阅读说明：本技术 基于自适应神经网络的磁悬浮垂直轴风电机组悬浮控制方法 (Magnetic suspension vertical axis wind turbine generator suspension control method based on adaptive neural network ) 是由 蔡彬 谌义喜 褚晓广 崔国栋 邱雅兰 于 2020-07-22 设计创作，主要内容包括：本发明涉及一种基于自适应神经网络的磁悬浮垂直轴风电机组悬浮控制方法,属电气工程技术领域。该方法采用滑模自适应神经网络控制策略,使磁悬浮垂直轴风电机组的悬浮系统实现稳定悬浮：在起浮阶段,采用滑模控制+PID控制策略,使旋转体上升至悬浮平衡点；然后改用自适应神经网络滑模控制+PID控制策略,利用RBF神经网络估算未知干扰项,输出至自适应神经网络滑模控制器,然后将此滑模控制器的输出求开方,得到悬浮气隙跟踪控制器的输出,即悬浮电流的参考值,减去其实际值,经内环悬浮电流跟踪控制器,实时调整悬浮电流,实现稳定悬浮。本发明自适应能力强、动态响应快、抗干扰能力强,稳定性好,可确保整个悬浮过程系统性能实时最优。(The invention relates to a magnetic suspension vertical axis wind turbine generator suspension control method based on a self-adaptive neural network, and belongs to the technical field of electrical engineering. The method adopts a sliding mode self-adaptive neural network control strategy to realize stable suspension of a suspension system of the magnetic suspension vertical axis wind turbine generator: in the floating stage, a sliding mode control and PID control strategy is adopted to enable the rotator to rise to a suspension balance point; and then, an adaptive neural network sliding mode control + PID control strategy is adopted, an unknown interference item is estimated by using an RBF neural network and is output to an adaptive neural network sliding mode controller, then the output of the sliding mode controller is solved to obtain the output of a suspension air gap tracking controller, namely the reference value of the suspension current, the actual value of the suspension current is subtracted, and the suspension current is adjusted in real time through an inner ring suspension current tracking controller to realize stable suspension. The invention has strong self-adaptive capacity, fast dynamic response, strong anti-interference capacity and good stability, and can ensure the real-time optimal system performance in the whole suspension process.)

1. The suspension control method of the magnetic suspension vertical axis wind turbine generator based on the adaptive neural network comprises the steps that the magnetic suspension vertical axis wind turbine generator comprises a magnetic suspension vertical axis wind turbine, a suspension control system, an air gap sensor, a wind wheel, a shell and a rotating shaft; the magnetic suspension vertical shaft wind driven generator comprises a permanent magnet direct drive type wind driven generator and a magnetic suspension disc type motor; the permanent magnet direct-drive wind driven generator comprises a stator and a rotor; the magnetic suspension disc type motor comprises a disc stator and a disc rotor; the disc stator consists of a disc stator iron core and a suspension winding, and the suspension winding is a direct-current excitation winding; the suspension control system consists of a suspension converter and a suspension controller thereof, the suspension converter is connected with the suspension winding, and the suspension controller comprises an outer ring suspension air gap tracking controller and an inner ring suspension current tracking controller; the rotor of the permanent-magnet direct-drive wind driven generator, the disc rotor of the magnetic suspension disc type motor, the wind wheel and the rotating shaft are collectively called as a rotating body; the method is characterized by comprising the following steps:

step 1, when the wind speed reaches the cut-in wind speed, an outer ring suspension air gap tracking controller of the suspension controller adopts a sliding mode control strategy, an inner ring suspension current tracking controller of the suspension controller adopts PID control to control the stator current of the magnetic suspension disc type motor, so that the rotating body is suspended upwards to and kept at a suspension balance point to realize stable suspension, and the specific method is as follows:

11) designing a slip form surface s as follows:

where e is the floating air gap tracking error: e ═ e^*-，^*The reference value of the suspension air gap at the suspension balance point is taken as the measurement value of the suspension air gap; c. C₀、c₁Is a positive real number;

by deriving equation (1) over time t, there is:

according to the mechanical equation of the rotating body in the vertical direction:

in the formula (I), the compound is shown in the specification,is the second derivative with respect to time t; m is the mass of the rotating body, and g is the acceleration of gravity; f. of_dIs an external random interference force; k is mu₀N²A/4, wherein₀The magnetic pole surface effective area of the disc stator is A, and the number of turns of the suspension winding is N; i.e. i_fThe stator current of the magnetic suspension disk type motor is obtained;

then there are:

wherein F is g + F_d/m，G＝-k/(m²)，u＝i_f ²F, G, u represent the uncertain parameter term, the system known term, and the output of the sliding mode controller, respectively;

when formula (4) is substituted for formula (2), it is possible to obtain:

12) calculating the output of the sliding mode controller:

the exponential approximation law is:

in the formula, μ and η are positive real numbers, and sigmoid function is a bipolar sigmoid function:

substituting formula (6) for formula (5) with u_SMCAnd replacing u, obtaining the output of the sliding mode controller as follows:

13) the output u of the sliding mode controller obtained in the step 12) in the formula (7) is processed_SMCThe absolute value of the outer ring suspension air gap tracking controller is obtained by calculating the evolution to obtain the output of the outer ring suspension air gap tracking controller, and the output is used as the stator current reference value of the magnetic suspension disk type motor

14) Reference value of stator current of the magnetic suspension disk type motor

step 2, after the rotating body reaches a balance point to achieve suspension, an outer ring suspension air gap tracking controller of the suspension controller is changed to a self-adaptive radial basis function neural network sliding mode control strategy, an inner ring suspension current tracking controller of the suspension controller adopts PID control to control the stator current of the magnetic suspension disc type motor, and the rotating body is enabled to keep stable suspension at the suspension balance point, and the specific method is as follows:

21) the method comprises the following steps of (1) approximating an uncertain parameter item F in an equation (4) by using a radial basis function neural network:

A. determining the number of layers of the radial basis function neural network:

the radial basis function neural network comprises 1 input layer, 1 hidden layer and 1 output layer, wherein the input vector of the input layer is as follows:

B. selecting a Gaussian function as an activation function of the hidden layer, the output of the hidden layer is

In the formula, h_jJ is 1,2, …, n is the number of nodes of the hidden layer, c is the output of the jth node of the hidden layer_j＝[c_j1,c_j2]^TIs the central vector, | | x-c, of the Gaussian basis function of the jth node of the hidden layer_jI is the Euclidean norm for measuring the input vector x and the j-th node center vector of the hidden layer, b_jIs the normalization constant of the jth node of the hidden layer;

C. calculating the output of the output layer by using the weighted value sum of the output values of the hidden layer, and enabling the output y of the output layer to be the estimated value of the uncertain parameter item F in the formula (4)Then there are:

in the formula (I), the compound is shown in the specification,

22) solving the output of the self-adaptive radial basis function neural network sliding mode controller:

according to formulae (5), (6) and (10), with u_ASMCReplacing u, and obtaining the output of the sliding mode controller of the adaptive radial basis function neural network as follows:

23) solving a weight self-adaptation law of the output layer of the radial basis function neural network:

let the uncertainty parameter term F be expressed as:

F＝W^*Th+ (12)

in the formula, the approximation error is obtained, and the condition that | | | is less than or equal to_N，_NIs a minimum supremum, is a bounded positive real number; w^*The ideal weight vector of the output layer of the radial basis function neural network is obtained;

by substituting formulae (11) and (12) for formula (5), it is possible to obtain:

in the formula (I), the compound is shown in the specification,

the Lyapunov function is constructed as:

wherein γ is a positive real number;

by taking the derivative of equation (14) in combination with equation (13), there are:

taking the weight adaptive law of the radial basis function neural network output layer as follows:

when formula (16) is substituted for formula (15), there are:

because the real number is very small, only eta is selected to be equal to or more than_NCan obtainAccording to the Lyapunov stability theory, the self-adaptive radial basis function neural network sliding mode controller is proved to be globally asymptotically stable;

24) the output u of the self-adaptive radial basis function neural network sliding mode controller obtained in the step 23) in the formula (11)_ASMCThe absolute value of the outer ring suspension air gap tracking controller is obtained by calculating the evolution to obtain the output of the outer ring suspension air gap tracking controller, and the output is used as the stator current reference value of the magnetic suspension disk type motor

25) Reference value of stator current of the magnetic suspension disk type motor

Technical Field

The invention relates to a control method, in particular to a magnetic suspension vertical axis wind turbine generator suspension control method based on an adaptive neural network, and belongs to the technical field of electrical engineering.

Background

At present, a high-power wind driven generator mainly takes a horizontal shaft wind driven generator as a main part. However, the horizontal axis wind turbine has inherent defects of large starting resistance moment, need of yawing to wind, difficulty in control, inconvenience in installation and the like, so that the healthy development of the horizontal axis wind turbine is influenced, and the requirement of a weak wind type wind power plant on low wind speed starting is particularly difficult to meet.

The magnetic suspension vertical axis wind driven generator has no mechanical friction, greatly reduces the starting resistance moment, can further reduce the starting wind speed, has the advantages of low starting wind speed, simple and convenient installation, no need of a yaw device and the like, can be used for a wind power plant with low wind speed and frequent wind direction change (the vertical axis wind driven generator does not need wind), and is the key direction of future wind power development. In an actual working environment, the suspension control of the magnetic suspension vertical axis wind turbine generator set must meet the requirements of strong self-adaptive capacity, fast dynamic response, strong anti-interference capacity and the like.

However, in practical applications, magnetic levitation technology presents many challenges, such as: open loop instability, strong non-linearity, inaccurate modeling, etc.; meanwhile, the randomness of wind interference seriously affects the suspension stability, so that the design of the suspension controller is very challenging. The conventional PID controller is simple in structure, but the parameters of the controller are difficult to adjust on line and difficult to automatically adjust to adapt to the change of the external environment, so that the ideal control effect on the wind power magnetic suspension system under random interference is difficult to achieve. The cascade PID control can improve the stability and response speed of the system by reducing the phase lag and equivalent time constant of the secondary loop closed-loop system, the control quality of the system is improved by increasing the damping frequency of the cascade control system through the gain of the secondary loop controller, the design is simple, the structure is flexible, the robustness is strong, but the cascade PID control depends on a determined object model, the parameters of the controller are fixed, and when the object model and the parameters are uncertain, the control effect is not obvious. The cascade PID controller based on the BP neural network can adjust the parameters of the outer ring PID controller of the suspension air gap in real time through the BP neural network, has high dynamic response speed and good anti-interference capability, but has low convergence speed and is easy to fall into a local minimum value. There is also a document that proposes an adaptive sliding mode controller based on a hybrid magnetic flux density observer to improve the system performance of the suspension system, but the magnetic flux density sensor is not easy to install in actual operation, and the method implementation difficulty is large. Some control strategies based on the adaptive neural fuzzy sliding mode controller are adopted to inhibit parameter perturbation, but due to the fact that single closed-loop control is used, when current is suddenly changed, system robustness is reduced. In addition, the traditional sliding mode control generally adopts an exponential approximation law containing a sign function, but the sign function is not a smooth function and is not favorable for weakening the buffeting of the sliding mode.

Disclosure of Invention

The main purposes of the invention are as follows: aiming at the defects and blanks of the prior art, the invention provides a suspension control method of a magnetic suspension vertical axis wind turbine generator based on an adaptive neural network, which is characterized in that the suspension control performance of the magnetic suspension vertical axis wind turbine generator is improved and stable suspension is realized under the conditions that the suspension system of the magnetic suspension vertical axis wind turbine generator is not accurately modeled and is subjected to random interference caused by wind speed change by controlling the adaptive neural network, combining sliding mode control and adopting a continuous and smooth bipolar S-shaped function to replace a symbolic function in the conventional sliding mode index approach law.

In order to achieve the above object, the magnetic suspension vertical axis wind turbine generator set of the present invention includes: the magnetic suspension vertical axis wind driven generator comprises a magnetic suspension vertical axis wind driven generator, a suspension control system, an air gap sensor, a wind wheel, an upper end bearing, a lower end bearing, a shell, a rotating shaft and the like; the magnetic suspension vertical shaft wind driven generator comprises a permanent magnet direct drive type wind driven generator and a magnetic suspension disc type motor.

The permanent magnet direct-drive wind driven generator comprises a stator and a rotor.

The magnetic suspension disc type motor is positioned below the permanent magnet direct-drive type wind driven generator and comprises a disc stator and a disc rotor; the disc stator is composed of a disc stator iron core and a suspension winding, and the suspension winding is a direct-current excitation winding.

The suspension control system consists of a suspension converter and a suspension controller thereof, and the suspension converter is connected with the suspension winding; the suspension controller comprises an outer ring suspension air gap tracking controller and an inner ring suspension current tracking controller.

The rotor of the permanent-magnet direct-drive wind driven generator, the disc rotor of the magnetic suspension disc type motor, the wind wheel and the rotating shaft are collectively called as a rotating body.

The invention relates to a magnetic suspension vertical axis wind turbine generator suspension control method based on an adaptive neural network, which comprises the following steps of:

step 1, when the wind speed reaches the cut-in wind speed, an outer ring suspension air gap tracking controller of the suspension controller adopts a sliding mode control strategy, an inner ring suspension current tracking controller of the suspension controller adopts PID control to control the stator current of the magnetic suspension disc type motor, so that the rotating body is suspended upwards to and kept at a suspension balance point to realize stable suspension, and the specific method is as follows:

11) designing a slip form surface s as follows:

where e is the floating air gap tracking error: e ═ e^*-，^*The reference value of the suspension air gap at the suspension balance point is taken as the measurement value of the suspension air gap; c. C₀、c₁Are positive real numbers.

By deriving equation (1) over time t, there is:

according to the mechanical equation of the rotating body in the vertical direction:

in the formula (I), the compound is shown in the specification,

is the second derivative with respect to time t; m is the mass of the rotating body, and g is the acceleration of gravity; f. of_dIs an external random interference force; k is mu₀N²A/4, wherein₀The magnetic pole surface effective area of the disc stator is A, and the number of turns of the suspension winding is N; i.e. i_fThe stator current of the magnetic suspension disk type motor.

Then there are:

wherein F is g + F_d/m，G＝-k/(m²)，F. G, u represent the uncertain parameter term, the system-known term, and the output of the sliding-mode controller, respectively.

When formula (4) is substituted for formula (2), it is possible to obtain:

12) calculating the output of the sliding mode controller:

the exponential approximation law is:

in the formula, μ and η are positive real numbers, and sigmoid function is a bipolar sigmoid function:

substituting formula (6) for formula (5) with u_SMCAnd replacing u, obtaining the output of the sliding mode controller as follows:

13) the output u of the sliding mode controller obtained in the step 12) in the formula (7) is processed_SMCThe absolute value of the outer ring suspension air gap tracking controller is obtained by calculating the evolution to obtain the output of the outer ring suspension air gap tracking controller, and the output is used as the stator current reference value of the magnetic suspension disk type motor

14) Reference value of stator current of the magnetic suspension disk type motorWith its actual measured value i_fAnd performing difference, sending the difference to a PWM module through the inner ring suspension current tracking controller, generating a driving signal of the suspension converter, and controlling the current i of the magnetic suspension disc stator_fAnd the rotating body is suspended upwards to and kept at the suspension balance point.

Step 2, after the rotating body reaches a balance point to achieve suspension, an outer ring suspension air gap tracking controller of the suspension controller is changed to a self-adaptive radial basis function neural network sliding mode control strategy, an inner ring suspension current tracking controller of the suspension controller adopts PID control to control the stator current of the magnetic suspension disc type motor, and the rotating body is enabled to keep stable suspension at the suspension balance point, and the specific method is as follows:

21) the method comprises the following steps of (1) approximating an uncertain parameter item F in an equation (4) by using a radial basis function neural network:

A. determining the number of layers of the radial basis function neural network:

the radial basis function neural network comprises 1 input layer, 1 hidden layer and 1 output layer, wherein the input vector of the input layer is as follows:

is the first derivative with respect to time t; the hidden layer has nA neuron; the output layer has 1 neuron.

B. Selecting a Gaussian function as an activation function of the hidden layer, and outputting the hidden layer as follows:

in the formula, h_jIs the output of the jth node of the hidden layer, j is 1,2, …, n is the number of nodes of the hidden layer, c_j＝[c_j1,c_j2]^TIs the central vector, | x-c, of the Gaussian basis function of the jth node of the hidden layer_jI is the Euclidean norm for measuring the input vector x and the j-th node center vector of the hidden layer, b_jIs the normalization constant of the jth node of the hidden layer.

C. Calculating the output of the output layer by using the weighted value sum of the output values of the hidden layer, and enabling the output y of the output layer to be the estimated value of the uncertain parameter item F in the formula (4)

Then there are:

in the formula (I), the compound is shown in the specification,a weight vector representing the output layer, h ═ h₁,h₂,…,h_n]^TAn output vector representing the hidden layer, wherein h_jThe value is obtained from the formula (9).

22) Solving the output of the self-adaptive radial basis function neural network sliding mode controller:

according to formulae (5), (6) and (10), with u_ASMCReplacing u, and obtaining the output of the sliding mode controller of the adaptive radial basis function neural network as follows:

23) solving a weight self-adaptation law of the output layer of the radial basis function neural network:

let the uncertainty parameter term F be expressed as:

F＝W^*Tin the h + (12) formula, the approximation error is obtained, and the condition that | | | is less than or equal to_N，_NIs a minimum supremum, is a bounded positive real number; w^*And outputting the ideal weight vector of the output layer of the radial basis function neural network.

By substituting formulae (11) and (12) for formula (5), it is possible to obtain:

in the formula (I), the compound is shown in the specification,

the Lyapunov function is constructed as:

wherein γ is a positive real number.

By taking the derivative of equation (14) in combination with equation (13), there are:

taking the weight adaptive law of the radial basis function neural network output layer as follows:

when formula (16) is substituted for formula (15), there are:

because the real number is very small, only eta is selected to be equal to or more than_NCan obtain

According to the Lyapunov stability theory, the self-adaptive radial basis function neural network sliding mode controller can be proved to be globally asymptotically stable.

24) The output u of the self-adaptive radial basis function neural network sliding mode controller obtained in the step 23) in the formula (11)_ASMCThe absolute value of the outer ring suspension air gap tracking controller is obtained by calculating the evolution to obtain the output of the outer ring suspension air gap tracking controller, and the output is used as the stator current reference value of the magnetic suspension disk type motor

25) Reference value of stator current of the magnetic suspension disk type motorWith its actual measured value i_fAnd performing difference, sending the difference to a PWM module through the inner ring suspension current tracking controller, generating a driving signal of the suspension converter, and controlling the stator current i of the magnetic suspension disk type motor_fAnd keeping the rotating body stably suspended at the suspension balance point.

The invention has the beneficial effects that: according to the method, a self-adaptive neural network sliding mode control strategy is adopted, and a smooth bipolar S-shaped function is adopted in a sliding mode index approaching law, so that on one hand, a sliding mode controller is adopted in a suspension starting stage, a rotating body of the magnetic suspension vertical axis wind turbine generator stably reaches a suspension balance point and keeps suspension; on the other hand, after the suspension balance point is reached, the suspension controller is automatically switched to adaptive radial basis function neural network sliding mode control, time-varying and nonlinear uncertain interference brought to the suspension system by the fluctuation and randomness of wind speed and wind direction is approached by using a radial basis function neural network (hereinafter abbreviated as RBF neural network) model, and meanwhile, the robustness and the dynamic performance of the suspension system are enhanced through the adjustment of the sliding mode controller, stable suspension is realized, and the real-time optimal performance of the magnetic suspension vertical axis wind turbine system is ensured.

Drawings

FIG. 1 is a schematic structural diagram of a magnetic suspension vertical axis wind turbine according to the present invention.

Fig. 2 is a schematic diagram of a suspension system structure and a mechanical analysis of the magnetic suspension disk type motor.

FIG. 3 is a structural block diagram of a suspension control system based on a sliding mode control strategy according to the present invention.

FIG. 4 is a structural block diagram of a suspension control system based on adaptive RBF neural network sliding mode control.

FIG. 5 is a model structure of RBF neural network of the present invention.

FIG. 6 is a structural block diagram of a suspension control system of a sliding mode control strategy without an RBF neural network.

FIG. 7 is a graph showing a comparison simulation of a suspended air gap under a constant-amplitude disturbance force according to the sliding mode control strategy without an RBF neural network.

FIG. 8 is a graph showing the comparative simulation of the levitation current under a constant-amplitude disturbance force according to the sliding mode control strategy without the RBF neural network.

Fig. 9 is a graph showing the variation of the non-linear disturbing force applied by the present invention.

FIG. 10 is a graph of a comparison simulation of a suspended air gap under a nonlinear disturbance force according to the sliding mode control strategy without an RBF neural network.

FIG. 11 is a graph showing the comparison simulation of the levitation current under the nonlinear disturbance force according to the sliding mode control strategy without adding the RBF neural network.

Reference numbers in the figures: the system comprises a 1-permanent magnet direct-drive type wind driven generator, a 11-permanent magnet direct-drive type wind driven generator stator, a 12-permanent magnet direct-drive type wind driven generator rotor, a 2-magnetic suspension disc type motor, a 21-magnetic suspension disc type motor stator, a 22-magnetic suspension disc type motor rotor, a 3-wind wheel, a 6-air gap sensor, a 7-lower end bearing, an 8-upper end bearing, a 9-shell, a 10-rotating shaft, an 18-suspension converter, a 211-disc type suspension iron core, a 212-suspension winding, a 221-disc type rotor iron core, a 222-disc type rotor winding, a 30-suspension controller, a 31-outer ring suspension air gap tracking controller and a 32-inner ring suspension current tracking controller.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings.

As shown in fig. 1 and 2, the magnetic suspension vertical axis wind turbine generator set of the present invention includes: the magnetic suspension vertical axis wind driven generator comprises a magnetic suspension vertical axis wind driven generator, a suspension control system, a wind wheel 3, an air gap sensor 6, an upper end bearing 7, a lower end bearing 8, a shell 9, a rotating shaft 10 and the like. The magnetic suspension vertical shaft wind driven generator consists of two motors, namely: a permanent magnet direct drive type wind driven generator 1 and a magnetic suspension disk type motor 2.

The permanent magnet direct drive type wind driven generator 1 comprises a stator 11 and a rotor 12; the magnetic suspension disc type motor 2 is positioned below the permanent magnet direct-drive type wind driven generator 1 and comprises a magnetic suspension disc type motor stator 21 and a magnetic suspension disc type motor rotor 22, the distance between the disc type stator 21 and the disc type rotor 22 is a suspension air gap, the magnetic suspension disc type motor stator 21 consists of a disc type stator iron core 211 and a suspension winding 212, the suspension winding 212 is a direct current excitation winding, and an air gap sensor 6 is attached to the surface of the disc type stator iron core 211 to measure the suspension air gap; the magnetic suspension disc type motor rotor 22 comprises a disc type rotor iron core 221 and a disc type rotor winding 222, the disc type rotor winding 222 is a three-phase winding, and the disc type rotor 22 is fixed with the bottom of the rotating shaft 10; the wind wheel 3 is fixed with the upper part of the rotating shaft 10.

As shown in fig. 1, all rotating parts of the rotor 12 of the permanent magnet direct drive type wind power generator 1, the disc rotor 22 of the magnetic levitation disc type motor 2, the wind wheel 3, the rotating shaft 10, and the like are collectively referred to as a rotating body.

As shown in fig. 3, 4 and 6, the levitation control system is composed of a levitation current transformer 18 and a levitation controller 30, wherein the levitation current transformer 18 is a DC/DC current transformer, and is connected to a levitation winding 212 for controlling levitation; the suspension controller 30 comprises an outer ring suspension air gap tracking controller 31 and an inner ring suspension current tracking controller 32, and the outer ring suspension air gap controller realizes suspension air gap tracking; the inner loop levitation current tracking controller 32 implements levitation current tracking.

The invention relates to a magnetic suspension vertical axis wind turbine generator suspension control method based on an adaptive neural network, which comprises the following steps of:

step 1, when the wind speed reaches the cut-in wind speed, as shown in fig. 3, an outer ring suspension air gap tracking controller 31 of a suspension controller 30 adopts a sliding mode control strategy, an inner ring suspension current tracking controller 32 adopts PID control, and controls the stator current of a magnetic suspension disc type motor 2, so that a rotating body of a wind turbine generator is suspended upwards to and kept at a suspension balance point to realize stable suspension, and the specific method is as follows:

11) the design slip form surface is:

where e is the floating air gap tracking error: e ═ e^*-，^*The reference value of the suspension air gap at the suspension balance point is the measured value of the suspension air gap, and the measured value is measured by an air gap sensor 6; c. C₀、c₁Are positive real numbers.

The derivation for equation (1) is:

according to a mechanical equation of a rotating body of the wind turbine generator in the vertical direction:

in the formula (I), the compound is shown in the specification,is the second derivative with respect to time t; m is the mass of the rotating body of the wind turbine generator, and g is the gravity acceleration; f. of_dIs an external random interference force; k is mu₀N²A/4, wherein₀For the vacuum permeability, a is the effective area of the magnetic pole surface of the disc stator 21, and N is the number of turns of the levitation winding 212; i.e. i_fIs the stator current of the disc motor 2.

Then there are:

wherein F is g + F_d/m，G＝-k/(m²)，

F. G, u represent the uncertain parameter term, the system-known term, and the output of the sliding-mode controller, respectively.

When formula (4) is substituted for formula (2), it is possible to obtain:

the mechanical equation (3) of the rotating body in the vertical direction is obtained through the following process:

as shown in FIG. 2, when the suspension winding 212 of the maglev disk type motor is electrified, an upward axial suspension attraction force f (i) is generated_fAnd) is:

the rotating body is subjected to upward levitation suction force f (i) in the axial direction_fB), the downward rotor weight mg and the external disturbance force f_dAccording to newton's second law, the mechanical equation of the rotating body in the vertical direction can be obtained as follows:

12) calculating the output of the sliding mode controller:

the exponential approximation law is:

in the formula, μ and η are positive real numbers, and sigmoid function is a bipolar sigmoid function:

substituting formula (6) for formula (5) with u_SMCInstead of u, the output of the sliding mode controller is found to be:

the Lyapunov function is constructed as:

the derivation of the above formula is:

according to the Lyapunov stability theory, the sliding mode controller is proved to be globally and gradually stable.

13) U obtained in the formula (7) in the step 12)_SMCThe absolute value of (a) is calculated to obtain the output of the outer ring air gap tracking controller 31, i.e. the stator current reference value of the disc motor 2

14) Referring to the stator current of the disc motor 2 as shown in fig. 3With its actual measured value i_fThe difference is sent to a PWM module through an inner ring suspension current tracking controller 32(PID controller) to generate a driving signal of the suspension converter 18, so that the stator current i of the disc type motor 2 is controlled_fI.e., the levitation current, levitates the rotating body upward to and remains at the levitation balance point.

Step 2, after the rotating body of the wind turbine generator reaches the balance point to achieve suspension, as shown in fig. 4, the outer-ring suspension air gap tracking controller 31 of the suspension controller 30 is changed to a self-adaptive radial basis function neural network sliding mode control strategyThe inner ring suspension current tracking controller 32 still adopts PID control to control the stator current i of the disc type motor 2_fThe method for keeping the rotating body stably suspended at the suspension balance point comprises the following steps:

21) the method comprises the following steps of (1) utilizing an uncertain parameter item F in an RBF neural network approximation formula (4):

A. determining the layer number of the RBF neural network:

as shown in fig. 5, the RBF neural network includes 1 input layer, 1 hidden layer, and 1 output layer, where the input vectors of the input layers are: is the first derivative with respect to time t; the hidden layer has 5 neurons (n-5); the output layer has 1 neuron y.

B. Selecting a gaussian function as the activation function of the hidden layer, the output of the hidden layer is:

in the formula, h_jIs the output of the jth node of the hidden layer, j is 1,2, …,5 is the number of nodes of the hidden layer, c_j＝[c_j1,c_j2]^TIs the central vector, | | x-c, of the jth node of the hidden layer Gaussian function_jI is the Euclidean norm measuring the input vector x and the j-th node center of the hidden layer, b_jIs the width of the jth neuron of the hidden layer gaussian function.

C. Calculating the output of the output layer by the weighted value sum of the output values of the hidden layer, and making the output y of the output layer be the estimated value of the uncertain parameter item F in the formula (4)Then there are:

in the formula (I), the compound is shown in the specification,weight vector representing output layer, h ═ h₁,h₂,…,h₅]^TAn output vector representing the hidden layer, where h_jThe value is obtained from the formula (9).

22) Solving the output of the self-adaptive RBF neural network sliding mode controller:

according to formulae (5), (6) and (10), with u_ASMCReplacing u, and solving the output of the sliding mode controller of the adaptive RBF neural network as follows:

23) solving a weight self-adaptation law of an output layer of the RBF neural network:

let the uncertainty parameter term F be expressed as:

F＝W^*Tin the h + (12) formula, the approximation error is obtained, and the condition that | | | is less than or equal to_N，_NIs a minimum supremum, is a bounded positive real number; w^*And the ideal weight vector is the output layer of the RBF neural network.

By substituting formulae (11) and (12) for formula (5), it is possible to obtain:

in the formula (I), the compound is shown in the specification,

the Lyapunov function is constructed as:

wherein γ is a positive real number.

By taking the derivative of equation (14) in combination with equation (13), there are:

taking the weight self-adaptation law of the output layer of the RBF neural network as follows:

when formula (16) is substituted for formula (15), there are:

because the real number is very small, only eta is selected to be equal to or more than_NCan obtain

Thus, according to the Lyapunov stability theory, the controller may prove to be globally asymptotically stable.

24) The output u obtained in the formula (11) in the step 23) is_ASMCThe absolute value of the outer ring suspension air gap tracking controller 31 is obtained, namely the stator current reference value of the disc type motor 2 is obtained

25) Referring to the stator current of the disc motor 2 as shown in fig. 4With its actual measured value i_fThe difference is sent to a PWM module through an inner ring suspension current tracking controller 32(PID controller) to generate a driving signal of the suspension converter 18, thereby controlling the stator current i of the motor 2_fAnd the rotating body is kept stably suspended at the suspension balance point.

The invention will be further described below with reference to a preferred embodiment.

In order to verify the effectiveness of the suspension control method based on the adaptive neural network, the suspension system of the magnetic suspension vertical axis wind turbine generator set is subjected to comparative simulation analysis by respectively adopting the adaptive neural network sliding mode control strategy and a sliding mode control strategy (SMC-PID) without an RBF neural network.

As shown in fig. 6, the main ideas of the SMC-PID method are: obtaining a sliding mode surface formula (1) by using a mechanical equation (3) of the rotating body in the vertical direction, then obtaining a derivative of the sliding mode surface and obtaining the output of the outer ring suspension air gap controller by using a formula (6), namely the output u of the sliding mode controller_SMC(formula (7)) and p_SMCThe absolute value of the absolute value is calculated to obtain a stator current reference value i of the magnetic suspension disk type motor 2_{f_y}Then i is_{f_y}With the stator current i of the actual magnetic levitation disc motor 2_fThe difference is sent to a PWM module through a PID controller to generate a driving signal of a suspension converter 18 and control the stator current i of the magnetic suspension disk type motor 2_fAnd the rotating body is kept stably suspended at the suspension balance point.

Specific simulation parameters are shown in tables 1 and 2.

TABLE 1 model parameters of magnetic levitation vertical axis wind turbine

Parameter name	Numerical value
		Rotating body mass m (kg)	500
Number of turns N of levitation winding 212	644
		Effective area a (mm) of magnetic pole surface of disk stator 212)	235050
Balance point suspension air gap reference value delta*(mm)	10
		Magnetic permeability mu in vacuum0(N/A2)	4π×10-7

TABLE 2 parameters of the levitation controller

The parameters of the RBF neural network are as follows:

c₁＝[c₁₁,c₁₂]^T＝[-1,-1]^T，c₂＝[c₂₁,c₂₂]^T＝[-0.5,-0.5]^T，c₃＝[c₃₁,c₃₂]^T＝[0,0]^T，

c₄＝[c₄₁,c₄₂]^T＝[0.5,0.5]^T，c₅＝[c₅₁,c₅₂]^T＝[1,1]^T；b_j＝0.1(j＝1,2,…,5)。

in order to verify the anti-jamming capability of the suspension system, constant amplitude jamming and non-linear jamming were applied separately.

1) Applying a constant amplitude disturbance: adding external interference with the amplitude of 1000N and 1500N at the 2s and 6s respectively, and removing the added external interference at the 4s and 8s respectively.

Fig. 7 shows the variation curve of the floating air gap of the adaptive neural network sliding mode control method and the SMC-PID method of the present invention under the effect of the disturbance, and fig. 8 shows the variation curve of the floating current of the adaptive neural network sliding mode control method and the SMC-PID method of the present invention under the effect of the disturbance.

As can be seen from FIGS. 7 and 8, at the equilibrium point, after the constant amplitude interference is added, the suspension air gap has smaller overshoot, the suspension current has faster dynamic response, and the system can be recovered to the stable state in a short time; and with the SMC-PID method, the suspended air gap has obvious steady-state error. Therefore, the self-adaptive neural network sliding mode control method can improve the dynamic response of the magnetic suspension vertical axis wind turbine generator.

2) Non-linear disturbances are applied as shown in fig. 9: fig. 10 is a graph showing the variation curve of the floating air gap of the adaptive neural network sliding mode control and SMC-PID method of the present invention under the effect of the disturbance, and fig. 11 is a graph showing the variation curve of the floating current of the adaptive neural network sliding mode control and SMC-PID method of the present invention under the effect of the disturbance.

As can be seen from fig. 10 and 11, after the nonlinear disturbance force is added, the suspension air gap of the present invention can be stabilized within 0.1mm, and no obvious buffeting phenomenon occurs; with the SMC-PID method, however, an oscillation occurs at the equilibrium point. Therefore, the self-adaptive neural network sliding mode control method can improve the anti-interference capability of the magnetic suspension vertical axis wind turbine generator.

In a word, the suspension control method based on the adaptive neural network can meet the requirements of the magnetic suspension vertical axis wind turbine generator on high dynamic response, high adaptive capacity, high robustness and the like of suspension control.