CN112701951A - Inverter voltage state prediction control method based on tolerant hierarchical sequence method - Google Patents

Abstract

The invention discloses a prediction control method for the voltage state of an inverter based on a tolerant hierarchical sequence method, wherein the inverter is a clamp type three-level three-phase inverter, and the prediction control method for the voltage state comprises the following steps: creating an MPC prediction model, and predicting the output capacitor voltage and the potential difference of the neutral point at the DC side at the K +2 moment according to the MPC prediction model at the K moment; then constructing a double-layer value objective function; then, solving a first layer of objective function in a double-layer value objective function by using a tolerant hierarchical sequence method to obtain a plurality of candidate voltage state vectors, substituting the plurality of candidate voltage state vectors obtained by solving into a second layer of objective function, and optimizing and solving the second layer of objective function to obtain an optimal voltage state vector; and finally, applying the inverter switching state corresponding to the optimal voltage state vector at the moment of K +1 to carry out prediction control on the inverter. The inverter voltage state prediction control method improves the modeling rate and the control stability.

Description

Inverter Voltage State Predictive Control Method Based on Tolerant Hierarchical Sequence Method

technical field

The invention relates to the technical field of inverters, in particular to an inverter voltage state prediction control method based on a tolerant hierarchical sequence method.

Background technique

The rapid development of new energy sources such as wind energy, solar energy, and bioenergy has made multi-level inverters widely concerned. As the link of photovoltaic grid connection, multi-level inverter needs to have excellent dynamic stability. The existing two-level inverter has the shortcomings of serious electromagnetic interference and low inverter efficiency, especially in high-voltage applications. Compared with the two-level inverter, the clamp-type three-level inverter relies on It has become the first choice for distributed power generation systems due to the advantages of low withstand voltage of the switch tube, low filter inductance loss, and low harmonic distortion.

Model Predictive Control (MPC) method, as a control strategy of clamped inverter, can evaluate each switching state sequence according to the value function, and make the switching state sequence that satisfies the minimum value function in the next one. The moment is used for the control of the power electronic inverter. Existing MPC methods generally use a value function with weight coefficients to achieve control of different objectives. Through the adjustment of the weight coefficient, the proportion of a single target in the total target is added or subtracted, so as to realize the limitation of the system and the nonlinear processing. However, the selection of the weight coefficient in this method is uncertain, and the selection of the weight coefficient is mostly carried out through the genetic algorithm. and empirical analysis, the lack of corresponding theoretical basis leads to cumbersome and randomness in the process of determining the weight coefficient, which reduces the modeling efficiency; and due to the invulnerability between different control objectives, balance the quantity between each objective. The class often requires a large weight coefficient, which will also lead to the amplification of noise, making the control strategy unstable and unable to meet the needs of use.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is to provide an inverter voltage state prediction control method based on the tolerance layered sequence method of the present invention, which can improve the modeling rate and improve the control stability.

In order to solve the above-mentioned technical problems, the technical solutions provided by the present invention are as follows:

A voltage state prediction control method for an inverter based on a tolerance hierarchical sequence method, wherein the inverter is a clamped three-level three-phase inverter, and the voltage state prediction control method comprises the following steps:

1) Create an MPC prediction model, and predict the output capacitor voltage and the neutral point potential difference of the DC side at time K+2 according to the MPC prediction model at time K;

2) Construct the first-layer value function according to the difference between the output capacitor voltage and the expected output capacitor voltage at the time of K+2, and construct the second-layer value function according to the potential difference of the neutral point of the DC side at the time of K+2. The first-layer value function and the second-layer value function constitute a double-layer value objective function;

3) Use the permissive hierarchical sequence method to solve the objective function of the first layer to obtain multiple candidate voltage state vectors, substitute the multiple candidate voltage state vectors obtained by solving the objective function of the first layer into the objective function of the second layer, and analyze the second layer of the objective function. The layer objective function is optimized and solved to obtain the optimal voltage state vector;

4) The inverter switch state corresponding to the optimal voltage state vector is applied at the time K+1 to perform predictive control on the inverter.

In one of the embodiments, the minimization principle is adopted when the objective function of the second layer is optimized and solved.

In one embodiment, the two-layer value objective function is:

表示第一层价值函数，

表示第二层价值函数，ΔV_PN(K+2)表示K+2时刻的直流侧中性点电位差，

表示K+2时刻的期望输出电容电压，

表示K+2时刻的输出电容电压，α、β分别表示两相静止坐标系中的α轴、β轴。in,

represents the first-level value function,

represents the value function of the second layer, ΔV _PN (K+2) represents the neutral point potential difference of the DC side at the moment of K+2,

represents the expected output capacitor voltage at time K+2,

Represents the output capacitor voltage at time K+2, and α and β represent the α-axis and β-axis in the two-phase static coordinate system, respectively.

In one of the embodiments, the following formula is used when the objective function of the second layer is optimized and solved to obtain the optimal voltage state vector:

表示最优电压状态矢量，n表示电压矢量的标号，ζ表示宽容值，

表示表示第一层候选价值函数，

表示第二层候选价值函数。

represents the optimal voltage state vector, n represents the label of the voltage vector, ζ represents the tolerance value,

represents the candidate value function of the first layer,

represents the second-level candidate value function.

In one embodiment,

表示K时刻的输出电容电压，T_s为采样时间，C为滤波电容，

表示K时刻的输出电感电流，

表示K时刻的输出负载电流，u_x(K)表示K时刻逆变器的输出电压，ΔV_PN(K)表示K时刻的直流侧中性点电位差，i_O(K)表示K时刻的直流侧中点电流。in,

represents the output capacitor voltage at time K, T _s is the sampling time, C is the filter capacitor,

represents the output inductor current at time K,

represents the output load current at time K, u _x (K) represents the output voltage of the inverter at time K, ΔV _PN (K) represents the neutral point potential difference of the DC side at time K, and i _O (K) represents the DC voltage at time K side midpoint current.

In one of the embodiments, the two-phase stationary coordinate system is converted from a three-phase stationary coordinate system.

In one embodiment, the two-phase stationary coordinate system is obtained by transforming the three-phase stationary coordinate system through Clark transformation or Park transformation.

The invention has the following beneficial effects: the inverter voltage state prediction control method based on the tolerance layered sequence method of the invention can avoid adjusting the weight coefficient in the value function, save the complicated and tedious trial and error process, and improve the modeling At the same time, it adjusts the invulnerability between different control targets, avoids the phenomenon of distortion caused by excessive noise caused by excessively amplifying the control targets, and has excellent steady-state performance and fast dynamic response.

Description of drawings

FIG. 1 is a schematic structural diagram of a clamped three-level three-phase inverter;

FIG. 2 is a voltage space vector relationship diagram of the inverter shown in FIG. 1;

Fig. 3 is the structural block diagram of the inverter voltage state prediction control method based on the tolerance hierarchical sequence method of the present invention;

Fig. 4 is the control strategy flow chart of the inverter voltage state prediction control method based on the tolerance hierarchical sequence method of the present invention;

Fig. 5 is a coordinate diagram of a three-phase stationary coordinate system and a two-phase stationary coordinate system;

Fig. 6 is the voltage state control waveform diagram that utilizes the predictive control method of the present invention to obtain;

7 is a distribution diagram of the number of candidate voltage state vectors entering the second-layer value function in the voltage vector;

Detailed ways

The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

As shown in FIGS. 3-4 , this embodiment discloses a method for predicting the voltage state of an inverter based on a tolerant hierarchical sequence method. The inverter is a clamped three-level three-phase inverter. Refer to Figure 1 to Figure 2 for the structure of the inverter. The clamped three-level three-phase inverter outputs AC power to the three-phase load. The inverter voltage state prediction control method based on the tolerance hierarchical sequence method includes the following steps:

1) Create an MPC prediction model, and predict the output capacitor voltage and the neutral point potential difference of the DC side at time K+2 according to the MPC prediction model at time K;

2) Construct the first-layer value function according to the difference between the output capacitor voltage and the expected output capacitor voltage at the time of K+2, and construct the second-layer value function according to the potential difference of the neutral point of the DC side at the time of K+2. The first-layer value function and the second-layer value function constitute a double-layer value objective function;

3) Use the permissive hierarchical sequence method to solve the objective function of the first layer to obtain multiple candidate voltage state vectors, substitute the multiple candidate voltage state vectors obtained by solving the objective function of the first layer into the objective function of the second layer, and analyze the second layer of the objective function. The layer objective function is optimized and solved to obtain the optimal voltage state vector;

4) The inverter switch state corresponding to the optimal voltage state vector is applied at the time K+1 to perform predictive control on the inverter.

In the above process, at time K, the output capacitor voltage and the DC side neutral point potential difference at the time K+2 are predicted according to the MPC prediction model, and the predicted output capacitor voltage and the DC side neutral point potential difference at the time K+2 are used. Construct the double-layer value function, and finally obtain the optimal voltage state vector, and apply the inverter switch state corresponding to the optimal voltage state vector at time K+1 to perform predictive control of the inverter, that is, at time K Bring in the prediction model to predict the output capacitor voltage and the DC side potential difference at the time of K+2, and apply it at the time of K+1, this method can effectively compensate the delay of the sampling step in the actual execution process, and realize the delay compensation, The control performance is optimized, which is more conducive to obtaining the best voltage state vector of the inverter at the current moment.

In one of the embodiments, the minimization principle is adopted when the objective function of the second layer is optimized and solved.

In one embodiment, the two-layer value objective function is:

表示第一层价值函数，

表示第二层价值函数，ΔV_PN(K+2)表示K+2时刻的直流侧中性点电位差，

表示K+2时刻的期望输出电容电压，

表示K+2时刻的输出电容电压，x＝α、β，α、β分别表示两相静止坐标系中的α轴、β轴,例如，x＝α时，

表示K+2时刻的输出电容电压在α轴上的坐标值。in,

represents the first-level value function,

represents the value function of the second layer, ΔV _PN (K+2) represents the neutral point potential difference of the DC side at the moment of K+2,

represents the expected output capacitor voltage at time K+2,

Represents the output capacitor voltage at time K+2, x=α, β, α, β represent the α-axis and β-axis in the two-phase static coordinate system, for example, when x=α,

Indicates the coordinate value of the output capacitor voltage on the α-axis at time K+2.

In one of the embodiments, the two-phase stationary coordinate system is converted from the three-phase stationary coordinate system.

In one of the embodiments, as shown in FIG. 5 , the two-phase stationary coordinate system is obtained by transforming the three-phase stationary coordinate system through Clark transformation (Clark transformation) or Park transformation (Pike transformation).

In one of the embodiments, the following formula is used when the objective function of the second layer is optimized and solved by using the minimization principle to obtain the optimal voltage state vector:

表示最优电压状态矢量，可以理解地，

分别表示A相、B相和C相的最优电压，n表示电压矢量的标号，ζ表示宽容值，

表示表示第一层候选价值函数，

表示第二层候选价值函数。

represents the optimal voltage state vector, understandably,

respectively represent the optimal voltages of A-phase, B-phase and C-phase, n represents the label of the voltage vector, ζ represents the tolerance value,

represents the candidate value function of the first layer,

represents the second-level candidate value function.

表示找出满足不大于

之和的所有候选电压状态矢量，以此候选电压状态矢量下的第一层价值函数记为第一层候选价值函数

然后将满足

的候选电压状态矢量代入第二层价值函数，将此时的第二层价值函数记为第二层候选价值函数

并最终找出第二层候选价值函数

的最小值

满足

的候选电压状态矢量则为最优电压矢量。Understandably,

Indicates to find out that the satisfaction is not greater than

The sum of all candidate voltage state vectors, the first layer value function under this candidate voltage state vector is recorded as the first layer candidate value function

then will satisfy

The candidate voltage state vector is substituted into the second-layer value function, and the second-layer value function at this time is recorded as the second-layer candidate value function

And finally find the second layer candidate value function

the minimum value of

Satisfy

The candidate voltage state vector of is the optimal voltage vector.

The objective function of the first layer is solved by the tolerant hierarchical sequence method, and the tolerance value ζ is introduced. By setting the appropriate tolerance value, the requirements of different performances can be met, and at least one optimal solution can be sent to the next layer. It solves the problem that "the optimal solution of the existing dictionary will fall into the only optimal solution of the next layer during the hierarchical calculation, resulting in that the control objective of the next layer cannot be realized as expected". The selection of the tolerance value is not as complicated as the selection of the weight coefficient. The size of the tolerance value determines the number of candidate voltage state vectors entering the second layer. The size of the tolerance value depends on the actual application site.

In one embodiment,

表示K时刻的输出电容电压，T_s为采样时间，C为滤波电容，

表示K时刻的输出电感电流，

表示K时刻的输出负载电流，u_x(K)表示K时刻逆变器的输出电压，ΔV_PN(K)表示K时刻的直流侧中性点电位差，i_O(K)表示K时刻的直流侧中点电流。in,

represents the output capacitor voltage at time K, T _s is the sampling time, C is the filter capacitor,

represents the output inductor current at time K,

represents the output load current at time K, u _x (K) represents the output voltage of the inverter at time K, ΔV _PN (K) represents the neutral point potential difference of the DC side at time K, and i _O (K) represents the DC voltage at time K side midpoint current.

In one embodiment, the MPC prediction model in step 1) is:

为输出电感电流，u_x为逆变器的输出电压，

为平衡电容电流，y＝P、N，分别表示直流侧上下两个平衡电容的标号，也即分别代表平衡电容P和平衡电容N，i_O为直流侧中点电流,ΔV_PN为直流侧中点电位差，C为滤波电容，

为输出电容电压，

为输出负载电流，L为滤波电感，C_y为平衡电容的电容值，V_y为平衡电容的电压,S_n′表示逆变器n′相的开关状态，i_n,表示逆变器n′相的相电流，n′＝a,b,c分别代表A相、B相和C相，

分别表示平衡电容P、平衡电容N的电容电流，V_P、V_N分别表示平衡电容P、平衡电容N的电容电压。In the formula,

is the output inductor current, u _x is the output voltage of the inverter,

is the balance capacitor current, y=P, N, which represent the labels of the upper and lower balance capacitors on the DC side respectively, that is, the balance capacitor P and the balance capacitor N respectively, i _O is the midpoint current of the DC side, and ΔV _PN is the middle point of the DC side. point potential difference, C is the filter capacitor,

is the output capacitor voltage,

is the output load current, L is the filter inductance, C _y is the capacitance value of the balance capacitor, V _y is the voltage of the balance capacitor, Sn _' represents the switching state of the n' phase of the inverter, i _n, represents the inverter n' The phase current of the phase, n′=a, b, c represent the A phase, the B phase and the C phase, respectively,

Represent the capacitance currents of the balance capacitor P and the balance capacitor N, respectively, and V _P and V _N represent the capacitor voltages of the balance capacitor P and the balance capacitor N, respectively.

Among them, the process of obtaining formula (3) from formula (4) is as follows:

S1) First derive the differential equation of the clamped three-phase three-level inverter in the continuous time domain in the αβ two-phase static coordinate system from the formula (4):

表示输出负载电流在α轴上的坐标值。Among them, x=α, β, representing the α-axis and β-axis in the α-β two-phase stationary coordinate system, for example, when x=α,

Indicates the coordinate value of the output load current on the α-axis.

S2) At time K, use the MPC prediction model to predict the output capacitor voltage and DC side potential difference of the inverter at time K+1. Considering that the microprocessor can complete the prediction calculation within tens of microseconds, these systems are discrete change. Let T _s be the sampling time, and obtain the output capacitor voltage and DC side potential difference at time K in the discrete time domain according to the Euler forward method:

表示K+1时刻的输出电容电压，ΔV_PN(K+1)表示K+1时刻的直流侧中性点电位差，

表示K时刻的输出电容电压，T_s为采样时间，C为滤波电容，

表示K时刻的输出电感电流，

表示K时刻的输出负载电流，ΔV_PN(K)表示K时刻的直流侧中性点电位差，i_O(K)表示K时刻的直流侧中点电流。in,

represents the output capacitor voltage at time K+1, ΔV _PN (K+1) represents the neutral point potential difference of the DC side at time K+1,

represents the output capacitor voltage at time K, T _s is the sampling time, C is the filter capacitor,

represents the output inductor current at time K,

Represents the output load current at time K, ΔV _PN (K) represents the neutral point potential difference of the DC side at time K, and i _O (K) represents the midpoint current of the DC side at time K.

S3) Iteratively predicts the output capacitor voltage and DC side potential difference at time K+2 according to the predicted value at time K+1 obtained in step S2), that is, formula (3) is obtained.

Among them, the obtaining process of formula (2) is as follows:

The mathematical model of the inverter in this embodiment can be identified by the following two coordinate systems, which are the abc three-phase static coordinate system and the αβ two-phase static coordinate system, as shown in FIG. 5 :

1) abc three-phase static coordinate system: abc are the axial directions of the three-phase windings of the stator, with an electrical angle difference of 120°;

2) αβ two-phase stationary coordinate system: α axis coincides with a axis, β axis is 90° ahead of α axis counterclockwise;

In order to transform the mathematical model of the abc three-phase static coordinate system into the αβ two-phase static coordinate system, it is necessary to perform Clark transformation (Clark transformation), which is referred to as 3/2 transformation, and the transformation matrix C _3/2 (equal amplitude coordinate transformation) )as follows:

The structure of the clamped three-phase three-level voltage source inverter is shown in Figure 2. The formula of the output voltage υ _s is as follows:

同时导通，

同时关断；S_a＝0代表

同时导通，

同时关断；例如S_a＝-1代表

同时导通，

同时关断。逆变器共有27种开关状态，分别对应27种电压矢量输出，电压空间矢量关系如图3。In the formula, V _dc is the voltage amplitude of the DC input terminal, and _Sa , _{Sb, and Sc respectively} identify the switching state of the inverter, for example, Sa ₌ 1 represents

turn on at the same time,

Turn off at the same time; Sa ₌ 0 means

turn on at the same time,

Turn off at the same time; for example, Sa ₌ -1 means

turn on at the same time,

simultaneously shut down. The inverter has a total of 27 switching states, corresponding to 27 voltage vector outputs. The voltage space vector relationship is shown in Figure 3.

According to the topology of the clamped three-phase three-level inverter in Fig. 2, the dynamic equation in the two-phase static coordinate system can be obtained:

(1) Output inductor current equation

分别表示输出电感电流在α轴、β轴上的坐标值，u_c为输出电容电压，C为滤波电容，

分别表示输出负载电流在在α轴、β轴上的坐标值。In the formula,

respectively represent the coordinate values of the output inductor current on the α-axis and β-axis, u _c is the output capacitor voltage, C is the filter capacitor,

They represent the coordinate values of the output load current on the α-axis and β-axis, respectively.

(2) Voltage equation:

In the formula, u _α and u _β are the coordinate values of the output voltage on the α-axis and β-axis, respectively, and L represents the filter inductance.

(3) DC side balance capacitor equation:

均为平衡电容电流，i_O为直流侧中点电流,ΔV_PN为直流侧中点电位差，C_P、C_N均为平衡电容容值，S_n′表示逆变器n′相的开关状态，i_n,表示逆变器n′相的相电流，n′＝a,b,c分别代表A相、B相和C相，

分别表示平衡电容P、平衡电容N的电容电流，V_P、V_N分别表示平衡电容P、平衡电容N的电容电压。In the formula,

are the balance capacitor current, i _O is the DC side midpoint current, ΔV _PN is the DC side midpoint potential difference, C _P and _CN are the balance capacitor value, and Sn _' represents the switching state of the n' phase of the inverter , i _n, represents the phase current of inverter n' phase, n'=a, b, c represent A phase, B phase and C phase respectively,

Represent the capacitance currents of the balance capacitor P and the balance capacitor N, respectively, and V _P and V _N represent the capacitor voltages of the balance capacitor P and the balance capacitor N, respectively.

According to formulas (8), (9) and (10), the differential equation of the clamped three-phase three-level inverter in the continuous time domain in the αβ coordinate system is derived:

表示输出负载电流在α轴上的坐标值；

为输出电容电压，

为输出负载电流，

为输出电感电流。Among them, x=α, β, representing the α-axis and β-axis in the α-β two-phase stationary coordinate system, for example, when x=α,

Indicates the coordinate value of the output load current on the α axis;

is the output capacitor voltage,

is the output load current,

is the output inductor current.

According to formula (11), two control objectives can be obtained, which are to consider the difference between the output capacitor voltage and the reference capacitor voltage and minimize the potential difference of the DC side balance capacitor. The specific value function equation is:

Among them, n=0, . . . , 26, representing different voltage state vector labels.

为说明需要，这里对以下定义进行说明：According to formula (12), in practice, there is almost no optimal voltage vector that satisfies the two control objectives at the same time, and there is only a set of equivalent Pareto optimal solutions. One of the set of Pareto optimal solutions is defined as

To illustrate the need, the following definitions are explained here:

Definition 1: When the label of the selected voltage state vector n=0,...,26, the corresponding solution is called a suitable solution

满足

或

该解称为帕累托最优解；Definition 2: Suitable solution

Satisfy

or

This solution is called the Pareto optimal solution;

满足

且

J₂代表该解下的第二层价值函数值，该解称为字典最优解；Definition 3: There are and only suitable solutions

Satisfy

and

J ₂ represents the value of the second layer value function under the solution, which is called the optimal solution of the dictionary;

The present invention solves the problem of multi-objective optimization by establishing a sequence structure. The value function is layered according to the priority level, and the input that meets the control objective of the current layer enters the next layer, and the layer calculation continues until the Pareto optimal solution of the last value function is obtained. Different value function levels have different Pareto optimal solutions. According to definition 3, the mathematical expression of the optimal solution of the dictionary can be obtained:

Equation (13) shows that the optimal solution of the dictionary is the only one that falls into the next layer during the hierarchical calculation, resulting in that the control objective of the next layer cannot be achieved as expected. In order to overcome this defect, the tolerance value ζ is introduced in the process of hierarchical calculation, and the formula (13) is improved to obtain

According to the above-mentioned method for predicting the voltage state of the inverter based on the tolerant hierarchical sequence method, a simulation model of the system is established in Matlab, and the above-mentioned method is used to obtain the optimal voltage vector control inverter operation. In order to verify that the control strategy can well balance the DC side midpoint voltage, the DC side capacitors are set to be unequal in the model, that is, a resistor R _P is connected in parallel with the capacitor C _P terminal. Simulation parameters: V _dc = 400V, R _dc = 0.001Ω, C _P = _CN = 500μF, R = 20Ω, L = 10mH, C = 10μF, R _P = 200Ω, the sampling period of the control system is 20kHz; the desired voltage is given Assign 150V, frequency 50Hz. The simulation comparison of traditional model predictive control and tolerance sequence model predictive voltage state control shows that the current and voltage performance of the two control strategies are roughly similar. 6 is a voltage state control waveform diagram obtained by using the predictive control method of the present embodiment, wherein (a) diagram represents the waveform diagram of the phase A phase current, and diagram (b) represents the fast Fourier transform of the phase A phase current. Figure (c) represents the line voltage fluctuation diagram, and (d) represents the DC side neutral point potential difference fluctuation diagram. From Figure 6 (b), it can be seen that the total harmonic distortion of the predicted voltage state control of this embodiment is 0.68%, the predicted voltage state control method still maintains excellent performance in current performance. Since the DC side capacitor C _P is connected in parallel with the resistor R _P , the overall neutral point voltage has a negative offset. The predicted voltage state control method of this embodiment controls the maximum neutral point voltage offset value to be -6.198V. When suppressing the neutral point voltage Volatility is also excellent. Figure 7 shows the number of candidate voltage state vectors entering the second layer of the double-layer cost function, of which the ratio of 1, 2, and 3 is the largest, which are 55%, 30%, and 11%, respectively, which represent the predicted voltage of this embodiment. The state control method can dynamically select the number of candidate voltage state vectors according to the current system situation.

The inverter voltage state prediction control method based on the tolerance layered sequence method in this embodiment solves the voltage state vector by establishing a two-layer value function, without adjusting the weight coefficient, avoiding the need for a single value function in the prior art. The tedious process of confirming the weight coefficients at the same time mediates the invulnerability between the two control targets, avoiding the phenomenon of excessively amplifying the control targets and causing distortion caused by excessive noise; at the same time, there is no candidate for entering the second layer in the fixed value function The number of voltage state vectors, but the dynamic adjustment of the number of candidate voltage vectors is realized by setting the tolerance value. According to the different voltage states at each moment, the control strategy independently selects the number of candidate voltage state vectors, and can dynamically select the candidate voltage state vectors. It increases the adaptability and flexibility of the control strategy; the overall method combines the dictionary method with tolerance value and model predictive control, which has excellent steady-state performance and fast dynamic response; greatly reduces the time spent on parameter design , which improves the modeling efficiency and computing efficiency.

The above-mentioned embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or transformations made by those skilled in the art on the basis of the present invention are all within the protection scope of the present invention. The protection scope of the present invention is subject to the claims.

Claims (7)

1. A voltage state prediction control method of an inverter based on a tolerant hierarchical sequence method is disclosed, the inverter is a clamp type three-level three-phase inverter, and the voltage state prediction control method is characterized by comprising the following steps:

1) creating an MPC prediction model, and predicting the output capacitor voltage and the potential difference of the neutral point at the DC side at the K +2 moment according to the MPC prediction model at the K moment;

2) constructing a first-layer cost function according to the difference value between the output capacitor voltage at the moment K +2 and the expected output capacitor voltage, constructing a second-layer cost function according to the potential difference of the neutral point at the direct current side at the moment K +2, wherein the first-layer cost function and the second-layer cost function form a double-layer cost target function;

3) solving the first layer of objective function by using a tolerant hierarchical sequence method to obtain a plurality of candidate voltage state vectors, substituting the plurality of candidate voltage state vectors obtained by solving the first layer of objective function into the second layer of objective function, and optimizing and solving the second layer of objective function to obtain an optimal voltage state vector;

4) and applying the inverter switching state corresponding to the optimal voltage state vector to the moment K +1 to perform prediction control on the inverter.

2. The inverter voltage state prediction control method based on the tolerant hierarchical sequence method according to claim 2, wherein a minimization principle is adopted when the second-layer objective function is optimized and solved.

3. The method for predictive control of inverter voltage states based on the tolerant hierarchical sequence method as set forth in claim 2, wherein the two-level cost objective function is:

wherein,

a first-level cost function is represented,

representing the second layer cost function, Δ V_PN(K +2) represents the DC-side neutral point potential difference at the time K +2,

representing the desired output capacitor voltage at time K +2,

the output capacitor voltage at the time K +2 is shown, and α and β respectively show an α axis and a β axis in the two-phase stationary coordinate system.

4. The inverter voltage state prediction control method based on the tolerant hierarchical sequence method according to claim 3, wherein the following formula is adopted when the optimal voltage state vector is obtained by optimizing and solving the second-layer objective function:

denotes an optimum voltage state vector, n denotes a reference numeral of the voltage vector, ζ denotes a wide tolerance value,

the representation represents a first-level candidate cost function,

representing a second tier candidate merit function.

5. The inverter voltage state prediction control method based on the tolerant hierarchical sequence method according to claim 3,

wherein,

indicating the output capacitor voltage at time K, T_sIs the sampling time, C is the filter capacitance,

representing the output inductor current at time K,

represents the output load current at time K, u_x(K) Representing the output voltage of the inverter at time K, Δ V_PN(K) Represents the potential difference of the neutral point on the DC side at the time K_O(K) The dc-side midpoint current at time K is shown.

6. The method for predictive control of inverter voltage states based on the tolerant hierarchical sequence method as set forth in claim 3, characterized in that the two-phase stationary coordinate system is converted from a three-phase stationary coordinate system.

7. The inverter voltage state prediction control method based on the tolerant hierarchical sequence method according to claim 6, wherein the two-phase static coordinate system is obtained by Clark conversion or Park conversion of a three-phase static coordinate system.

Images