Optimization method, device, equipment and storage medium of magnetic levitation planar motor

Abstract

The present invention relates to the field of structural optimization technology, and in particular to an optimization method, device, equipment, and storage medium for a magnetic levitation planar motor. The method comprises: pre-training a linear force prediction model, measuring an initial quantum code generated based on parameters to be optimized to determine a search direction, updating a preset left antenna code and a right antenna code based on the search direction and then measuring again, obtaining a left fitness value and a right fitness value based on the measurement results to update the initial quantum code and obtain a corresponding updated fitness value; if the updated fitness value is less than the preset fitness value, obtaining an updated position corresponding to the updated quantum code to replace a preset optimal position, returning to execute the search direction determined according to the initial binary code and the preset optimal position, and repeating the iteration to obtain an optimal solution; the method disclosed in the present application can quickly obtain an optimal solution for the parameters to be optimized, effectively improving the energy utilization efficiency of the magnetic levitation planar motor.

Description

Technical Field

[0001] The present invention relates to the technical field of structural optimization, and in particular to an optimization method, device, equipment and storage medium for a magnetically levitated planar motor. Background Art

[0002] The magnetic levitation planar motor is a high-precision motion control device that uses electromagnetic force to achieve non-contact suspension and drive. By precisely controlling the electromagnetic field, the device can suspend the moving body on the working plane and achieve multi-degree-of-freedom motion control. Multi-degree-of-freedom motion usually covers x- and y-direction movement within the plane and z-axis rotation. Thanks to its high precision, frictionless, and low-noise characteristics, the magnetic levitation planar motor has been widely used in many fields such as semiconductor manufacturing, precision machining, medical equipment, aerospace, etc.

[0003] In the typical structure of a magnetic levitation planar motor, electrical energy is converted into kinetic energy of the mover through the interaction between the motor rotor coil array and the magnet array; depending on the structure, it can be divided into two designs: moving magnet steel type and planar moving coil type; the moving magnet steel type design installs the permanent magnet array on the mover, and the coil array on the stator; on the contrary, the moving coil type design installs the coil array on the mover; in the moving magnet steel type planar motor, the mover magnet is the core component, and its structural design directly affects the performance of the equipment; in addition to the mover magnet, the design of the stator coil is also crucial to the performance of the planar motor; the geometric parameters of the coil and permanent magnet (such as size ratio and gap size) directly determine the utilization rate of the magnetic field and the driving efficiency.

[0004] However, current research and innovation on magnetic levitation planar motors mainly focus on structural design, weight reduction, position detection, and optimization of control methods, while research on the optimization of the geometric parameter coordination between coils and permanent magnets is still insufficient. This has led to the failure to fully optimize the geometric coordination between coils and permanent magnets, resulting in low magnetic field utilization, inability to fully realize the performance potential of the levitation system, and limiting further improvement of equipment performance. Summary of the Invention

[0005] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide an optimization method for a magnetic levitation planar motor, which can quickly obtain the optimal solution for the coil parameters to be optimized and the permanent magnet parameters to be optimized, thereby achieving the goals of improving energy utilization and optimizing the structural dimensions of the planar motor.

[0006] The first aspect of the present invention provides an optimization method for a magnetic levitation planar motor, comprising: pre-training a linear force prediction model based on coil parameters and permanent magnet parameters; generating an initial quantum code based on the coil parameters to be optimized and the permanent magnet parameters to be optimized, and measuring the initial quantum code to obtain an initial binary code; obtaining a preset optimal position, determining a search direction according to the initial binary code and the preset optimal position, and updating a preset left antenna code and a right antenna code based on the determined search direction using a preset update rule; measuring the updated left antenna code and the updated right antenna code respectively to obtain a left binary code and a right binary code, and performing an optimization on the left antenna code and the right binary code based on the left binary code and the right binary code. The method comprises the following steps: obtaining a left fitness value and a right fitness value by using the encoding and the linear force prediction model; updating the initial quantum code based on the left fitness value, the right fitness value, the initial binary code and the preset optimal position using a preset update rule to obtain an updated quantum code, and obtaining an updated fitness value based on the updated quantum code and the linear force prediction model; if the updated fitness value is less than the preset fitness value, obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, and returning to execute to determine the search direction according to the initial binary code and the preset optimal position; obtaining the optimal fitness value when a preset iterative stop condition is met, and outputting an optimal solution corresponding to the optimal fitness value.

[0007] Optionally, in a first implementation method of the first aspect of the present invention, the pre-training linear force prediction model based on coil parameters and permanent magnet parameters includes: obtaining simulation analysis data, the independent variables of the simulation analysis data are the number of coils, coil spacing, the number of permanent magnets, the permanent magnet spacing and motion time, and the dependent variable of the simulation analysis data is linear force; performing data cleaning processing, feature scaling processing and data partitioning processing on the simulation analysis data to obtain a training set, a validation set and a test set; constructing a linear force prediction model based on a neural network architecture, the linear force prediction model includes an input layer, a hidden layer and an output layer, the activation function of the hidden layer is a ReLU activation function, and the activation function of the output layer is a linear activation function; the training set, validation set and test set are respectively input into the linear force prediction model to train the model, and the training process of the linear force prediction model is optimized by combining regularization technology and random search algorithm; and the trained linear force prediction model is deployed into the optimization system.

[0008] Optionally, in a second implementation manner of the first aspect of the present invention, generating an initial quantum code based on the coil parameters to be optimized and the permanent magnet parameters to be optimized, and measuring the initial quantum code to obtain an initial binary code, includes: using quantum bits to encode the coil parameters to be optimized and the permanent magnet parameters to be optimized, each coil parameter to be optimized and each permanent magnet parameter to be optimized corresponds to a quantum bit, and integrating all quantum bits to obtain the initial quantum code; mapping the quantum bits to the Bloch sphere respectively, and performing a measurement operation on the mapped quantum bits to obtain an initial binary code corresponding to the initial quantum code; performing a denormalization process on the initial binary code to obtain an initial solution corresponding to the initial binary code, the initial solution including three feasible solutions; and selecting any feasible solution from the initial solution as an initial search point according to a preset selection rule.

[0009] Optionally, in a third implementation of the first aspect of the present invention, the preset optimal position is obtained, the search direction is determined according to the initial binary code and the preset optimal position, and based on the determined search direction, the preset left antenna code and the right antenna code are updated using a preset update rule, including: obtaining the preset optimal position, determining the search direction according to the initial binary code and the preset optimal position; obtaining the initial longicorn whisker search distance, determining the longicorn whisker rotation angle based on the initial longicorn whisker search distance and the determined search direction; based on the longicorn whisker rotation angle and the initial binary code, updating the preset left antenna code and the right antenna code using a preset update rule.

[0010] Optionally, in a fourth implementation method of the first aspect of the present invention, the updated left antenna code and the updated right antenna code are measured respectively to obtain a left binary code and a right binary code, and a left fitness value and a right fitness value are obtained based on the left binary code, the right binary code and the linear force prediction model, including: measuring the updated left antenna code and the updated right antenna code respectively to obtain a left binary code and a right binary code; performing denormalization processing on the left binary code and the right binary code respectively to obtain a left solution and a right solution; substituting the left solution and the right solution into the linear force prediction model respectively to obtain a left fitness value and a right fitness value.

[0011] Optionally, in a fifth implementation manner of the first aspect of the present invention, the updating of the initial quantum code based on the left fitness value, the right fitness value, the initial binary code, and the preset optimal position using a preset update rule to obtain an updated quantum code, and obtaining an updated fitness value based on the updated quantum code and the linear force prediction model includes: obtaining an initial movement distance, and updating the initial quantum code based on the left fitness value, the right fitness value, the initial binary code, the preset optimal position, and the initial movement distance using a preset update rule to obtain an updated quantum code; measuring the updated quantum code to obtain an updated binary code; performing denormalization processing on the updated binary code to obtain an updated solution; and substituting the updated solution into the linear force prediction model to obtain an updated fitness value.

[0012] Optionally, in a sixth implementation manner of the first aspect of the present invention, if the updated fitness value is less than the preset fitness value, obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, and returning to determine the search direction according to the initial binary code and the preset optimal position, includes: if the updated fitness value is less than the preset fitness value, obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, updating the initial search distance and the initial moving distance based on a preset update rule, and then returning to determine the search direction according to the initial binary code and the preset optimal position; if the updated fitness value is greater than or equal to the preset fitness value, updating the initial search distance and the initial moving distance based on the preset update rule, and returning to determine the search direction according to the initial binary code and the preset optimal position.

[0013] The second aspect of the present invention provides an optimization device for a magnetic levitation planar motor, comprising: a training module for pre-training a linear force prediction model based on coil parameters and permanent magnet parameters; a first measurement module for generating an initial quantum code based on the coil parameters to be optimized and the permanent magnet parameters to be optimized, and measuring the initial quantum code to obtain an initial binary code; a first update module for obtaining a preset optimal position, determining a search direction according to the initial binary code and the preset optimal position, and updating the preset left antenna code and the right antenna code based on the determined search direction using a preset update rule; a second measurement module for measuring the updated left antenna code and the updated right antenna code respectively to obtain a left binary code and a right binary code, and updating the left antenna code based on the left binary code. , right binary code and linear force prediction model to obtain left fitness value and right fitness value; a second updating module, used to update the initial quantum code based on the left fitness value, the right fitness value, the initial binary code and the preset optimal position, using a preset update rule to obtain an updated quantum code, and obtain an updated fitness value based on the updated quantum code and the linear force prediction model; a judgment module, used to obtain an updated position corresponding to the updated quantum code if the updated fitness value is less than the preset fitness value, replace the preset optimal position with the updated position, and return to execute to determine the search direction according to the initial binary code and the preset optimal position; an output module, used to obtain the optimal fitness value when a preset iteration stop condition is met, and output the optimal solution corresponding to the optimal fitness value.

[0014] The third aspect of the present invention provides an optimization device for a magnetic levitation planar motor, which includes: a memory and at least one processor, wherein the memory stores instructions; at least one processor calls the instructions in the memory so that the optimization device for the magnetic levitation planar motor performs each step of the optimization method for the magnetic levitation planar motor described in any one of the above items.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored thereon, and when the instructions are executed by a processor, the steps of any of the above-mentioned methods for optimizing a magnetically levitated planar motor are implemented.

[0016] In the technical solution of the present invention, the coil parameters and permanent magnet parameters are optimized in combination with quantum coding technology, and the optimal solutions for the coil parameters to be optimized and the permanent magnet parameters to be optimized can be quickly obtained. The optimization solutions for the coil size structure and the permanent magnet size structure can be quickly obtained, so that the optimized magnetic levitation planar motor can significantly reduce the working energy consumption while meeting the suspension force and high-precision control requirements, and improve the energy utilization efficiency of the magnetic levitation positioning platform, that is, achieve the goal of reducing working power consumption and optimizing the motor structure size, thereby improving the overall performance of the magnetic levitation planar motor in many aspects. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A logic flow chart of an optimization method for a magnetically levitated planar motor provided by an embodiment of the present invention;

[0018] Figure 2 A schematic structural diagram of an optimization device for a magnetically levitated planar motor according to an embodiment of the present invention;

[0019] Figure 3 A schematic structural diagram of an optimization device for a magnetically levitated planar motor according to an embodiment of the present invention;

[0020] Figure 4 This is an exploded structural diagram of the magnetic levitation planar motor provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The present invention provides an optimization method, device, apparatus and storage medium for a magnetically levitated planar motor. In the present invention, the terms "first", "second", "third", "fourth", etc. (if any) in the description and claims of the present invention 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 apparatus 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 apparatus.

[0022] To more clearly illustrate the optimization process of the present invention, Figure 4An example of a magnetically levitated planar motor is demonstrated. The motor is structurally divided into two main parts: a stator and a mover. A permanent magnet array is embedded in the mover, while a coil array is provided on the surface of the stator. This realizes the functions of magnetic levitation and planar motion, relying on the precise collaboration between the permanent magnet array on the mover and the inductor coil on the stator. By finely controlling the on-off state of the current in the inductor coil and its periodic changes, the mover can be effectively driven to perform a predetermined motion on the stator surface. In terms of structural design, both the stator and the mover adopt an innovative three-layer structure layout. Specifically, the mover consists of covers on both sides and a mounting assembly in the middle. This design prevents the permanent magnets from being directly exposed to the external environment, thereby improving the durability and safety of the motor. The structure of the stator includes an upper cover, a mounting assembly and a base. The upper cover not only protects the coil array, but also ensures the integrity of the overall structure. Furthermore, a spacious cavity area is designed between the base of the stator and the middle mounting assembly to simplify the wiring process, improve assembly efficiency and facilitate maintenance.

[0023] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the optimization method of the magnetic levitation planar motor in the embodiment of the present invention includes:

[0024] 101. Pre-training linear force prediction model based on coil parameters and permanent magnet parameters;

[0025] In this embodiment, by applying the output results of the linear force prediction model, the levitation force can be further maximized and optimized to determine the optimal structural parameter combination, thereby significantly improving the energy efficiency performance of the magnetic levitation planar motor and providing precise guidance for the optimization of the magnetic levitation system.

[0026] 102. Generate an initial quantum code based on the coil parameters to be optimized and the permanent magnet parameters to be optimized, and measure the initial quantum code to obtain an initial binary code;

[0027] 103. Obtain a preset optimal position, determine a search direction according to the initial binary code and the preset optimal position, and update the preset left antenna code and right antenna code based on the determined search direction using a preset update rule;

[0028] 104. Measure the updated left antenna code and the updated right antenna code respectively to obtain a left binary code and a right binary code, and obtain a left fitness value and a right fitness value based on the left binary code, the right binary code, and a linear force prediction model;

[0029] 105. Based on the left fitness value, the right fitness value, the initial binary code, and the preset optimal position, the initial quantum code is updated using a preset update rule to obtain an updated quantum code, and an updated fitness value is obtained based on the updated quantum code and the linear force prediction model;

[0030] 106. If the updated fitness value is less than the preset fitness value, obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, and returning to determine the search direction according to the initial binary code and the preset optimal position;

[0031] 107. When the preset iteration stop condition is met, the optimal fitness value is obtained, and the optimal solution corresponding to the optimal fitness value is output;

[0032] In this embodiment, the preset iteration stop condition can be a preset maximum number of iterations, which can be pre-set by the designer according to the accuracy requirements of the solution; the optimal fitness value is the smaller value after the last iteration comparison, that is, in the last iteration, if the updated fitness value is less than the fitness value of the previous iteration, then the optimal fitness value is the updated fitness value; if the updated fitness value is greater than or equal to the fitness value of the previous iteration, then the optimal fitness value is the fitness value of the previous iteration.

[0033] The present application discloses an optimization method for a magnetic levitation planar motor. By combining quantum coding technology, the coil parameters and permanent magnet parameters are optimized, and the optimal solutions for the coil parameters to be optimized and the permanent magnet parameters to be optimized can be quickly obtained. The optimization solutions for the coil size structure and the permanent magnet size structure can be quickly obtained. The optimized magnetic levitation planar motor can significantly reduce the operating energy consumption while meeting the requirements of suspension force and high-precision control, thereby improving the energy utilization efficiency of the magnetic levitation positioning platform, that is, achieving the goals of reducing operating power consumption and optimizing the motor structure size, thereby improving the overall performance of the magnetic levitation planar motor in many aspects.

[0034] In this embodiment, the pre-training of the linear force prediction model based on the coil parameters and the permanent magnet parameters specifically includes:

[0035] 201. Acquire simulation analysis data, where the independent variables of the simulation analysis data are the number of coils, the coil spacing, the number of permanent magnets, the permanent magnet spacing, and the motion time, and the dependent variable of the simulation analysis data is the linear force;

[0036] In this embodiment, the simulation analysis data is based on the finite element analysis method, and the COMSOL Simulation platform output: During the simulation process, the movement of the mover is set to a constant speed. Based on this, by adjusting the number of coils, the spacing between the coils, the number of permanent magnets, and the spacing between the permanent magnets, the linear force changes of the mover and the stator throughout the entire motion trajectory are calculated and analyzed, thereby obtaining multiple sets of analysis results under different conditions, and integrating multiple sets of analysis results to obtain simulation analysis data.

[0037] 202. Perform data cleaning, feature scaling, and data partitioning on the simulation analysis data to obtain a training set, a validation set, and a test set;

[0038] In this embodiment, missing values ​​and outliers in the data set are checked, and data cleaning processing is performed (for example, filling, deletion or interpolation); then, all features (number of coils, number of permanent magnets, permanent magnet spacing, coil spacing, movement time) are standardized or normalized to eliminate the influence of different dimensions and achieve feature scaling processing; finally, the data set is divided into training set, validation set and test set, for example, 70% training set, 15% validation set, and 15% test set, to complete the data division processing.

[0039] 203. Construct a linear force prediction model based on a neural network architecture, wherein the linear force prediction model includes an input layer, a hidden layer, and an output layer, wherein the activation function of the hidden layer is a ReLU activation function, and the activation function of the output layer is a linear activation function;

[0040] In this embodiment, the input layer is responsible for receiving simulation analysis data, and the hidden layer processes this data through a series of neurons. The ReLU activation function is selected because it can effectively deal with the gradient vanishing problem and accelerate convergence during training. The output layer uses a linear activation function to ensure that the predicted value output by the model maintains a linear relationship with the actual force value.

[0041] 204. Input the training set, validation set, and test set into the linear force prediction model respectively to train the model, and optimize the training process of the linear force prediction model by combining regularization technology and random search algorithm;

[0042] In this embodiment, the training set is used for model learning and parameter adjustment, the validation set is used to evaluate model performance and fine-tune hyperparameters, and the test set is used to finally evaluate the generalization ability of the model. To ensure that the model does not overfit on the training data, regularization techniques such as L1 or L2 regularization are introduced. Regularization techniques can limit model complexity and thus improve the generalization ability of the model. In addition, a random search algorithm is used to optimize hyperparameters. The random search algorithm is an efficient parameter optimization method that finds the optimal model configuration by randomly selecting parameter combinations and evaluating their performance.

[0043] 205. Deploy the trained linear force prediction model to the optimization system.

[0044] In this embodiment, the initial quantum code is generated based on the coil parameters to be optimized and the permanent magnet parameters to be optimized, and the initial quantum code is measured to obtain the initial binary code, specifically including:

[0045] 301. Use quantum bits to encode the coil parameters to be optimized and the permanent magnet parameters to be optimized, where each coil parameter to be optimized and each permanent magnet parameter to be optimized corresponds to one quantum bit, and integrate all quantum bits to obtain an initial quantum code;

[0046] In this embodiment, quantum bits are used to encode design variables. Each design variable corresponds to a quantum bit. The variables are expressed using the superposition property of quantum states. Each quantum bit can simultaneously represent the superposition of two basic quantum states, as shown in formula (1):

[0047]

[0048] Where, It represents the angle between the line connecting the studied position on the xy plane and the projection of the positive x-axis, ranging from 0 to 2π; represents the relative phase. Quantum algorithms use this phase relationship between quantum bits to make the correct results of the calculation constructively interfere and be amplified, while the incorrect results will be suppressed due to destructive interference; θ represents the angle between the line connecting the research position and the origin of the coordinate system and the projection on the positive z-axis, ranging from 0 to π; |0> and |1> represent two basic quantum states.

[0049] 302. Mapping the quantum bits onto the Bloch sphere respectively, and performing measurement operations on the mapped quantum bits to obtain initial binary codes corresponding to the initial quantum codes;

[0050] In this embodiment, the Bloch sphere is a three-dimensional unit sphere, and any point on it can be expressed by spherical coordinates, as shown in formula (2); the state of the quantum bit can be simultaneously represented at three positions in space, and the codes corresponding to these states are obtained through a single measurement operation. The code corresponding to the i-th dimension is recorded as [p i,x ,p i,y ,p i,z ], as shown in formula (3):

[0051]

[0052] The Bloch quantum coding corresponding to n design variables can be written as formula (4):

[0053]

[0054] In this embodiment, there are four design variables, namely the number of coils, the coil spacing, the number of permanent magnets, and the permanent magnet spacing.

[0055] 303. Perform a denormalization process on the initial binary code to obtain an initial solution corresponding to the initial binary code, wherein the initial solution includes three feasible solutions;

[0056] In this embodiment, since each dimension of the Bloch coordinates is in the range of [-1, 1], in order to meet the actual needs of the specific design problem, these coordinates need to be appropriately transformed and mapped to the target solution space; assuming that the Bloch coordinates of the i-th quantum bit are [p i,x ,p i,y ,p i,z ], which can be mapped to the specified solution [x i,x ,x i,y ,x i,z ], this step can be called denormalization:

[0057]

[0058] Among them, lb i and ub i They represent the lower limit and upper limit of the i-th dimension variable in the solution space respectively.

[0059] 304. Select any feasible solution from the initial solution as an initial search point according to a preset selection rule;

[0060] In this embodiment, three feasible solutions can be generated by formula (5). This step realizes the selection and optimization of feasible solutions by introducing parameters such as random variables and the number of iterations. This method improves the distribution quality of the initial population in the search space, thereby improving the solution efficiency and accuracy. The selection of feasible solutions during the entire search process is shown in formula (6):

[0061]

[0062] Where T is the maximum number of iterations, rand(·) generates a random variable in the range [0,1], and t is the current number of iterations.

[0063] In this embodiment, when returning to execute the search direction determined according to the initial binary code and the preset optimal position, an initial solution corresponding to the initial binary code is obtained, and according to the preset selection rules, a feasible solution is reselected from the initial solution as the initial search point to guide the next iterative search process.

[0064] In this embodiment, the step of obtaining a preset optimal position, determining a search direction according to the initial binary code and the preset optimal position, and updating the preset left antenna code and the right antenna code based on the determined search direction using a preset update rule specifically includes:

[0065] 401. Obtain a preset optimal position, and determine a search direction according to the initial binary code and the preset optimal position;

[0066] In this embodiment, in the first iteration, the position q corresponding to the currently searched optimal solution is best is the preset optimal position; in each iteration, according to the current code q t The position q corresponding to the currently searched optimal solution best , dynamically determine the search direction of the longicorn whiskers The specific calculation formula is shown in formula (7):

[0067]

[0068] Among them, rs, rg, and rnd(n,1) are n-dimensional random variables in the range [0,1].

[0069] 402. Obtain an initial beetle whisker search distance, and determine a beetle whisker rotation angle based on the initial beetle whisker search distance and the determined search direction;

[0070] 403. Based on the rotation angle of the longicorn antennae and the initial binary code, a preset update rule is used to update the preset left antenna code and the right antenna code;

[0071] In this embodiment, the quantum encoding of the two antennae is calculated by the update rules of equations (8) and (9). and Perform iterative updates:

[0072]

[0073]

[0074] Among them, d t is the distance that the beetle must search at time t, φ t Represents the rotation angle of the beetle's whiskers at time t.

[0075] In this embodiment, the updated left antenna code and the updated right antenna code are measured respectively to obtain the left binary code and the right binary code, and the left fitness value and the right fitness value are obtained based on the left binary code, the right binary code and the linear force prediction model, specifically including:

[0076] 501. Measure the updated left antenna code and the updated right antenna code respectively to obtain a left binary code and a right binary code;

[0077] In this embodiment, the left binary code is The right binary code is

[0078] 502. Denormalize the left binary code and the right binary code to obtain a left solution and a right solution respectively;

[0079] In this embodiment, the left solution and the right solution are respectively and

[0080] 503. Substitute the left solution and the right solution into the linear force prediction model respectively to obtain a left fitness value and a right fitness value;

[0081] In this embodiment, the left fitness value and the right fitness value are respectively set to and

[0082] In this embodiment, based on the left fitness value, the right fitness value, the initial binary code and the preset optimal position, the initial quantum code is updated using a preset update rule to obtain an updated quantum code, and the updated fitness value is obtained based on the updated quantum code and the linear force prediction model, specifically including:

[0083] 601. Obtain an initial moving distance, and based on the left fitness value, the right fitness value, the initial binary code, the preset optimal position, and the initial moving distance, update the initial quantum code using a preset update rule to obtain an updated quantum code;

[0084] In this embodiment, the update rules of the initial quantum code are shown in Equations (8) and (10):

[0085]

[0086] Among them, δ t is the distance moved by the longicorn at time t, and sign(·) is the sign function.

[0087] 602. Measure the updated quantum code to obtain an updated binary code;

[0088] In this embodiment, let the updated binary code be p t+1 .

[0089] 603. Denormalize the updated binary code to obtain an updated solution.

[0090] In this embodiment, let the updated solution be x t+1 .

[0091] 604. Substitute the updated solution into the linear force prediction model to obtain an updated fitness value;

[0092] In this embodiment, the updated fitness value is G(x t+1 ).

[0093] In this embodiment, if the updated fitness value is less than the preset fitness value, then obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, and returning to determine the search direction according to the initial binary code and the preset optimal position, specifically includes:

[0094] 701. If the updated fitness value is less than the preset fitness value, obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, and updating the initial search distance and the initial moving distance based on a preset update rule, and then returning to determine the search direction based on the initial binary code and the preset optimal position;

[0095] In this embodiment, when returning to execute the search direction determined according to the initial binary code and the preset optimal position, a feasible solution is first re-selected as the initial solution from the initial solution corresponding to the initial binary code based on the initial binary code, and then the search direction is re-determined according to the initial binary code and the preset optimal position to re-execute the search process.

[0096] 702. If the updated fitness value is greater than or equal to the preset fitness value, the initial search distance and the initial moving distance are updated based on a preset update rule, and the process returns to determining the search direction according to the initial binary code and the preset optimal position.

[0097] In this embodiment, the initial moving distance and the initial search distance are updated based on equations (11) and (12) respectively; specifically, the distance δ moved by the longicorn at time t is updated based on the cosine theorem. t Perform nonlinear update, δ 0 is the initial step size, as shown in formula (11); the enhanced step size follows the curve based on the cosine theorem, maintaining a large step size in the early stage of iteration, quickly covering the search space and improving the global search efficiency; in the later stage of iteration, the step size gradually decreases, accurately locating the optimal solution and improving the local search accuracy; this dynamic step size adjustment method effectively balances the global search capability and local convergence performance of the algorithm, thereby significantly improving the optimization efficiency and solution accuracy;

[0098]

[0099] Among them, η δis the decay rate of the beetle's whisker search distance, c is a constant, and δ t+1 and d t+1 , to ensure the gradualness and stability of the search process, T is the maximum number of iterations, and t is the current number of iterations.

[0100] In this embodiment, since the improved beetle whisker algorithm can dynamically adjust the optimal position to dynamically adjust the search direction, it can converge to the global optimal solution more quickly during the search process; this fast convergence feature not only improves the solution speed of the algorithm, but also ensures the efficiency and stability of the solution process; further, the present application proposes an improved beetle whisker algorithm based on quantum state coding, variable normalization and cosine theorem, which realizes iterative updates through quantum bit coding and quantum rotating gates, thereby having a more powerful global search capability, effectively avoiding the defect that the traditional beetle whisker algorithm is prone to falling into local optimality.

[0101] The above describes the optimization method of the magnetic levitation planar motor in the embodiment of the present invention. The following describes the optimization device of the magnetic levitation planar motor in the embodiment of the present invention. Figure 2 An embodiment of an optimization device for a magnetically levitated planar motor according to an embodiment of the present invention includes:

[0102] A training module 801 is used to pre-train a linear force prediction model based on coil parameters and permanent magnet parameters;

[0103] A first measurement module 802 is configured to generate an initial quantum code based on the coil parameters to be optimized and the permanent magnet parameters to be optimized, and measure the initial quantum code to obtain an initial binary code;

[0104] A first updating module 803 is configured to obtain a preset optimal position, determine a search direction according to the initial binary code and the preset optimal position, and update the preset left antenna code and the preset right antenna code based on the determined search direction using a preset updating rule;

[0105] A second measurement module 804 is configured to measure the updated left antenna code and the updated right antenna code respectively to obtain a left binary code and a right binary code, and obtain a left fitness value and a right fitness value based on the left binary code, the right binary code, and a linear force prediction model;

[0106] A second updating module 805 is configured to update the initial quantum code using a preset updating rule based on the left fitness value, the right fitness value, the initial binary code, and the preset optimal position to obtain an updated quantum code, and obtain an updated fitness value based on the updated quantum code and the linear force prediction model;

[0107] A judgment module 806 is configured to obtain an updated position corresponding to the updated quantum code if the updated fitness value is less than a preset fitness value, replace the preset optimal position with the updated position, and return to determine a search direction based on the initial binary code and the preset optimal position;

[0108] Output module 807 is used to obtain the best fitness value when the preset iteration stop condition is met, and output the optimal solution corresponding to the best fitness value

[0109] Based on the same idea as the method in the above embodiment, the device provided in this application can implement the method in the above embodiment.

[0110] above Figure 2 The optimization device of the magnetic levitation planar motor in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The optimization device of the magnetic levitation planar motor in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0111] Figure 3 Schematic diagram of the structure of a magnetic levitation planar motor optimization device provided in an embodiment of the present invention. The magnetic levitation planar motor optimization device 900 may vary significantly due to different configurations or performance. It may include one or more processors (central processing units, CPUs) 910 (for example, one or more processors) and memory 920, and one or more storage media 930 (for example, one or more mass storage devices) storing application programs 933 or data 932. The memory 920 and storage medium 930 may be either short-term or persistent storage. The program stored in the storage medium 930 may include one or more modules (not shown), each of which may include a series of instruction operations in the magnetic levitation planar motor optimization device 900. Furthermore, the processor 910 may be configured to communicate with the storage medium 930, and the magnetic levitation planar motor optimization device 900 may execute the series of instruction operations in the storage medium 930 to implement the steps of the magnetic levitation planar motor optimization method provided in the above-mentioned method embodiments.

[0112] The optimization device 900 for the magnetically levitated planar motor may further include one or more power supplies 940, one or more wired or wireless network interfaces 950, one or more input / output interfaces 960, and / or one or more operating systems 931, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be understood by those skilled in the art that Figure 3The optimized device structure of the magnetic levitation planar motor shown does not constitute a limitation on the optimized device of the magnetic levitation planar motor, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0113] 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 optimization method for a magnetically levitated planar motor.

[0114] 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, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0115] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc., various media that can store program code.

[0116] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for optimizing a magnetically levitated planar motor, characterized in that: include: Pre-training a linear force prediction model based on coil parameters and permanent magnet parameters; An initial quantum code is generated based on the coil parameters to be optimized and the permanent magnet parameters to be optimized, and the initial quantum code is measured to obtain an initial binary code. Specifically, quantum bits are used to encode the coil parameters to be optimized and the permanent magnet parameters to be optimized, with each coil parameter to be optimized and each permanent magnet parameter to be optimized corresponding to one quantum bit, and all quantum bits are integrated to obtain the initial quantum code. Mapping the quantum bits onto the Bloch sphere respectively, and performing measurement operations on the mapped quantum bits to obtain the initial binary code corresponding to the initial quantum code; Performing a denormalization process on the initial binary code to obtain an initial solution corresponding to the initial binary code, wherein the initial solution includes three feasible solutions; Select any feasible solution from the initial solution as an initial search point according to a preset selection rule; Obtaining a preset optimal position, determining a search direction according to the initial binary code and the preset optimal position, and updating the preset left antenna code and the right antenna code based on the determined search direction using a preset update rule; The updated left antenna coding and the updated right antenna coding are measured respectively to obtain a left binary coding and a right binary coding, and a left fitness value and a right fitness value are obtained based on the left binary coding, the right binary coding and the linear force prediction model; specifically, the updated left antenna coding and the updated right antenna coding are measured respectively to obtain a left binary coding and a right binary coding; the left binary coding and the right binary coding are denormalized respectively to obtain a left solution and a right solution; the left solution and the right solution are substituted into the linear force prediction model respectively to obtain a left fitness value and a right fitness value; Based on the left fitness value, the right fitness value, the initial binary code and the preset optimal position, the initial quantum code is updated using a preset update rule to obtain an updated quantum code, and an updated fitness value is obtained based on the updated quantum code and the linear force prediction model; If the updated fitness value is less than the preset fitness value, obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, and returning to determine the search direction according to the initial binary code and the preset optimal position; When the preset iteration stopping condition is met, the optimal fitness value is obtained and the optimal solution corresponding to the optimal fitness value is output.

2. The optimization method of the magnetic levitation planar motor according to claim 1, characterized in that: The pre-trained linear force prediction model based on coil parameters and permanent magnet parameters includes: Acquiring simulation analysis data, wherein the independent variables of the simulation analysis data are the number of coils, the coil spacing, the number of permanent magnets, the permanent magnet spacing, and the motion time, and the dependent variable of the simulation analysis data is the linear force; The simulation analysis data are processed by data cleaning, feature scaling and data partitioning to obtain the training set, validation set and test set; Constructing a linear force prediction model based on a neural network architecture, wherein the linear force prediction model includes an input layer, a hidden layer, and an output layer, wherein the activation function of the hidden layer is a ReLU activation function, and the activation function of the output layer is a linear activation function; Inputting the training set, validation set, and test set into the linear force prediction model respectively to train the model, and optimizing the training process of the linear force prediction model by combining regularization technology and random search algorithm; Deploy the trained linear force prediction model to the optimization system.

3. The optimization method of the magnetic levitation planar motor according to claim 1, characterized in that: The step of obtaining a preset optimal position, determining a search direction according to the initial binary code and the preset optimal position, and updating the preset left antenna code and the right antenna code using a preset update rule based on the determined search direction includes: Obtaining a preset optimal position, and determining a search direction according to the initial binary code and the preset optimal position; Obtaining an initial beetle whisker search distance, and determining a beetle whisker rotation angle based on the initial beetle whisker search distance and the determined search direction; Based on the rotation angle of the longicorn antennae and the initial binary code, the preset left antennae code and the right antennae code are updated using a preset update rule.

4. The optimization method of the magnetic levitation planar motor according to claim 3, characterized in that: The method of updating the initial quantum code based on the left fitness value, the right fitness value, the initial binary code, and the preset optimal position using a preset update rule to obtain an updated quantum code, and obtaining an updated fitness value based on the updated quantum code and the linear force prediction model includes: Obtaining an initial moving distance, and updating the initial quantum code using a preset update rule based on the left fitness value, the right fitness value, the initial binary code, the preset optimal position, and the initial moving distance to obtain an updated quantum code; Measure the updated quantum code to obtain the updated binary code; Denormalize the updated binary code to obtain the updated solution; Substitute the updated solution into the linear force prediction model to obtain an updated fitness value.

5. The optimization method of the magnetic levitation planar motor according to claim 4, characterized in that: If the updated fitness value is less than the preset fitness value, obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, and returning to determine the search direction according to the initial binary code and the preset optimal position, including: If the updated fitness value is less than the preset fitness value, then obtaining an updated position corresponding to the updated quantum code, replacing the preset optimal position with the updated position, and updating the initial search distance and the initial moving distance based on the preset update rule, and then returning to determine the search direction according to the initial binary code and the preset optimal position; If the updated fitness value is greater than or equal to the preset fitness value, the initial search distance and the initial moving distance are updated based on a preset update rule, and the process returns to determining the search direction according to the initial binary code and the preset optimal position.

6. An optimization device for a magnetically levitated planar motor, characterized in that: include: A training module for pre-training a linear force prediction model based on coil parameters and permanent magnet parameters; A first measurement module is configured to generate an initial quantum code based on the coil parameters to be optimized and the permanent magnet parameters to be optimized, and measure the initial quantum code to obtain an initial binary code. Specifically, quantum bits are used to encode the coil parameters to be optimized and the permanent magnet parameters to be optimized, with each coil parameter to be optimized and each permanent magnet parameter to be optimized corresponding to one quantum bit, and all quantum bits are integrated to obtain the initial quantum code. Mapping the quantum bits onto the Bloch sphere respectively, and performing measurement operations on the mapped quantum bits to obtain the initial binary code corresponding to the initial quantum code; Performing a denormalization process on the initial binary code to obtain an initial solution corresponding to the initial binary code, wherein the initial solution includes three feasible solutions; Select any feasible solution from the initial solution as an initial search point according to a preset selection rule; a first updating module, configured to obtain a preset optimal position, determine a search direction according to the initial binary code and the preset optimal position, and update the preset left antenna code and the right antenna code based on the determined search direction using a preset updating rule; A second measurement module is configured to measure the updated left antenna coding and the updated right antenna coding respectively to obtain a left binary code and a right binary code, and obtain a left fitness value and a right fitness value based on the left binary code, the right binary code, and the linear force prediction model; specifically, the updated left antenna coding and the updated right antenna coding are measured respectively to obtain a left binary code and a right binary code; the left binary code and the right binary code are denormalized respectively to obtain a left solution and a right solution; the left solution and the right solution are substituted into the linear force prediction model respectively to obtain a left fitness value and a right fitness value; a second updating module, configured to update the initial quantum code using a preset updating rule based on the left fitness value, the right fitness value, the initial binary code, and the preset optimal position to obtain an updated quantum code, and obtain an updated fitness value based on the updated quantum code and the linear force prediction model; a judgment module, configured to obtain an updated position corresponding to the updated quantum code if the updated fitness value is less than a preset fitness value, replace the preset optimal position with the updated position, and return to determine a search direction based on the initial binary code and the preset optimal position; The output module is used to obtain the optimal fitness value when a preset iteration stopping condition is met, and output the optimal solution corresponding to the optimal fitness value.

7. An optimization device for a magnetically levitated planar motor, characterized in that: The optimization device for the magnetically levitated planar motor includes: a memory and at least one processor, wherein the memory stores instructions; At least one of the processors calls the instructions in the memory to enable the magnetic levitation planar motor optimization device to perform each step of the magnetic levitation planar motor optimization method according to any one of claims 1 to 5.

8. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by a processor, the various steps of the method for optimizing the magnetic levitation planar motor according to any one of claims 1 to 5 are implemented.

Images