Method for coordinated optimization of output current and neutral point potential of three-level inverter based on predictive control

Abstract

The application discloses a three-level inverter output current and midpoint potential collaborative optimization method based on predictive control, comprising: 1, establishing a predictive model of an NPC three-level inverter in a discrete system, which is used for predicting midpoint potential offset and output voltage; 2, after determining the sector where the output voltage predictive value is located, a double-vector modulation strategy is adopted, only four rolling optimization calculations are needed in each control period to determine the double-vector switch switching sequence, the switch frequency is ensured to be constant, the calculation amount is greatly reduced, and the dependence of traditional single-vector model prediction on high sampling frequency is eliminated; 3, the midpoint potential balance control is realized by directly adjusting the action time of a redundant small vector. Through the double-vector modulation strategy, the switch frequency is constant, and two control targets of output current harmonic and midpoint potential balance are uniformly optimized, so that the dynamic response speed and the steady-state control precision of the system are enhanced, and thus the overall performance can be improved.

Description

TECHNICAL FIELD

[0001] The application relates to a three-level inverter output current and midpoint potential collaborative optimization method based on predictive control, and belongs to the fields of energy storage and variable flow control technology and intelligent control. BACKGROUND

[0002] With the development of control theory and software and hardware technology, the application of finite control set model predictive control (FCS-MPC) in high-power grid-connected technology has attracted more and more attention. The traditional FCS-MPC contains the discrete characteristics and nonlinearity of the inverter, does not need a modulation unit, realizes collaborative optimization control of multiple control objectives through a weight factor, and the core lies in calculating the predicted values of control variables corresponding to different switching states, and relying on a value function to accurately evaluate the error between the command values and the predicted values of the control variables, and finally selecting a group of switching states with the minimum absolute value of the value function as the output of the controller in a control period, so that the system has strong robustness and dynamic response capability. However, each control period needs to calculate the optimal switching state through the enumeration method for 27 groups of switching states, and the amount of calculation will increase significantly. In order to realize the collaborative control of multiple objectives, a weight factor needs to be introduced in the value function, and the weight factor is generally obtained by trial and error method. However, due to the coupling effect between different control objectives, it is difficult to set the weight factor. In addition, the optimal switching state is selected independently based on the current system state and the model prediction in each control period, and the continuity and correlation of the switching state selection between different control periods are ignored, resulting in an unfixed switching frequency of the inverter, an increased current harmonic content, and an increased difficulty in filter design. In order to ensure the dynamic performance and stability of the system, a high sampling frequency is usually required, but in high-power grid-connected inverters, the efficiency of the inverter is closely related to the switching loss. In the single loading mode, the sampling frequency is usually equal to the switching frequency. Therefore, in order to improve the operating efficiency of the inverter, the switching frequency needs to be reduced to reduce the energy loss, and in order to enhance the stability of the system, the switching frequency needs to be fixed, which limits the application range of FCS-MPC to some extent. In summary, although FCS-MPC has significant advantages in high-power grid-connected technology, the problems still need to be further studied and solved. SUMMARY

[0003] In view of the above-mentioned shortcomings of the prior art, the application provides a three-level inverter output current and midpoint potential optimization method based on predictive control, so as to realize constant switching frequency through a double-vector modulation strategy, and uniformly optimize the two control objectives of output current harmonic and midpoint potential balance, so as to enhance the dynamic response speed and steady-state control accuracy of the system, thereby improving the overall performance.

[0004] The application achieves the above-mentioned application purposes by using the following technical scheme.

[0005] The three-level inverter output current and midpoint potential collaborative optimization method based on predictive control has the characteristics that the method comprises the following steps.

[0006] Step 1, a continuous state mathematical model of the NPC three-level inverter in the two-phase stationary coordinate system is established, and after the continuous state mathematical model is discretized, an expression of a midpoint potential offset prediction value and an expression of an inverter output voltage prediction value are obtained.

[0007] Step 2, a correspondence between a switching state and a basic synthesis voltage vector is established, different types of basic synthesis voltage vectors corresponding to the switching state are determined, then a 60° region between every two adjacent center vectors is divided into a sector, so that the two-phase stationary coordinate system is divided into A-F sectors, and there are six basic synthesis voltage vectors in each sector.

[0008] Step 3, according to the voltage prediction value output by the three-level inverter at the k time , the angle θ(k) between the voltage prediction value and the α positive half axis of the two-phase stationary coordinate system is determined. , the sector in which the angle θ(k) is located is determined.

[0009] Step 4, a redundant small vector in each sector is taken as a center small vector, and the remaining four vectors in each sector are taken as candidate vectors, the optimal candidate vector in the candidate vector of each sector is determined through a value function based on a voltage constraint variable, so as to form a double vector combination of each sector together with the center small vector of the corresponding sector.

[0010] Step 5, the respective action time of each vector in the double vector combination is calculated.

[0011] Step 6, when the midpoint potential offset prediction value is 0, an expression of a time distribution factor is constructed, and the respective action time of the center small vector in two switching states is calculated.

[0012] The three-level inverter output current and midpoint potential collaborative optimization method based on predictive control has the characteristics that the step 1 comprises:

[0013] A continuous time state model of the three-level inverter in the two-phase stationary coordinate system is obtained by using formula (1):

[0014] (1)

[0015] In formula (1), is a three-phase output voltage of the inverter , , The expression in the two-phase stationary coordinate system, L is the filter inductance, R is the parasitic resistance of the filter inductance, is the inductor current in the two-phase stationary coordinate system, is the grid voltage in the two-phase stationary coordinate system;

[0016] The dynamic constraint equation of the upper and lower capacitor voltages and the neutral current on the DC side is obtained by using formula (2):

[0017] (2)

[0018] In formula (2), is the neutral current, the direction of the neutral current is positive when the neutral current flows out of the midpoint of the DC side, and vice versa; C represents the DC side capacitor; is the offset of the midpoint potential, is the voltage of the upper capacitor on the DC side, is the voltage of the lower capacitor on the DC side;

[0019] The switching function of the three-phase bridge arm is obtained by using formula (3):

[0020] (3)

[0021] In formula (3), represents the switching function of the phase bridge arm, P represents the switching function of the phase bridge arm in the on state of the first switch tube of the upper bridge arm , the on state of the second switch tube of the upper bridge arm , O represents the switching function of the phase bridge arm in the on state of the second switch tube of the upper bridge arm , the on state of the first switch tube of the lower bridge arm , N represents the switching function of the phase bridge arm in the on state of the first switch tube of the lower bridge arm , the on state of the second switch tube of the lower bridge arm ;

[0022] After discretization of formula (1) by forward Euler, the inductor current prediction value is obtained as shown in formula (4):

[0023] (4)

[0024] In formula (4), is the inductor current prediction value in the two-phase stationary coordinate system at k+1, is the inductor current value in the two-phase stationary coordinate system at k; is the control period; is the grid voltage value in two-phase stationary coordinate system at k moment;

[0025] After discretization of equation (2) by forward Euler, the expression of midpoint potential offset prediction value is obtained as shown in equation (5):

[0026] (5)

[0027] In equation (5), ,are respectively the offset prediction value of midpoint potential at k+1 moment and the offset value of midpoint potential at k moment; is the current value of the middle line at k moment; , are respectively the voltage value of the upper and lower capacitors on the DC side at k+1 moment; , are respectively the voltage value of the upper and lower capacitors on the DC side at k moment;

[0028] The reference prediction value of inductance current at k+1 moment in two-phase stationary coordinate system is obtained by equation (6) :

[0029] (6)

[0030] In equation (6), is the reference prediction value of inductance current at k moment in two-phase stationary coordinate system, is the reference prediction value of inductance current at k-1 moment in two-phase stationary coordinate system, is the reference prediction value of inductance current at k-2 moment in two-phase stationary coordinate system;

[0031] After equation (4) and equation (6) are combined, the expression of voltage prediction value of inverter output is obtained as shown in equation (7):

[0032] (7)

[0033] In equation (7), is the output voltage prediction value of inverter at k moment.

[0034] Further, the correspondence between switching state and basic synthesis voltage vector is established by equation (8) in step 2:

[0035] (8)

[0036] In equation (8), is the DC bus voltage, , , indicates the switching function of phase bridge arm; , These are the projection components of the voltage vector corresponding to the switching state on the α-axis and β-axis, respectively.

[0037] Let the types of basic composite voltage vectors include: zero vector, redundant small vector, medium vector, and large vector; determine the three-phase switch states corresponding to different types of basic composite voltage vectors according to equation (8):

[0038] The three-phase switch state corresponding to the zero vector is OOO;

[0039] The redundant small vectors correspond to the P-type three-phase switch states as POO, PPO, OPO, OPP, OOP, and POP.

[0040] The N-type three-phase switch states corresponding to the redundant small vector are ONN, OON, NON, NOO, NNO, ONO;

[0041] The three-phase switch states corresponding to the medium vector are PON, PNO, OPN, NPO, NOP, and PNP;

[0042] The three-phase switch states corresponding to the large vector are PNN, PPN, NPN, NPP, NNP, and PNP.

[0043] The three-phase switch states corresponding to the six basic composite voltage vectors in each sector are as follows:

[0044] Within sector A, the redundant small vectors correspond to the P-type and N-type three-phase switch states POO and ONN, respectively; the two medium vectors correspond to the three-phase switch states PON and PNO, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state PNN.

[0045] Within sector B, the redundant small vectors correspond to the P-type and N-type three-phase switch states PPO and OON, respectively; the two medium vectors correspond to the three-phase switch states OPN and PON, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state PPN.

[0046] Within sector C, the redundant small vectors correspond to the P-type and N-type three-phase switch states OPO and NON, respectively; the two medium vectors correspond to the three-phase switch states NPO and OPN, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state NPN.

[0047] Within sector D, the redundant small vectors correspond to the P-type and N-type three-phase switch states OPP and NOO, respectively; the two medium vectors correspond to the three-phase switch states NOP and NPO, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state NPP.

[0048] Within sector E, the redundant small vectors correspond to the P-type and N-type three-phase switch states OOP and NNO, respectively; the two medium vectors correspond to the three-phase switch states ONP and NOP, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state NNP.

[0049] Within sector F, the redundant small vectors correspond to the P-type and N-type three-phase switch states POP and ONO, respectively; the two medium vectors correspond to the PNO and ONP, respectively; the zero vector corresponds to the OOO three-phase switch state; and the large vector corresponds to the NPN three-phase switch state.

[0050] Furthermore, step 3 includes:

[0051] when hour, Located in sector A;

[0052] when hour, Located in sector B;

[0053] when hour, Located in sector C;

[0054] when ∪ hour, Located in sector D;

[0055] when hour, Located in sector E;

[0056] when hour, Located in sector F.

[0057] Furthermore, step 4 includes:

[0058] Step 4.1: Determine the optimal candidate vector using equation (9):

[0059] (9)

[0060] In equation (9), The value function is based on the optimal candidate vector under voltage constraint variables. When i=1, it indicates the selection of... The zero vector of the three-phase switch in the sector where the state is OOO is used as a candidate vector and participates in the determination of the optimal candidate vector. When i=2, it indicates that the optimal candidate vector is selected. One of the median vectors in the sector is selected as a candidate vector and participates in the determination of the optimal candidate vector. When i=3, it indicates that the optimal candidate vector is selected. the remaining one of the vectors in the sector as a candidate vector and participates in the determination of the optimal candidate vector, and when i = 4, it means that the large vector in the sector is selected as a candidate vector and participates in the determination of the optimal candidate vector. the remaining one of the vectors in the sector as a candidate vector and participates in the determination of the optimal candidate vector, and when i = 4, it means that the large vector in the sector is selected as a candidate vector and participates in the determination of the optimal candidate vector.

[0061] Step 4.2, in each control period, the value functions of the four candidate vectors are calculated in turn, the candidate vector that makes the value function minimum is selected as the optimal candidate vector, and is paired with the central small vector to form a double vector combination in each control period.

[0062] Further, the step 5 includes:

[0063] Step 5.1, when n = 1, the value of the central small vector in the sector is determined by formula (10): the value of the central small vector in the sector :

[0064] (10)

[0065] Step 5.2, the action time of each double vector is obtained by formula (11):

[0066] (11)

[0067] In formula (11), is the action time of the central small vector in the sector in the control period, is the action time of the optimal candidate vector in the corresponding control period.

[0068] Further, the step 6 includes:

[0069] The relationship between the neutral line current and the three-phase output current is established by formula (12):

[0070] (12)

[0071] In formula (12), is the value of the neutral line current at time k; is the output current of the A-phase bridge arm at time k; is the output current of the B-phase bridge arm at time k; is the output current of the C-phase bridge arm at time k;

[0072] The expression of the average value of the neutral line current is obtained by formula (13):

[0073] (13)

[0074] In formula (13), is the average value of the neutral line current at time k; is the corresponding midline current when the center small vector at time k acts; is the corresponding midline current when the optimal candidate vector at time k acts; is the center small vector time allocation factor;

[0075] When the midpoint potential offset prediction value is 0, the expression of the center small vector time allocation factor is established by formula (14):

[0076] (14)

[0077] The action time of the center small vector P-type three-phase switch state and the action time of the N-type three-phase switch state are obtained by formula (15):

[0078] (15).

[0079] The electronic device comprises a memory and a processor, and the memory is used to store a program supporting the processor to execute the three-level inverter output current and midpoint potential collaborative optimization method based on predictive control, and the processor is configured to execute the program stored in the memory.

[0080] The computer readable storage medium stores a computer program, and when the computer program is run by a processor, the steps of the three-level inverter output current and midpoint potential collaborative optimization method based on predictive control are executed.

[0081] Compared with the prior art, the present application has the following advantages:

[0082] 1. The present application selects a double vector combination in each control cycle by a double vector modulation strategy, realizes constant switching frequency control, overcomes the limitation of the traditional single vector model prediction depending on high sampling frequency, and further reduces output current harmonics and reduces filter design difficulty.

[0083] 2. The present application effectively reduces inverter switching loss by reducing sampling frequency, and improves inverter operation efficiency and controller processing efficiency.

[0084] 3. The present application fully utilizes the redundancy characteristics of small vectors, is conducive to reducing the coupling effect between grid-connected current and capacitor voltage equalization control two control targets, so that the value function only needs to evaluate the current command tracking control target, and the unified optimization control of multiple control targets can be realized without introducing a weight factor, which improves the accuracy and efficiency of prediction and is helpful for engineering practical application. ​​​

[0085] 4、The application can determine the double vector combination through 4 times of rolling optimization calculation of the value function in each control cycle by optimizing the algorithm structure, effectively avoids repeated invalid calculation, reduces the complexity of calculation, and helps to improve the dynamic response performance and steady state control accuracy of the system. BRIEF DESCRIPTION OF DRAWINGS

[0086] Figure 1 It is a topology diagram of the NPC type three-level grid-connected inverter of the application;

[0087] Figure 2 It is a sector diagram of the basic voltage vector of the application;

[0088] Figure 3 It is a switching sequence switching and vector action time diagram of the first A sector

[0089] Figure 4 It is a switching sequence diagram of the 24 double vector combinations of the application;

[0090] Figure 5 It is a control framework diagram of the three-level grid-connected inverter based on the predictive control of the application;

[0091] Figure 6 It is a DC side midpoint potential offset voltage waveform diagram of the application;

[0092] Figure 7 It is a DC side midpoint potential offset voltage waveform diagram of the traditional single vector model prediction;

[0093] Figure 8 It is a three-phase inductance current output waveform diagram of the application;

[0094] Figure 9 It is a three-phase inductance current output waveform diagram of the traditional single vector model prediction. DETAILED DESCRIPTION

[0095] In this embodiment, in order to solve the limitation of traditional finite set model prediction technology (FCS-MPC) applied in high-power grid-connected inverter, a three-level grid-connected inverter output current and midpoint potential collaborative optimization method based on predictive control is proposed. First, the continuous state mathematical model of NPC three-level grid-connected inverter in two-phase stationary coordinate system is established, and then the expression of midpoint potential offset and inverter output voltage prediction value is obtained by discretizing the control system. Secondly, the double vector modulation strategy is adopted, which eliminates the limitation of traditional single vector model prediction relying on high sampling frequency. After determining the sector where the reference voltage is located, only 4 times of rolling optimization calculation is needed in each control period to determine the switch switching sequence of the selected double vector, which ensures the constant of switch frequency and greatly reduces the calculation amount. Finally, after the action time of the center small vector and the optimal candidate vector in each control period is obtained by modulating the model prediction, the time allocation factor formula is used to adjust the action time of the redundant small vector, so as to realize the balanced collaborative optimization control of output current and midpoint potential, and avoid the increase of the complexity of the weight factor. Specifically, the steps include:

[0096] Step 1, the topology of NPC three-level grid-connected inverter is as shown in Figure 1 , first, the continuous time state mathematical model in two-phase stationary coordinate system is as shown in formula (1):

[0097] (1)

[0098] In formula (1), is the inverter output voltage , , is the expression in two-phase stationary coordinate system, L is the filter inductance, R is the filter inductance parasitic resistance, is the inductance current in two-phase stationary coordinate system, is the grid voltage in two-phase stationary coordinate system.

[0099] Step 2, define the midpoint potential offset as shown in formula (2):

[0100] (2)

[0101] In formula (2) is the midpoint potential offset, is the upper DC side capacitor voltage, is the lower DC side capacitor voltage.

[0102] Step 3, construct the dynamic constraint equation of upper and lower DC side capacitor voltage and neutral current as shown in formula (3):

[0103] (3)

[0104] in formula (3), is the neutral current, the neutral current is defined as flowing out of the DC side midpoint, the direction of the neutral current is positive, and vice versa is negative; is the upper DC side capacitor; is the lower DC side capacitor; for the convenience of analysis .

[0105] Step 4, construct the three-phase bridge arm switch function as shown in formula (4):

[0106] (4)

[0107] in formula (3), represents the phase switch function, , P represents the phase switch function in the state when the first switch tube of the upper bridge arm , the second switch tube of the upper bridge arm is turned on, O represents the phase switch function in the state when the second switch tube of the upper bridge arm , the first switch tube of the lower bridge arm is turned on, N represents the phase switch function in the state when the first switch tube of the lower bridge arm , the second switch tube of the lower bridge arm is turned on;

[0108] Step 5, use formula (4) to establish the relationship between the neutral current and the three-phase output current as formula (5):

[0109] (5)

[0110] wherein, is the inductor current, .

[0111] Step 6, use forward Euler to discretize formula (3), formula (1), and obtain the midpoint potential offset prediction value and the inductor current prediction value as shown in formula (6), formula (7):

[0112] (6)

[0113] (7)

[0114] wherein, , are the midpoint potential offset prediction value at k+1 moment and the midpoint potential offset value at k moment respectively; is the neutral current value at k moment; , are respectively the upper and lower DC side capacitor voltage values at time k+1; , are respectively the upper and lower DC side capacitor voltage values at time k; is the inductor current prediction value in two-phase stationary coordinate system at time k+1, is the inductor current value in two-phase stationary coordinate system at time k; is the sampling period of the control system;

[0115] is the grid voltage value in two-phase stationary coordinate system at time k.

[0116] Step 7, the inductor current reference prediction value is obtained by the Lagrange trend extrapolation method as shown in formula (8):

[0117] (8)

[0118] In formula (8), is the inductor current reference prediction value in two-phase stationary coordinate system at time k+1, is the inductor current reference value in two-phase stationary coordinate system at time k, is the inductor current reference value in two-phase stationary coordinate system at time k-1, is the inductor current reference value in two-phase stationary coordinate system at time k-2.

[0119] Step 8, the prediction equation based on voltage constraint variable at time k is obtained by the formula (7), formula (8) from the deadbeat control as shown in formula (9):

[0120] (9)

[0121] In formula (9), is the inverter output voltage prediction value at time k.

[0122] Step 9, according to formula (4), the NPC type three-level topology has 27 different switching states, each switching state corresponds to a basic synthesis voltage vector, the projection calculation method of the voltage vector corresponding to each switching state on the αβ axis is shown in formula (10):

[0123] (10)

[0124] In formula (10), is the DC bus voltage.

[0125] The basic synthetic vectors are divided into four categories according to the modulus size: zero vector, small vector, medium vector and large vector; wherein, the small vectors with the same polarity and modulus value of output voltage have redundant states, and are divided into P-type switching state and N-type switching state according to the different switching states, the generated neutral line current values are the same and the directions are opposite.

[0126] Table 1 switching state corresponding to different types of vectors and generated neutral line current

[0127]

[0128] Step 10, the 60° region between each adjacent two medium vectors is divided into a sector, so that The coordinate system is divided into six sectors, and the output voltage prediction value is determined according to the sector The sector in which the sector is located is determined according to the sector The sector is divided as shown in Figure 2 .

[0129] The double vector modulation strategy is adopted, in order to reduce the complexity of zero vector selection and avoid the influence of switching states PPN and NNN on common mode voltage, the zero vector of OOO is selected; in order to realize constant switching frequency control and fully utilize the redundancy characteristics of small vectors to suppress the midpoint potential offset, the double vector modulation strategy is adopted Taking the first sector as an example, the pair of redundant small vectors in the sector are taken as the center small vectors, the remaining one zero vector (OOO), two medium vectors and one large vector in the sector are taken as candidate vectors, the center small vector and the optimal candidate vector are selected in each control period to form the double vector modulation, and the five-stage switching sequence is adopted, so that the four groups of double vector combination switching sequences selectable in the first sector in each control period are as follows:

[0130] The first group of double vector combination switching sequence: ONN→OOO→POO→OOO→ONN;

[0131] The second group of double vector combination switching sequence: ONN→PON→POO→PON→ONN;

[0132] The third group of double vector combination switching sequence: ONN→PNO→POO→PNO→ONN;

[0133] The fourth group of double vector combination switching sequence: ONN→PNN→POO→PNN→ONN;

[0134] Step 11, the optimal candidate vector is determined by the value function based on the voltage constraint variable, as shown in formula (11):

[0135] (11)

[0136] In formula (11), is a value function based on a voltage constraint variable, in the A sector, i=1 indicates that the voltage vector with the switch state of OOO is selected to participate in evaluation, i=2 indicates that the voltage vector with the switch state of PON is selected to participate in evaluation, i=3 indicates that the voltage vector with the switch state of PNO is selected to participate in evaluation, and i=4 indicates that the voltage vector with the switch state of PNN is selected to participate in evaluation.

[0137] Step 12, in each control period, the value functions of the four candidate vectors are evaluated in turn, the candidate vector that makes the value function minimum is selected as the optimal candidate vector, and is paired with the center small vector to form a double vector combination.

[0138] The application aims to realize the collaborative optimization control of reducing the output current harmonic and balancing the midpoint potential on the basis of fixed switching frequency. The double vector modulation strategy described above helps to select the double vector combination through 4 times of rolling optimization calculation in each control period after determining the sector where the reference voltage is located, optimizes the algorithm structure, greatly reduces the complexity of calculation, improves the dynamic response performance and steady control accuracy of the system, and ensures the constant control of the switching frequency, reduces the output current harmonic, and reduces the filter design difficulty. Meanwhile, the application fully utilizes the redundancy characteristics of the small vector, directly realizes the balancing control of the midpoint potential of the direct current side by adjusting the action time of the redundant small vector, thereby avoiding introducing the weight factor in the value function to reflect the balancing control of the midpoint potential, and causing the complex weight factor optimization problem.

[0139] Step 13, also taking the A sector as an example, after the center small vector and the optimal candidate vector are determined in each control period, the vector action time is obtained through the modulation model predictive control, that is, the vector action time is calculated according to the reciprocal of the value function, the greater the value function, the shorter the vector action time in the corresponding control period, and vice versa. The value function of the center small vector (POO, ONN) based on the voltage constraint variable is shown in formula (12), and the parallel equations (11), (12) obtain the center small vector and the optimal candidate vector time shown in formula (13):

[0140] (12)

[0141] (13)

[0142] Wherein, is the value function of the center small vector in the A sector based on the voltage constraint variable; is the action time of the center small vector in the I sector control period, This refers to the application time of the optimal candidate vector within the corresponding control cycle.

[0143] Step 14, as follows Figure 3 The diagram shows the dual-vector time allocation for sector A. This invention uses the midpoint potential shift prediction value... The time allocation factor is obtained to adjust the action time of the two switching states of the center small vector, so as to realize the DC side midpoint potential balance control. The relationship between the neutral current and the three-phase output current is established using equation (14):

[0144] (14)

[0145] In equation (14): Let k be the value of the midline current. Let k be the output current of phase A bridge arm; Let k be the output current of phase B bridge arm; Let k be the output current of phase C bridge arm.

[0146] Step 15: Use equation (15) to obtain the expression for the neutral current value:

[0147] (15)

[0148] In equation (15): Let be the average value of the midline current at time k; Let be the neutral current corresponding to the action of the small vector at time k; The midline current corresponds to the action of the optimal candidate vector at time k. The time allocation factor for the central small vector.

[0149] Step 16: Use equation (16) to establish the central small vector time allocation factor. The expression:

[0150] (16)

[0151] Step 17: Use equation (17) to obtain the action time of the central small vector P-type switch state and N-type switch state:

[0152] (17)

[0153] In equation (17): The duration of the P-type switch state of the central small vector; The duration of the N-type switch state of the central small vector.

[0154] Similarly, the 20 sets of dual-vector combinations for sector BF can be obtained, and the dual-vector combination switch sequence is as follows: Figure 4As shown, the time allocation factor is obtained by the same method, and after determining the sector where the output voltage is located, the center small vector of each sector can be directly determined, and only 4 rolling optimization calculations are needed in each control period to determine the optimal candidate vector, and then determine the dual vector combination of the participating action in each control period, the dual vector action time is obtained by using the modulation model prediction, and the redundant small vector action time is adjusted by the time allocation factor, to realize the capacitor voltage balancing control. The three-level grid-connected inverter control framework based on predictive control is as shown in the figure. Figure 5 As shown in the figure.

[0155] The following are the main parameters in this embodiment:

[0156]

[0157] The control frequency used in the application is 3kHz, which on the one hand reduces the switching loss and improves the efficiency of the inverter, and on the other hand verifies that the application can still maintain the steady-state control accuracy of the system at a low control frequency. Figure 6 The midpoint potential offset waveform diagram using the modulation strategy of the application, Figure 7 The midpoint potential offset waveform diagram using the traditional single vector model prediction under the same circuit parameter condition, the maximum deviation of the midpoint potential using the method proposed in the application is about 0.5%, and the maximum offset of the midpoint potential using the traditional method is about 2.3%, which shows that the application can realize the balancing control of the capacitor voltage on the DC side by adjusting the redundant small vector action time through the time allocation factor. Figure 8 The inductance current waveform diagram using the modulation strategy of the application, the inductance current amplitude is about 2063A, and the THD is about 2.83%, Figure 9 The inductance current waveform diagram using the traditional control strategy, the inductance current amplitude is about 2049A, and the THD is about 5.21%, which shows that the method used in the application makes the harmonic content of the inverter output current smaller, and Figure 4 It can be inferred that the switching frequency of the inverter is fixed, so the application can realize the collaborative optimization control of the two control targets of small output current harmonic and DC side capacitor voltage balancing control on the basis of fixed switching frequency.

Claims

1. A method for co-optimizing the output current and midpoint potential of a three-level inverter based on predictive control, characterized in that, Includes the following steps: Step 1: Establish a continuous state mathematical model of the NPC-type three-level inverter in a two-phase stationary coordinate system, and after discretizing the continuous state mathematical model, obtain the expression for the predicted value of the midpoint potential offset and the expression for the predicted value of the inverter output voltage. Step 2: Establish the correspondence between the switching state and the basic composite voltage vector, determine the switching state corresponding to different types of basic composite voltage vectors, and then divide the 60° region between each two adjacent vectors into a sector, thereby dividing the two-phase stationary coordinate system into AF sectors, and each sector has a total of 6 basic composite voltage vectors. Step 3: Based on the predicted voltage output value of the three-level inverter at time k. The angle θ(k) between the two stationary coordinate systems and the positive α-axis is determined. The sector it belongs to; Step 4: Take the redundant small vectors in each sector as the central small vectors and the remaining four vectors in each sector as candidate vectors. Determine the optimal candidate vector among the candidate vectors of each sector through the value function based on the voltage constraint variables, so as to form a two-vector combination of each sector with the central small vector of the corresponding sector. Step 5: Calculate the individual action time of each vector in the two-vector combination; Step 6: When the predicted value of the midpoint potential offset is 0, construct the expression for the time allocation factor and calculate the time of action of the center small vector in each of the two switching states.

2. The method for co-optimizing the output current and midpoint potential of a three-level inverter based on predictive control according to claim 1, characterized in that, Step 1 includes: Using equation (1), the continuous-time state model of the three-level inverter in the two-phase stationary coordinate system is obtained: (1) In equation (1), The three-phase output voltage of the inverter , , In the expression for a two-phase stationary coordinate system, L is the filter inductance, and R is the parasitic resistance of the filter inductance. The inductor current is in a two-phase stationary coordinate system. The grid voltage is in a two-phase stationary coordinate system. Using equation (2), the dynamic constraint equations for the DC-side upper and lower capacitor voltages and the neutral current are obtained: (2) In equation (2), The neutral current is defined as positive when it flows out of the midpoint of the DC side, and negative otherwise; C represents the DC side capacitance. This represents the offset of the midpoint potential. This is the voltage across the capacitor on the DC side. This is the voltage of the capacitor on the DC side; The switching function of the three-phase bridge arm is obtained using equation (3): (3) In formula (3): express The switching function of the phase bridge arm, P represents The switching function of the phase bridge arm is in the first switching transistor of the upper bridge arm. The second switch of the upper bridge arm is turned on. The switch state when it is on, O indicates The switching function of the phase bridge arm is in the second switching transistor of the upper bridge arm. The first switching transistor of the lower bridge arm is turned on. The switching state when the circuit is on, N represents The switching function of the phase bridge arm is in the first switching transistor of the lower bridge arm. The second switch of the lower bridge arm is turned on. The switch state when it is on; After discretizing equation (1) using forward Euler, the predicted inductor current value is obtained as shown in equation (4): (4) In equation (4), The value of the inductor current at time k+1 is the predicted value in the two-phase stationary coordinate system. Let k be the inductor current value in the two-phase stationary coordinate system at time k; To control the cycle; Let k be the grid voltage value in the two-phase stationary coordinate system at time k; After discretizing equation (2) using forward Euler, we obtain the expression for the midpoint potential shift prediction value as shown in equation (5): (5) In equation (5), , These are the predicted offset of the midpoint potential at time k+1 and the offset of the midpoint potential at time k, respectively. Let k be the current value of the midline at time k; , These are the voltage values ​​of the upper and lower capacitors on the DC side at time k+1, respectively. , These are the voltage values ​​of the upper and lower capacitors on the DC side at time k, respectively; The reference predicted value of the inductor current at time k+1 in the two-phase stationary coordinate system is obtained using equation (6). : (6) In equation (6), This is the reference predicted value of the inductor current at time k in a two-phase stationary coordinate system. This is the reference predicted value of the inductor current at time k-1 in a two-phase stationary coordinate system. This is the reference predicted value of the inductor current at time k-2 in the two-phase stationary coordinate system; Combining equations (4) and (6), we obtain the expression for the predicted voltage value of the inverter output as shown in equation (7): (7) In equation (7), Let k be the predicted output voltage of the inverter at time k.

3. The method for co-optimizing the output current and midpoint potential of a three-level inverter based on predictive control according to claim 1, characterized in that, In step 2, the correspondence between the switching state and the basic composite voltage vector is established using equation (8): (8) In equation (8), This is the DC bus voltage. , , express Switching functions of the phase bridge arms; , These are the projection components of the voltage vector corresponding to the switching state on the α-axis and β-axis, respectively; The types of basic composite voltage vectors include: zero vector, redundant small vector, medium vector, and large vector; The three-phase switch states corresponding to different types of basic composite voltage vectors are determined according to equation (8): The three-phase switch state corresponding to the zero vector is OOO; The redundant small vectors correspond to the P-type three-phase switch states as POO, PPO, OPO, OPP, OOP, and POP. The N-type three-phase switch states corresponding to the redundant small vector are ONN, OON, NON, NOO, NNO, ONO; The three-phase switch states corresponding to the medium vector are PON, PNO, OPN, NPO, NOP, and PNP; The three-phase switch states corresponding to the large vector are PNN, PPN, NPN, NPP, NNP, and PNP. The three-phase switch states corresponding to the six basic composite voltage vectors in each sector are as follows: Within sector A, the redundant small vectors correspond to the P-type and N-type three-phase switch states POO and ONN, respectively; the two medium vectors correspond to the three-phase switch states PON and PNO, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state PNN. Within sector B, the redundant small vectors correspond to the P-type and N-type three-phase switch states PPO and OON, respectively; the two medium vectors correspond to the three-phase switch states OPN and PON, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state PPN. Within sector C, the redundant small vectors correspond to the P-type and N-type three-phase switch states OPO and NON, respectively; the two medium vectors correspond to the three-phase switch states NPO and OPN, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state NPN. Within sector D, the redundant small vectors correspond to the P-type and N-type three-phase switch states OPP and NOO, respectively; the two medium vectors correspond to the three-phase switch states NOP and NPO, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state NPP. Within sector E, the redundant small vectors correspond to the P-type and N-type three-phase switch states OOP and NNO, respectively; the two medium vectors correspond to the three-phase switch states ONP and NOP, respectively; the zero vector corresponds to the three-phase switch state OOO; and the large vector corresponds to the three-phase switch state NNP. Within sector F, the redundant small vectors correspond to the P-type and N-type three-phase switch states POP and ONO, respectively; the two medium vectors correspond to the PNO and ONP, respectively; the zero vector corresponds to the OOO three-phase switch state; and the large vector corresponds to the NPN three-phase switch state.

4. The method for co-optimizing the output current and midpoint potential of a three-level inverter based on predictive control according to claim 1, characterized in that, Step 3 includes: when hour, Located in sector A; when hour, Located in sector B; when hour, Located in sector C; when ∪ hour, Located in sector D; when hour, Located in sector E; when hour, Located in sector F.

5. The method for co-optimizing the output current and midpoint potential of a three-level inverter based on predictive control according to claim 3, characterized in that, Step 4 includes: Step 4.1: Determine the optimal candidate vector using equation (9): (9) In equation (9), The value function is based on the optimal candidate vector under voltage constraint variables. When i=1, it indicates the selection of... The zero vector of the three-phase switch in the sector where the state is OOO is used as a candidate vector and participates in the determination of the optimal candidate vector. When i=2, it indicates that the optimal candidate vector is selected. One of the median vectors in the sector is selected as a candidate vector and participates in the determination of the optimal candidate vector. When i=3, it indicates that the optimal candidate vector is selected. The remaining median vector in the sector is used as a candidate vector and participates in the determination of the optimal candidate vector. When i=4, it indicates that the optimal candidate vector is selected. Large vectors in the sector are used as candidate vectors and participate in the determination of the optimal candidate vector; Step 4.2: In each control cycle, calculate the value function of the four candidate vectors in sequence, select the candidate vector that minimizes the value function as the optimal candidate vector, and pair it with the central small vector to form a two-vector combination in each control cycle.

6. The method for co-optimizing the output current and midpoint potential of a three-level inverter based on predictive control according to claim 4, characterized in that, Step 5 includes: Step 5.1: When n=1, use equation (10) to determine the position of the n-th cell. The value of the center vector of the sector : (10) Step 5.2: Use equation (11) to obtain the action time of each of the two vectors: (11) In equation (11), for The duration of the central small vector's influence within the control cycle of the sector in question. This refers to the duration of action of the optimal candidate vector within the corresponding control cycle.

7. The method for co-optimizing the output current and midpoint potential of a three-level inverter based on predictive control according to claim 6, characterized in that, Step 6 includes: Using equation (12), the relationship between the neutral current and the three-phase output current can be established: (12) In equation (12): Let k be the value of the midline current. Let k be the output current of phase A bridge arm; Let k be the output current of phase B bridge arm; The output current of phase C bridge arm at time k; The expression for the average value of the neutral current is obtained using equation (13): (13) In equation (13): Let be the average value of the midline current at time k; Let be the neutral current corresponding to the action of the small vector at time k; The midline current corresponds to the action of the optimal candidate vector at time k. The time allocation factor for the central small vector; When the predicted value of the midpoint potential offset is 0, the time allocation factor of the center small vector is established using equation (14). The expression: (14) The state of the central small vector P-type three-phase switch is obtained using equation (15). The duration of action time and the duration of action time of N-type three-phase switch state : (15)。 8. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports the processor in executing the predictive control-based three-level inverter output current and midpoint potential co-optimization method according to any one of claims 1-7, and the processor is configured to execute the program stored in the memory.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, it executes the steps of the method for co-optimizing the output current and midpoint potential of a three-level inverter based on predictive control as described in any of claims 1-7.

Images