Three-phase hybrid rectifier model predictive current control method based on variable index sliding mode

Abstract

The three-phase hybrid rectifier model prediction current control method based on variable index sliding mode has the following steps: a fast optimal three-vector model prediction controller is used for active rectifier in the inner loop; a three-order extended state observer-model prediction controller is used for passive rectifier in the inner loop; and a variable index approaching law sliding mode control is used for voltage outer loop. In the method, the variable index approaching law sliding mode control can effectively inhibit the violent chattering caused by a large gain. The fast optimal three-vector model prediction controller can determine the optimal vector combination through only two optimizations, so that the calculation complexity is greatly reduced. The input voltage observation method based on the three-order extended state observer reduces the number of sensors used and improves the input current quality. The research of the method has important significance for improving the reliability of industrial equipment power supply.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-phase hybrid rectifier control, and particularly relates to a three-phase hybrid rectifier model predictive current control method based on variable exponential sliding mode. BACKGROUND

[0002] Three-phase hybrid rectifiers have attracted extensive attention due to their simple structure, unity power factor operation, high power density and low harmonic content. Compared with single active rectifiers, three-phase hybrid rectifiers can load more power and are more efficient, reliable and economical.

[0003] A three-phase hybrid rectifier is composed of an active rectifier module and a passive rectifier module in parallel. When the three-phase hybrid rectifier is running, the input currents of the passive rectifier module (Three-phase Diode Rectifier, TDR) and the active rectifier module (Three-phase Active Rectifier, TAR) are combined to form a sinusoidal current synchronized with the grid voltage. The load power is shared by the two rectifier modules, which not only reduces the rated load and loss of each device, but also reduces the system volume by sharing the DC side capacitor, thereby improving the efficiency and power density.

[0004] In the traditional single vector model predictive control, the current values corresponding to the six non-zero switching vectors are calculated by traversal, and the switching vector with the minimum error is selected by using a cost function to act on the system. However, the problem of this method is that the error between the selected action vector and the reference voltage vector is large, which affects the control effect. In addition, when the prediction equation is used for control in the actual control system, there is a one-step delay problem. This problem will cause errors in model prediction and affect the control effect. SUMMARY

[0005] To improve the stability and dynamic performance of the three-phase hybrid rectifier, the present application provides a three-phase hybrid rectifier model predictive current control method based on variable exponential sliding mode. The variable exponential reaching law sliding mode control can effectively suppress the severe chattering caused by large gain. The fast optimal three-vector model predictive controller can determine the optimal vector combination through only two optimizations, greatly reducing the computational complexity. The input voltage observation method based on a third-order extended state observer reduces the number of sensors used and improves the quality of the input current. The research of this method has important significance for improving the reliability of industrial equipment power supply.

[0006] The technical scheme adopted by the present application is as follows:

[0007] a three-phase hybrid rectifier model predictive current control method based on variable exponential sliding mode,

[0008] The inner loop active rectifier adopts a fast optimal three-vector model predictive controller.

[0009] The inner loop passive rectifier adopts a three-order extended state observer-model predictive controller.

[0010] The outer loop voltage adopts a variable exponential reaching law sliding mode control.

[0011] The fast optimal three-vector model predictive controller comprises the following calculation steps:

[0012] Step (1): Calculate the corresponding optimal voltage vector through the reference current, as shown in the following formula:

[0013]

[0014] In the formula: u d (k) is the optimal voltage vector of the d-axis in the dq coordinate system at time k; u q (k) is the optimal voltage vector of the q-axis at time k; i d1 (k) and i d1 (k+1) are the calculated values of the active rectifier d-axis current at time k and k+1, respectively, where i d1 (k+1) is replaced by the outer loop control signal; i q1 (k) and i q1 (k+1) are the calculated values of the active rectifier q-axis current at time k and k+1, respectively, where i q1 (k+1) is set to 0; T S is the sampling period; u gd (k) and u gq (k) are the grid voltages of the d-axis and q-axis, respectively; ω is the grid angular frequency; L1 represents the filter inductance in front of the active rectifier.

[0015] Step (2): Transform the optimal voltage vector to the αβ coordinate system, and the transformation formula is:

[0016]

[0017] In the formula: θ is the phase information of the grid voltage; u α (k) and u β (k) are the values of the voltage vectors transformed to the αβ coordinate system calculated in step (1); u d (k) and u q (k) are the optimal vectors in the dq coordinate system calculated in step (1).

[0018] Step (3): Determine the sector position of the optimal voltage vector through u α (k) and u β (k):

[0019] Let the angle between the optimal voltage vector and the alpha axis be η:

[0020]

[0021] Counting from 0° anticlockwise, each 60° is a sector, and there are a total of 6 sectors.

[0022] Step (4): Calculate the action time of the three voltage vectors V i , V j and V0 in the sector position of the optimal voltage vector;

[0023]

[0024] In the formula: S xi , S xj and S x0 are the current change rates when the voltage vectors V i , V j and V0 act, x = d, q; t i , t j and t0 are the action times of the three voltage vectors; i dr and i qr are the d-axis and q-axis current reference values, respectively; T s is the sampling period. Step (5): Switching control is performed through SVPWM modulation.

[0025] The weight function of the third-order extended state observer-model predictive controller is:

[0026]

[0027] In the formula: F is the weight function; is the predicted inductance current reference value of the passive rectifier at the k+1 moment; i D (k+1) is the inductance current prediction value of the passive rectifier at the k+1 moment; ζ is the duty cycle weight coefficient, which is 0.6 here; d(k+1) is the duty cycle at the k+1 moment; d prev (k+1) is the duty cycle reference value at the k+1 moment. The weight function determines the switching state of the passive rectifier at the next moment by calculating the difference between the inductance current and the reference value under two switching conditions of the passive rectifier, and combining the duty cycle weight coefficient.

[0028] In the third-order extended state observer-model predictive controller:

[0029] The inductance current prediction value i D (k+1) of the passive rectifier at the k+1 moment is calculated by the following formula:

[0030] i D (k+1) = TS / L 22 ×(0.5U D (k)-R 22 i D (k)-0.5(1-d)u dc (k))+i D (k);

[0031] In the formula: L 22 R is the boost inductance value of the passive rectifier; 22 U is the equivalent resistance of the boost inductor; D (k) represents the grid voltage after rectification by the uncontrolled rectifier bridge, determined by the third-order extended state observer. Get; i D (k) represents the inductor current of the passive rectifier; d represents the duty cycle of the passive rectifier; u dc (k) DC side voltage sampling value.

[0032] The equations for the third-order extended state observer are:

[0033]

[0034] In the formula: L represents the observed values ​​of the passive rectifier inductor current, the observed value of the input voltage, and the bounded disturbance, respectively; l1, l2, and l3 are the third-order extended observer gains; L 22 R is the boost inductance value of the passive rectifier; 22 The equivalent resistance of the boost inductor in a passive rectifier. dc This represents the DC-side voltage. d represents the duty cycle. i D Inductor current.

[0035] The gain of the third-order extended observer is:

[0036]

[0037] In the formula: ω is the cutoff angular frequency of the observer, ω=628rad / s.

[0038] In the voltage outer loop variable exponential approach law sliding mode control:

[0039] The sliding mode controller is designed as follows:

[0040] i = u dc / R L +C(-k0S-((2 |S| / (2 |S| +1)-0.5)ε0+λρ)sign(S)-e) / λ

[0041] Where: i represents the output signal of the sliding mode controller. dcis the DC side voltage sampling value. R is the load resistance value. C is the DC side capacitance value. k0 is the exponential gain, and ε0 is the switching gain. λ is the control gain. 0<ρ<1 is the error boundary. e is the difference between the DC side voltage and the reference voltage. L is the DC side voltage sampling value. R is the load resistance value. C is the DC side capacitance value. k0 is the exponential gain, and ε0 is the switching gain. λ is the control gain. 0<ρ<1 is the error boundary. e is the difference between the DC side voltage and the reference voltage.

[0042] sign(S) is a sign function, and S represents a sliding mode surface. When S≥0, sign(S)=1, and when S≤0, sign(S)=0. The output of the sliding mode controller is multiplied by control coefficients k1 and k2 as the reference value of the current inner loop.

[0043] The variable exponential reaching law is:

[0044]

[0045] wherein: represents the derivative of S with respect to time; k0 and ε0 both represent control gains, and k0>0 and ε0>0; e is the difference between the DC bus voltage and the reference voltage; S is the sliding mode surface; sign(S) is a sign function, and the variable in the parentheses is greater than 0, sign(S)=1; the variable in the parentheses is less than 0, sign(S)=-1; and the variable in the parentheses is equal to 0, sign(S)=0.

[0046] In the variable exponential reaching law sliding mode control of the voltage outer loop:

[0047] When the difference between the load voltage and the reference voltage increases, the main role of the parameter k0 is to make the DC load voltage quickly approach the reference value; when reaching the vicinity of the reference value, k0S is approximately 0, at which time the part of the variable exponential reaching law plays a leading role.

[0048] Since the 2 |S| / (2 |S| +1)-0.5 in the second part of the variable exponential reaching law is variable, when the error between the DC load voltage and the reference voltage is zero, the part of the variable exponential reaching law will also be equal to zero. Therefore, the variable exponential reaching law can reduce the chattering existing in the output signal of the voltage outer loop sliding mode controller, especially when the gain ε0 is large. The three-phase hybrid rectifier control method based on the current model prediction-sliding mode control has the following technical effects:

[0049] 1) The three-phase hybrid rectifier control method can reduce the change of the DC voltage to a certain extent when the load suddenly changes, has a fast response speed, and makes the DC side output voltage quickly recover to a stable state.

[0050] 2) The three-phase hybrid rectifier control method can effectively reduce the THD of the total input current on the AC side and make the system stably operate under a unit power factor.

[0051] 3) The fast optimal three-vector model predictive control in the application can effectively reduce the calculation loss.

[0052] 4) The third-order extended state observer in the application can reduce the use of sensors and effectively reduce the cost. BRIEF DESCRIPTION OF DRAWINGS

[0053] The application will be further described below in combination with the drawings and examples;

[0054] Figure 1 It is a topology structure diagram of a three-phase hybrid rectifier.

[0055] Figure 2 It is a process diagram of fast optimal three-vector model predictive control of an active rectifier.

[0056] Figure 3 It is a principle block diagram of a sliding mode controller.

[0057] Figure 4 It is a model predictive current control block diagram of a three-phase hybrid rectifier based on variable exponential sliding mode.

[0058] Figure 5 It is a schematic diagram of a running power factor of an example implementation of the application.

[0059] Figure 6 It is a schematic diagram of an input current waveform of an example implementation of the application.

[0060] Figure 7 It is a schematic diagram of a DC side load mutation voltage waveform of an example implementation of the application. DETAILED DESCRIPTION

[0061] Figure 1 It is a topology structure diagram of a three-phase hybrid rectifier. Wherein, u ga , u gb , u gc is a three-phase input voltage of a power grid side; i a , i b , i c is a three-phase input current of the power grid side; L1 is a filter inductance in front of an active rectifier, R1 is an equivalent resistance of the filter inductance; S j (j=1, 2,..., 6) is a switch tube of the active rectifier; L 21 is a filter inductance in front of a passive rectifier, R 21 is an equivalent resistance of the filter inductance; L 22 is a boost inductance of the passive rectifier, R 22 is an equivalent resistance of the boost inductance; S7 is a switch tube of the passive rectifier; u dc is a DC side load voltage; i dc is a DC side load current; RL C is the load resistor; C is the DC-side filter capacitor.

[0062] The active rectifier is connected to the power grid through three inductors. According to Kirchhoff's voltage law, the mathematical model of the active rectifier can be obtained by writing the circuit equations:

[0063]

[0064] Transforming equation (1) into a two-phase rotating coordinate system yields:

[0065]

[0066] In equation (2): i d1 i q1 i a1 i b1 i c1 The input current of the active rectifier on the dq0 axis; u gd u gq u ga u gb u gc The grid voltage on the dq0 axis; u d =S d u dc u q =S q u dc ω is the angular frequency of the power grid.

[0067] The mathematical model of a passive rectifier in a three-phase stationary coordinate system is as follows:

[0068] L 22 di D / dt=0.5U D -R 22 i D -0.5(1-d)u dc (3);

[0069] In equation (3), U D This is the equivalent DC power supply output of the diode rectifier bridge; i D d represents the output current of the uncontrolled rectifier bridge; d represents the duty cycle.

[0070] The DC side shares a common capacitor, which can be obtained from Kirchhoff's current law.

[0071] i = i1 + i2 = Cdu dc / dt+u dc / R L (4);

[0072] Figure 2The process chart of the fast optimal three-vector model predictive control of the active rectifier is shown in Figure 1. After the discretization of formula (2) and formula (3), the following formula is obtained:

[0073]

[0074] i D (k+1) = T S / L 22 × (0.5U D (k) - R 22 i D (k) - 0.5 (1 - d) u dc (k) + i D (k) (6).

[0075] In the formula, T s is the sampling time. In addition, the sampling period is much smaller than the grid voltage period, so the discrete values of the grid voltage in adjacent periods are approximately equal;

[0076] u g (k+1) = u g (k) (7).

[0077] In formula (5), i d1 (k) and i q1 (k) are the sampling values of the input current of the active rectifier after the dq transformation. In this process, the main function of the active rectifier is to control the sinusoidalization and in-phase of the input current of the active rectifier. However, the input current of the passive rectifier presents a rectangular wave. At this time, the total input current can be represented by formula (8):

[0078] i x = i x1 + i x2 (8).

[0079] The total input current will be obviously distorted after the superposition of the input currents of the two parallel rectifiers. In order to solve this problem, the current sampling value of the active rectifier is adjusted to the total input current in the present application, so that the control target of the active rectifier is changed to realize the sinusoidalization of the total input current. That is, the input current of the active rectifier is superimposed with the input current of the passive rectifier, and the total input current can still maintain the ideal sinusoidal waveform.

[0080] For the active rectifier, the three-vector model predictive control can significantly reduce the error between the output control voltage vector and the reference voltage vector in one control period, thereby improving the control accuracy and performance of the system.

[0081] Moving u d (k) and u q (k) in formula (5) to the left side of the equation is:

[0082]

[0083] In formula (9), u d (k) is the optimal voltage vector of the d-axis in the dq coordinate system at time k, u q (k) is the optimal voltage vector of the q-axis at time k. d1 (k) and i d1 (k+1) are the calculated values of the active rectifier d-axis current at times k and k+1, respectively, where i d1 (k+1) is replaced by the outer loop control signal. Similarly, i q1 (k) and i q1 (k+1) are the calculated values of the active rectifier q-axis current at times k and k+1, respectively, where i q1 (k+1) is set to 0. S T is the sampling period, u gd (k) and u gq (k) are the grid voltages of the d-axis and q-axis, respectively, and ω is the grid angular frequency.

[0084] The optimal voltage vector u α (k) and u β (k) can be obtained by transforming formula (9) to the αβ coordinate system, and the sector in which the optimal vector is located can be obtained through a comparison, and two non-zero voltage vectors can be obtained. When three vectors act, the action time of the three vectors needs to be solved, and the two constant vectors are set as V i and V j , and the zero vector is V0, and the action times are t i , t i , and t0. The predicted current formula is:

[0085]

[0086] In formula (10), S xi , S xj , and S x0 (x=d, q) are the current change rates when the vectors V i , V j , and V0 act. The action times of the three vectors can be obtained by substituting the three vectors into formula (2) and simultaneously solving formula (10).

[0087]

[0088] Finally, switching control is performed in combination with modulation, which can effectively reduce the calculation time.

[0089] In formula (6), U D(k) is a three-phase uncontrolled rectifier bridge rectified voltage, so the information of input voltage is needed in the predictive model of passive rectifier, therefore, a third-order extended state observer is designed to observe the input voltage, reducing the use of additional sensors. The equation of the third-order extended state observer is:

[0090]

[0091] wherein: are the observed values of passive rectifier inductor current, input voltage and bounded disturbance respectively. l1, l2, l3 are observer gains, and the observer gains are:

[0092]

[0093] wherein: ω is the observer cut-off angular frequency.

[0094] After discretization of equation (12), the inductor current prediction formula after the observer can be obtained from equation (6) and equation (12).

[0095]

[0096] The weight function of passive rectifier is:

[0097]

[0098] wherein: ζ is the duty ratio weight coefficient, which is 0.6 here. D is the predicted inductor current of passive rectifier.

[0099] In addition, when using the prediction equation for control in the actual control system, a one-step delay problem will occur. This problem will cause errors in model prediction and affect the control effect. In order to solve this delay problem, the k+2 prediction value obtained through the model is usually used instead of the k+1 prediction value.

[0100] Fast optimal three-vector model predictive control reduces the number of optimization times of the system by directly solving u α (k) and u β (k), and only two optimization times are needed to obtain the optimal vector. Moreover, the three vectors acting on the control can synthesize a vector in any direction, which is superior to single-vector model predictive control in optimization times and control effect of acting voltage vector, greatly reducing the calculation loss. The third-order extended state observer model predictive control can reduce the use of sensors and improve the quality of input current quality to a certain extent.

[0101] Figure 3 is the principle block diagram of the sliding mode controller. The control variable is defined as the DC bus voltage error, and the expression of the sliding surface is as follows:

[0102] e = u dc -U dc.ref (16);

[0103] S V = a1e + a2∫edt (17);

[0104] In the formula: U dc.ref is the DC side voltage reference value. When reaching the sliding mode surface, we can get:

[0105] S V = λe + ∫edt = 0 (18);

[0106] In the formula: λ = a1 / a2, and λ is a normal number, which can adjust the control performance such as reaching steady state time, steady state error and overshoot.

[0107] Taking the derivative of formula (17) can get:

[0108]

[0109] In the formula: λ = a1 / a2;

[0110] From formula (4), we can get:

[0111]

[0112] In the formula: d represents the uncertain disturbance in u dc . Suppose the boundary of d is |d|≤ρ<1, and ρ is a given normal number.

[0113] In order to reduce the chattering near the sliding mode surface, the variable exponential reaching law is used to design the sliding mode controller, which can get:

[0114]

[0115] When the system is far away from the sliding mode surface, the main role of the parameter k0 is to make the system quickly approach the reference value. When reaching the vicinity of the reference value, k0S is approximately 0, at this time the second term plays a leading role. Since the parameter of the second term is variable, when the error of the system is zero, the parameter of the second term will also equal to zero. Therefore, the variable exponential reaching law can prevent the system from chattering near the sliding surface.

[0116] Therefore, the sliding mode controller can be set as:

[0117] i = u dc / R L +C(-k0S-((2 |S| / (2 |S| +1)-0.5)ε0+λρ)sign(S)-e) / λ (22);

[0118] In the formula, ε0 represents control gain, and ε0>0; sign(S) is a sign function, sign(S)=1 when S is greater than or equal to 0, and sign(S)=0 when S is less than or equal to 0.

[0119] Figure 4 The application is a model predictive current control diagram of a three-phase hybrid rectifier based on a variable exponential sliding mode. The main circuit parameters for testing the application are as follows: the filter inductance in front of the active rectifier is 5 mH; the filter inductance in front of the passive rectifier is 0.4 mH, the DC side boost inductance is 20 mH; the DC side reference voltage is 700 V, the capacitance is 1100 mu F, and the load is 120 ohms. The equivalent resistance values of the inductors are all small. The power distribution of the application is modified by modifying the control coefficients k1 and k2 multiplied by the output of the outer loop sliding mode control.

[0120] Figure 5 The application is a schematic diagram of the operating power factor of an example embodiment of the application. The method of the application can enable the three-phase hybrid rectifier to continuously operate at a power factor cos phi greater than or equal to 0.998.

[0121] Figure 6 The application is a schematic diagram of an input current waveform of an example embodiment of the application. a The total input current is composed of the input currents of the active rectifier and the passive rectifier. a1 The input current of the active rectifier is i a2 The input current of the passive rectifier is i. The filter inductance in series in front of the passive rectifier can effectively suppress the periodic current spikes of the input current of the passive rectifier.

[0122] Figure 7 The application is a schematic diagram of the DC side load mutation voltage waveform of an example embodiment of the application. At t=0.3 s, the load changes from 120 ohms at the beginning of the simulation to 60 ohms, at which time the bus voltage drops by about 12 V, and it takes 0.6 s to recover to a stable state. At t=0.5 s, the load changes from 60 ohms to 120 ohms, and the voltage rises by 12 V, and it takes 0.6 s to recover to a stable state.

[0123] The application is a model predictive current control method for a three-phase hybrid rectifier based on a variable exponential sliding mode. Model predictive control can effectively improve the control accuracy, stability and robustness of the system by predicting the future behavior of the system and optimizing the control strategy. In the rectifier, model prediction can accurately regulate the output voltage and current, reduce fluctuations and ensure stable operation of the system under load changes or input fluctuations. Model prediction can also handle voltage and current constraints to ensure that the equipment operates within a safe range, while coordinating multi-variable control to optimize the overall performance of the rectifier. In addition, model prediction can dynamically adapt to load changes and external disturbances, improving the adaptability and anti-interference ability of the system, thereby significantly improving the efficiency and reliability of the rectifier.

Claims

1. A variable exponential sliding mode based model predictive current control method for three-phase hybrid rectifier, characterized in that: The inner loop active rectifier adopts a fast optimal three-vector model predictive controller; The inner loop passive rectifier adopts a third-order extended state observer-model predictive controller; The voltage outer loop adopts a variable exponential reaching law sliding mode control; The fast optimal three-vector model predictive controller comprises the following calculation steps: Step (1): Calculate the corresponding optimal voltage vector through the reference current, as shown in the following formula: ； wherein: is the optimal voltage vector of d-axis in dq coordinate system at k moment; is the optimal voltage vector of q-axis at k moment; and are respectively k and k+1 are respectively and are respectively k and k+1 are respectively is the sampling period; and are respectively grid voltages of d-axis and q-axis; is the grid angular frequency; represents the filter inductance in front of the active rectifier. Step (2): Transform the optimal voltage vector to the coordinate system, the transformation formula is: ； where: is the phase information of the grid voltage; and is the value of the voltage vector calculated in step (1) transformed to the coordinate system; and is the optimal vector in the dq coordinate system calculated in step (1). Step (3): determining the sector position where the optimal voltage vector is located by and judging the sector position where the optimal voltage vector is located Let the optimal voltage vector and the angle between the optimal voltage vector and the axis of the stator is : ； Step (4): calculating three voltage vectors of sector position where optimal voltage vector is located , and acting time; ； wherein: is the voltage vector , and is the rate of change of current when the voltage vector ; , and are the time of action of the three voltage vectors, respectively; and are the d-axis and q-axis current references, respectively; is the sampling period; Step (5): Perform switching control through SVPWM modulation.

2. The model predictive current control method for three-phase hybrid rectifier based on variable exponential sliding mode according to claim 1, characterized in that: The weight function of the third-order extended state observer-model predictive controller is: ； In the formula: is a weight function; is a predicted inductance current reference value of the passive rectifier at the moment k+ 1; is a predicted inductance current value of the passive rectifier at the moment k+ 1; is a weight coefficient of the duty cycle; is a duty cycle at the moment k+ 1; is a duty cycle reference value at the moment k+ 1.

3. The model predictive current control method for three-phase hybrid rectifier based on variable exponential sliding mode according to claim 1, characterized in that: The equation of the third-order extended state observer is: ； wherein: , , are the observed value of the passive rectifier inductor current, the observed value of the input voltage and the bounded disturbance, respectively; , , is the third order extended observer gain; is the boost inductor value of the passive rectifier; is the equivalent resistance of the passive rectifier boost inductor; denotes the DC side voltage; denotes the duty cycle; inductor current; In the third-order extended state observer-model predictive controller: Passive rectifier k+1th moment inductor current prediction value Is calculated from the following equation: ； In the formula: is the boost inductance value of the passive rectifier; is the equivalent resistance of the boost inductance; is the grid voltage rectified by the uncontrolled rectifier bridge, obtained from in the third-order extended state observer; is the inductance current of the passive rectifier; is the duty cycle of the passive rectifier; is the DC side voltage sampling value.

4. The model predictive current control method for three-phase hybrid rectifier based on variable exponential sliding mode according to claim 3, characterized in that: The third-order extended observer gain is: ； In the formulae: is the observer cut-off corner frequency.

5. The model predictive current control method of three-phase hybrid rectifier based on variable exponential sliding mode according to claim 1, characterized in that: In the variable exponential reaching law sliding mode control of the voltage outer loop: The sliding mode controller is designed as: ； wherein: represents an output signal of the sliding mode controller; is a DC side voltage sampling value; is a load resistance value; is a DC side capacitance value; is an exponential gain, is a switching gain; is a control gain; is an error boundary; is a difference between the DC side voltage and a reference voltage; is a sign function, represents a sliding surface, when , when , ; The outputs of the sliding mode controllers are multiplied by control coefficients and as reference values for the current inner loop.

6. The model predictive current control method for three-phase hybrid rectifiers based on variable exponent sliding mode according to claim 5, characterized in that: The variable exponential reaching law is: ； wherein: represents the derivative of ; , both represent control gains, , ; is the difference between the DC bus voltage and the reference voltage; is the sliding mode surface.

7. The model predictive current control method for three-phase hybrid rectifier based on variable exponent sliding mode according to claim 6, characterized in that: In the variable exponential reaching law sliding mode control of the voltage outer loop: When the difference between the load voltage and the reference voltage increases, the parameter acts to make the DC load voltage quickly approach the reference value; when it reaches the vicinity of the reference value, is approximately 0, at which time the of the variable exponential approach law plays a leading role; Since the second part of the variable exponent reaching law is variable, when the error between the DC load voltage and the reference voltage is zero, the second part of the variable exponent reaching law is also equal to zero; therefore, the variable exponent reaching law can reduce the chattering existing in the output signal of the voltage outer loop sliding mode controller.

Images