阅读说明：本技术 一种阻抗源逆变器的鲁棒预测控制方法 (Robust prediction control method of impedance source inverter ) 是由 谭天诚 刘平 张桂斌 谭天利 吴刚 吴文昊 于 2020-12-28 设计创作，主要内容包括：本发明提供了一种阻抗源逆变器的鲁棒预测控制方法,该方法中对电容电压和输出电流使用两个单独的代价函数,对其代价函数值进行比较,分别选取使得电容电压、输出电流代价函数值较小的三个电压矢量,将两者公共的电压矢量作为最优电压矢量,并用于控制逆变器。在参数鲁棒性方面,根据逆变器的状态量建立参数观测器模型,在线修正预测模型中的电感和电容值,增强控制器的鲁棒性能。最终实现权重因子优化的阻抗源逆变器鲁棒预测控制,在不增加系统开关频率的同时消除代价函数中权重系数的设计,降低代价函数计算的复杂度,且在模型参数变化时具有较强的鲁棒性。(The invention provides a robust prediction control method of an impedance source inverter, which uses two independent cost functions for capacitor voltage and output current, compares the cost function values of the two independent cost functions, respectively selects three voltage vectors which enable the capacitor voltage and the output current to have smaller cost function values, and takes the voltage vector which is common to the two voltage vectors as an optimal voltage vector to be used for controlling the inverter. In the aspect of parameter robustness, a parameter observer model is established according to the state quantity of the inverter, the inductance and the capacitance in the prediction model are corrected on line, and the robust performance of the controller is enhanced. Finally, the robust predictive control of the impedance source inverter optimized by the weight factor is realized, the design of the weight coefficient in the cost function is eliminated while the switching frequency of the system is not increased, the complexity of the calculation of the cost function is reduced, and the robust predictive control method has stronger robustness when the model parameters are changed.)

1. A robust predictive control method of an impedance source inverter is characterized by specifically comprising the following steps:

step 1: sampling the output current at time k, the capacitor voltage and the inductor current:

at the time k, three-phase output currents i of qZSI are respectively measured_a，i_bAnd i_cVoltage v of capacitor_c1And the inductor current i_L1Sampling, and obtaining current components i of the output current on alpha and beta axes in a static coordinate system_αAnd i_βRecording the output current i_o＝i_α+ji_β：

Step 2: calculating the inductive current and the corresponding cost function in the direct connection state:

discretizing the formula (2) by a first-order Euler discretization method to obtain the predicted value of the inductive current at the moment of k +1

In the formula i_L1(k +1) is the predicted value of the inductive current in the (k +1) th sampling period, i_L1(k) Is the inductor current of the kth sampling period, v_c1(k) Is the capacitor voltage of the kth sampling period, T_sIs a sampling period, L₁Inductance of qZSI, R_L1Is the internal resistance of the input inductor.

Cost function g of the inductor current in the shoot-through state_{iL_ST}Comprises the following steps:

g_{iL_ST}＝|i_{L1_ref}-i_{L1_ST}(k+1)| (4)

in the formula i_{L1_ref}Is an inductor current reference value, i_{L1_ST}(k +1) is an inductive current predicted value of the (k +1) th sampling period in a direct-connection state;

and step 3: calculating the inductive current and the corresponding cost function in the non-direct-through state:

discretizing the formula (5) by a first-order Euler discretization method to obtain the predicted value of the inductive current at the moment of k +1

Cost function g of inductive current in non-through state_{iL_ns}Can be designed as follows:

g_{iL_ns}＝|i_{L1_ref}-i_{L1_ns}(k+1)| (7)

in the formula i_{L1_ns}(k +1) is an inductive current predicted value of the (k +1) th sampling period in the non-through state;

when g is_{iL_ST}Less than g_{iL_ns}If the state is judged to be a through state, the direct output is performedA straight-through vector; when g is_{iL_ST}Greater than g_{iL_ns}Judging the capacitor to be in a non-straight-through state, and calculating a cost function of the capacitor voltage and the output current in the next step;

and 4, step 4: calculating cost function of capacitor voltage and output current

Discretizing the formula (8) by a first-order Euler discretization method to obtain a predicted value of the capacitance voltage at the k +1 moment in a non-through state as

In the formula i_inv(k +1) is the inverter current;

the phase voltage output equation of the three-phase inverter is

Discretizing the formula (10) by a first-order Euler discretization method to obtain the final product

In the formula: r and L are respectively a load resistance and an inductance, i_o(k +1) is the predicted value of the output current in the (k +1) th sampling period; i.e. i_o(k) The sampling value of the output current in the kth sampling period can be obtained by the formula (1); v_x(k) The output voltage value of the kth sampling period;

two separate cost functions are used for the capacitor voltage and the output current, and the cost functions of the capacitor voltage and the output current are respectively designed as follows:

g_vc1(i)＝|v_{c1_ref}-v_c1(k+1)| (12)

g_io(i)＝|i_{o_ref}-i_o(k+1)| (13)

in the formula, v_{c1_ref}As a reference value of the capacitor voltage, i_{o_ref}For outputting voltage reference value, 7 voltage vectors V output by the inverter_iSubstitution of the formulae (10), (11) in this order and obtaining 7 g according to formulae (12) and (13)_vc1(i) And g_io(i) Wherein i is 1, 2, …, 7;

and 5: inductance capacitance observation

Correcting the established qZSI prediction model through inductance-capacitance observation, and assuming that the inductance value of the qZSI in the prediction model is L_mA capacitance value of C_mAt this time, the prediction expressions of the inductive current and the capacitor voltage in the non-through state are

The prediction expressions of the inductive current and the capacitor voltage in the direct-through state are

When the load inductance value in the prediction model is L_pAt this time, the predicted expression of the output current is

And (3) deriving expressions of the inductance equivalent value of the quasi-Z source network in the non-through state and the through state according to the expressions (3), (6), (15) and (17):

deriving an expression of the capacitance equivalent value of the quasi-Z source network in the non-through state according to the expressions (9) and (16):

from equations (11) and (19), an expression for the load inductance equivalent is derived:

correcting the proposed prediction model according to the quasi-Z source network inductance, the capacitance and the load inductance observed in the formulas (20) to (22);

step 6: obtaining an optimal voltage vector:

seven errors of the capacitor voltage and the output current can be calculated according to the formulas (12) and (13), i takes a value of 0-7, and the errors of the two are arranged in ascending order

r₁ ⁱ＝rank[g_vc1(i)] (23)

r₂ ⁱ＝rank[g_io(i)] (24)

As can be seen from the equations (23) and (24), the smaller the error between the capacitor voltage and the output current, the higher the corresponding switch state rank, and the final selection is made so that g is obtained according to the switch states arranged by the equations (23) and (24)_vc1(i) And g_io(i) The first 3 voltage vectors corresponding to the smaller ones respectively, and g is selected from them_vc1(i) And g_io(i) The common voltage vector of (a) is used as an optimal voltage vector, thereby obtaining a set of optimal voltage vectors.

2. According to claimThe robust predictive control method of an impedance source inverter as claimed in claim 1, wherein said i_{L1_ref}Is calculated by the formula i_{L1_ref}＝P_{out_ref}/v_inIn which P is_{out_ref}For the output power reference value, v_inIs a dc input voltage.

3. The robust predictive control method of an impedance source inverter as claimed in claim 1, wherein the selection principle of the optimal voltage vector is: when cost function g_vc1(i) And g_io(i) When more than one common voltage vector exists in the three selected voltage vectors, the voltage vector with the top rank is selected as the optimal voltage vector; when cost function g_vc1(i) And g_io(i) In the absence of a common voltage vector, g is chosen such that_vc1(i) The minimum voltage vector is used as the optimal voltage vector, so that the robustness of the system is improved.

4. A robust predictive control system for an impedance source inverter, comprising:

at least one processor and at least one memory communicatively coupled to the processor, wherein: the memory stores program instructions executable by the processor, the processor invoking the program instructions to enable execution of the robust predictive control method of an impedance source inverter of any of claims 1 to 3.

5. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the robust predictive control method of an impedance source inverter according to any one of claims 1 to 3.

Technical Field

The invention belongs to the field of improved control methods of quasi-Z source inverters, and particularly relates to a robust prediction control method of an impedance source inverter.

Background

Compared with the traditional voltage source inverter, the impedance source inverter has the advantages of boosting and reducing voltage, improving reliability, generating continuous input current, having smaller size of a passive device and the like, and has wide prospect in the fields of motor driving, renewable energy systems and battery energy storage. Commonly used impedance source inverter control methods are mainly based on conventional Proportional Integral (PI) control. However, this method not only requires a complicated pulse width modulator, but also its control performance is greatly affected by the PI regulator. With the rapid increase of the computation performance of the digital signal processor, in recent years, many new control theories are also gradually applied to the control of the impedance source inverter, such as fuzzy control, neural network control, Model Predictive Control (MPC), and the like. The finite control set model predictive control has the advantages of simple principle, flexible control, realization of multi-target optimization control, no need of Pulse Width Modulation (PWM) and the like, and is widely researched and applied.

The cost function of the finite control set model predictive control is a key link for realizing the selection of the optimal voltage vector, and must be constructed according to a control target, a predictive variable based on a system model and a reference variable. The cost function may include a plurality of control objectives, variables, and constraints, and may enable control to be achieved simultaneously. Each term in the cost function needs to be assigned a weight coefficient, which has the effect of adjusting the importance between the term and other control objectives or determining the weight relationship in the cost function. Correctly designed and reasonable weight coefficients are very important for selecting voltage vectors, realizing control targets and ensuring good dynamic and steady-state performance of the system. The model predictive control of the impedance source inverter at least comprises more than three control targets, the control targets play roles independently and are restricted with each other, and the comprehensive coordination of the relationship is the key for realizing good model predictive control, so that corresponding weight coefficients are required to be designed to adjust the weight occupied by the control targets in a cost function, but the design of the weight coefficients is mainly trial and error at present, and a complete theoretical system and a general effective solution are not provided. In addition, the existing model prediction control technology is too dependent on an accurate mathematical model of a controlled object, and once the model parameters are mismatched due to external disturbance, the control effect is adversely affected.

However, the following drawbacks exist in the current control method of the impedance source inverter:

1. the conventional Proportional Integral (PI) based PWM control method requires a complicated pulse width modulation unit, and is usually used in combination with a PI controller, the control performance of which is greatly affected by the PI regulator, and PI parameter design is required.

2. The dynamic response of the fuzzy control and the neural network control is fast, but the structure is complex, and the design difficulty is higher compared with the model prediction control.

3. The invention provides a permanent magnet synchronous motor non-weight coefficient prediction torque control method based on per unit, 201910399476.X, which provides a non-weight coefficient prediction torque control method taking torque and flux linkage as objective functions. However, the invention is only suitable for the permanent magnet synchronous motor driven by the traditional voltage source inverter and is not suitable for the quasi-Z source inverter.

4. Most of the existing impedance source inverter model prediction control methods need to design a plurality of weight coefficients, and the traditional method determines fixed weight distribution on the basis of a large number of trial and error or experiments, so that the global optimization of dynamic performance, steady-state performance and switching performance cannot be realized, and the robustness is poor.

5. The document "a powerfull finish Control Set-Model Predictive Control for Quasi Z-Source Inverter" uses an inductive current sub-cost function to judge whether the state is a direct-through state, and proposes a Quasi-Z Source Inverter Model Predictive Control method for simplifying the cost function, but the method still has the difficulty in designing the weight coefficient in the traditional cost function. In addition, the method does not consider the influence of parameter mismatch on the system performance, and the system robustness is poor.

6. The invention patent CN202010572451.8 "a model predictive control method for quasi Z-source inverter without weight coefficient" adopts a cascade model predictive control method for quasi Z-source inverter to determine the priority of the control object to eliminate the weight coefficient. This patent relates to 2 methods: a method of calculating the capacitor voltage and then the output current, denoted as S-MPC 1; a method of calculating the output current first and then the capacitor voltage is shown as S-MPC 2. In the S-MPC1 method, in the non-cut-through state, only two better voltage vectors are selected from the capacitance voltage cost function, and then the two voltage vectors are substituted into the output current cost function to select an optimal voltage vector. The number of voltage vectors selected from the capacitance-voltage cost function by the S-MPC1 method is small, the optimal voltage vector is only selected from two voltage vectors, so that a capacitance voltage item and an output current item are not effectively optimized, and the S-MPC2 method is the same. In addition, the invention finds that the S-MPC1 method and the S-MPC2 do not actually consider the influence of parameter mismatch on the system performance, and once external disturbance occurs, each variable of an actual system may change in the operation process, so that the patent cascade model predicts the failure of the controller, and the robustness is poor.

Therefore, it is urgently needed to design a robust predictive control method of the impedance source inverter to simultaneously solve the problems of poor robustness and large data calculation amount.

Disclosure of Invention

Technical problem to be solved

Based on the method, two cost functions of the capacitor voltage and the output current are established, seven different voltage vectors are used for evaluating the respective cost functions, then three voltage vectors which enable the cost functions of the capacitor voltage and the output current to be smaller are respectively selected, and finally the voltage vector which is common to the capacitor voltage and the output current is selected as the optimal voltage vector, so that the weight coefficient is eliminated. In the aspect of parameter robustness, a parameter observer is established to carry out online observation on the inductance and the capacitance, so that the robustness of the system is improved. The improved method can reduce the complexity of cost function calculation and has strong adaptability when the model parameters change.

(II) technical scheme

According to an aspect of the present invention, a robust predictive control method for an impedance source inverter is provided, where the control method specifically includes:

step 1: sampling the output current at time k, the capacitor voltage and the inductor current:

at the time k, three-phase output currents i of qZSI are respectively measured_a，i_bAnd i_cVoltage v of capacitor_c1And the inductor current i_L1Sampling, and obtaining current components i of the output current on alpha and beta axes in a static coordinate system_αAnd i_βRecording the output current i_o＝i_α+ji_β：

Step 2: calculating the inductive current and the corresponding cost function in the direct connection state:

discretizing the formula (2) by a first-order Euler discretization method to obtain the predicted value of the inductive current at the moment of k +1

In the formula i_L1(k +1) is the predicted value of the inductive current in the (k +1) th sampling period, i_L1(k) Is the inductor current of the kth sampling period, v_c1(k) Is the capacitor voltage of the kth sampling period, T_sIs a sampling period, L₁Inductance of qZSI, R_L1Is the internal resistance of the input inductor.

Cost function g of the inductor current in the shoot-through state_{iL_ST}Comprises the following steps:

g_{iL_ST}＝|i_{L1_ref}-i_{L1_ST}(k+1)| (4)

in the formula i_{L1_ref}Is an inductor current reference value, i_{L1_ST}(k +1) is an inductive current predicted value of the (k +1) th sampling period in a direct-connection state;

and step 3: calculating the inductive current and the corresponding cost function in the non-direct-through state:

discretizing the formula (5) by a first-order Euler discretization method to obtain the predicted value of the inductive current at the moment of k +1

Cost function g of inductive current in non-through state_{iL_ns}Can be designed as follows:

g_{iL_ns}＝|i_{L1_ref}-i_{L1_ns}(k+1)| (7)

in the formula i_{L1_ns}(k +1) is an inductive current predicted value of the (k +1) th sampling period in the non-through state;

when g is_{iL_ST}Less than g_{iL_ns}If the direct vector is judged to be in a direct state, the direct vector is directly output; when g is_{iL_ST}Greater than g_{iL_ns}Judging the capacitor to be in a non-straight-through state, and calculating a cost function of the capacitor voltage and the output current in the next step;

and 4, step 4: calculating cost function of capacitor voltage and output current

Discretizing the formula (8) by a first-order Euler discretization method to obtain a predicted value of the capacitance voltage at the k +1 moment in a non-through state as

In the formula i_inv(k +1) is the inverter current;

the phase voltage output equation of the three-phase inverter is

Discretizing the formula (10) by a first-order Euler discretization method to obtain the final product

In the formula: r and L are respectively a load resistance and an inductance, i_o(k +1) is the predicted value of the output current in the (k +1) th sampling period; i.e. i_o(k) The sampling value of the output current in the kth sampling period can be obtained by the formula (1); v_x(k) The output voltage value of the kth sampling period;

two separate cost functions are used for the capacitor voltage and the output current, and the cost functions of the capacitor voltage and the output current are respectively designed as follows:

g_vc1(i)＝|v_{c1_ref}-v_c1(k+1)| (12)

g_io(i)＝|i_{o_ref}-i_o(k+1)| (13)

in the formula, v_{c1_ref}As a reference value of the capacitor voltage, i_{o_ref}For outputting voltage reference value, 7 voltage vectors V output by the inverter_iSubstitution of the formulae (10), (11) in this order and obtaining 7 g according to formulae (12) and (13)_vc1(i) And g_io(i) Wherein i is 1, 2, …, 7;

and 5: inductance capacitance observation

Correcting the established qZSI prediction model through inductance-capacitance observation, and assuming that the inductance value of the qZSI in the prediction model is L_mA capacitance value of C_mAt this time, the prediction expressions of the inductive current and the capacitor voltage in the non-through state are

The prediction expressions of the inductive current and the capacitor voltage in the direct-through state are

When the load inductance value in the prediction model is L_pAt this time, the predicted expression of the output current is

And (3) deriving expressions of the inductance equivalent value of the quasi-Z source network in the non-through state and the through state according to the expressions (3), (6), (15) and (17):

deriving an expression of the capacitance equivalent value of the quasi-Z source network in the non-through state according to the expressions (9) and (16):

from equations (11) and (19), an expression for the load inductance equivalent is derived:

correcting the proposed prediction model according to the quasi-Z source network inductance, the capacitance and the load inductance observed in the formulas (20) to (22);

step 6: obtaining an optimal voltage vector:

seven errors of the capacitor voltage and the output current can be calculated according to the formulas (12) and (13), i takes a value of 0-7, and the errors of the two are arranged in ascending order

r₁ ⁱ＝rank[g_vc1(i)] (23)

r₂ ⁱ＝rank[g_io(i)] (24)

As can be seen from the equations (23) and (24), the smaller the error between the capacitor voltage and the output current, the higher the corresponding switch state rank, and the final selection is made so that g is obtained according to the switch states arranged by the equations (23) and (24)_vc1(i) And g_io(i) The first 3 voltage vectors corresponding to the smaller ones respectively, and g is selected from them_vc1(i) And g_io(i) The common voltage vector of (a) is used as an optimal voltage vector, thereby obtaining a set of optimal voltage vectors.

Further, said i_{L1_ref}Is calculated by the formula i_{L1_ref}＝P_{out_ref}/v_inIn which P is_{out_ref}For the output power reference value, v_inIs a dc input voltage.

Further, the selection principle of the optimal voltage vector is as follows: when cost function g_vc1(i) And g_io(i) When more than one common voltage vector exists, selecting the voltage vector with the top rank as the optimal voltage vector; when cost function g_vc1(i) And g_io(i) In the absence of a common voltage vector, g is chosen such that_vc1(i) The minimum voltage vector is used as the optimal voltage vector, so that the robustness of the system is improved.

The invention also discloses a robust predictive control system of the impedance source inverter, which comprises the following components:

at least one processor and at least one memory communicatively coupled to the processor, wherein: the memory stores program instructions executable by the processor to invoke a robust predictive control method of an impedance source inverter as in any above.

Furthermore, a non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the robust predictive control method of an impedance source inverter as described in any one of the above is disclosed.

(III) advantageous effects

Compared with the prior art, the robust prediction control method has the advantages that the parameter prediction method for calculating the cost function parameters and the optimization method for the optimal vector are strong in interaction, in the aspect of parameter robustness, a parameter observer model is established according to the state quantity of the inverter, resistance, inductance and capacitance values in the prediction model are specially corrected on line aiming at the impedance source inverter, the robust performance of the controller is enhanced, two independent cost functions are used for capacitor voltage and output current, cost function values of the cost functions are compared, three voltage vectors enabling the cost function values of the capacitor voltage and the output current to be smaller are respectively selected, and the voltage vector common to the capacitor voltage and the output current is used as the optimal voltage vector and used for controlling the inverter. Finally, the quasi-Z source inverter robust prediction control of weight factor optimization is realized, the design of weight coefficients in a cost function is eliminated while the system switching frequency is not increased, the complexity of cost function calculation is reduced, and stronger robustness can be ensured when model parameters are changed.

Drawings

Fig. 1 is a topology diagram of an impedance source inverter according to the present invention.

Fig. 2 is an equivalent circuit diagram of an impedance source inverter qZSI in the present invention, wherein the left diagram is a pass-through state diagram and the right diagram is a non-pass-through state diagram.

Fig. 3 is a flowchart of a robust predictive control method of an impedance source inverter according to the present invention.

FIG. 4 is a schematic block diagram of a parameter observer used in the case of a parameter mismatch, in which the inductance L is shown in the diagram (a)₁A view, graph (b) is a view of the capacitance C, graph (C)) Is a load inductance L₂And (6) observing the map.

Detailed Description

The present invention will be described more fully hereinafter with reference to the accompanying drawings and examples, in which the technical problems and advantages of the present invention are solved, wherein the described examples are only intended to facilitate the understanding of the present invention, and are not to be construed as limiting in any way.

The impedance source inverter (qZSI) topology used in the present invention is shown in FIG. 1, and comprises a quasi-Z source network, a three-phase conventional inverter and a resistive-inductive load, and the equivalent circuit diagram of qZSI in FIG. 1 is shown in FIG. 2. The left diagram in fig. 2 is an equivalent circuit diagram for the qZSI through state, with both switches of the same bridge arm open simultaneously; the right diagram in fig. 2 is an equivalent circuit diagram for the qZSI non-through state, where the three-phase inverter functions the same as the conventional inverter.

The flow chart of the prediction control method of the invention is shown in fig. 3, and the concrete implementation steps are as follows:

1. sampling the output current at time k, the capacitor voltage and the inductor current:

at the time k, three-phase output currents i of qZSI are respectively measured_a，i_bAnd i_cVoltage v of capacitor_c1And the inductor current i_L1Sampling, and obtaining current components i on alpha and beta axes of the output current in a static coordinate system according to Clarke transformation_αAnd i_βRecording the output current i_o＝i_α+ji_β：

2. Calculating the inductive current and the corresponding cost function in the direct connection state:

from the left image of FIG. 2, the

Discretizing the formula (2) by a first-order Euler discretization method to obtain the predicted value of the inductive current at the moment of k +1

In the formula i_L1(k +1) is the predicted value of the inductive current in the (k +1) th sampling period, i_L1(k) Is the inductor current of the kth sampling period, v_c1(k) Is the capacitor voltage of the kth sampling period, T_sIs a sampling period, L₁Inductance of qZSI, R_L1Is the internal resistance of the input inductor.

Cost function g of the inductor current in the shoot-through state_{iL_ST}Can be designed as follows:

g_{iL_ST}＝|i_{L1_ref}-i_{L1_ST}(k+1)| (4)

in the formula i_{L1_ref}For the inductor current reference value, i can be calculated by the formula_{L1_ref}＝P_{out_ref}/v_inIn which P is_{out_ref}For the output power reference value, v_inThe reference value of the inductive current, i, can be determined for the DC input voltage by specifying the specific values of the two_{L1_ST}And (k +1) is an inductance current predicted value of the (k +1) th sampling period in the through state.

3. Calculating the inductive current and the corresponding cost function in the non-direct-through state:

from the right drawing of FIG. 2, the

Discretizing the formula (5) by a first-order Euler discretization method to obtain the predicted value of the inductive current at the moment of k +1

Cost function g of inductive current in non-through state_{iL_ns}Can be designed as follows:

g_{iL_ns}＝|i_{L1_ref}-i_{L1_ns}(k+1)| (7)

in the formula i_{L1_ns}And (k +1) is an inductance current predicted value of the (k +1) th sampling period in the non-through state.

To obtain g_{iL_ST}And g_{iL_ns}Then, judging the state of the next control period according to the predicted value of the inductive current, and when g is_{iL_ST}Less than g_{iL_ns}If the direct vector is judged to be in a direct state, the direct vector is directly output; when g is_{iL_ST}Greater than g_{iL_ns}And judging the capacitor to be in a non-through state, and calculating the cost function of the capacitor voltage and the output current in the next step.

4. Calculating a cost function of the capacitor voltage and the output current:

from the right drawing of FIG. 2, the

Discretizing the formula (8) by a first-order Euler discretization method to obtain a predicted value of the capacitance voltage at the k +1 moment in a non-through state as

In the formula i_inv(k +1) is the inverter current and can be calculated by the formula, i_inv(k+1)＝S₁i_a(k)+S₃i_b(k)+S₅i_c(k) Wherein i is_a(k)，i_b(k) And i_c(k) The current at the time of k is respectively phase a, phase b and phase c; s₁、S₃、S₅Respectively A, B, C phase.

From FIG. 1, the phase voltage output equation of a three-phase inverter is given by

Discretizing the formula (10) by a first-order Euler discretization method to obtain the final product

In the formula: r and L are respectively a load resistance and an inductance, i_o(k +1) is the predicted value of the output current in the (k +1) th sampling period; i.e. i_o(k) The sampling value of the output current in the kth sampling period can be obtained by the formula (1); v_x(k) Is the output voltage value of the k-th sampling period, a-e^j2π/3，；x＝[0～7]：

This can be obtained from table 1 below:

TABLE 1 output voltages at different switching states of qZSI

Two separate cost functions are used for the capacitor voltage and the output current, and the cost functions of the capacitor voltage and the output current are respectively designed as follows:

g_vc1(i)＝|v_{c1_ref}-v_c1(k+1)| (12)

g_io(i)＝|i_{o_ref}-i_o(k+1)| (13)

in the formula, v_{c1_ref}As a reference value of the capacitor voltage, i_{o_ref}For the output voltage reference value, note i_{o_ref}＝i_{α_ref}+ji_{β_ref}Wherein

In the formula i_{α_ref}And i_{β_ref}Respectively, the components of the output current reference values in the α β coordinate system, i_{a_ref}，i_{b_ref}And i_{c_ref}The unit output currents of the phase a, the phase b and the phase c of the inverter are respectively, and R is a load resistor. v. of_{c1_ref}For the reference value of the capacitor voltage, v can be calculated by the formula_{c1_ref}＝(V_{dc_ref}+v_in) /2 wherein V_{dc_ref}Is a reference value of DC bus voltage, v_inThe reference value of the capacitor voltage can be obtained when the specific values of the direct current input voltage and the direct current input voltage are given. As can be seen from FIG. 3, when the algorithm enters the loop, excluding the shoot-through state, the voltage vectors have values as shown in Table 1, and the specific values of the i voltage vectors are V_iThe subscript i ═ 1, 2, …, 7, where V is_dcIs the dc bus voltage. 7 voltage vectors V to be output by the inverter_iSubstitution of the formulae (10), (11) in this order and obtaining 7 g according to formulae (12) and (13)_vc1(i) And g_io(i) Where i is 1, 2, …, 7.

5. And (3) inductance and capacitance observation:

when the parameters are mismatched (of course, when the parameters are not mismatched, the robustness can also be improved by using the observation method), the established qZSI prediction model can be corrected by inductance-capacitance observation, and the inductance value of the qZSI in the prediction model is assumed to be L_mA capacitance value of C_mAt this time, the prediction expressions of the inductive current and the capacitor voltage in the non-through state are

The prediction expressions of the inductive current and the capacitor voltage in the direct-through state are

When the load inductance value in the prediction model is L_pAt this time, the predicted expression of the output current is

According to the expressions (3), (6), (15) and (17), the expressions of the inductance equivalent value of the quasi-Z source network in the non-through state and the through state can be deduced:

according to the expressions (9) and (16), an expression of the capacitance equivalent value of the quasi-Z source network in the non-direct-through state can be deduced:

from equations (11) and (19), an expression for the load inductance equivalent can be derived:

and correcting the proposed prediction model according to the quasi-Z source network inductance, the capacitance and the load inductance observed in the formulas (20) to (22) so as to avoid the parameter mismatch to deteriorate the system performance. And the proposed observer structure takes into account a low-pass filter to reduce various disturbances in practice. The proposed parameter observer is shown in block diagram form in fig. 4.

6. Obtaining an optimal voltage vector:

seven errors of the capacitor voltage and the output current can be calculated according to the formulas (12) and (13), i takes a value of 0-7, and the errors of the two are arranged in ascending order

r₁ ⁱ＝rank[g_vc1(i)] (23)

r₂ ⁱ＝rank[g_io(i)] (24)

As can be seen from equations (23) and (24), the smaller the error between the capacitor voltage and the output current, the higher the corresponding switch state rank, and some unreasonable switch states can be discarded according to the switch states arranged by equations (23) and (24). The invention finally selects the formula g_vc1(i) And g_io(i) The first 3 voltage vectors corresponding to the smaller ones respectively, and g is selected from them_vc1(i) And g_io(i) The common voltage vector is used as an optimal voltage vector, so that an optimal voltage vector (generally 3-6 optimal voltage vectors) is obtained.

In addition, the optimal voltage vector selection principle is as follows: when cost function g_vc1(i) And g_io(i) When more than one common voltage vector exists, selecting the voltage vector with the top rank as the optimal voltage vector; when cost function g_vc1(i) And g_io(i) In the absence of a common voltage vector, g is chosen such that_vc1(i) The minimum voltage vector is used as the optimal voltage vector, so that the robustness of the system is improved.

For example: consider a special case when g_vc1(i) And g_io(i) When there are multiple common voltage vectors, e.g. such that g_vc1(i) The 3 voltage vectors corresponding to smaller are V₁，V₃，V₄So that g is_io(i) The 3 voltage vectors corresponding to smaller are V₁，V₃，V₅At this time, the voltage vector V is selected₁As an optimal voltage vector and acts on the inverter.

It is noted that in the control method proposed by the present invention, two cost functions g of the capacitor voltage and the output current are calculated separately and simultaneously_vc1(i) And g_io(i) The choice of only considering the first 3 voltage vectors is the conclusion that the inventors have made experiments and analyses, since if g is the case_vc1(i) And g_io(i) If the number of the respectively selected voltage vectors is less than 3, the capacitance voltage item and the load current item cannot be effectively optimized; on the contrary, if the number of voltage vectors exceeds 3, the amount of calculation increases greatly.

Finally, it should be noted that, the technical problem that the robustness is poor is found based on the defect of the prior art "CN 202010572451.8, a quasi-Z source inverter model predictive control method without weight coefficients" (since it requires professional experiment and analysis capability, the technical problem is found and proposed without universality); in addition, the inventor also considers that organic selection uses an inductance-capacitance observation model aiming at the Z-source inverter, independent two independent cost functions for capacitance voltage and output current, and selects three voltage vectors with smaller cost function values of the capacitance voltage and the output current respectively, and the voltage vector common to the three voltage vectors is used as an optimal voltage vector to be used for controlling the inverter, so that the robustness of the system is greatly improved (namely the characteristics in steps 4-6), and the improved technical means of the invention also obviously do not belong to the conventional means in the field.

The robust predictive control method of the impedance source inverter of the present invention described above can be executed as a software program or computer instructions in a non-transitory computer-readable storage medium or in a control system with a memory and a processor, and the calculation procedure thereof is simple and fast. Each functional unit in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit. The integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium. The software functional unit is stored in a storage medium and includes several instructions to enable a computer device (which may be a personal computer, a server, or a network device) or a processor (processor) to execute some steps of the methods according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.