals, gene sequence length, constraint satisfaction rate, and diversity index. This information is used for parameter adjustment and performance evaluation in subsequent genetic algorithm operations.

[0046] The encoding initialization process is illustrated using a magnetically coupled thruster for under-ice operations in polar regions. When the thruster is required to operate in the low-temperature, high-pressure polar environment, the hysteresis compensation coefficient conversion first determines the constraint boundary value based on the changes in magnetic material properties in low-temperature environments. Low temperatures increase the coercivity of magnetic materials, and the corresponding magnetic pole phase difference value range needs to be expanded to overcome the increased hysteresis resistance. The phase difference encoding interval is re-divided into encoding segments based on the expanded value range, and the encoding accuracy is selected to take into account the special control accuracy requirements of the polar environment. The binary encoding of the excitation current amplitude and magnetic field direction angle also considers the impact of low temperatures on electrical performance. The current encoding range is adjusted based on the change in conductor resistance at low temperatures, and the direction angle encoding takes into account the special effects of the polar magnetic field environment. Triple combination random initialization generates one hundred candidate individuals, each representing a magnetic pole configuration suitable for polar environments. The encoding matrix contains parameter combinations optimized for the low-temperature, high-pressure polar environment. The magnetic saturation constraint verification specifically focuses on the changes in magnetic saturation characteristics in low-temperature environments. The verification standard is calibrated based on actual magnetic performance test data from polar environments. The replenishment generation process after effective individual screening ensures that the final initial population not only meets the physical constraints of the polar environment but also maintains sufficient configuration diversity.

[0047] In a specific embodiment, the process of executing step S104 may specifically include the following steps: The torque control command is decomposed into three magnetic coupling units using the optimal magnetic pole phase difference as the reference configuration to obtain the expected propulsion force and steering torque components of each unit. Based on the expected propulsion force and steering torque components, a convex optimization problem is constructed to solve the magnetic saturation constraint and the adjacent magnetic field interference threshold, and the constrained optimization objective function is obtained. Input the constrained optimization objective function into the enhanced Lagrange multiplier method for numerical solution processing to obtain the initial Lagrange multiplier value; The initial Lagrangian multiplier value is corrected by a magnetic field sensing factor according to the actual magnetic field strength fed back by the Hall sensor to obtain a corrected Lagrangian multiplier value; Based on the corrected Lagrange multiplier value, current distribution calculation processing is performed on the excitation demand of the magnetic coupling unit to obtain a target excitation current value.

[0048] Specifically, the three-magnetic coupling unit decomposition process decomposes the torque control command issued by the upper-level motion controller into the specific control requirements of three independent magnetic coupling units based on the optimal magnetic pole phase difference configuration. The torque control command includes the desired total propulsion force vector and the total steering torque vector. The propulsion force vector describes the magnitude and direction of the thrust in the forward direction of the propeller, while the steering torque vector describes the rotational torque about the vertical and horizontal axes. The decomposition algorithm uses a geometric projection method to determine the weight of each unit's contribution to the total propulsion force and steering torque based on the spatial layout of the three magnetic coupling units and the optimal magnetic pole phase difference. The geometric projection calculation decomposes the total propulsion force vector along the axial direction of the three units. The desired propulsion force component of each unit is equal to the projection of the total propulsion force along the unit's axis multiplied by the unit's weight coefficient. The weight coefficient is determined by the optimal magnetic pole phase difference; units with larger phase differences contribute more thrust in a specific direction. The steering torque decomposition process uses the torque balance principle. The length of the moment arm generating the steering torque of each magnetic coupling unit is calculated based on its position vector relative to the propeller's center of gravity. The desired steering torque component is obtained by solving the torque balance equation. The decomposition results include the expected propulsion force, propulsion force direction and steering torque of each of the three magnetic coupling units. These component parameters serve as the target constraints for subsequent optimization calculations.

[0049] The convex optimization problem construction process converts the desired propulsion force and steering torque components into the constraints and objective function of a mathematical optimization problem. The magnetic saturation constraint ensures that the excitation current of each magnetic coupling unit does not cause the magnetic material to enter a saturated state. The constraint is expressed as upper and lower bounds on the excitation current, with the upper bound determined by the saturation current of the magnetic material and the lower bound determined by the minimum effective current. The adjacent magnetic field interference threshold constraint prevents the magnetic fields of adjacent magnetic coupling units from interfering with each other beyond an acceptable range. The degree of interference is quantified by the cross-coupling coefficient of the magnetic field strength. The convex optimization objective function aims to minimize the energy consumption difference between each magnetic coupling unit. Small energy consumption differences indicate uniform load distribution and avoid overloading of individual units. The objective function adopts a quadratic function form, which has a unique global optimal solution and good numerical stability. The constrained optimization objective function construction expresses all constraints as inequalities or equations, forming a standard mathematical model for a convex optimization problem. The convexity of the problem ensures that the optimization algorithm converges to the global optimal solution, avoiding the dilemma of local optimality.

[0050] The enhanced Lagrangian multiplier method (ALM) numerical solution transforms a constrained optimization problem into an unconstrained optimization problem for iterative solution. The LMM incorporates constraints into the objective function by introducing Lagrangian multipliers, forming an augmented Lagrangian function. The LMM adds a penalty term to the traditional Lagrangian function, imposing an additional penalty on solutions that violate the constraints, enhancing the algorithm's convergence and numerical stability. The numerical solution is iteratively calculated using gradient descent. Each iteration determines the search direction by calculating the gradient of the augmented Lagrangian function with respect to the decision variables. The gradient calculation includes the objective function gradient, the equality constraint gradient, and the inequality constraint gradient. The combined gradient vectors determine the optimal search direction. The iterative update rule simultaneously adjusts the values of the decision variables and the Lagrangian multipliers. The decision variables move in the direction of gradient descent, while the Lagrangian multipliers adjust based on the degree of constraint violation. The initial Lagrangian multiplier values are set using heuristic methods and are typically set to zero or empirically determined based on the characteristics of the problem. Convergence judgment is based on two indicators: gradient norm and constraint violation degree. The algorithm stops iterating when the gradient norm is less than the preset threshold and the constraint violation degree is acceptable.

[0051] The magnetic field perception factor correction process dynamically adjusts the initial Lagrangian multiplier value based on real-time magnetic field strength data fed back by Hall sensors. Hall sensors are placed at key locations on the three magnetic coupling units, monitoring the actual magnetic field strength generated by each unit and the degree of magnetic field interference between adjacent units in real time. The magnetic field perception factor is defined as the ratio of the actual magnetic field strength to the theoretically calculated magnetic field strength. This ratio reflects the degree of deviation between the theoretical model and the actual system. A perception factor greater than one indicates that the actual magnetic field strength exceeds theoretical expectations, requiring a reduction in the Lagrangian multiplier to reduce control intensity. A perception factor less than one indicates that the actual magnetic field strength is insufficient, requiring an increase in the Lagrangian multiplier to increase control intensity. The correction algorithm utilizes a proportional-integral control strategy. The proportional term makes an immediate correction based on the current perception factor deviation, while the integral term makes a long-term correction based on accumulated historical deviations. The correction calculation performs a correlation analysis between the perception factor of each magnetic coupling unit and the corresponding Lagrangian multiplier to determine the direction and magnitude of the correction. The corrected Lagrangian multiplier value maintains the mathematical consistency of the optimization algorithm while reflecting the dynamic characteristics of the actual magnetic field environment, ensuring accurate and real-time control allocation.

[0052] The current allocation calculation process calculates the specific excitation current required for each magnetic coupling unit based on the modified Lagrange multiplier values. The current allocation algorithm uses the product of the Lagrange multiplier and the control gain, which describes the sensitivity of excitation current changes to changes in magnetic field strength. The calculation process first calculates the control strength corresponding to each constraint based on the modified Lagrange multiplier values. Then, the control strength is converted into excitation current requirements through the inverse transformation of the control gain matrix. The elements of the control gain matrix are obtained through magnetic field simulation or experimental measurement. The matrix describes the coupling relationship and independent control capability between the magnetic coupling units. The current allocation process also considers practical constraints such as power supply power limitations and current regulation accuracy to ensure that the calculated current values are within the achievable range of the hardware system. The target excitation current value includes the current amplitude and current direction of each of the three magnetic coupling units. The current amplitude determines the magnetic field strength, and the current direction determines the magnetic field polarity. The allocation results are verified for consistency, using forward calculation and deviation analysis to ensure that the current combination of the three units can produce the desired total propulsion force and steering torque.

[0053] In a specific embodiment, the process of executing step S105 may specifically include the following steps: The target excitation current value and the actual magnetic field strength detected in real time by the Hall sensor are calculated and processed to obtain a magnetic field control error signal; Based on the magnetic field control error signal, the hysteresis compensation lookup table is queried for feedforward compensation to obtain the hysteresis pre-compensation current value; The magnetic field control error signal is input into the PID controller for proportional-integral-differential operation to obtain the basic PID control quantity; The basic PID control quantity is superimposed and corrected according to the hysteresis pre-compensation current value to obtain the hysteresis compensation PID control output; The hysteresis compensation PID control output drives the electromagnet to adjust the magnetic field strength to obtain a magnetic coupling torque output.

[0054] Specifically, the deviation calculation process converts the target excitation current value into a desired magnetic field strength using a numerical comparison algorithm. This difference is then calculated against the actual magnetic field strength detected in real time by the Hall effect sensor. The target excitation current value is first calculated using a current-magnetic field conversion model to determine the theoretical magnetic field strength. This conversion model is based on Ampere's circuit law and the magnetization properties of magnetic materials. Model parameters include physical parameters such as the number of coil turns, core permeability, and geometric dimensions. The Hall effect sensor uses high-frequency sampling to acquire the actual magnetic field strength data around the magnetic coupling unit in real time. The sampling frequency is set to at least ten times the control frequency to ensure data timeliness and accuracy. The deviation calculation uses vector subtraction, subtracting the desired magnetic field strength vector from the actual magnetic field strength vector component by component to obtain magnetic field error components in three spatial directions. The magnetic field control error signal contains three dimensions: error magnitude, direction, and trend. The error magnitude reflects the severity of the control deviation, the error direction indicates the magnetic field direction requiring adjustment, and the trend indicates the development of the error. The error signal is also filtered to eliminate sensor noise and high-frequency interference. The filtering algorithm uses a low-pass filter to retain the effective frequency components required for control.

[0055] The feedforward compensation query process in the hysteresis compensation lookup table rapidly retrieves the corresponding hysteresis compensation data based on the characteristic parameters of the magnetic field control error signal. The lookup table is organized into a two-dimensional index based on the magnitude and rate of change of the magnetic field error. The table stores experimentally calibrated hysteresis compensation current values under different error conditions. The query algorithm first calculates the amplitude and rate of change of the magnetic field control error signal. The amplitude is calculated by the modulus of the error vector, and the rate of change is calculated by the difference between the current error and the previous error. A two-dimensional interpolation algorithm handles situations where the query parameters do not fully match the table index. When the combination of the error amplitude and rate of change does not fall within a predefined index point, the algorithm calculates a weighted average of adjacent index points using bilinear interpolation. The hysteresis pre-compensation current value is derived based on the directional characteristics of the hysteresis loop. Different compensation strategies are required for forward and reverse magnetization processes. The lookup table distinguishes the magnetization directions and provides corresponding compensation parameters. The compensation query also includes a temperature correction step. The compensation value in the lookup table is corrected based on the current temperature detected by the temperature sensor. The correction algorithm calculates the temperature compensation factor based on the temperature coefficient of the magnetic material.

[0056] The PID controller's proportional-integral-derivative (PI-D) operation converts the magnetic field control error signal into the basic feedback control variable. The proportional operation directly multiplies the current magnetic field error signal by the proportional gain coefficient. The proportional term aims to quickly respond to the current error and reduce control deviation. The selection of the proportional gain requires a balance between response speed and system stability. The integral operation accumulates historical error signals. The integral term improves control accuracy by eliminating steady-state errors. The integral calculation uses numerical integration to sum discrete error samples. The integral gain determines the contribution of the integral term to the control output. The differential operation calculates the rate of change of the error signal to predict the error's development trend. The differential term provides a proactive control function, reducing system overshoot and oscillation. The differential operation estimates the derivative of the error using forward or backward differencing. The weighted sum of the three operational components forms the basic PID control variable. The weight coefficients are the PID parameters, which are optimized through experimental methods or automatic tuning algorithms. The basic PID control variable is also limited to prevent the control output from exceeding the actuator's operating range. The limiter value is determined by the maximum drive current of the electromagnet.

[0057] The superposition correction process numerically combines the hysteresis pre-compensation current value with the basic PID control variable to generate the final control output. The superposition algorithm uses a weighted summation approach, with the hysteresis pre-compensation current value serving as the feedforward compensation component and the basic PID control variable serving as the feedback control component. The weights of the two components are assigned based on the required control accuracy. The feedforward compensation component primarily mitigates the effects of hysteresis nonlinearity on control performance, and its effectiveness is directly related to the accuracy of the hysteresis compensation lookup table. The feedback control component eliminates control errors caused by various uncertainties, including model errors, external interference, and parameter variations. The superposition correction also includes phase compensation. Since hysteresis can cause a phase lag in the magnetic field response, phase compensation offsets this lag by adjusting the timing of the feedforward compensation. The correction algorithm monitors the numerical relationship between the two control components and prioritizes system stability when a conflict arises between feedforward compensation and feedback control. The hysteresis-compensated PID control output undergoes digital filtering and smoothing to eliminate sudden changes and glitches in the control signal, ensuring the continuity and stability of the drive signal.

[0058] The magnetic field strength regulation process converts the hysteresis-compensated PID control output into the actual current driving the electromagnet through a power amplifier circuit. The power amplifier circuit uses linear amplification or pulse-width modulation to amplify the control signal to a power level sufficient to drive the electromagnet. The amplification gain is determined by the electromagnet's current demand and the amplitude range of the control signal. Precise control of the electromagnet's drive current is achieved through a current feedback loop. A current sensor monitors the actual electromagnet current, and the controller performs closed-loop regulation based on the deviation between the set current and the actual current. Adjusting the magnetic field strength directly affects the torque transmission capability of the magnetic coupling unit. The relationship between magnetic field strength and torque output is nonlinear. When the magnetic field is weak, the torque increases rapidly with magnetic field strength, but the increase becomes more gradual as magnetic saturation approaches. The calculation of magnetic coupling torque output is based on the geometric relationship between magnetic flux density and the magnetic coupling gap. The torque is proportional to the square of the magnetic flux and inversely proportional to the size of the magnetic coupling gap. Torque output is also affected by load impedance and speed. The control algorithm adjusts the magnetic field strength control strategy based on the current load conditions, increasing the magnetic field strength to ensure sufficient torque output under high load conditions and reducing the magnetic field strength to reduce energy consumption under low load conditions.

[0059] The above describes the intelligent control method of the underwater magnetic coupling thruster in the embodiment of the present application. The following describes the intelligent control system of the underwater magnetic coupling thruster in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the underwater magnetic coupling thruster intelligent control system includes: A detection module is used to detect and process the magnetic field of the magnetic coupling unit through a Hall sensor array to obtain magnetic flux density distribution data; A modeling module, configured to perform hysteresis characteristic modeling on the permanent magnet array according to the magnetic flux density distribution data to obtain hysteresis compensation parameters; A processing module, configured to input the hysteresis compensation parameters into a magnetic field genetic algorithm to optimize the magnetic pole configuration and obtain an optimal magnetic pole phase difference; a distribution module, configured to perform magnetic field-aware Lagrangian distribution processing on the torque control command according to the optimal magnetic pole phase difference to obtain a target excitation current value; The regulating module is used to perform magnetic field regulation processing on the target excitation current value through a hysteresis compensation PID controller to obtain a magnetic coupling torque output.

[0060] above Figure 2 The intelligent control system of the underwater magnetic coupling thruster in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The intelligent control device of the underwater magnetic coupling thruster in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0061] Reference Figure 3In the embodiment of the present invention, there is also provided an underwater magnetic coupling thruster intelligent control device, which can be a server, and its internal structure can be as follows Figure 3 As shown. The underwater magnetic coupling thruster intelligent control device includes a processor, a memory, a display screen, an input device, a network interface and a database connected via a system bus. Among them, the computer-designed processor is used to provide computing and control capabilities. The memory of the underwater magnetic coupling thruster intelligent control device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the underwater magnetic coupling thruster intelligent control device is used to store the corresponding data in this embodiment. The network interface of the underwater magnetic coupling thruster intelligent control device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0062] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the underwater magnetic coupling thruster intelligent control device to which the solution of the present invention is applied.

[0063] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the underwater magnetic coupling thruster intelligent control method.

[0064] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0065] If the integrated unit is implemented as 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, or the portion that contributes to the prior art, or all or part of the 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 enabling an underwater magnetic coupling thruster intelligent control device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0066] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An intelligent control method for underwater magnetic coupling thrusters, characterized in that: The method comprises: The magnetic field of the magnetic coupling unit is detected and processed by the Hall sensor array to obtain magnetic flux density distribution data; Performing hysteresis characteristic modeling on the permanent magnet array according to the magnetic flux density distribution data to obtain hysteresis compensation parameters; Inputting the hysteresis compensation parameters into a magnetic field genetic algorithm to optimize the magnetic pole configuration to obtain an optimal magnetic pole phase difference; Performing magnetic field sensing Lagrangian distribution processing on the torque control command according to the optimal magnetic pole phase difference to obtain a target excitation current value; The target excitation current value is subjected to magnetic field regulation processing by a hysteresis compensation PID controller to obtain a magnetic coupling torque output.

2. The intelligent control method for underwater magnetic coupling thrusters according to claim 1, characterized in that: The method of performing magnetic field detection processing on the magnetic coupling unit by using the Hall sensor array to obtain magnetic flux density distribution data includes: Performing a full-scale scanning detection process on the magnetic field strength of the three magnetic coupling units to obtain initial magnetic field strength data; Inputting the initial magnetic field strength data into a three-dimensional coordinate mapping algorithm for spatial positioning processing to obtain a corresponding relationship between magnetic field strength and position coordinates; Performing vector decomposition processing on the magnetic flux density based on the corresponding relationship to obtain magnetic flux density components in the X, Y, and Z directions; Performing pressure detection processing on the oil-filled pressure compensation structure according to the magnetic flux density component to obtain a current water depth pressure parameter; The magnetic flux density component is subjected to data fusion processing with the water depth pressure parameter to obtain magnetic flux density distribution data.

3. The intelligent control method for underwater magnetic coupling thrusters according to claim 1, characterized in that: The hysteresis characteristic modeling process is performed on the permanent magnet array according to the magnetic flux density distribution data to obtain the hysteresis compensation parameters, including: Performing hysteresis loop feature extraction processing on the magnetic flux density distribution data to obtain coercive force, remanence and saturation magnetic induction intensity parameters; Different excitation current sequences are input into the permanent magnet array to collect and process the magnetic field response, thereby obtaining a magnetic field response data set; Performing least squares fitting processing on the hysteresis loop based on the magnetic field response data set to obtain a hysteresis loop mathematical model; Calculating and processing the hysteresis compensation coefficient for different working conditions according to the hysteresis loop mathematical model to obtain a hysteresis compensation lookup table; The magnet temperature detected by the temperature sensor is input into the temperature compensation algorithm for hysteresis correction processing to obtain the hysteresis compensation parameter.

4. The intelligent control method for underwater magnetic coupling thrusters according to claim 1, characterized in that: The step of inputting the hysteresis compensation parameter into a magnetic field genetic algorithm to optimize the magnetic pole configuration to obtain an optimal magnetic pole phase difference comprises: Using the hysteresis compensation parameter as a constraint condition, a ternary gene encoding process is performed on the magnetic pole phase difference, the excitation current amplitude, and the magnetic field direction angle to obtain an initial population containing 100 individuals; Performing a magnetic coupling transfer efficiency simulation calculation on the magnetic pole configuration scheme of each individual in the initial population to obtain a corresponding fitness function value; Based on the fitness function value, probability selection, magnetic field physical constraint crossover and dynamic mutation probability mutation processing are performed on the individuals of the population to obtain a new generation of population that meets the magnetic saturation constraint; Inputting the new generation population into an iterative evolutionary algorithm for 50 generations of cyclic evolution processing to obtain a convergent individual with maximized magnetic coupling efficiency; The convergent individuals are subjected to ternary gene decoding processing to obtain an optimal magnetic pole phase difference.

5. The intelligent control method for underwater magnetic coupling thrusters according to claim 4, characterized in that: The hysteresis compensation parameter is used as a constraint condition to perform ternary gene encoding processing on the magnetic pole phase difference, the excitation current amplitude and the magnetic field direction angle to obtain an initial population of 100 individuals, including: Converting the hysteresis compensation coefficient in the hysteresis compensation parameter into a magnetic field constraint boundary value, limiting the range of the magnetic pole phase difference value, and obtaining a phase difference encoding interval; Binary encoding is performed on the excitation current amplitude and the magnetic field direction angle based on the phase difference encoding interval to obtain the gene string length of each individual; The gene string length is randomly initialized according to a ternary combination method to obtain a coding matrix of magnetic pole phase difference, excitation current amplitude and magnetic field direction angle; Performing a magnetic saturation constraint check on each gene position in the coding matrix to obtain a valid individual that satisfies the physical constraint; The effective individuals are supplemented and generated according to the population size requirement to obtain an initial population containing 100 individuals.

6. The intelligent control method for underwater magnetic coupling thrusters according to claim 1, characterized in that: The performing magnetic field sensing Lagrangian distribution processing on the torque control instruction according to the optimal magnetic pole phase difference to obtain a target excitation current value includes: Decomposing the torque control command into three magnetic coupling units using the optimal magnetic pole phase difference as a reference configuration to obtain the desired propulsion force and steering torque components of each unit; Based on the desired propulsion force and steering torque components, a convex optimization problem is constructed for the magnetic saturation constraint and the adjacent magnetic field interference threshold to obtain a constrained optimization objective function; Inputting the constraint optimization objective function into the enhanced Lagrange multiplier method for numerical solution processing to obtain an initial Lagrange multiplier value; Performing magnetic field sensing factor correction processing on the initial Lagrangian multiplier value according to the actual magnetic field strength fed back by the Hall sensor to obtain a corrected Lagrangian multiplier value; Based on the corrected Lagrange multiplier value, current distribution calculation processing is performed on the excitation demand of the magnetic coupling unit to obtain a target excitation current value.

7. The intelligent control method for underwater magnetic coupling thrusters according to claim 1, characterized in that: The target excitation current value is subjected to magnetic field regulation processing by a hysteresis compensation PID controller to obtain a magnetic coupling torque output, including: Performing a deviation calculation on the target excitation current value and the actual magnetic field strength detected in real time by the Hall sensor to obtain a magnetic field control error signal; Performing feedforward compensation amount query processing on a hysteresis compensation lookup table based on the magnetic field control error signal to obtain a hysteresis pre-compensation current value; The magnetic field control error signal is input into the PID controller for proportional-integral-differential operation to obtain a basic PID control variable; Performing superposition correction processing on the basic PID control variable according to the hysteresis pre-compensation current value to obtain a hysteresis compensation PID control output; The hysteresis compensation PID control output drives the electromagnet to adjust the magnetic field strength to obtain a magnetic coupling torque output.

8. An intelligent control system for underwater magnetic coupling thrusters, characterized in that: The method for realizing the intelligent control of an underwater magnetic coupling thruster according to any one of claims 1 to 7, wherein the intelligent control system of the underwater magnetic coupling thruster comprises: A detection module is used to detect and process the magnetic field of the magnetic coupling unit through a Hall sensor array to obtain magnetic flux density distribution data; A modeling module, configured to perform hysteresis characteristic modeling on the permanent magnet array according to the magnetic flux density distribution data to obtain hysteresis compensation parameters; A processing module, configured to input the hysteresis compensation parameters into a magnetic field genetic algorithm to optimize the magnetic pole configuration and obtain an optimal magnetic pole phase difference; a distribution module, configured to perform magnetic field-aware Lagrangian distribution processing on the torque control command according to the optimal magnetic pole phase difference to obtain a target excitation current value; The regulating module is used to perform magnetic field regulation processing on the target excitation current value through a hysteresis compensation PID controller to obtain a magnetic coupling torque output.

9. An intelligent control device for underwater magnetic coupling thrusters, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the intelligent control method for an underwater magnetic coupling thruster according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the processor is enabled to perform the underwater magnetic coupling thruster intelligent control method according to any one of claims 1 to 7.

Images