Anti-electromagnetic interference magnetic suspension controller equipment and filtering method

Abstract

The invention belongs to the technical field of motors, and discloses anti-electromagnetic interference magnetic suspension controller equipment and a filtering method. The method comprises the following steps: preparing a quantum confinement magnetic permeability unit, acquiring magnetic flux and temperature data in a magnetic core in real time through an embedded sensor array, and building a corresponding magnetic permeability regulation and control model; the output of the magnetic conductivity regulation and control model is obtained, and real-time inductance is obtained through analysis; on the basis of the real-time inductance, in combination with a preset target inductance, compensating the real-time inductance through a grid voltage regulation and inductance compensation cooperation strategy; according to the compensated real-time inductance, a PID controller is adopted for filtering; under the strong magnetic field and high temperature coupling environment, the nonlinear decrease amplitude of the magnetic conductivity can be effectively reduced, the drift amount of the inductance value, the cut-off frequency offset of the filter and the current ripple can be reduced, the rotor suspension precision can be improved, and the filtering stability and the control reliability of a magnetic suspension system under the extreme environment can be remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric machines, and more particularly, to a magnetic suspension controller device and a filtering method resistant to electromagnetic interference. BACKGROUND

[0002] A magnetic suspension system realizes non-contact suspension through magnetic force, and its core control unit needs to process weak feedback signals to maintain stable suspension of the rotor, so it is extremely sensitive to electromagnetic interference. The filter module of the controller is the key to suppressing EMI. The filter composed of specific inductance and capacitance can attenuate high-frequency interference signals and ensure the signal-to-noise ratio of the control loop.

[0003] However, the magnetic core material of the filter inductance has significant defects in the strong magnetic field and high-temperature coupling environment of the magnetic suspension system. On the one hand, the strong magnetic field generated during system operation can cause the magnetic domain of the magnetic core to arrange disorderly, leading to a nonlinear decrease in magnetic permeability. On the other hand, the high temperature (such as -40℃ to 125℃) during long-term operation of the device can exacerbate the thermal magnetic loss of the magnetic core, further deteriorating the stability of the magnetic permeability. Under the coupling action of the two, the inductance value will fluctuate significantly with the drift of the magnetic permeability, directly affecting the cutoff frequency of the filter.

[0004] Chinese patent application No. CN116317717A discloses an anti-interference power supply device, a motor control device, and a control method thereof. In the case where a filter module body is arranged on the output side of a power supply and a control module body is arranged between the output side of the filter module body and the power supply side of a load, it is determined whether the load needs to be powered on to start and work. If the load needs to be powered on to start and work, the control module body is controlled to be connected to connect the power supply path between the power supply and the load. If the load does not need to be powered on to start and work, the control module body is controlled to be disconnected to disconnect the power supply path between the power supply and the load. The application sets a modular filter module and a switching control module between the power supply and the motor, which facilitates users to flexibly select the filter module and the switching control module according to actual needs, and greatly improves the applicability of the filter module and the switching control module between the power supply and the motor.

[0005] Although the above method can meet most scenarios, research and actual application of the above method and prior art have found that the above method and prior art at least have the following defects:

[0006] The above method focuses on the suppression of external electromagnetic interference and cannot solve the nonlinear decrease in magnetic permeability of the magnetic core of the filter inductance under the coupling action of the strong magnetic field and high temperature, which leads to the drift of the inductance value, causes the cutoff frequency of the filter to shift, and the resonant point to drift, and the interference signal to invade the magnetic suspension controller through the control loop, resulting in an increase in current ripple and a decrease in rotor suspension precision.

[0007] In view of this, the application proposes an electromagnetic interference resistant maglev controller device and filtering method to solve the above problems. SUMMARY

[0008] In order to overcome the above-mentioned defects of the prior art, in order to achieve the above-mentioned purpose, the application provides the following technical scheme: an electromagnetic interference resistant maglev controller filtering method, comprising the following steps:

[0009] A quantum confinement magnetic guide unit is prepared, the internal magnetic flux and temperature data of the magnetic core are collected in real time through an embedded sensor array, and a corresponding permeability regulation model is constructed;

[0010] The output of the permeability regulation model is obtained, and real-time inductance is obtained by analysis;

[0011] Based on the real-time inductance, in combination with a preset target inductance, the real-time inductance is compensated through a gate voltage regulation and inductance compensation cooperative strategy;

[0012] According to the compensated real-time inductance, a PID controller is used for filtering.

[0013] Further, the method for preparing the quantum confinement magnetic guide unit comprising an embedded sensor array comprises:

[0014] An ultralattice composed of a preset material is selected as a magnetic core main body;

[0015] The pretreated SiO2 is used as a substrate of the ultralattice, the ultralattice is controlled to grow alternately on the SiO2 substrate through a preset growth process, and an ultralattice with a preset period is formed; during the growth of the ultralattice, a layer gap with a preset size is reserved at the interface of the ultralattice with a preset reserved period by shielding through a mask plate, a preset chip is embedded in the layer gap, and a conductive polymer is used for connection, thereby forming an embedded sensor array;

[0016] An Au layer is evaporated and deposited on the end face of the ultralattice through electron beam lithography, and a gold wire gate with a preset line width is obtained by etching, thereby forming a substrate-ultralattice-gold electrode structure and obtaining a quantum confinement magnetic guide unit.

[0017] Further, the method for constructing the permeability regulation model comprises:

[0018] The basic quantum tunneling probability is calculated;

[0019] The generated vertical electric field is calculated based on the gate voltage and the single-layer barrier thickness, the single-layer barrier thickness is corrected based on the vertical electric field, the single-layer barrier thickness, and the electronic charge, and the corrected barrier thickness is obtained;

[0020] The quantum tunneling probability is calculated based on a basic quantum tunneling probability, a single-layer barrier thickness, a modified barrier thickness, a tunneling probability decay factor correction value, and a temperature correction term; and the tunneling probability decay factor correction value is obtained based on a magnetic core internal magnetic flux density and a tunneling probability decay factor correction.

[0021] The magnetic permeability regulation model is constructed by obtaining a zero magnetic field reference permeability, a quantum tunneling probability, a coupling coefficient, a modified barrier thickness, a gate voltage sensitivity coefficient, a gate voltage, a magnetic field decay coefficient, and a magnetic core internal magnetic flux density.

[0022] Further, the method for obtaining the basic quantum tunneling probability comprises:

[0023] The barrier electron wave vector is calculated based on an electron energy, a barrier thickness, a Dirac constant, and an electron effective mass; the graphene layer electron wave vector is calculated based on the electron energy, the Dirac constant, and the electron effective mass; and the basic quantum tunneling probability is calculated based on the barrier electron wave vector, the graphene layer electron wave vector, and the single-layer barrier thickness.

[0024] Further, the method for obtaining the gate voltage sensitivity coefficient comprises:

[0025] The gate voltage sensitivity coefficient is obtained by fixing the magnetic field and the temperature, changing the gate voltage according to a preset step size through the GaN high electron mobility transistor, measuring and obtaining the corresponding magnetic permeability, calculating the difference between the corresponding magnetic permeability and the zero magnetic field reference magnetic permeability, combining the modified barrier thickness and the gate voltage, fitting the coupling relationship of the difference with respect to the gate voltage sensitivity coefficient, and fitting the gate voltage sensitivity coefficient through the least square method.

[0026] Further, the method for compensating the real-time inductance comprises:

[0027] A K-group of air core coils are set, and the GaN switch is switched;

[0028] When the change difference between the real-time inductance and the reference value is not greater than a preset drift threshold, the magnetic flux error change rate in a preset period is calculated, the state vector is defined according to the real-time magnetic flux density, the real-time temperature, the magnetic flux error, and the error change rate; the PID parameter used for adjusting the gate voltage is taken as the action vector; the reward function composed of the quantum tunneling probability term, the multi-physical field coupling penalty term, and the gate voltage energy consumption penalty is constructed; the optimal action is obtained through iterative training and autonomous learning; and the gate voltage is adjusted according to the PID parameter in the optimal action.

[0029] When the change difference between the real-time inductance and the reference value is greater than the preset drift threshold, the corresponding coil is activated according to the change difference, and the real-time inductance is adjusted until the change difference is lower than the preset safety threshold.

[0030] Further, the method for obtaining the reward function comprises:

[0031] The quantum tunneling probability term is calculated according to the quantum tunneling probability, the target magnetic flux density and the magnetic flux error;

[0032] The multi-physical field coupling penalty term is calculated according to the current temperature, the current magnetic flux density, the temperature reference value and the magnetic flux density reference value.

[0033] The gate voltage energy consumption penalty is calculated according to the gate voltage and the maximum gate voltage.

[0034] Further, the method for obtaining the optimal action comprises:

[0035] The agent randomly extracts samples from the experience pool, and outputs actions through an actor network built by a W-layer fully connected neural network; the experience replay pool is composed of Y groups of sample data; each group of samples includes a state vector at the current time, an action, a reward and a state vector at the next time;

[0036] The PID parameters corresponding to the action are substituted into the gate voltage control formula and applied to the superlattice;

[0037] The next state and the reward are collected, and the Q value of the critic network is updated; the Q value is calculated according to a reward function and a maximum Q value of the corresponding action selected in the next state, and a discount factor, wherein the discount factor is obtained through experimental tuning;

[0038] The actor network parameters are optimized through gradient descent with the goal of maximizing the Q value.

[0039] The target network is updated every preset step length until the reward converges to a preset reward interval, and the action corresponding to the maximum Q value when the reward converges to the preset reward interval is obtained as the optimal action.

[0040] Further, the method for obtaining the inductance drift quantization model comprises:

[0041] The initial inductance, the real-time permeability output by the permeability regulation model, the real-time magnetic flux density, the real-time temperature, the corrected barrier thickness, the total length of the magnetic core and the quantum tunneling probability are obtained, an inductance drift quantization model is established, and the real-time inductance is calculated.

[0042] Further, the method for filtering using a PID controller comprises:

[0043] In combination with a fixed capacitance value, the filter cutoff frequency is calculated according to the compensated real-time inductance;

[0044] The PWM carrier frequency is filtered by tracking the filter cutoff frequency using a PID controller.

[0045] The electromagnetic interference-resistant magnetic suspension controller device is used to implement the electromagnetic interference-resistant magnetic suspension controller filtering method, and comprises:

[0046] The acquisition processing module: prepare a quantum limited magnetic guide unit, acquire the magnetic flux and temperature data inside the magnetic core in real time through the embedded sensor array, and construct a corresponding permeability regulation model;

[0047] The inductance analysis module: obtains the output of the permeability regulation model and analyzes the real-time inductance;

[0048] The inductance compensation module: based on the real-time inductance, combines the preset target inductance, and compensates the real-time inductance through a cooperative strategy;

[0049] The compensation filtering module: according to the compensated real-time inductance, a PID controller is used for filtering.

[0050] The technical effects and advantages of the anti-electromagnetic interference magnetic suspension controller device and filtering method of the application are as follows:

[0051] The application selects a preset material superlattice as the main body of the magnetic core, enhances the anti-interference ability of the magnetic core to strong magnetic field and high temperature from the root by means of quantum confinement effect, and combines the preset growth process and the real-time magnetic flux and temperature acquisition of the embedded sensor array to provide double inputs of microscopic quantum state and macroscopic environmental parameters for permeability regulation; through the inductance drift quantization model of the fusion of quantum tunneling probability, corrected barrier thickness and real-time permeability, the nonlinear decline law of permeability is accurately captured, and full-dimensional tracing from microscopic quantum state fluctuation to macroscopic inductance value drift is realized; K sets of air core coils and GaN switches are also used for rapid switching to cope with large inductance drift, and a reinforcement learning PID strategy based on quantum state stability term is combined to realize fine adjustment of the grid voltage; finally, the PID controller drives the PWM carrier frequency to track the cutoff frequency corresponding to the compensated inductance in real time, effectively prevents the cutoff frequency and resonance point of the filter from deviating, and blocks the path of interference signals invading the magnetic bearing controller; the application can effectively reduce the nonlinear decline amplitude of the permeability, reduce the inductance value drift, filter cutoff frequency deviation and current ripple, improve the rotor suspension precision, and significantly improve the filtering stability and control reliability of the magnetic suspension system in extreme environments. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 The filtering method flowchart of the anti-electromagnetic interference magnetic suspension controller of the application;

[0053] Figure 2 The data flowchart of the application;

[0054] Figure 3 The method flowchart of building the permeability regulation model of the application;

[0055] Figure 4 The structure diagram of the anti-electromagnetic interference magnetic suspension controller device of the application. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0057] Embodiment 1

[0058] Please refer to Figure 1 , Figure 2 The magnetic suspension controller filtering method against electromagnetic interference provided in the embodiment includes the following steps:

[0059] A quantum confinement magnetic guide unit containing an embedded sensor array is prepared to collect the magnetic flux and temperature data inside the magnetic core in real time, and a corresponding magnetic permeability regulation model is built.

[0060] The method for preparing the quantum confinement magnetic guide unit containing the embedded sensor array includes:

[0061] An ultralattice composed of a preset material is selected as the main body of the magnetic core, such as an ultralattice composed of graphene / hexagonal boron nitride.

[0062] The pretreated SiO2 is used as the substrate of the ultralattice, and a preset growth process, such as low-pressure chemical vapor deposition, is used to grow graphene layers and hexagonal boron nitride layers alternately using a graphene precursor, such as CH4, and a hexagonal boron nitride precursor, such as NH2, as the gas source at 800℃ and a pressure of 50Pa. The ultralattice is controlled to be grown on the SiO2 substrate to form an ultralattice with a preset period, such as 5 periods. During the growth of the ultralattice, a mask plate is used to shield a layer gap with a preset size, such as a 1μm×1μm layer gap, at the interface of a preset reserved period, such as the 3rd period, of the ultralattice. A preset chip, specifically a miniaturized magnetic flux sensor chip and a temperature sensor chip, is embedded in the layer gap. Only the magnetic flux density and temperature inside the magnetic core are measured, without the need to integrate other redundant functions, thereby avoiding occupying the layer gap of the ultralattice. The measurement accuracy of the chip is set in advance in combination with the calculation requirements of the magnetic permeability regulation model, such as a magnetic flux density accuracy of μT level and a temperature accuracy of ±0.5℃, to ensure that the collected data can reflect the subtle changes in the microstate of the magnetic core. A conductive polymer is used for connection to form an embedded sensor array.

[0063] A 50nm thick Au layer is evaporated and deposited on the end face of the ultralattice by electron beam lithography, and a gold wire grid with a preset line width, such as a 200nm gold wire grid, is obtained by electron beam lithography etching to form a substrate-ultralattice-gold electrode structure and obtain a quantum confinement magnetic guide unit.

[0064] The above method selects graphene / hexagonal boron nitride superlattice as the main body of the magnetic core, alternately grows into a multi-period layered structure, enhances the anti-interference ability of the magnetic core to strong magnetic field and high temperature by means of this quantum confinement characteristic, and suppresses the nonlinear downward trend of the magnetic permeability μ from the material level; at the same time, the interlayer gap is reserved during the growth of the superlattice, the preset chip is embedded by focused ion beam deposition, and the embedded sensor array is formed by connecting with conductive polymer, which can accurately collect the magnetic flux and temperature data inside the magnetic core in real time, provide reliable input for the subsequent magnetic permeability regulation model, and realize real-time monitoring of the inductance value L drift; in addition, a gold grid is prepared on the end face of the superlattice by electron beam lithography, and the substrate-superlattice-gold electrode structure constructed can efficiently respond to gate voltage regulation, dynamically adjust the magnetic permeability combined with real-time monitoring data, thereby stabilizing the inductance value, avoiding the shift of the filter cutoff frequency and the resonance point, blocking the interference signal from invading the magnetic bearing controller through the control loop, ultimately reducing the current ripple and improving the rotor suspension precision, and effectively solving the filtering and control stability problem of the magnetic suspension system in the strong magnetic field and high temperature coupling environment.

[0065] Reference Figure 3 The method for building the magnetic permeability regulation model comprises:

[0066] Calculate the basic quantum tunneling probability;

[0067] Calculate the vertical electric field generated based on the gate voltage and the single-layer barrier thickness, correct the single-layer barrier thickness based on the vertical electric field, the single-layer barrier thickness and the electronic charge to obtain the corrected barrier thickness; wherein the single-layer barrier thickness specifically refers to the thickness of a single material layer constituting the barrier, and if the superlattice composed of graphene / hexagonal boron nitride is used, the thickness of the single-layer hexagonal boron nitride can be obtained; if the vertical electric field The corrected barrier thickness , wherein, is the gate voltage; is the single-layer barrier thickness; is the electronic charge; is the dielectric constant of the barrier material;

[0068] Calculate the quantum tunneling probability based on the basic quantum tunneling probability, the single-layer barrier thickness, the corrected barrier thickness, the tunneling probability decay factor correction value and the temperature correction term; the tunneling probability decay factor correction value is obtained based on the magnetic flux density inside the magnetic core, and the tunneling probability decay factor is obtained by experiment fitting; if the temperature correction term The quantum tunneling probability , wherein, is the barrier height; is the Boltzmann constant; is the absolute temperature; is the reference temperature; is the basic quantum tunneling probability; is a tunneling probability decay factor correction value, is a magnetic flux correction coefficient, which can be obtained by experimental fitting, is the internal magnetic flux density of the magnetic core; is a tunneling probability decay factor;

[0069] Obtain the zero magnetic field reference permeability, quantum tunneling probability, coupling coefficient, corrected barrier thickness, gate voltage sensitivity coefficient, gate voltage, magnetic field decay coefficient, and internal magnetic flux density of the magnetic core, and build a permeability control model; wherein the zero magnetic field reference permeability is the initial permeability of the quantum confinement magnetic unit at zero external magnetic field, a predetermined reference temperature, such as 25℃, which can be obtained by averaging multiple tests.

[0070] The method for obtaining the basic quantum tunneling probability includes:

[0071] Based on the electron energy, barrier thickness, Dirac constant, and electron effective mass, the electron wave vector in the barrier is calculated; based on the electron energy, Dirac constant, and electron effective mass, the electron wave vector in the graphene layer is calculated; and based on the electron wave vector in the barrier, the electron wave vector in the graphene layer, and the single-layer barrier thickness, the basic quantum tunneling probability is calculated. For example, the electron wave vector in the barrier is the electron wave vector in the graphene layer is the basic quantum tunneling probability is is the electron effective mass in the barrier region; is the electron energy; is the Dirac constant; is the characteristic velocity of graphene electrons;

[0072] The method for obtaining the gate voltage sensitivity coefficient includes:

[0073] Fix the magnetic field and temperature, change the gate voltage according to the preset step size through the gallium nitride high electron mobility transistor, measure and obtain the corresponding permeability, calculate the difference between the corresponding permeability and the zero magnetic field reference permeability, combine the corrected barrier thickness and the gate voltage, fit the coupling relationship of the difference with respect to the gate voltage sensitivity coefficient, and obtain the gate voltage sensitivity coefficient by least squares fitting.

[0074] The method for obtaining the gate voltage sensitivity coefficient is similar to the method for obtaining the gate voltage sensitivity coefficient, that is, by fixing the gate voltage and temperature, changing the magnetic field to collect the permeability, and then combining the zero magnetic field reference permeability to fit and calculate the magnetic field decay coefficient.

[0075] ​​The method of building the permeability regulation model is to systematically integrate the quantum tunneling effect and the quantitative relationship of multi-physical field coupling, and to build a dynamic model that can accurately reflect the nonlinear change of permeability under strong magnetic field and high temperature, thereby solving the series of problems caused by inductance drift from the root. The above method breaks through the limitations of traditional models that ignore quantum effects and multi-field coupling, and can accurately capture the nonlinear decline law of permeability μ under the coupling of strong magnetic field and high temperature: both the quantum tunneling probability and the modified barrier thickness are used to quantify the influence of microscopic quantum states on permeability, and the grid voltage sensitivity coefficient and the magnetic field decay coefficient are used to reflect the effects of macroscopic regulation and environmental factors, thereby providing reliable permeability input for accurate calculation of inductance value. Based on the permeability regulation model, the system can predict the trend of permeability change in advance, provide accurate basis for subsequent compensation strategy, and then stabilize the inductance value, avoid the shift of filter cutoff frequency and resonance point, block the path of interference signal invading the magnetic bearing controller, ultimately reduce current ripple, improve rotor suspension accuracy, and ensure the stable operation of the magnetic suspension system in extreme environments.

[0076] Obtaining the output of the permeability regulation model, and analyzing to obtain the real-time inductance;

[0077] The method for obtaining the inductance drift quantitative model comprises:

[0078] Obtaining the initial inductance, the real-time permeability output by the permeability regulation model, the real-time magnetic flux density, the real-time temperature, the modified barrier thickness, the total length of the magnetic core, and the quantum tunneling probability, establishing the inductance drift quantitative model, and calculating to obtain the real-time inductance. For example, the real-time inductance , is the initial inductance; is the real-time permeability, which is the real-time permeability output by the permeability regulation model; is the initial permeability; is the real-time magnetic flux density; is the reference magnetic flux density; is the inductance temperature correction term, , is the temperature coefficient, which can be obtained by experimental fitting, is the real-time temperature, is the reference temperature; is the magnetic flux sensitivity coefficient, which can be obtained by experimental fitting; is the barrier thickness correction value; is the reference barrier thickness.

[0079] The method for obtaining the inductance drift quantification model integrates the initial inductance, the real-time magnetic permeability output by the magnetic permeability regulation model, the real-time magnetic flux density, the real-time temperature, the corrected barrier thickness, the total length of the magnetic core and the quantum tunneling probability and other key parameters, and constructs a quantification model that can accurately reflect the multi-physical field coupling and quantum effect, thereby realizing dynamic calculation of the real-time inductance. The method breaks through the limitations of traditional inductance calculation, which ignores quantum characteristics and multi-physical field nonlinear coupling, and can accurately capture the inductance value drift rule caused by the nonlinear decline of magnetic permeability under the action of a strong magnetic field and high temperature. The method reflects the influence of the environment on the magnetic core through macroscopic parameters such as real-time magnetic permeability, magnetic flux density and temperature, and quantifies the modulation effect of microscopic quantum state changes on inductance by means of quantum parameters such as corrected barrier thickness and quantum tunneling probability, thereby realizing full-dimensional inductance drift tracing from the macroscopic to the microscopic. The real-time inductance calculated based on the inductance drift quantification model can provide accurate input for subsequent compensation strategies, so that the system can predict the inductance drift trend in advance and adjust accordingly, thereby suppressing the shift of the filter cutoff frequency and the drift of the resonance point, reducing the risk of interference signals invading the magnetic bearing controller through the control loop, ultimately reducing current ripple and improving rotor suspension accuracy, and effectively solving the stability problem of the magnetic suspension system caused by inductance drift in a strong magnetic field and high temperature coupling environment.

[0080] Based on the real-time inductance, in combination with a preset target inductance, the real-time inductance is compensated through a gate voltage regulation and inductance compensation collaborative strategy.

[0081] The method for compensating the real-time inductance comprises the following steps:

[0082] K groups of air core coils are set, and GaN switches are used for switching;

[0083] When the change difference between the real-time inductance and the reference value is not greater than a preset drift threshold, the preset drift threshold is set based on the inductance fluctuation range allowed by the normal operation of the system, the magnetic flux error change rate in a preset period is calculated, and the magnetic flux error change rate is , is the magnetic flux error change amount; is the preset period; is the real-time magnetic flux density at time t; a state vector is defined according to the real-time magnetic flux density, the real-time temperature, the magnetic flux error and the error change rate; PID parameters for adjusting the gate voltage are used as an action vector; a reward function composed of a quantum tunneling probability term, a multi-physical field coupling penalty term and a gate voltage energy consumption penalty is constructed; the optimal action is obtained through iterative training and autonomous learning; and the gate voltage is adjusted according to the PID parameters in the optimal action.

[0084] When the change difference between the real-time inductance and the reference value is greater than the preset drift threshold, the corresponding coil is activated according to the change difference, and the real-time inductance is adjusted until the change difference is lower than a preset safety threshold, and the preset safety threshold is set based on the minimum requirement for safe operation of the system.

[0085] The method for compensating real-time inductance sets K groups of hollow coils which can be quickly switched by GaN switches, combines two targeted compensation strategies, forms a full-range inductance stabilization mechanism covering different degrees of drift, and effectively solves the problem of inductance drift under the coupling of strong magnetic field and high temperature. When the change difference between the real-time inductance and the reference value is not greater than the preset drift threshold, by defining a state vector containing the real-time magnetic flux density, temperature, magnetic flux error and error change rate, taking the PID parameters of the gate voltage as the action vector, constructing a reward function integrating quantum tunneling probability term, multi-physical field coupling penalty term and gate voltage energy consumption penalty term, and through iterative training, the intelligent agent autonomously learns the optimal PID parameters to realize precise regulation of the gate voltage; This fine adjustment based on quantum characteristics and multi-field perception can adapt to the slight nonlinear fluctuations of magnetic permeability and suppress small inductance drift from the quantum level; When the change difference is greater than the preset threshold, the corresponding hollow coil is quickly activated by the GaN switch, and the fixed inductance of the coil is used to realize immediate correction of large inductance drift, ensuring that the inductance quickly returns to the safe range. The two strategies of gate voltage regulation and inductance compensation work together, which can not only cope with the dynamic and subtle drift of quantum confined magnetic cores through gate voltage regulation, but also compensate for large fluctuations under extreme conditions through hollow coils, thereby stabilizing the inductance value, avoiding the shift of filter cutoff frequency and resonance point, blocking the path of interference signals invading the magnetic bearing controller, ultimately reducing current ripple, improving rotor suspension accuracy, and ensuring the stable operation of the magnetic suspension system in the coupling environment of strong magnetic field and high temperature.

[0086] The method for obtaining the reward function comprises:

[0087] The quantum tunneling probability term is calculated according to the quantum tunneling probability, the target magnetic flux density and the magnetic flux error; for example, the quantum tunneling probability term is , is the magnetic flux error weight coefficient, which can be obtained by experimental fitting; is the target magnetic flux density;

[0088] The multi-physical field coupling penalty term is calculated according to the current temperature, the current magnetic flux density, the temperature reference value and the magnetic flux density reference value; for example, the multi-physical field coupling penalty term is , and is the coupling penalty weight, which can be obtained by experimental fitting;

[0089] The gate voltage energy consumption penalty is calculated according to the gate voltage and the maximum gate voltage; for example, the gate voltage energy consumption penalty is , is the real-time gate voltage; is the maximum gate voltage; is the energy consumption penalty coefficient;

[0090] The method for obtaining the reward function provides a precise optimization target for reinforcement learning by constructing a comprehensive evaluation mechanism that integrates quantum tunneling probability items, multi-physical field coupling penalty items and gate voltage energy consumption penalty items, thereby specifically solving the inductance drift problem under the coupling of strong magnetic field and high temperature. Among them, the quantum tunneling probability item is directly related to the stability of the microscopic quantum state of the quantum limited magnetic core; by quantifying the deviation of the tunneling probability and the magnetic flux error, it ensures that the control strategy can accurately respond to the quantum level permeability fluctuation, and from the root, it inhibits the small inductance drift caused by quantum effects; the multi-physical field coupling penalty item combines the deviation of the current temperature, magnetic flux density and the reference value, and is specifically aimed at the cross-coupling effect of strong magnetic field and high temperature, avoids the adjustment error caused by ignoring this coupling, and ensures that the control strategy adapts to the nonlinear characteristics in complex environments; the gate voltage energy consumption penalty item calculates according to the ratio of the gate voltage and the maximum value, limits the gate voltage energy consumption while ensuring the adjustment effect, prevents excessive regulation from causing additional electromagnetic interference or energy loss, and balances stability and economy. The reward function formed by the three can guide the reinforcement learning to autonomously learn the optimal control strategy that not only adapts to quantum state fluctuations, but also is compatible with multi-field coupling and energy consumption, and promotes the inductance value to be stable within the target range, thereby avoiding the shift of the filter cutoff frequency and the drift of the resonance point, blocking the path of interference signals invading the magnetic bearing controller, ultimately reducing current ripple and improving rotor suspension precision, and achieving stable operation of the magnetic suspension system in extreme environments.

[0091] The method for obtaining the optimal action comprises:

[0092] The agent randomly extracts samples from the experience pool and outputs actions through the actor network built by the W-layer fully connected neural network; the experience replay pool is composed of Y groups of sample data; each group of samples includes a state vector at the current time, an action, a reward and a state vector at the next time;

[0093] Substitute the PID parameters corresponding to the action into the gate voltage control formula and apply it to the superlattice;

[0094] Collect the next state and reward, and update the Q value of the critic network; the Q value is obtained according to the reward function and the maximum Q value of the corresponding action selected at the next state, as well as the discount factor, wherein the discount factor is obtained by experimental tuning;

[0095] Maximize the Q value as the target, and optimize the actor network parameters through gradient descent;

[0096] Update the target network every preset step size until the reward converges to the preset reward interval, and obtain the action corresponding to the maximum Q value when the reward converges to the preset reward interval as the optimal action.

[0097] The method of obtaining the optimal action realizes the precise optimization of the gate voltage adjustment PID parameter through the dynamic interaction of the agent and the quantum confined superlattice system in the reinforcement learning framework, thereby solving the inductance drift problem under the coupling of strong magnetic field and high temperature. The Y group of samples stored in the experience replay pool covers the state vector, action and reward of the current and next time, ensuring that the agent can learn the multi-working condition characteristics under strong magnetic field and high temperature; the PID parameter action output by the actor network built by the W layer full connection neural network is applied to the superlattice after the gate voltage control formula, and the Q value updated by the critic network based on the reward function, and the network parameters are continuously optimized through gradient descent, and finally the optimal action obtained when the reward converges to the preset interval can accurately adapt to the nonlinear decline characteristics of the magnetic permeability; both through learning the influence of quantum tunneling effect and multi-field coupling, the PID parameter dynamically responds to the change of microscopic quantum state to stabilize the magnetic permeability, and through the discount factor to balance the current adjustment effect and long-term inductance stability, to ensure that the action can suppress the immediate drift of inductance and avoid excessive regulation to cause new fluctuations. The optimal action obtained through autonomous learning is applied to the gate voltage adjustment, which can stabilize the inductance value, prevent the cutoff frequency and resonance point of the filter from shifting, block the path of interference signals invading the magnetic bearing controller, and thus reduce the current ripple and improve the rotor suspension accuracy, ensuring the stable operation of the magnetic suspension system in extreme environments.

[0098] According to the compensated real-time inductance, a PID controller is used for filtering;

[0099] The method of filtering by using a PID controller includes:

[0100] Combined with the fixed capacitance value, the filter cutoff frequency is calculated according to the compensated real-time inductance; if the filter cutoff frequency , is the compensated real-time inductance; is the fixed capacitance value;

[0101] The PWM carrier frequency is tracked by using the PID controller to filter the filter cutoff frequency.

[0102] According to the compensated real-time inductance, a PID controller is used for filtering. By combining the fixed capacitance value and the compensated real-time inductance, the precise filter cutoff frequency is obtained, and the PID controller is used to drive the PWM carrier frequency to dynamically track the cutoff frequency, forming a closed-loop frequency stabilization mechanism, which can solve the problem of filter failure under the coupling of strong magnetic field and high temperature. The compensated real-time inductance has corrected the inductance drift caused by the nonlinear decline of magnetic permeability. The cutoff frequency calculated on this basis can truly reflect the current working frequency band of the filter. The PID controller continuously adjusts the PWM carrier frequency to keep it consistent with the cutoff frequency, which can offset the cutoff frequency deviation caused by inductance drift in real time, and suppress the drift of the resonance point caused by inductance change. It directly blocks the path of interference signals invading the magnetic bearing controller due to mismatched filtering frequency band, and avoids the decline of filtering efficiency caused by frequency deviation in traditional filtering. Finally, by stabilizing the cutoff frequency and resonance point of the filter, the current ripple in the control loop is effectively reduced, the rotor suspension precision is improved, and the control performance of the magnetic suspension system in the extreme environment of the coupling of strong magnetic field and high temperature is still maintained.

[0103] Embodiment 2

[0104] The embodiment provides a method for obtaining a reward function applied to the embodiment 1, comprising the following steps:

[0105] The standard deviation of the quantum tunneling probability in the preset time window and the corresponding mean value of the quantum tunneling probability are counted, the relative volatility of the quantum tunneling probability is calculated, the preset target volatility is combined, the quantum state stability term is calculated, and the quantum tunneling probability term in the reward function of the embodiment 1 is replaced, that is, the reward function is composed of the quantum state stability term, the multi-physical field coupling penalty term and the gate voltage energy consumption penalty.

[0106] The method for obtaining the reward function in the embodiment calculates the relative volatility by the standard deviation and the mean value of the quantum tunneling probability in a preset time window, obtains a quantum state stability term in combination with a preset target volatility, and replaces the quantum tunneling probability term in the original reward function, so that the reward function is composed of the quantum state stability term, a multi-physical field coupling penalty term and a gate voltage energy consumption penalty. The reward function is more directly focused on the dynamic stability of the quantum state; the quantum state stability accurately captures the microscopic quantum state disturbance of the quantum confined magnetic core under the coupling of the strong magnetic field and the high temperature by quantifying the fluctuation degree of the quantum tunneling probability, and the disturbance is the deep root of the nonlinear decline of the magnetic permeability; in combination with the inhibition of the temperature and magnetic field cross interference by the multi-physical field coupling penalty term and the limitation of the gate voltage energy consumption penalty on excessive regulation, the new reward function can guide the reinforcement learning to autonomously learn a control strategy that is more suitable for the stability requirements of the quantum state, suppress the magnetic permeability drift caused by the quantum state fluctuation from the microscopic level, and further reduce the inductance value fluctuation. Finally, the quantum state is indirectly stabilized to stabilize the magnetic permeability and the inductance, avoid the shift of the filter cutoff frequency and the drift of the resonance point, block the path of the interference signal invading the magnetic suspension controller, effectively reduce the current ripple and improve the rotor suspension precision, and solve the series of stability problems of the magnetic suspension system caused by the instability of the quantum state in the coupling environment of the strong magnetic field and the high temperature.

[0107] Embodiment 3:

[0108] Referring to Figure 4 The magnetic suspension controller device provided by the embodiment is shown in the figure, and the magnetic suspension controller device comprises:

[0109] The acquisition and processing module: a quantum confined magnetic permeability unit is prepared, the internal magnetic flux and temperature data of the magnetic core are acquired in real time through the embedded sensor array, and a corresponding magnetic permeability regulation model is built;

[0110] The inductance analysis module: the output of the magnetic permeability regulation model is obtained, and the real-time inductance is obtained by analysis;

[0111] The inductance compensation module: based on the real-time inductance, in combination with the preset target inductance, the real-time inductance is compensated through a cooperative strategy;

[0112] The compensation filtering module: according to the compensated real-time inductance, a PID controller is used for filtering.

[0113] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited to this, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0114] Finally: the above only for the preferred embodiments of the present application, and not for limiting the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application, should be included in the scope of protection of the present application.

Claims

1. A method of filtering electromagnetic interference from a magnetic levitation controller, characterized by, The application relates to a quantum confinement magnetic guide unit and a preparation method thereof. The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

2. The electromagnetic interference resistant magnetic levitation controller filtering method of claim 1, wherein, The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

3. The electromagnetic interference resistant magnetic levitation controller filtering method of claim 1, wherein, The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

4. The electromagnetic interference resistant magnetic levitation controller filtering method of claim 3, wherein, The method comprises the following steps: The method comprises the following steps:

5. The electromagnetic interference immune magnetic levitation controller filtering method of claim 1, wherein, The method comprises the following steps: The method comprises the following steps:

6. The electromagnetic interference resistant magnetic levitation controller filtering method of claim 5, wherein, The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method When the difference between the real-time inductance and the reference value is not greater than the preset drift threshold, a magnetic flux error change rate in a preset period is calculated, a state vector is defined according to the real-time magnetic flux density, the real-time temperature, the magnetic flux error and the error change rate, PID parameters for adjusting the gate voltage are taken as an action vector, a reward function composed of a quantum tunneling probability term, a multi-physical field coupling penalty term and a gate voltage energy consumption penalty is constructed, and the optimal action is obtained through iterative training and autonomous learning; and the gate voltage is adjusted according to the PID parameters in the optimal action. When the difference between the real-time inductance and the reference value is greater than the preset drift threshold, the corresponding coil is activated according to the difference, and the real-time inductance is adjusted until the difference is less than the preset safety threshold.

7. The electromagnetic interference resistant magnetic levitation controller filtering method of claim 6, wherein, The method for obtaining the reward function comprises: The quantum tunneling probability term is calculated according to the quantum tunneling probability, the target magnetic flux density and the magnetic flux error. The multi-physical field coupling penalty term is calculated according to the current temperature, the current magnetic flux density, the temperature reference value and the magnetic flux density reference value. The gate voltage energy consumption penalty is calculated according to the gate voltage and the maximum gate voltage.

8. The electromagnetic interference immune magnetic levitation controller filtering method of claim 1, wherein, The method for obtaining the optimal action comprises: The agent randomly extracts samples from the experience pool, and outputs actions through an actor network built by a W-layer fully connected neural network; the experience replay pool is composed of Y groups of sample data; each group of samples includes a state vector, an action, a reward and a state vector at the next moment; The PID parameters corresponding to the action are substituted into the gate voltage control formula and applied to the superlattice; The next moment state and the reward are collected, and the Q value of the critic network is updated; the Q value is calculated according to the reward function, the maximum Q value of the corresponding action in the next state and a discount factor, wherein the discount factor is obtained through experimental tuning; The actor network parameters are optimized through gradient descent with the goal of maximizing the Q value; The target network is updated every preset step length until the reward converges to a preset reward interval, and the action corresponding to the maximum Q value when the reward converges to the preset reward interval is taken as the optimal action.

9. The electromagnetic interference immune magnetic levitation controller filtering method of claim 1, wherein, The method for obtaining the inductance drift quantization model comprises: The initial inductance, the real-time magnetic permeability output by the magnetic permeability regulation and control model, the real-time magnetic flux density, the real-time temperature, the corrected barrier thickness, the total length of the magnetic core and the quantum tunneling probability are obtained, the inductance drift quantization model is established, and the real-time inductance is calculated.

10. The electromagnetic interference immune magnetic levitation controller filtering method of claim 1, wherein, The method for filtering by using the PID controller comprises: The filter cutoff frequency is calculated according to the compensated real-time inductance in combination with a fixed capacitance value; The PWM carrier frequency is filtered by tracking the filter cutoff frequency by using the PID controller.

11. A magnetic levitation controller device for electromagnetic interference immunity, for implementing the magnetic levitation controller filtering method for electromagnetic interference immunity according to any one of claims 1 to 10, characterized in that, It comprises: The acquisition and processing module: preparing a quantum confinement magnetic permeability unit, collecting the internal magnetic flux and temperature data of the magnetic core in real time through an embedded sensor array, and constructing a corresponding magnetic permeability regulation and control model; The inductance analysis module: obtaining the output of the magnetic permeability regulation and control model, and analyzing the real-time inductance; The inductance compensation module: compensating the real-time inductance by using a cooperative strategy based on the real-time inductance and in combination with the preset target inductance; The compensation and filtering module: filtering by using the PID controller according to the compensated real-time inductance.

Images