CN106357192A - Method and system for lowering torque pulsation of switched reluctance motor by current adaptive control - Google Patents

Abstract

The invention relates to a method and system for lowering torque pulsation of a switched reluctance motor by current adaptive control; the method comprises: preprocessing a deviation: subjecting the torque deviation to nonlinear conversion; solving torque estimated output and adaptive PID (proportion integration differentiation) control coefficient by using double-weight neural network; acquiring current set total current by PID control calculation, and acquiring each phase control current via current allocation; predicting current feed-forward compensation control by finite difference extended Kalman filter, and effectively inhibiting and lowering torque pulsation of the switched reluctance motor by joint action of adaptive PID control and prediction-based current feed-forward compensation control. Current, torque and rotor position sensors are connected with a signal processor, the signal processor executes modules of the method, compensated three-phase reference current is output to control a power converter of the motor via a current hysteresis controller, and torque pulsation of the switched reluctance motor is significantly and effective inhibited.

Description

Method and system for reducing torque ripple of switched reluctance motor through current self-adaptive control

Technical Field

The invention relates to the field of motor power driving of electric automobiles, in particular to a method and a system for reducing torque pulsation of a switched reluctance motor through current self-adaptive control.

Background

The Switched Reluctance Motor (SRM) has neither windings nor permanent magnets, and has many advantages of simple structure, high efficiency, stable and reliable operation, low cost, easy maintenance, etc. However, the switched reluctance motor has a special double salient pole structure, adopts a switched power supply mode, has high nonlinearity and strong coupling property of electromagnetic characteristics, and accordingly has large torque pulsation during operation, and severely restricts application and development of the switched reluctance motor due to noise and vibration caused by the torque pulsation, so that a research method for reducing the torque pulsation of the switched reluctance motor has extremely important significance.

In order to suppress and reduce the torque ripple of the switched reluctance motor SRM, various studies have been made by domestic and foreign scholars, and many results have been obtained.

The torque distribution function method is a control strategy which is provided by combining the characteristics of the SRM, is mostly applied to inhibiting and reducing the torque pulsation of the SRM, enables the torque to be smoothly transited during the phase change through a proper distribution function, and then is added with a proper control strategy to enable the total output torque of the motor to smoothly track the given reference torque.

Another document proposes an off-line torque distribution control strategy and compares it with conventional linear, sinusoidal, cubic, exponential torque distribution control strategies. In order to improve the accuracy of the model, a non-linear model of torque-current is provided on the basis of a linear model, but a specific identification process of unknown parameters is not provided.

There are also documents that propose current sharing and flux linkage sharing control strategies on the basis of torque sharing. The two modes avoid directly searching the nonlinear relation of torque-current, the control sensitivity is higher, but at present, the two modes are only limited to the theoretical research of simulation.

In addition, an iterative learning controller is introduced on the basis of a traditional torque distribution control strategy, iterative learning control is carried out on reference current by taking the minimum torque deviation as a control target, current variation compensation reference current is output, a partition refining hysteresis current controller with 1, 0 and-1 working states is designed on the basis of a traditional hysteresis controller with only 1 state and-1 state, and a simulation result proves that the controller is effective in inhibiting torque pulsation.

In addition, a speed ring is added to the traditional direct torque control, and direct torque PID (proportional integral derivative) control is provided to adjust the voltage duty ratio aiming at the defect of small allowable error fluctuation of hysteresis control, so that the torque control of the SRM is realized.

However, due to the extremely strong nonlinear characteristics of the switched reluctance motor, the conventional motor torque control methods based on the simplified mathematical model are all prone to cause severe torque ripple, and thus cannot be practically applied.

Disclosure of Invention

The invention aims to design a method for reducing the torque ripple of a switched reluctance motor by current self-adaptive control, which comprises the steps of firstly carrying out deviation preprocessing and carrying out nonlinear conversion on torque deviation; solving parameters self-adaptive to PID control by using a dual-weight neural network; and then the current feedforward compensation control predicted by the finite difference extended Kalman filter is adopted, so that the one-step prediction capability of the control system is improved. And the adaptive PID control and the current feedforward compensation control based on prediction act together to effectively inhibit and reduce the SRM torque ripple.

Another objective of the present invention is to design a system for reducing the torque ripple of the switched reluctance motor by current adaptive control according to the above method for reducing the torque ripple of the switched reluctance motor by current adaptive control of the present invention, and combine the adaptive PID control with the current feedforward compensation control based on prediction to form a control system of the switched reluctance motor.

The invention discloses a method for reducing torque pulsation of a switched reluctance motor by current self-adaptive control, which mainly comprises the following steps:

i, torque deviation preprocessing

The invention introduces a nonlinear function to preprocess the torque deviation, and preprocesses small error, large gain, large error and small gain. The specific pretreatment is as follows:

$fal\left(\mathrm{\&Delta;}T\right)=\left\{\begin{array}{cc}|\mathrm{\&Delta;}T{|}^{\mathrm{\&alpha;}}sign\left(\mathrm{\&Delta;}T\right),& \left|\mathrm{\&Delta;}T\right|\&GreaterEqual;\mathrm{\&delta;}\\ \frac{\mathrm{\&Delta;}T}{{\mathrm{\&delta;}}^{(1-\mathrm{\&alpha;})}},& \left|\mathrm{\&Delta;}T\right|<\mathrm{\&delta;}\end{array}\right.---\left(1\right)$

wherein the transformation range of the feedback deviation is shown, and the value range is 0.01T_d～0.1T_d，T_dFor setting the torque, fal represents a preprocessing function, sign represents a sign function, namely, Δ T is greater than zero, and takes a value of 1, is less than zero, and takes a value of-1. Δ T is a set torque T_dAnd the measured instantaneous torque T_eα is the adjustment coefficient range 0 ~ 1.

II self-adaptive PID torque control based on double-weight neural network

II-1 double weight neural network

The SRM is a strong nonlinear system with variable structure and variable parameters, and an accurate mathematical model is difficult to obtain. The invention realizes the self-adaptive PID control of the SRM by online adjusting three parameters of PID control (proportional, integral and differential control) through a Double Weight Neural Network (DWNN).

The dual weight neural network DWNN includes an input layer, a hidden layer, and an output layer. Instantaneous torque T detected by torque sensor mounted on switch reluctance motor_eInstantaneous torque T at the previous moment_{e_1}And setting the current i_dIncrement △ i_dPrevious time value ofAs an input amount of DWNN, torque T is set_dThe DWNN predicts a torque estimation output T after learning the target value of DWNN_outAnd proportional coefficient K of PID control_pIntegral coefficient K_iiAnd a differential coefficient K_d。

PID control is carried out according to a torque deviation preprocessing result fal (delta T) obtained in the step I and a proportionality coefficient K obtained by a double-weight neural network DWNN_pIntegral coefficient K_iiAnd a differential coefficient K_dObtaining the current set total current i through PID calculation_dObtaining control currents of each phase through current distributionAnd

the invention adds the adaptive adjustment of the power, and the improved DWNN is described as follows:

$\left\{\begin{array}{c}h\left(j\right)=f\left(\underset{z=1}{\overset{3}{\&Sigma;}}{w_{zj}}^m{\left(x_z-q_{zj}\right)}^{b_{zj}}\right)\\ f\left(x\right)=\frac{1}{1+e^{-ax}}\\ T_{out}=\underset{j=1}{\overset{3}{\&Sigma;}}{v}_j*h\left(j\right)\end{array}\right.---\left(2\right)$

where h (j) is the hidden layer output function, f (x) is the activation function, T_outFor the torque estimation output of DWNN, w_zjIs a directional weight, q_zjIs a core weight, v_jIs an output weight value of 0<a<1, m ranges from 1 to 10, b_zjFor adaptive adjustment of the power, e is the base of the natural logarithm. The hidden layer is a z layer, and z is 1,2 and 3; the output layer is j layers, j equals 1,2,3.

To obtain the parameters in equation (2), the performance indicator function is taken as:

$\left\{\begin{array}{c}{\mathrm{\&Delta;T}}_1\left(k\right)=T_e\left(k\right)-T_{out}\left(k\right)\\ J=\frac{1}{2}{\left({\mathrm{\&Delta;T}}_1\right(k\left)\right)}^2\end{array}\right.---\left(3\right)$

wherein, △ T₁Defined as the measured instantaneous torque T_eAnd double weight neural network DWNN torque estimation output T_outThe deviation therebetween.

According to the gradient descent method, the direction weight w_zjCore weight q_zjAnd the output weight v_jAnd the power b_zjThe iterative algorithm of increments of (1) is as follows:

$\left\{\begin{array}{c}{\mathrm{\&Delta;v}}_j\left(k\right)=-\mathrm{\&eta;}\frac{\&part;J}{\&part;v_j}={\mathrm{\&eta;\&Delta;T}}_1\left(k\right)h_j\left(k\right)\\ {\mathrm{\&Delta;w}}_{zj}=-\mathrm{\&eta;}\frac{\&part;J}{\&part;w_{zj}}={\mathrm{\&eta;ma\&Delta;T}}_1\left(k\right)v_j{h}_j(1-h_j){w_{zj}}^{m-1}{(x_z-q_{zj})}^{b_{zj}}\\ {\mathrm{\&Delta;q}}_{zj}=-\mathrm{\&eta;}\frac{\&part;J}{q_{zj}}={\mathrm{\&eta;ab}}_{zj}{\mathrm{\&Delta;T}}_1\left(k\right)v_j{h}_j(1-h_j){w_{zj}}^m{(x_z-q_{zj})}^{(b_{zj}-1)}\\ {\mathrm{\&Delta;b}}_{zj}=-{\mathrm{\&eta;}}_b\frac{\&part;J}{\&part;b_{zj}}={\mathrm{\&eta;}}_b{\mathrm{a\&Delta;T}}_1\left(k\right)v_j{h}_j(1-h_j){w_{zj}}^m{(x_z-q_{zj})}^{b_{zj}}\ln (x_z-q_{zj})\end{array}\right.---\left(4\right)$

η learning rate of weight value 0-1, η_bThe learning rate is in the range of 0-1.

Adaptive PID control of II-2 dual weight neural network

The set current output by the adaptive PID control is as follows:

i_d(k)＝i_d(k-1)+K_pxc(1)+K_iixc(2)+K_dxc(3) (5)

wherein,

the performance indicator function is selected as follows,

$E\left(k\right)=\frac{1}{2}{\left(T_d\right(k)-T_e(k\left)\right)}^2=\frac{1}{2}\mathrm{\&Delta;}T{\left(k\right)}^2---\left(6\right)$

according to the gradient descent method, K_P,K_ii,K_dThe iterative algorithm of (1) is as follows:

$\left\{\begin{array}{c}{\mathrm{\&Delta;K}}_p\left(k\right)=-{\mathrm{\&eta;}}_{k_p}\frac{\&part;E}{\&part;k_p}={\mathrm{\&eta;}}_{k_p}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(1\right)\\ {\mathrm{\&Delta;K}}_{ii}\left(k\right)=-{\mathrm{\&eta;}}_{k_{ii}}\frac{\&part;E}{\&part;k_{ii}}={\mathrm{\&eta;}}_{k_{ii}}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(2\right)\\ {\mathrm{\&Delta;K}}_d\left(k\right)=-{\mathrm{\&eta;}}_{k_d}\frac{\&part;E}{\&part;k_d}={\mathrm{\&eta;}}_{k_d}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(3\right)\end{array}\right.---\left(7\right)$

η_kp，η_kz，η_kdthe learning coefficients are respectively proportional, integral and differential coefficients, the value range is 0-1, and the values are all 0.2.The value is identified by a dual weighted neural network.

Is i_dIncrement △ i_dIs approximately taken at the previous time of dayObtaining:

$\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d^{*}}\&ap;\frac{\&part;T_{out}}{\&part;{\mathrm{\&Delta;i}}_d}=\underset{j=1}{\overset{3}{\&Sigma;}}{\mathrm{ab}}_{1j}{v}_j{h}_j(1-h_j){w_{1j}}^m{({\mathrm{\&Delta;i}}_d-q_{1j})}^{(b_{1j}-1)}---\left(8\right)$

wherein T is_outTorque estimate output, w, for dual weight neural network identification_1jQ is the direction weight of the z ═ 1 layer neuron from the jth output layer to the hidden layer_1jAs its core weight, b_1jTo which the power is adjusted. These parameters are obtained by equation (4).

The output of PID is the set total current i_d. Three-phase control currents are respectively i_A ^*And i_B ^*And i_C ^*In the following, taking phase B as an off-phase and phase C as an on-phase as an example, the current distribution of two phase commutation is as follows:

$\left\{\begin{array}{c}{i_C}^{*}={i_d}^{*}f\left(\mathrm{\&theta;}\right)\\ {i_B}^{*}={i_d}^{*}(1-f(\mathrm{\&theta;}\left)\right)\end{array}\right.---\left(9\right)$

wherein: the distribution function:

$f\left(\mathrm{\&theta;}\right)=\frac{3}{{\mathrm{\&theta;}}_{ov}^2}\&CenterDot;{(\mathrm{\&theta;}-{\mathrm{\&theta;}}_{on})}^2-\frac{2}{{\mathrm{\&theta;}}_{ov}^3}\&CenterDot;{(\mathrm{\&theta;}-{\mathrm{\&theta;}}_{on})}^3,$

where θ is the motor rotor position angle, θ_onThe opening angle is within the range of 10-15 degrees; theta_ovThe rotor position overlap angle is in a range of 2-5 degrees.

III current feedforward compensation control based on Kalman filtering prediction

Current prediction for III-1 finite difference extended Kalman filtering

The current feedforward compensation control of the invention firstly predicts the output current of a motor in one step through finite difference extended Kalman filtering, then compensates the reference current in real time according to the difference between the predicted current value and the reference current, so that the reference current is corrected in one step in advance before the error comes, and the torque pulsation of an SRM is indirectly inhibited. Switched reluctance motor n is the nth phase voltage U in 1,2,3 phases, namely A, B, C three phases_nRotor position angle theta and compensated output three-phase current I_A、I_BAnd I_CFDEKF predicted three-phase current I for FDEKF input_A ^*、I_B ^*And I_C ^*Inputting a relative error processing module, and simultaneously controlling the output three-phase current i by using a DWNN self-adaptive PID_A ^*、i_B ^*And i_C ^*Also input into a relative error processing module which outputs a three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current i output by DWNN self-adaptive PID control_A ^*、i_B ^*And i_C ^*The sum of the three-phase reference currents i is compensated_A、i_BAnd i_C。

N-th phase current I of switched reluctance motor_nThe equation of state for the rotational speed w and the rotor position angle θ is as follows:

$\begin{array}{c}\left[\begin{array}{c}{I}_n\left(k\right)\\ w\left(k\right)\\ \mathrm{\&theta;}\left(k\right)\end{array}\right]=\left[\begin{array}{ccc}1-aaRT& -aacT& 0\\ 0& 1& 0\\ 0& T& 1\end{array}\right]\left[\begin{array}{c}{I}_n(k-1)\\ w(k-1)\\ \mathrm{\&theta;}(k-1)\end{array}\right]\\ \left[\begin{array}{cc}aa& 0\\ 0& \frac{T}{J}(T_e-T_l)\\ 0& 0\end{array}\right]\left[\begin{array}{c}{U}_n(k-1)\\ 1\end{array}\right]\end{array}---\left(9\right)$

wherein,psi is the flux linkage of the switched reluctance motor, T is a sampling period, the range is 1-3 seconds, and the value is 1 second. J is selected moment of inertia, T_eFor instantaneous torque, T_lFor load torque, U_nThe n-th phase voltage is R, and the resistance value of the winding of the switched reluctance motor is R.

The invention adopts the FDEKF with mature technology to realize the output three-phase current I after compensation_A、I_BAnd I_CTo yield I_A ^*、I_B ^*And I_C ^*. The resistance R, the parameters aa and c in equation (9) are estimated by FDEKF.

Feedforward compensation control of III-2 current

FDEKF predicted three-phase current I_A ^*And I_B ^*And I_C ^*And three-phase reference current i obtained by DWNN self-adaptive PID control_A ^*And i_B ^*And i_C ^*Obtaining the error e of the three-phase current to be compensated after the relative error processing_iA ^*And e_iB ^*And e_iC ^*Used for feedforward compensation to obtain compensated three-phase reference current i_AAnd i_BAnd i_C。

The relative error processing process is as follows:

$\left\{\begin{array}{c}{e_{i_A}}^{*}=({i_A}^{*}-{I_A}^{*})*\frac{{i_A}^{*}}{{I_A}^{*}}\\ {e_{i_B}}^{*}=({i_B}^{*}-{I_B}^{*})*\frac{{i_B}^{*}}{{I_B}^{*}}\\ {e_{i_C}}^{*}=({i_C}^{*}-{I_C}^{*})*\frac{{i_C}^{*}}{{I_C}^{*}}\end{array}\right.---\left(10\right)$

three-phase reference current i after feedforward compensation_A、i_BAnd i_CThe following were used:

$\left\{\begin{array}{c}{i}_A={i_A}^{*}+{e_{i_A}}^{*}\\ i_B={i_B}^{*}+{e_{i_B}}^{*}\\ i_C={i_C}^{*}+{e_{i_C}}^{*}\end{array}\right.---\left(11\right)$

the three-phase reference current after feedforward compensation is obtained by the formula (11) and is used as a set value of a technically mature current hysteresis controller, the output of the current hysteresis controller drives the switched reluctance motor through a power converter, and the torque pulsation of the switched reluctance motor is effectively inhibited.

The invention designs a system for reducing the torque ripple of a switched reluctance motor by current self-adaptive control according to the method for reducing the torque ripple of the switched reluctance motor by current self-adaptive control, which comprises a signal processor, an analog-to-digital conversion module, a current hysteresis controller, a power converter, a three-phase current sensor, a torque sensor and a rotor position sensor.

The three current sensors are respectively arranged on a three-phase power line of the switched reluctance motor and used for detecting each phase of current, the torque sensor is arranged on an output shaft of the switched reluctance motor and used for detecting the output torque of the motor, and the rotor position sensor is arranged on a rotor of the switched reluctance motor and used for detecting the position angle of the rotor. The position sensor, the torque sensor and the current sensor are connected with the signal processor through the analog-to-digital conversion module.

The signal processor comprises a torque deviation preprocessing module, a PID self-adaptive control module of a double-weight neural network and a current feedforward compensation module based on finite expansion Kalman filtering prediction.

Torque deviation preprocessing module for setting torque T_dAnd the instantaneous torque T detected by the torque sensor_eThe torque deviation (2) is preprocessed by a nonlinear function, and the result fal (Δ T) is used as the input of the PID adaptive control module.

Instantaneous torque T detected by torque sensor through double-weight neural network DWNN in PID adaptive control module of double-weight neural network_eInstantaneous torque T at the previous moment_{e_1}And setting the current i_dIncrement △ i_dPrevious time value ofAs an input amount, a torque T is set_dThe torque estimation output T is predicted after learning the target value of DWNN_outAnd proportional coefficient K of PID control_pIntegral coefficient K_iiAnd a differential coefficient K_d。

PID control is based on the result fal (delta T) of the torque deviation preprocessing module and the proportionality coefficient K obtained by the dual-weight neural network_pIntegral coefficient K_iiAnd a differential coefficient K_dObtaining the current set total current i through PID calculation_dObtaining control currents of each phase through current distributionAnd

based on limitationsThe current feedforward compensation module for the extended Kalman filtering prediction comprises a finite difference extended Kalman filter FDEKF and a relative error processing module. FDEKF phase voltage U_nRotor position angle theta and compensated output three-phase current I_A、I_BAnd I_CPredicted three-phase current I for input_A ^*、I_B ^*And I_C ^*Inputting the difference between the predicted current value and the PID output control current into a relative error processing module as the three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current i output by DWNN self-adaptive PID control_A ^*、i_B ^*And i_C ^*The sum of the three-phase reference currents i is compensated_A、i_BAnd i_C. The three-phase reference current is input into the current hysteresis controller as a set value, the output of the current hysteresis controller is connected into the power converter, and the output of the power converter is used for driving a three-phase power supply of the switched reluctance motor, so that the torque pulsation of the switched reluctance motor is effectively inhibited.

The system compensates the reference current in real time according to the current torque and current value and the method for inhibiting the torque ripple of the switched reluctance motor in a current self-adaptive mode, corrects the reference current in one step in advance before an error comes, and indirectly inhibits the torque ripple of the SRM.

The signal processor is connected with the display screen and displays the current, the torque information and the torque pulse rate information on line.

The signal processor is equipped with a CAN interface for connection to a CAN controller of the vehicle and other electrical communication connections.

Compared with the prior art, the method and the system for reducing the torque ripple of the switched reluctance motor by current self-adaptive control have the advantages that: 1. preprocessing the nonlinear transformation of the torque deviation of the switched reluctance motor to adapt to the nonlinear characteristic of the SRM; 2. parameters in the self-adaptive PID control are adjusted on line through an improved double-weight neural network; 3. the current feedforward compensation control is realized through the prediction current of the finite extended Kalman filtering, and the prediction capability of a control system is improved; 4. the DWNN self-adaptive PID torque control is combined with FDEKF predicted instantaneous current to realize current feedforward compensation, and the two controls simultaneously act to effectively inhibit the torque pulsation of the switched reluctance motor; by adopting the traditional open-loop torque distribution control, the pulse rate of the switched reluctance motor reaches 46.42 percent, and the pulse rate of the switched reluctance motor is restrained from being reduced to 1.65 percent by adopting the invention.

Drawings

FIG. 1 is a schematic diagram of a current adaptive adjustment structure based on a dual-weight neural network in step II-1 of the current adaptive control method for reducing the torque ripple of the switched reluctance motor;

FIG. 2 is a schematic diagram of a dual weight neural network learning structure in step II-1 of the present method for reducing the torque ripple of the switched reluctance motor through current adaptive control;

FIG. 3 is a block diagram of a current feedforward compensation based on finite extended Kalman filter prediction in step III-1 of an embodiment of the current adaptive control method for reducing torque ripple of a switched reluctance motor;

FIG. 4 is a schematic diagram of an embodiment of a system for reducing torque ripple of a switched reluctance motor by current adaptive control;

fig. 5 is a flow chart illustrating the operation of the present system for current adaptive control to reduce the torque ripple of a switched reluctance motor.

Detailed Description

Method embodiment for reducing torque ripple of switched reluctance motor through current self-adaptive control

The embodiment of the method for reducing the torque ripple of the switched reluctance motor by current self-adaptive control mainly comprises the following steps:

i, torque deviation preprocessing

The introduction of the non-linear function of this example preconditions the torque offset as follows:

$fal\left(\mathrm{\&Delta;}T\right)=\left\{\begin{array}{cc}|\mathrm{\&Delta;}T{|}^{\mathrm{\&alpha;}}sign\left(\mathrm{\&Delta;}T\right),& \left|\mathrm{\&Delta;}T\right|\&GreaterEqual;\mathrm{\&delta;}\\ \frac{\mathrm{\&Delta;}T}{{\mathrm{\&delta;}}^{(1-\mathrm{\&alpha;})}},& \left|\mathrm{\&Delta;}T\right|<\mathrm{\&delta;}\end{array}\right.---\left(1\right)$

wherein the transformation range of the feedback deviation is shown, and the value range is 0.01T_d～0.1T_d，T_dFor setting the torque, fal represents a preprocessing function, sign represents a sign function, namely, Δ T is greater than zero, and takes a value of 1, is less than zero, and takes a value of-1. Δ T is a set torque T_dAnd the measured instantaneous torque T_eα is the adjustment factor, which is 1 in this example.

II self-adaptive PID torque control based on double-weight neural network

II-1 double weight neural network

As shown in fig. 2, the dual weight neural network DWNN includes an input layer, a hidden layer, and an output layer. As shown in fig. 1, the instantaneous torque T detected by a torque sensor mounted on a switched reluctance motor_eInstantaneous torque T at the previous moment_{e_1}And setting the current i_dIncrement △ i_dPrevious time value ofAs an input amount of DWNN, torque T is set_dThe DWNN predicts a torque estimation output T after learning the target value of DWNN_outAnd proportional coefficient K of PID control_pIntegral coefficient K_iiAnd a differential coefficient K_d。

PID control is carried out according to a torque deviation preprocessing result fal (delta T) obtained in the step I and a proportionality coefficient K obtained by a double-weight neural network DWNN_pIntegral coefficient K_iiAnd a differential coefficient K_dObtaining the current set total current i through PID calculation_dObtaining control currents of each phase through current distributionAnd

the adaptive adjustment of the added power to the modified DWNN is described as follows:

$\left\{\begin{array}{c}h\left(j\right)=f\left(\underset{z=1}{\overset{3}{\&Sigma;}}{w_{zj}}^m{\left(x_z-q_{zj}\right)}^{b_{zj}}\right)\\ f\left(x\right)=\frac{1}{1+e^{-ax}}\\ T_{out}=\underset{j=1}{\overset{3}{\&Sigma;}}{v}_j*h\left(j\right)\end{array}\right.---\left(2\right)$

wherein h (j) is the hidden layer outputFunction, f (x) is the activation function, T_outFor the torque estimation output of DWNN, w_zjIs a directional weight, q_zjIs a core weight, v_jFor the output weight, a is 0.1, m is 1, b_zjFor adaptive adjustment of the power, e is the base of the natural logarithm. The hidden layer is a z layer, and z is 1,2 and 3; the output layer is j layers, j equals 1,2,3.

To obtain the parameters in equation (2), the performance indicator function is taken as:

$\left\{\begin{array}{c}{\mathrm{\&Delta;T}}_1\left(k\right)=T_e\left(k\right)-T_{out}\left(k\right)\\ J=\frac{1}{2}{\left({\mathrm{\&Delta;T}}_1\right(k\left)\right)}^2\end{array}\right.---\left(3\right)$

wherein, △ T₁Defined as the measured instantaneous torque T_eAnd double weight neural network DWNN torque estimation output T_outThe deviation therebetween.

According to the gradient descent method, the direction weight w_zjCore weight q_zjAnd the output weight v_jAnd the power b_zjThe iterative algorithm of increments of (1) is as follows:

$\left\{\begin{array}{c}{\mathrm{\&Delta;v}}_j\left(k\right)=-\mathrm{\&eta;}\frac{\&part;J}{\&part;v_j}={\mathrm{\&eta;\&Delta;T}}_1\left(k\right)h_j\left(k\right)\\ {\mathrm{\&Delta;w}}_{zj}=-\mathrm{\&eta;}\frac{\&part;J}{\&part;w_{zj}}={\mathrm{\&eta;ma\&Delta;T}}_1\left(k\right)v_j{h}_j(1-h_j){w_{zj}}^{m-1}{(x_z-q_{zj})}^{b_{zj}}\\ {\mathrm{\&Delta;q}}_{zj}=-\mathrm{\&eta;}\frac{\&part;J}{q_{zj}}={\mathrm{\&eta;ab}}_{zj}{\mathrm{\&Delta;T}}_1\left(k\right)v_j{h}_j(1-h_j){w_{zj}}^m{(x_z-q_{zj})}^{(b_{zj}-1)}\\ {\mathrm{\&Delta;b}}_{zj}=-{\mathrm{\&eta;}}_b\frac{\&part;J}{\&part;b_{zj}}={\mathrm{\&eta;}}_b{\mathrm{a\&Delta;T}}_1\left(k\right)v_j{h}_j(1-h_j){w_{zj}}^m{(x_z-q_{zj})}^{b_{zj}}\ln (x_z-q_{zj})\end{array}\right.---\left(4\right)$

η is the learning rate of weight, this example takes η ═ 0.2, η_bFor the power learning rate, this example takes η_b＝0.4。

Adaptive PID control of II-2 dual weight neural network

The set current output by the adaptive PID control is as follows:

i_d(k)＝i_d(k-1)+K_pxc(1)+K_iixc(2)+K_dxc(3) (5)

wherein,

the performance indicator function is selected as follows,

$E\left(k\right)=\frac{1}{2}{\left(T_d\right(k)-T_e(k\left)\right)}^2=\frac{1}{2}\mathrm{\&Delta;}T{\left(k\right)}^2---\left(6\right)$

according to the gradient descent method, K_P,K_ii,K_dThe iterative algorithm of (1) is as follows:

$\left\{\begin{array}{c}{\mathrm{\&Delta;K}}_p\left(k\right)=-{\mathrm{\&eta;}}_{k_p}\frac{\&part;E}{\&part;k_p}={\mathrm{\&eta;}}_{k_p}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(1\right)\\ {\mathrm{\&Delta;K}}_{ii}\left(k\right)=-{\mathrm{\&eta;}}_{k_{ii}}\frac{\&part;E}{\&part;k_{ii}}={\mathrm{\&eta;}}_{k_{ii}}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(2\right)\\ {\mathrm{\&Delta;K}}_d\left(k\right)=-{\mathrm{\&eta;}}_{k_d}\frac{\&part;E}{\&part;k_d}={\mathrm{\&eta;}}_{k_d}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(3\right)\end{array}\right.---\left(7\right)$

η_kp，η_kz，η_kdlearning coefficients, which are proportional, integral and differential coefficients, respectively, are taken to be 0.2 in this example.The value is identified by a dual weighted neural network.

Is i_dIncrement △ i_dIs approximately taken at the previous time of dayObtaining:

$\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d^{*}}\&ap;\frac{\&part;T_{out}}{\&part;{\mathrm{\&Delta;i}}_d}=\underset{j=1}{\overset{3}{\&Sigma;}}{\mathrm{ab}}_{1j}{v}_j{h}_j(1-h_j){w_{1j}}^m{({\mathrm{\&Delta;i}}_d-q_{1j})}^{(b_{1j}-1)}---\left(8\right)$

wherein T is_outTorque estimate output, w, for dual weight neural network identification_1jQ is the direction weight of the z ═ 1 layer neuron from the jth output layer to the hidden layer_1jAs its core weight, b_1jAdjust it forA power. These parameters are obtained by equation (4).

The output of PID is the set total current i_d. Three-phase control currents are respectively i_A ^*And i_B ^*And i_C ^*In the following, taking phase B as an off-phase and phase C as an on-phase as an example, the current distribution of two phase commutation is as follows:

$\left\{\begin{array}{c}{i_C}^{*}=i_d*f\left(\mathrm{\&theta;}\right)\\ {i_B}^{*}=i_d*(1-f(\mathrm{\&theta;}\left)\right)\end{array}\right.---\left(9\right)$

wherein: the distribution function:

$f\left(\mathrm{\&theta;}\right)=\frac{3}{{\mathrm{\&theta;}}_{ov}^2}\&CenterDot;{(\mathrm{\&theta;}-{\mathrm{\&theta;}}_{on})}^2-\frac{2}{{\mathrm{\&theta;}}_{ov}^3}\&CenterDot;{(\mathrm{\&theta;}-{\mathrm{\&theta;}}_{on})}^3,$

where θ is the motor rotor position angle, θ_onFor opening angle, e.g. of_on11.25 degrees, θ_ovFor rotor position overlap angle, this example theta_ov3.5 degrees.

III current feedforward compensation control based on Kalman filtering prediction

Current prediction for III-1 finite difference extended Kalman filtering

As shown in FIG. 3, the finite difference extended Kalman filter pre-current compensation of the present example comprises two parts, namely a finite difference extended Kalman filter FDEKF and a relative error processing module. Switched reluctance motor n is 1,2,3 phases, namely A, B, C three phasesThe nth phase voltage U_nRotor position angle theta and compensated output three-phase current I_A、I_BAnd I_CFDEKF predicted three-phase current I for FDEKF input_A ^*、I_B ^*And I_C ^*Inputting a relative error processing module, and simultaneously controlling the output three-phase current i by using a DWNN self-adaptive PID_A ^*、i_B ^*And i_C ^*Also input into a relative error processing module which outputs a three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current i output by DWNN self-adaptive PID control_A ^*、i_B ^*And i_C ^*The sum of the three-phase reference currents i is compensated_A、i_BAnd i_C。

N-th phase current I of switched reluctance motor_nThe equation of state for the rotational speed w and the rotor position angle θ is as follows:

$\begin{array}{c}\left[\begin{array}{c}{I}_n\left(k\right)\\ w\left(k\right)\\ \mathrm{\&theta;}\left(k\right)\end{array}\right]=\left[\begin{array}{ccc}1-aaRT& -aacT& 0\\ 0& 1& 0\\ 0& T& 1\end{array}\right]\left[\begin{array}{c}{I}_n(k-1)\\ w(k-1)\\ \mathrm{\&theta;}(k-1)\end{array}\right]\\ +\left[\begin{array}{cc}aa& 0\\ 0& \frac{T}{J}(T_e-T_l)\\ 0& 0\end{array}\right]\left[\begin{array}{c}{U}_n(k-1)\\ 1\end{array}\right]\end{array}---\left(9\right)$

wherein,psi is the flux linkage of the switched reluctance motor, T is the sampling period, and T is 1 second in the example. J is selected moment of inertia, T_eFor instantaneous torque, T_lFor load torque, U_nThe n-th phase voltage is R, and the resistance value of the winding of the switched reluctance motor is R.

The embodiment adopts FDEKF to realize the output three-phase current I after compensation_A、I_BAnd I_CTo yield I_A ^*、I_B ^*And I_C ^*. The resistance R, the parameters aa and c in equation (9) are estimated by FDEKF.

Feedforward compensation control of III-2 current

FDEKF predicted three-phase current I_A ^*And I_B ^*And I_C ^*And three-phase reference current i obtained by DWNN self-adaptive PID control_A ^*And i_B ^*And i_C ^*Obtaining the error e of the three-phase current to be compensated after the relative error processing_iA ^*And e_iB ^*And e_iC ^*Used for feedforward compensation to obtain compensated three-phase reference current i_AAnd i_BAnd i_C。

The relative error processing process is as follows:

$\left\{\begin{array}{c}{e_{i_A}}^{*}=({i_A}^{*}-{I_A}^{*})*\frac{{i_A}^{*}}{{I_A}^{*}}\\ {e_{i_B}}^{*}=({i_B}^{*}-{I_B}^{*})*\frac{{i_B}^{*}}{{I_B}^{*}}\\ {e_{i_C}}^{*}=({i_C}^{*}-{I_C}^{*})*\frac{{i_C}^{*}}{{I_C}^{*}}\end{array}\right.---\left(10\right)$

three-phase reference current i after feedforward compensation_A、i_BAnd i_CThe following were used:

$\left\{\begin{array}{c}{i}_A={i_A}^{*}+{e_{i_A}}^{*}\\ i_B={i_B}^{*}+{e_{i_B}}^{*}\\ i_C={i_C}^{*}+{e_{i_C}}^{*}\end{array}\right.---\left(11\right)$

the three-phase reference current after feedforward compensation is obtained by the formula (11) and is used as a set value of a technically mature current hysteresis controller, the output of the current hysteresis controller drives the switched reluctance motor through a power converter, and the torque pulsation of the switched reluctance motor is effectively inhibited.

System embodiment for reducing torque ripple of switched reluctance motor by current self-adaptive control

According to the embodiment of the method for reducing the torque ripple of the switched reluctance motor through the current adaptive control, an embodiment of a system for reducing the torque ripple of the switched reluctance motor through the current adaptive control is designed, and as shown in fig. 4 and 5, the system comprises a signal processor, an analog-to-digital conversion module, a current hysteresis controller, a power converter, a three-phase current sensor, a torque sensor and a rotor position sensor.

The three current sensors are respectively arranged on a three-phase power line of the switched reluctance motor and used for detecting each phase of current, the torque sensor is arranged on an output shaft of the switched reluctance motor and used for detecting the instant torque of the motor, and the rotor position sensor is arranged on a rotor of the switched reluctance motor and used for detecting the position angle of the rotor. The rotor position sensor, the torque sensor and the current sensor are connected with the signal processor through the analog-to-digital conversion module.

The signal processor comprises a torque deviation preprocessing module, a PID self-adaptive control module of a double-weight neural network and a current feedforward compensation module based on finite expansion Kalman filtering prediction.

Torque deviation preprocessing module for setting torque T_dAnd the instantaneous torque T detected by the torque sensor_eThe torque deviation (2) is preprocessed by a nonlinear function, and the result fal (Δ T) is used as the input of the PID adaptive control module.

Instantaneous torque T detected by torque sensor through double-weight neural network DWNN in PID adaptive control module of double-weight neural network_eInstantaneous torque T at the previous moment_{e_1}And setting the current i_dIncrement △ i_dPrevious time value ofAs an input amount, a torque T is set_dThe torque estimation output T is predicted after learning the target value of DWNN_outAnd proportional coefficient K of PID control_pIntegral coefficient K_iiAnd a differential coefficient K_d。

PID control is based on the result fal (delta T) of the torque deviation preprocessing module and the proportionality coefficient K obtained by the dual-weight neural network_pIntegral coefficient K_iiAnd a differential coefficient K_dObtaining the current set total current i through PID calculation_dObtaining control currents of each phase through current distributionAnd

the current feedforward compensation module based on the finite extended Kalman filter prediction comprises a finite difference extended Kalman filter FDEKF and a relative error processing module. FDEKF phase voltage U_nRotor position angle theta and compensated output three-phase current I_A、I_BAnd I_CPredicted three-phase current I for input_A ^*、I_B ^*And I_C ^*Inputting the difference between the predicted current value and the PID output control current into a relative error processing module as the three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current i output by DWNN self-adaptive PID control_A ^*、i_B ^*And i_C ^*The sum of the three-phase reference currents i is compensated_A、i_BAnd i_C. The three-phase reference current is input into the current hysteresis controller as a set value, the output of the current hysteresis controller is connected into the power converter, and the output of the power converter is used for driving a three-phase power supply of the switched reluctance motor, so that the torque pulsation of the switched reluctance motor is effectively inhibited.

The system compensates the reference current in real time according to the current torque and current value and the method for inhibiting the torque ripple of the switched reluctance motor in a current self-adaptive mode, corrects the reference current in one step in advance before an error comes, and indirectly inhibits the torque ripple of the SRM.

The signal processor is connected with the display screen and displays the current, the torque information and the torque pulse rate information on line.

The signal processor is provided with a CAN interface and is connected with a CAN controller of the automobile and other electrical communication connections.

The above-described embodiments are only specific examples for further explaining the object, technical solution and advantageous effects of the present invention in detail, and the present invention is not limited thereto. Any modification, equivalent replacement, improvement and the like made within the scope of the disclosure of the present invention are included in the protection scope of the present invention.

Claims (4)

1. A method for reducing torque ripple of a switched reluctance motor by current self-adaptive control mainly comprises the following steps:

i, torque deviation preprocessing

Introducing a nonlinear function to preprocess the torque deviation:

$fal\left(\mathrm{\&Delta;}T\right)=\left\{\begin{array}{cc}|\mathrm{\&Delta;}T{|}^{\mathrm{\&alpha;}}sign\left(\mathrm{\&Delta;}T\right),& \left|\mathrm{\&Delta;}T\right|\&GreaterEqual;\mathrm{\&delta;}\\ \frac{\mathrm{\&Delta;}T}{{\mathrm{\&delta;}}^{(1-\mathrm{\&alpha;})}},& \left|\mathrm{\&Delta;}T\right|<\mathrm{\&delta;}\end{array}\right.---\left(1\right)$

wherein the transformation range of the feedback deviation is shown, and the value range is 0.01T_d～0.1T_d，T_dFor setting the torque, fal represents the preprocessing function and sign representsA sign function, namely, delta T is greater than zero, the value is 1, is less than zero, and the value is-1; Δ T is a set torque T_dAnd the measured instantaneous torque T_eα is the adjustment coefficient range of 0-1;

II self-adaptive PID torque control based on double-weight neural network

II-1 double weight neural network

The double-weight neural network DWNN comprises an input layer, a hidden layer and an output layer; instantaneous torque T detected by torque sensor mounted on switch reluctance motor_eInstantaneous torque T at the previous moment_{e_1}And setting the current i_dIncrement △ i_dPrevious time value ofAs an input amount of DWNN, torque T is set_dThe DWNN predicts a torque estimation output T after learning the target value of DWNN_outAnd proportional coefficient K of PID control_pIntegral coefficient K_iiAnd a differential coefficient K_d；

PID control is carried out according to a torque deviation preprocessing result fal (delta T) obtained in the step I and a proportionality coefficient K obtained by a double-weight neural network DWNN_pIntegral coefficient K_iiAnd a differential coefficient K_dObtaining the current set total current i through PID calculation_dObtaining control currents of each phase through current distributionAnd

the invention adds the adaptive adjustment of the power, and the improved double-weight neural network DWNN is described as follows:

$\left\{\begin{array}{c}h\left(j\right)=f\left(\underset{z=1}{\overset{3}{\mathrm{\&Sigma;}}}{w_{zj}}^m{\left(x_z-q_{zj}\right)}^{b_{zj}}\right)\\ f\left(x\right)=\frac{1}{1+e^{-ax}}\\ T_{out}=\underset{j=1}{\overset{3}{\mathrm{\&Sigma;}}}{v}_j*h\left(j\right)\end{array}\right.---\left(2\right)$

where h (j) is the hidden layer output function, f (x) is the activation function, T_outFor the torque estimation output of DWNN, w_zjIs a directional weight, q_zjIs a core weight, v_jIs an output weight value of 0<a<1, taking a as 0.1, taking m as 1-10, taking m as 1, b_zjE is the base of the natural logarithm for adaptive adjustment of the power; the hidden layer is a z layer, and z is 1,2 and 3; the output layer is j layers, j is 1,2 and 3;

taking the performance index function as:

$\left\{\begin{array}{c}{\mathrm{\&Delta;T}}_1\left(k\right)=T_e\left(k\right)-T_{out}\left(k\right)\\ J=\frac{1}{2}{\left({\mathrm{\&Delta;T}}_1\left(k\right)\right)}^2\end{array}\right.---\left(3\right)$

wherein, △ T₁Defined as the measured instantaneous torque T_eAnd double weight neural network DWNN torque estimation output T_outThe deviation therebetween;

according to the gradient descent method, the direction weight w_zjCore weight q_zjAnd the output weight v_jAnd the power b_zjIncrement of (2)The iterative algorithm of (1) is as follows:

$\left\{\begin{array}{c}{\mathrm{\&Delta;v}}_j\left(k\right)=-\mathrm{\&eta;}\frac{\&part;J}{\&part;v_j}={\mathrm{\&eta;\&Delta;T}}_1\left(k\right)h_j\left(k\right)\\ {\mathrm{\&Delta;w}}_{zj}=-\mathrm{\&eta;}\frac{\&part;J}{\&part;w_{zj}}={\mathrm{\&eta;ma\&Delta;T}}_1\left(k\right)v_j{h}_j\left(1-h_j\right){w_{zj}}^{m-1}{\left(x_z-q_{zj}\right)}^{b_{zj}}\\ {\mathrm{\&Delta;q}}_{zj}=-\mathrm{\&eta;}\frac{\&part;J}{q_{zj}}={\mathrm{\&eta;ab}}_{zj}{\mathrm{\&Delta;T}}_1\left(k\right)v_j{h}_j\left(1-h_j\right){w_{zj}}^m{\left(x_z-q_{zj}\right)}^{\left(b_{zj}-1\right)}\\ {\mathrm{\&Delta;b}}_{zj}=-{\mathrm{\&eta;}}_b\frac{\&part;J}{\&part;b_{zj}}={\mathrm{\&eta;}}_b{\mathrm{a\&Delta;T}}_1\left(k\right)v_j{h}_j\left(1-h_j\right){w_{zj}}^m{\left(x_z-q_{zj}\right)}^{b_{zj}}\ln \left(x_z-q_{zj}\right)\end{array}\right.---\left(4\right)$

η learning rate of weight value 0-1, η_bThe value range is 0-1 for the learning rate of the next power;

adaptive PID control of II-2 dual weight neural network

The set current output by the adaptive PID control is as follows:

i_d(k)＝i_d(k-1)+K_pxc(1)+K_iixc(2)+K_dxc(3) (5)

wherein,

the performance indicator function is selected as follows,

$E\left(k\right)=\frac{1}{2}{\left(T_d\right(k)-T_e(k\left)\right)}^2=\frac{1}{2}\mathrm{\&Delta;}T{\left(k\right)}^2---\left(6\right)$

according to the gradient descent method, K_P,K_ii,K_dThe iterative algorithm of (1) is as follows:

$\left\{\begin{array}{c}{\mathrm{\&Delta;K}}_p\left(k\right)=-{\mathrm{\&eta;}}_{k_p}\frac{\&part;E}{\&part;k_p}={\mathrm{\&eta;}}_{k_p}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(1\right)\\ {\mathrm{\&Delta;K}}_{ii}\left(k\right)=-{\mathrm{\&eta;}}_{k_{ii}}\frac{\&part;E}{\&part;k_{ii}}={\mathrm{\&eta;}}_{k_{ii}}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(2\right)\\ {\mathrm{\&Delta;K}}_d\left(k\right)=-{\mathrm{\&eta;}}_{k_d}\frac{\&part;E}{\&part;k_d}={\mathrm{\&eta;}}_{k_d}\mathrm{\&Delta;}T\left(k\right)\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d}xc\left(3\right)\end{array}\right.---\left(7\right)$

η_kp，η_kz，η_kdlearning coefficients of proportional, integral and differential coefficients are respectively, the value range is 0-1,the value is obtained by identifying a double-weight neural network;

is i_dIncrement △ i_dIs approximately taken at the previous time of dayObtaining:

$\frac{\&part;T_e}{\&part;{\mathrm{\&Delta;i}}_d^{*}}\&ap;\frac{\&part;T_{out}}{\&part;{\mathrm{\&Delta;i}}_d}=\underset{j=1}{\overset{3}{\&Sigma;}}{\mathrm{ab}}_{1j}{v}_j{h}_j(1-h_j){w_{1j}}^m{({\mathrm{\&Delta;i}}_d-q_{1j})}^{(b_{1j}-1)}---\left(8\right)$

wherein T is_outTorque estimate output, w, for dual weight neural network identification_1jQ is the direction weight of the z ═ 1 layer neuron from the jth output layer to the hidden layer_1jAs its core weight, b_1jAdjusting the power thereof; these parameters are obtained by formula (4);

the output of PID is the set total current i_d(ii) a Three-phase control currents are respectively i_A ^*And i_B ^*And i_C ^*In the following, taking phase B as an off-phase and phase C as an on-phase as an example, the current distribution of two phase commutation is as follows:

$\left\{\begin{array}{c}{i_C}^{*}=i_d*f\left(\mathrm{\&theta;}\right)\\ {i_B}^{*}=i_d*(1-f(\mathrm{\&theta;}\left)\right)\end{array}\right.---\left(9\right)$

wherein: the distribution function:

$f\left(\mathrm{\&theta;}\right)=\frac{3}{{\mathrm{\&theta;}}_{ov}^2}\&CenterDot;{(\mathrm{\&theta;}-{\mathrm{\&theta;}}_{on})}^2-\frac{2}{{\mathrm{\&theta;}}_{ov}^3}\&CenterDot;{(\mathrm{\&theta;}-{\mathrm{\&theta;}}_{on})}^3,$

where θ is the motor rotor position angle, θ_onThe opening angle is in the range of 10-15 degrees theta_ovThe rotor position overlap angle is in a range of 2-5 degrees;

III current feedforward compensation control based on Kalman filtering prediction

Current prediction for III-1 finite difference extended Kalman filtering

The finite difference extended Kalman filtering pre-current compensation comprises two parts, namely a finite difference extended Kalman filter FDEKF and a relative error processing module; switched reluctance motor n is the nth phase voltage U in 1,2,3 phases, namely A, B, C three phases_nRotor position angle theta and compensated output three-phase current I_A、I_BAnd I_CFDEKF predicted three-phase current I for FDEKF input_A ^*、I_B ^*And I_C ^*Inputting a relative error processing module, and simultaneously controlling the output three-phase current i by using a DWNN self-adaptive PID_A ^*、i_B ^*And i_C ^*Also input into a relative error processing module which outputs a three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current i output by DWNN self-adaptive PID control_A ^*、i_B ^*And i_C ^*The sum of the three-phase reference currents i is compensated_A、i_BAnd i_C；

N-th phase current I of switched reluctance motor_nThe equation of state for the rotational speed w and the rotor position angle θ is as follows:

$\begin{array}{c}\left[\begin{array}{c}{I}_n\left(k\right)\\ w\left(k\right)\\ \mathrm{\&theta;}\left(k\right)\end{array}\right]=\left[\begin{array}{ccc}1-aaRT& -aacT& 0\\ 0& 1& 0\\ 0& T& 1\end{array}\right]\left[\begin{array}{c}{I}_n(k-1)\\ w(k-1)\\ \mathrm{\&theta;}(k-1)\end{array}\right]\\ +\left[\begin{array}{cc}aa& 0\\ 0& \frac{T}{J}(T_e-T_l)\\ 0& 0\end{array}\right]\left[\begin{array}{c}{U}_n(k-1)\\ 1\end{array}\right]\end{array}---\left(9\right)$

wherein,psi is switched reluctance motor flux linkage, T is sampling period, the value is 1-3 seconds, J is selected moment of inertia, and T_eFor instantaneous torque, T_lFor load torque, U_nThe voltage is the nth phase voltage, and R is the winding resistance value of the switched reluctance motor; the resistance R, the parameters aa and c are obtained by estimating through FDEKF;

feedforward compensation control of III-2 current

FDEKF predicted three-phase current I_A ^*And I_B ^*And I_C ^*And three-phase reference current i obtained by DWNN self-adaptive PID control_A ^*And i_B ^*And i_C ^*Obtaining the error e of the three-phase current to be compensated after the relative error processing_iA ^*And e_iB ^*And e_iC ^*Used for feedforward compensation to obtain compensated three-phase reference current i_AAnd i_BAnd i_C；

The relative error processing process is as follows:

$\left\{\begin{array}{c}{e_{i_A}}^{*}=({i_A}^{*}-{I_A}^{*})*\frac{{i_A}^{*}}{{I_A}^{*}}\\ {e_{i_B}}^{*}=({i_B}^{*}-{I_B}^{*})*\frac{{i_B}^{*}}{{I_B}^{*}}\\ {e_{i_C}}^{*}=({i_C}^{*}-{I_C}^{*})*\frac{{i_C}^{*}}{{I_C}^{*}}\end{array}\right.---\left(10\right)$

three-phase reference current i after feedforward compensation_A、i_BAnd i_CThe following were used:

$\left\{\begin{array}{c}{i}_A={i_A}^{*}+{e_{i_A}}^{*}\\ i_B={i_B}^{*}+{e_{i_B}}^{*}\\ i_C={i_C}^{*}+{e_{i_C}}^{*}\end{array}\right.---\left(11\right)$

the three-phase reference current after feedforward compensation is obtained by the formula (11) and is used as a set value of the current hysteresis controller, the output of the current hysteresis controller drives the switched reluctance motor through the power converter, and the torque pulsation of the switched reluctance motor is effectively inhibited.

2. The method for reducing the torque ripple of the switched reluctance motor through the adaptive current control according to claim 1, wherein the designed system for reducing the torque ripple of the switched reluctance motor through the adaptive current control comprises a signal processor, an analog-to-digital conversion module, a current hysteresis controller, a power converter, a three-phase current sensor, a torque sensor and a rotor position sensor; the method is characterized in that:

the three current sensors are respectively arranged on a three-phase power line of the switched reluctance motor and used for detecting each phase of current, the torque sensor is arranged on an output shaft of the switched reluctance motor and used for detecting the output torque of the motor, and the rotor position sensor is arranged on a rotor of the switched reluctance motor and used for detecting a rotor position angle; the position sensor, the torque sensor and the current sensor are connected with the signal processor through the analog-to-digital conversion module;

the signal processor comprises a torque deviation preprocessing module, a PID self-adaptive control module of a double-weight neural network and a current feedforward compensation module based on finite expansion Kalman filtering prediction;

torque deviation preprocessing module for setting torque T_dAnd the instantaneous torque T detected by the torque sensor_eThe torque deviation of the control module is subjected to nonlinear function preprocessing, and the result fal (delta T) is used as the input of the PID adaptive control module;

instantaneous torque T detected by torque sensor through double-weight neural network DWNN in PID adaptive control module of double-weight neural network_eInstantaneous torque T at the previous moment_{e_1}And setting the current i_dIncrement △ i_dPrevious time value ofAs an input amount, a torque T is set_dThe torque estimation output T is predicted after learning the target value of DWNN_outAnd proportional coefficient K of PID control_pIntegral coefficient K_iiAnd a differential coefficient K_d；

PID control is based on the result fal (delta T) of the torque deviation preprocessing module and the proportionality coefficient K obtained by the dual-weight neural network_pIntegral coefficient K_iiAnd a differential coefficient K_dObtaining the current set total current i through PID calculation_dObtaining control currents of each phase through current distributionAnd

the current feedforward compensation module based on the finite extended Kalman filtering prediction comprises a finite difference extended Kalman filter FDEKF and a relative error processing module; FDEKF phase voltage U_nRotor position angle theta and compensated output three-phase current I_A、I_BAnd I_CPredicted three-phase current I for input_A ^*、I_B ^*And I_C ^*Inputting the difference between the predicted current value and the PID output control current into a relative error processing module as the three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current deviation e to be compensated_iA ^*、e_iB ^*And e_iC ^*Three-phase current i output by DWNN self-adaptive PID control_A ^*、i_B ^*And i_C ^*The sum of the three-phase reference currents i is compensated_A、i_BAnd i_C(ii) a The three-phase reference current is input into the current hysteresis controller as a set value, the output of the current hysteresis controller is connected into the power converter, and the output of the power converter is used for driving a three-phase power supply of the switched reluctance motor.

3. The system for reducing the torque ripple of the switched reluctance motor in the current adaptive control mode according to claim 2, wherein:

and the signal processor is connected with the display screen and displays the current, the torque information and the torque pulse rate information on line.

4. The system for reducing the torque ripple of the switched reluctance motor in the current adaptive control mode according to claim 2, wherein:

the signal processor is provided with a CAN interface.