Voice coil motor hysteresis modeling method and related products thereof

Abstract

The embodiment of the application discloses a voice coil motor hysteresis modeling method and a related product, wherein the steps of the method comprise: determining a transfer function relationship between an output displacement and an input current, determining a phase frequency response and an amplitude frequency response relationship according to the transfer function relationship between the output displacement and the input current; and based on the phase frequency response and the amplitude frequency response relationship, fitting a voice coil motor hysteresis trajectory by using a neural network to obtain a voice coil motor hysteresis model. The scheme disclosed in the application can effectively describe the hysteresis phenomenon exhibited by the voice coil motor when the output millimeter-level displacement is output, and the description effect of the model basically remains unchanged with the change of the frequency, and the scheme has a more stable dynamic trajectory tracking effect than a pure NARX neural network model.

Description

Technical Field

[0001] This solution relates to the field of computer processing technology. More specifically, it relates to a method for modeling hysteresis of voice coil motors based on NARX neural networks. Background Technology

[0002] The latest developments in precision manufacturing and nanoscale scientific research have created a huge demand for high-precision positioning systems with long stroke ranges. In recent years, the continuous improvement in the performance of ultra-precision positioning systems has made it imperative to resolve the contradiction between the accuracy and speed of actuators and their long stroke range.

[0003] Traditional precision positioning platforms use piezoelectric actuators, which typically have a stroke range of no more than 150 μm. Even with complex lever mechanisms to amplify the displacement, it is still difficult to achieve a stroke range greater than 1 mm. Furthermore, this often reduces the platform's mechanical bandwidth. This has led researchers to search for new actuators that meet application requirements.

[0004] A voice coil motor (VCO) is a special type of DC linear motor, its structure resembling the moving coil speaker element in a loudspeaker. Since it lacks a transduction process and energy transfer device, it can theoretically provide infinitesimal resolution. It also boasts advantages such as high acceleration, high thrust, high frequency response, easy control, and small size. Due to these advantages, VCO motors are considered capable of resolving the trade-off between accuracy, speed, and large stroke. This makes them highly attractive for applications requiring precise positioning and rapid response. Therefore, VCO motors are currently gaining attention not only in high-excitation motion systems such as vibration damping systems, vibrating screen platforms, and medical devices, but also in precision positioning systems such as semiconductor manufacturing equipment, camera lenses, hard disk heads, high-end CNC machine tools, and microscopes. Summary of the Invention

[0005] The purpose of this invention is to provide a method for modeling hysteresis in voice coil motors and related products, so as to solve the problem that hysteresis in voice coil motors can cause large errors.

[0006] To achieve the above objectives, the following technical solution is adopted:

[0007] Firstly, this solution provides a method for modeling the hysteresis of a voice coil motor, the steps of which include:

[0008] Determine the transfer function relationship between the output displacement and the input current, and determine the phase frequency response and amplitude frequency response relationship based on the transfer function relationship between the output displacement and the input current;

[0009] Based on the relationship between phase frequency response and amplitude frequency response, a neural network is used to fit the hysteresis trajectory of the voice coil motor to obtain the hysteresis model of the voice coil motor.

[0010] In a preferred embodiment,

[0011] The determination of the transfer function relationship between the output displacement and the input current includes...

[0012] Establish the dynamic equations:

[0013] F e -F d -F s =ma

[0014]

[0015] Where N is the number of turns of the coil, B is the magnetic field strength around the coil, l is the effective length of the coil in the magnetic field, and k e F is the electromagnetic force constant of the voice coil motor; e It is electromagnetic force; F d It is a damping force, which includes air damping force and eddy current damping force, with a damping coefficient of k. d ;F s It is the elastic force of the spring, and the spring constant of the spring is k. s m is the mass of the entire coil and load, a is the acceleration of the coil during movement, v is the velocity of the coil, x is the displacement of the coil, and i is the current in the coil.

[0016] According to Kirchhoff's voltage law, the voltage equation for the equivalent circuit of the voice coil motor is as follows:

[0017]

[0018] Where L is the inductance of the coil, R is the resistance of the coil, and u is the voltage applied to the coil;

[0019] The relationship between the coil displacement x and the current i in the coil can be summarized as follows:

[0020]

[0021] Obtained through Laplace transform

[0022]

[0023] The transfer function relationship between the output displacement and the input current is obtained as follows:

[0024]

[0025] In a preferred embodiment, the relationship between the phase frequency response and the amplitude frequency response is as follows:

[0026]

[0027]

[0028]

[0029] A = (Real) 2 +Imaginary 2 ) 0.5

[0030]

[0031] In a preferred embodiment, the neural network includes an input layer, a hidden layer, and an output layer.

[0032] In a preferred embodiment, the activation function of the neurons in the hidden layer is a selected hyperbolic tangent function.

[0033] In a preferred embodiment, the activation functions of neurons in the input and output layers are linear mapping functions.

[0034] In a preferred embodiment, the neural network is trained using the LM method.

[0035] Secondly, this solution provides a voice coil motor hysteresis modeling system, which includes:

[0036] The module is constructed to determine the phase frequency response and amplitude frequency response relationship based on the transfer function relationship between the output displacement and the input current;

[0037] The fitting module, based on the relationship between phase frequency response and amplitude frequency response, uses a neural network to fit the hysteresis trajectory of the voice coil motor, thereby obtaining the hysteresis model of the voice coil motor.

[0038] Thirdly, this solution provides a computer storage medium, characterized in that it stores a computer program thereon, which, when executed by a processor, implements the method described in any of the preceding claims.

[0039] Fourthly, this solution provides a voice coil motor hysteresis modeling device, including: a processor; and a memory for storing executable instructions of the processor;

[0040] The processor is configured to execute the method described in any of the preceding methods by executing the executable instructions.

[0041] The beneficial effects of this invention are as follows:

[0042] The solution described in this application can effectively describe the hysteresis phenomenon exhibited by the voice coil motor when outputting millimeter-level displacement, and the model's descriptive effect remains basically unchanged with the change of frequency, providing a more stable dynamic trajectory tracking effect than simply using the NARX neural network model. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A schematic diagram illustrating an example of the voice coil motor hysteresis modeling method described in this scheme is shown.

[0045] Figure 2 This diagram illustrates the structure of the TF-NARX model described in this scheme.

[0046] Figure 3 A schematic diagram of the equivalent circuit described in this scheme is shown;

[0047] Figure 4 A schematic diagram of the NARX neural network model with the serial-parallel architecture described in this scheme is shown;

[0048] Figure 5 The diagram shows the effect of hysteresis modeling for the initial sample at 30Hz.

[0049] Figure 6 The comparison chart of modeling results at different frequencies is shown.

[0050] Figure 7 A schematic diagram illustrating an example of the voice coil motor hysteresis modeling system described in this scheme is shown.

[0051] Figure 8 This diagram illustrates an example of the voice coil motor hysteresis modeling device described in this scheme. Detailed Implementation

[0052] To make the present invention, its technical solutions, and advantages clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0053] Analysis and research of existing technologies have shown that the hysteresis of voice coil motors can cause significant errors, severely reducing their positioning performance. Furthermore, machine learning-based intelligent algorithms have attracted widespread attention due to their strong nonlinear fitting capabilities; however, few studies have yet applied these algorithms to compensate for the hysteresis characteristics of voice coil motors with large stroke ranges.

[0054] Therefore, this solution aims to provide a method for modeling hysteresis in voice coil motors. This method utilizes a modified NARX neural network model to attempt to combine the transfer function model of the voice coil motor with the traditional NARX neural network (referred to as the TF-NARX model) to achieve dynamic rate-dependent hysteresis modeling. Specifically, rate-dependent hysteresis modeling is achieved using fixed parameters within a certain frequency range, rather than repeatedly tuning the parameters for specific frequencies. This significantly improves the model's versatility.

[0055] The following section will describe in detail the hysteresis modeling method for voice coil motors proposed in this scheme, with reference to the accompanying drawings. Figure 1 As shown, specifically, the method includes:

[0056] Step S1: Determine the transfer function relationship between the output displacement and the input current, and determine the phase frequency response and amplitude frequency response relationship based on the transfer function relationship between the output displacement and the input current;

[0057] Step S2: Based on the relationship between phase frequency response and amplitude frequency response, a neural network is used to fit the hysteresis trajectory of the voice coil motor to obtain the hysteresis model of the voice coil motor.

[0058] Specifically, the voice coil battery hysteresis model construction described in this solution is divided into two parts, as follows: Figure 2 As shown, the first part, consisting of a transfer function, is responsible for initially describing the phase and amplitude frequency responses, transforming the input current signal into an intermediate variable. The second part, the NARX neural network, is responsible for further nonlinear fitting, improving the model's accuracy.

[0059] The working principle of a voice coil motor follows the left-hand rule: when a current *i* is applied to the coil, the coil experiences an electromagnetic force from the surrounding magnetic field. The dynamic equation for the entire process can be expressed as:

[0060] F e -F d -F s =ma

[0061]

[0062] Where N is the number of turns of the coil, B is the magnetic field strength around the coil, l is the effective length of the coil in the magnetic field, and k e F is the electromagnetic force constant of the voice coil motor; e It is electromagnetic force; F d This refers to the damping force, which includes air damping and eddy current damping. Here, air damping and eddy current damping are considered as a single factor, with a damping coefficient of k. d ;F s It is the elastic force of the spring, and the spring constant of the spring is k. sWhere m is the mass of the entire coil and load, a is the acceleration of the coil during movement, v is the coil velocity, and x is the coil displacement. It is understandable that, under the same current conditions, the maximum thrust and the fastest response acceleration that the motor can provide both increase with the electromagnetic force constant k. e The current increases with the increase of ; i is the current in the coil;

[0063] Ignoring magnetic leakage at the magnet edges and yoke gaps, the electrical equivalent circuit of a traditional voice coil motor is as follows: Figure 3 As shown. According to Lenz's law, the motion of a coil in a magnetic field generates a back electromotive force e, which opposes the motion in the corresponding direction, thus inhibiting the motion. The coil itself also has a certain resistance and self-inductance. Therefore, according to Kirchhoff's voltage law, the voltage equation for the equivalent circuit of the voice coil motor can be obtained as follows:

[0064]

[0065] Where L is the inductance of the coil, R is the resistance of the coil, u is the voltage applied to the coil, i is the current in the coil, v is the speed of the coil, and x is the displacement of the coil.

[0066] The relationship between displacement x and current i can be summarized as follows:

[0067]

[0068] Through Laplace transform, we can obtain...

[0069]

[0070] Hysteresis is essentially a phase shift. This invention aims to predict the relationship between frequency and phase, starting from the physical model of a voice coil motor. The transfer function relationship between output displacement and input current can be derived from the working principle of the voice coil motor.

[0071]

[0072] The relationship between the phase frequency response and the amplitude frequency response can be derived theoretically as follows:

[0073]

[0074]

[0075]

[0076] A = (Real) 2 +Imaginary 2 ) 0.5

[0077]

[0078] However, due to issues such as magnetic leakage and model approximation errors, directly substituting physical parameters does not yield ideal results. Therefore, in determining the parameters of the transfer function sub-model, a phenomenological model approach is adopted. The PSO (Particle Swarm Optimization) parameter identification algorithm is used to determine a, b, c, and d in the formula.

[0079] NARX (Nonlinear Auto-Regressive with Exogenous Inputs) model is a dynamic model with feedback, primarily used to handle time series problems. It automatically identifies the functional relationship between the target variable and input variables within a time series. NARX models exist in various forms, with neural network forms being the most common.

[0080] The NARX neural network model used in this solution is specifically constructed as follows: Figure 4 As shown.

[0081] Specifically, the proposed NARX neural network sub-model consists of three layers: an input layer, a hidden layer, and an output layer. The hyperbolic tangent function is chosen as the activation function for neurons in the hidden layer, while the linear mapping function is used for the activation functions of neurons in both the input and output layers. Therefore, the input-output relationship of this neural network is as follows:

[0082]

[0083] Where q = 1 + m a +m b and n represent the number of neurons in the input layer and hidden layer, respectively; and These are the weight coefficients for neurons in the previous layer to the hidden layer and the output layer, respectively. and b o This is the corresponding bias; z i (t) is the generalized input representation of the entire neural network, specifically the correspondence: {z1(t), ..., z...} i (t)} is σ(·) represents the hyperbolic tangent activation function.

[0084]

[0085] Because the Levenberg-Marquardt method (LM method) exhibits excellent fast convergence and stability in solving least squares problems, it is used as the default method for parameter tuning of NARX neural networks in MATLAB. This invention employs the LM method to train the neural network during the parameter identification process.

[0086] Comparing the fitted performance of the modified model with that of the simple NARX neural network model, it can be seen that at the reference frequency (30Hz), the error difference between the NARX neural network model and the TF-NARX neural network model is very small (e.g., Figure 5 (As shown). The errors between the TF-NARX model (represented by the blue dashed line), the NARX model (represented by the red dashed line), and the actual displacement (represented by the black line) are all within 1 μm, and the blue and red lines are very close. For parameter identification at a fixed frequency, this process is a static lag modeling, mainly reflecting the nonlinear fitting capability of the NARX neural network. Therefore, the trajectory prediction errors of both the NARX and TF-NARX neural network models are very small, less than 0.3 μm at 30 Hz.

[0087] like Figure 6 As shown, the measured displacement data of the voice coil motor at various frequencies are compared with the modeling results of the NARX neural network model and the TF-NARX model: (a) 15Hz; (b) 20Hz; (c) 25Hz; (d) 35Hz; (e) 40Hz; (f) 45Hz; (g) 50Hz; (h) 55Hz; (A0 = 0.3A). When the frequency is far from 30Hz, the absolute error values ​​of the two models change from submicron to micron level. Furthermore, due to the presence of a transfer function sub-model in the TF-NARX neural network model, the initial prediction of amplitude and phase shift, which are key information for hysteresis, is dynamically achieved. As the input current frequency increases, the relative root mean square error and relative maximum error of the TF-NARX neural network model are consistently less than 3.265‰ and 6.157‰, respectively. The errors of the TF-NARX neural network model are all smaller than those of the NARX neural network model.

[0088] Based on the trajectory fitting results described above, the proposed model for voice coil drivers effectively describes the hysteresis phenomenon exhibited by voice coil motors when outputting millimeter-level displacements. Furthermore, the model's descriptive performance remains largely unchanged with frequency variations, demonstrating more stable dynamic trajectory tracking compared to simply using the NARX neural network model. Its relative root mean square error is less than 3.265‰ within the experimental frequency band, and the maximum relative error is consistently less than 6.157‰, making it suitable for feedforward and hybrid control of voice coil drivers. It also shows great promise for applications in large-scale, high-precision positioning.

[0089] Based on the above-described implementation method for voice coil motor hysteresis modeling, this solution further provides a voice coil motor hysteresis modeling system 201. For example... Figure 7 As shown, the system includes a construction module 202 and a fitting module 203. The construction module 202 determines the phase frequency response and amplitude frequency response relationship based on the transfer function relationship between the output displacement and the input current; the fitting module 203, based on the phase frequency response and amplitude frequency response relationship, uses a neural network to fit the hysteresis trajectory of the voice coil motor to obtain the hysteresis model of the voice coil motor.

[0090] Based on the above-described implementation of the voice coil motor hysteresis modeling method, this solution further provides a computer-readable storage medium. This computer-readable storage medium is used to implement the program product of the above-described logistics sorting and scheduling method. It can be a portable compact disc read-only memory (CD-ROM) and include program code, and can run on a device, such as a personal computer. However, the program product of this solution is not limited to this. In this document, the readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0091] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0092] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0093] Program code contained on a computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0094] Program code for performing the operations of this solution can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing devices can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or they can be connected to external computing devices (e.g., via the Internet using an Internet service provider).

[0095] Based on the above-described implementation method for modeling voice coil motor hysteresis, this solution further provides an electronic device. For example... Figure 8 The electronic device shown is merely an example and should not impose any limitations on the functionality and scope of use of this embodiment.

[0096] like Figure 8 As shown, the voice coil motor hysteresis modeling device 301 is represented in the form of a general-purpose computing device. The components of the electronic device 301 may include, but are not limited to: at least one storage unit 302, at least one processing unit 303, a display unit 304, and a bus 305 for connecting different system components.

[0097] The storage unit 302 stores program code that can be executed by the processing unit 303, causing the processing unit 303 to perform the steps of the various exemplary embodiments described in the voice coil motor hysteresis modeling method. For example, the processing unit 303 can perform actions such as... Figure 1 The steps are shown in the figure.

[0098] Storage unit 302 may include volatile storage units, such as random access memory (RAM) and / or cache storage units, and may further include read-only memory (ROM).

[0099] Storage unit 302 may also include programs / utilities with program modules, such program modules including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0100] Bus 305 may include a data bus, an address bus, and a control bus.

[0101] Electronic device 301 can also communicate with one or more external devices 307 (e.g., keyboard, pointing device, Bluetooth device, etc.) via input / output (I / O) interface 306. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 301, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0102] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. For those skilled in the art, other variations or modifications can be made based on the above description. It is impossible to exhaustively list all the implementation methods here. All obvious variations or modifications derived from the technical solutions of the present invention are still within the protection scope of the present invention.

Claims

1. A method for modeling hysteresis in a voice coil motor, characterized in that, The steps of this method include: Determine the transfer function relationship between the output displacement and the input current, and determine the phase frequency response and amplitude frequency response relationship based on the transfer function relationship between the output displacement and the input current; Based on the relationship between phase frequency response and amplitude frequency response, the hysteresis trajectory of the voice coil motor is fitted using a NARX neural network to obtain the hysteresis model of the voice coil motor. The determination of the transfer function relationship between the output displacement and the input current includes... Establish the dynamic equations: in, It is the number of turns of the coil. It is the magnetic flux density around the coil. It is the effective length of the coil in the magnetic field. It is the electromagnetic force constant of the voice coil motor; It is electromagnetic force; It is a damping force, which includes air damping force and eddy current damping force, with a damping coefficient of . ; It is the elastic force of the spring, and the spring constant of the spring is... ; It refers to the mass of the entire coil and load. It is the acceleration of the coil during its motion. It is the coil speed. It is the displacement of the coil; It is the current in the coil; According to Kirchhoff's voltage law, the voltage equation for the equivalent circuit of the voice coil motor is as follows: in, It is the inductance value of the coil. It is the resistance value of the coil. It is the voltage applied to the coil; Summarize the coil displacement With the current in the coil The relationship between them is The transfer function relationship between the output displacement and the input current is obtained through Laplace transform. ； This represents the output displacement after the Laplace transform; The input current after Laplace transformation; The transfer function after Laplace transformation; Time-domain parameters Frequency domain variables after Laplace transform; a, b, c, and d are determined using the PSO parameter identification algorithm; The relationship between the phase frequency response and the amplitude frequency response is as follows: ω is the angular frequency, and j is the imaginary unit; The sub-model of the NARX neural network includes an input layer, hidden layers, and an output layer; the activation function of neurons in the hidden layer is the hyperbolic tangent function, and the activation functions of neurons in the input and output layers are linear mapping functions, as detailed below: in, and These represent the number of neurons in the input layer and the hidden layer, respectively. and These are the weight coefficients for neurons in the previous layer to the hidden layer and the output layer, respectively. and This is the corresponding bias; It is the generalized input representation of the entire neural network. This represents the activation function of the hyperbolic tangent function.

2. The hysteresis modeling method for voice coil motors according to claim 1, characterized in that, The neural network includes an input layer, a hidden layer, and an output layer.

3. The hysteresis modeling method for voice coil motors according to claim 2, characterized in that, The activation function of neurons in the hidden layer is the hyperbolic tangent function.

4. The hysteresis modeling method for voice coil motors according to claim 1 or 3, characterized in that, The activation functions of neurons in the input and output layers are linear mapping functions.

5. The hysteresis modeling method for voice coil motors according to claim 1, characterized in that, The neural network is trained using the LM method.

6. A voice coil motor hysteresis modeling system based on the voice coil motor hysteresis modeling method as described in any one of claims 1-5, characterized in that, The system includes: The module is constructed to determine the phase frequency response and amplitude frequency response relationship based on the transfer function relationship between the output displacement and the input current; The fitting module, based on the relationship between phase frequency response and amplitude frequency response, uses a neural network to fit the hysteresis trajectory of the voice coil motor, thereby obtaining the hysteresis model of the voice coil motor.

7. A computer storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1-5.

8. A hysteresis modeling device for voice coil motors, characterized in that, include: processor; and memory for storing the executable instructions of the processor; The processor is configured to perform the method as described in any one of claims 1-5 by executing the executable instructions.

Images