A power point tracking control method for NPC rectifiers with simplified computational complexity

Abstract

This invention discloses a simplified power point tracking (PPT) control method for NPC rectifiers, comprising: calculating the grid-side complex power S and the differential expression of the negative conjugate complex power based on the mathematical model of the three-level NPC rectifier and instantaneous power theory; discretizing the negative conjugate complex power at time k+1 in the dq rotating coordinate system using the forward Euler method; calculating the power error of the negative conjugate complex power under zero vector action at time k+1; performing preliminary vector pre-selection based on the sector where the power error is located; reducing the number of pre-selected vectors in each group in step four to four; cyclically calculating the predicted active and reactive power values ​​and the cost function J at time k+1, and determining the optimal vector. This invention simplifies the computational workload in the vector selection process of a three-level midpoint clamp rectifier by reducing polling pre-selection vectors, and introduces power point tracking error correction on the basis of traditional PI, thereby achieving accurate power point tracking control.

Description

TECHNICAL FIELD

[0001] The application relates to a rectifier model prediction control method, in particular to a power tracking control method with simplified calculation amount of an NPC rectifier. BACKGROUND

[0002] The three-level rectifier is a PWM rectifier which can be used for high-voltage and high-power and has the characteristics of a power factor close to 1, and has wide application in high-power traction transmission.

[0003] The control strategy and method of the three-level NPC rectifier include traditional voltage tracking double-closed-loop PI control and direct power control based on a switch table, and the control methods are mature in the industry. The model prediction control strategy is a control strategy with simple structure, fast response and no time-voltage constraint. The traditional model prediction control is based on the optimization of an objective function, and the optimal vector is selected through polling to act on the rectifier, so that the input constraint problem is solved. However, for the three-level rectifier, the calculation amount of the optimal vector obtained through the polling method is large, especially in the multi-step prediction field, and the processing problem of the weight factor in the cost function is complex, the power prediction accuracy is not enough, and the expected effect cannot be achieved. SUMMARY

[0004] In view of the problems in the prior art, the application provides a power tracking control method with simplified calculation amount of an NPC rectifier. According to the position of the negative conjugate complex power error in the space region under the action of the zero vector, the preselected vector is screened out, and further improvement is carried out, so that the model prediction power accurate control with simplified calculation amount and no weight value is realized.

[0005] In order to achieve the above purpose, the technical scheme adopted by the application is as follows: a power tracking control method with simplified calculation amount of an NPC rectifier, specifically comprising:

[0006] Step one: according to the mathematical model of the three-level NPC rectifier in the alpha-beta axis coordinate and the instantaneous power theory, the grid-side complex power S is calculated, and the differential expression of the negative conjugate complex power under ideal conditions is derived;

[0007] Step two: according to the forward Euler method, the prediction value of the model negative conjugate complex power in the dq rotating coordinate system at k+1 time is obtained;

[0008] Step three: the power error of the negative conjugate complex power under the action of the zero vector at k+1 time is calculated

[0009] Step four: according to the power error The sector where the sector is located, the preliminary vector pre-selection is carried out, and 12 groups of 7 pre-selected vectors are pre-selected, including 3 zero vectors, 2 small vectors, 1 medium vector and 1 large vector;

[0010] Step five: The small vectors pre-selected in step four are grouped according to the difference between the upper and lower capacitor values on the DC bus side, and the small vectors that make the difference between the two capacitors close to 0 are selected, and the small vectors that make the difference between the two capacitors far from 0 are discarded, and then the number of pre-selected vectors in each group is reduced from 7 to 6.

[0011] Step six: Remove the two zero vectors in each group of pre-selected vectors from the polling process, so that the number of pre-selected vectors in each group is reduced to 4;

[0012] Step seven: Calculate the active and reactive power prediction values at time k+1 and the value of the cost function J, and after 4 cycles, select the vector that makes the cost function J minimum, which is the optimal vector; When the optimal vector is a zero vector, according to the switching state at time k and the principle of minimum switching action, different zero vectors are selected as output; When the optimal vector is a non-zero vector, the optimal non-zero vector is output.

[0013] Further, in step five, the small vectors in the 12 groups of pre-selected vectors are divided into two groups according to whether the difference between the upper and lower capacitor voltages on the DC side is greater than or equal to 0.

[0014] Further, in step seven, when the optimal vector is a zero vector, it is divided into four cases:

[0015] a. When the switching state S abc When there are 2 1s, v0(111) is selected as output;

[0016] b. When the switching state S abc When there are 2 0s, v0(000) is selected as output;

[0017] c. When the switching state S abc When there are 2 -1s, v0(-1-1-1) is selected as output;

[0018] d. The remaining cases are selected to prevent jumps and output v0(000).

[0019] Further, in step seven, the active and reactive power prediction values at time k+1 are corrected by using the power reference value update function to update the power reference value, and the power reference value update function is

[0020] The application greatly simplifies the calculation amount in the vector selection process of the three-level neutral point clamped rectifier by reducing the polling preselected vector, and introduces the repetitive control thought on the basis of the traditional PI, continuously corrects the control input signal of each switching period according to the previous power tracking error signal, so as to realize the accurate power tracking control, has good power tracking performance, and is practical. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is the control method flow chart of the application;

[0022] Figure 2 is the vector sector division diagram of the three-level NPC rectifier of the application;

[0023] Figure 3 is the repetitive control power reference value updating schematic diagram in the application. DETAILED DESCRIPTION

[0024] The NPC rectifier simplified calculation amount power tracking control method of the application will be further described below in combination with the drawings and specific examples. It should be noted that the drawings are simplified and the proportions are not accurate, and are only used to facilitate and clarify the purpose of assisting in the description of the embodiments of the application.

[0025] The application will be further described below, as shown in a kind of NPC rectifier simplified calculation amount power tracking control method flow chart: Figure 1

[0026] In αβ stationary coordinate system, the mathematical model of three-level NPC rectifier is:

[0027]

[0028] According to instantaneous power theory, the complex power at grid side is:

[0029]

[0030] Under ideal conditions, the differential of grid voltage vector can be expressed as:

[0031]

[0032] Convert (3) to dq rotating coordinate system, and change the control variable to negative conjugate complex power:

[0033]

[0034] In formula (4), E is the peak value of grid voltage vector, R s is the equivalent resistance of power switching device internal resistance and inductance internal resistance, L s is grid inductance, u​dq The output voltage of the rectifier on the AC side in the dq rotating coordinate system.

[0035] The formula (4) is discretized by forward Euler method, and the predicted value of the model negative conjugate complex power at the k+1 time in the dq rotating coordinate system is obtained:

[0036]

[0037] When the zero vector acts, the negative conjugate complex power is:

[0038]

[0039] Therefore, the formula (5) can be changed to:

[0040]

[0041] It can be known from the formula (7) that the negative conjugate complex power at the k+1 time under the action of the zero vector is only related to the negative conjugate complex power at the k time, and is irrelevant to the output voltage of the rectifier on the AC side. According to the reference complex power, the power error of the negative conjugate complex power at the k+1 time can be solved:

[0042]

[0043] Wherein is the power error of the negative conjugate complex power at the k+1 time under the action of the zero vector:

[0044]

[0045] According to the sector where the negative conjugate complex power error at the k+1 time under the action of the zero vector is located, the vector pre-selection is performed, as shown in the figure (2), the sector judgment is divided into 12 sectors, and the judgment formula is:

[0046]

[0047] else

[0048]

[0049] Since the error of the negative conjugate complex power at k+1 moment under the action of the zero vector rotates with the dq axis in the dq synchronous rotating coordinate system, when the sector is determined, the angle of the error vector plus the angle of the grid voltage vector needs to be determined. The function of the floor function is to take the integer part, and the function of the angle function is to take the angle of the complex vector. Since the value range of angle in MATLAB is [-π, π], when the angle of the error vector plus the angle of the grid voltage vector is greater than or equal to 0, it is divided by pi / 6, then the integer part is taken after being rounded down, and 1 is added to obtain the sector; otherwise, when the angle of the error vector plus the angle of the grid voltage vector is less than 0, it is divided by pi / 6, then the integer part is taken after being rounded down, and 13 is added to obtain the sector.

[0050] As can be seen from FIG. (2), according to the sector of the negative conjugate complex power error at k+1 moment under the action of the zero vector, it can be seen that there are 3 zero vectors, 2 small vectors, 1 medium vector and 1 large vector in each sector. According to the position of different vectors and the corresponding switching state, 12 groups can be obtained, each group has 7 preselected vectors, and the specific preselected vectors are shown in the following table:

[0051]

[0052] Different small vectors have different effects on the midpoint potential under different conditions of the upper and lower capacitor voltages on the DC bus side, that is, when the difference between the upper and lower capacitor voltages on the DC side is greater than or equal to 0, a part of the small vectors will make the difference between the two capacitors close to 0, while another part of the small vectors will make the difference between the two capacitors away from 0, thereby causing the midpoint potential to be unbalanced. At this time, in order to ensure the balance of the midpoint potential, when the difference between the upper and lower capacitor voltages on the DC side is greater than 0, the small vector that makes the difference between the two capacitors close to 0 is selected and the small vector that makes the difference between the two capacitors away from 0 is discarded. Similarly, when the difference between the upper and lower capacitor voltages on the DC side is less than 0, the small vector that makes the difference between the two capacitors close to 0 is also selected and the small vector that makes the difference between the two capacitors away from 0 is discarded. The specific small vector grouping is shown in the following table:

[0053]

[0054] The midpoint potential control term in the cost function can be removed, and the overall control method becomes a weightless control. According to the data obtained by analyzing each cycle calculation, the parameter calculation values of the three zero vectors are equal in the polling process, and the cost function is also equal. Therefore, two zero vectors are removed from the polling process, the number of preselected vectors becomes four, and the polling times also become four, which greatly reduces the calculation amount of the control algorithm. The predictive values of the active power and the reactive power at the k+1 moment and the value of the cost function J are calculated in each cycle process, and the vector that makes the cost function minimum is selected after four cycles, which is the optimal vector. When the optimal vector is a zero vector, the output zero vector is selected according to the switching state at the last moment (k moment) and the principle of minimum switching action: when the switching state S abc of the optimal vector at the last moment has two 1s, the output v0(111) is selected; when the switching state S abc of the optimal vector at the last moment has two 0s, the output v0(000) is selected; when the switching state S abc of the optimal vector at the last moment has two -1s, the output v0(-1-1-1) is selected; and in other cases, the output v0(000) is selected to prevent jump. When the optimal vector is a non-zero vector, the optimal non-zero vector is output.

[0055] The cost function formula is:

[0056]

[0057] In the formula, P ref * is the reference active power updated by repeated control, Q ref * is the reactive power reference value, P(k+1) is the active power predictive value at the k+1 moment, and Q(k+1) is the reactive power predictive value at the k+1 moment.

[0058] In the calculation of the active power and reactive power predictive values at the k+1 moment, in view of the problem of inaccurate power tracking in power control, the invention introduces the repeated control idea on the basis of the traditional PI outer ring. The input signal is continuously corrected according to the previous power tracking error signal in each switching period, so as to realize power tracking and eliminate periodic disturbance. The control block diagram is shown in FIG. (3), where Z N represents the delay of N sampling periods. At the end of each moment, the error is accumulated to correct the power reference value, and the specific expression is as follows:

[0059]

[0060] In the formula, k1 and k2 are two coefficients, P ref *(k+1) is the reference active power updated by repetitive control at k+1 moment, P(k) is the active power prediction value at k moment, and P(i) is the active power prediction value at i moment.

[0061] At k moment, the power reference value at k+1 moment is first corrected. The main idea is that the power prediction value at k moment is superimposed with the error of previous sampling. After the power reference value is corrected, the power at k+1 moment is calculated.

[0062] The above embodiments are only for illustrating the technical concept and characteristics of the present application, and the purpose is to enable those skilled in the art to understand the content of the present application and implement it, and cannot limit the protection scope of the present application. Any equivalent changes or modifications made according to the spirit and essence of the present application shall be covered within the protection scope of the present application.

Claims

1. A power tracking control method for reducing the calculation amount of an NPC rectifier, characterized by: The method comprises the following steps: Step one: according to the mathematical model of three-level NPC rectifier in αβ coordinate and the instantaneous power theory, the complex power S of the grid side is calculated, and the differential expression of the negative conjugate complex power under ideal conditions is derived where E is the peak value of the grid-side voltage vector, R s is the equivalent resistance of the power switch device internal resistance and inductance internal resistance, L s is the grid-side inductance, u dq is the rectifier AC side output voltage in the dq rotating coordinate system; Step two: the forward Euler method is used for discretization to obtain the predicted value of the negative conjugate complex power in the dq rotating coordinate system at the k+1 moment; Step three: find the power error of the negative conjugate complex power under the zero vector at k+1 time Step four: Based on the power error Step 1: The sector where the mobile station is located is determined, and a preliminary vector preselection is made, preselecting 12 groups, each group having 7 preselected vectors: 3 zero vectors, 2 small vectors, 1 medium vector, and 1 large vector; Step five: the small vectors preselected in step four are grouped according to the difference between the upper and lower capacitors on the DC bus, and the small vectors that make the difference between the two capacitors close to 0 are selected, while the small vectors that make the difference between the two capacitors far from 0 are discarded, and then the number of preselected vectors in each group is reduced from 7 to 6; Step six: two zero vectors in each group of preselected vectors are removed from the polling process, so that the number of preselected vectors in each group is reduced to 4; Step seven: the predicted values of the active power and the reactive power at the k+1 moment and the value of the cost function J are calculated in a loop, and after four loops, the vector that makes the cost function J minimum is selected as the optimal vector; when the optimal vector is a zero vector, different zero vectors are selected according to the switching state at the k moment and the principle of minimum switching action; when the optimal vector is a non-zero vector, the optimal non-zero vector is output.

2. The power tracking control method with simplified calculation amount for an NPC rectifier according to claim 1, characterized in that: In step five, the small vectors in the 12 groups of preselected vectors are divided into two groups according to whether the difference between the upper and lower capacitor voltages on the DC side is greater than or equal to 0.

3. The method of claim 1, wherein the method is a simplified power tracking control method for an NPC rectifier. In step seven, when the optimal vector is a zero vector, it is divided into four cases: a. Switching state S of the optimal vector at time instant k abc When there are 2 ones, the output v0(111) is selected; b. Switching state S of the optimal vector at time instant k abc When there are 2 zeros, select output v0(000); c. Switching state S of the optimal vector at time k abc When there are 2 -1s, select output v0(-1-1-1); d. In order to prevent jump, v0(000) is selected as the output.

4. The power tracking control method with simplified calculation amount for an NPC rectifier according to claim 1, characterized in that: In step seven, the predicted values of the active power and the reactive power at the k+1 moment are calculated by using the power reference value updating function to correct the power reference value, and the power reference value updating function is In the formula, k1, k2 are two coefficients, P(k+1) is the reference active power updated by repetitive control at k+1 moment, P(k) is the active power prediction value at k moment, and P(i) is the active power prediction value at i moment.

Images