Three-phase cascaded h-bridge inter-phase battery charge equalization method based on model predictive control

Abstract

This invention discloses a method for interphase battery charge balancing in a three-phase cascaded H-bridge based on model predictive control. A simulation circuit model of a 7-level cascaded H-bridge grid-connected inverter based on model predictive control was built in Matlab / Simulink simulation software, and the simulation results were recorded. The DC-side voltage vo of each H-bridge unit of the inverter is specified. dc The value is 150V, and the mains voltage is v. g The values ​​are set to 220V, the fundamental frequency to 50Hz, the sampling time T for model predictive control to 0.1ms, the grid-connected filter inductor L to 1mH, and the equivalent resistance R of the grid-connected filter inductor to 0.1Ω. The SOC is set... a =70.1%, SOC b =70%, SOC c =69.9%, ε=0.01, the grid-connected filter inductor L is 1mH, the equivalent resistance R of the grid-connected filter inductor is 0.1Ω, the grid-connected current is controlled by model predictive control, and a reward function is introduced to select redundant voltage vectors to balance the phase SoC without affecting the current performance. After the error of the SoC is less than a certain given value, the SoC balancing is stopped to achieve the minimum common mode voltage.

Description

Technical Field

[0001] This invention relates to the field of three-phase cascaded H-bridge interphase battery technology, specifically a charge balancing method for three-phase cascaded H-bridge interphase batteries based on model predictive control. Background Technology

[0002] To alleviate the energy crisis, green new energy sources have received unprecedented attention. However, wind and solar power generation suffer from intermittent and intermittent issues, necessitating the addition of battery energy storage systems. Cascaded H-bridge topologies can integrate distributed energy storage and deliver it to the grid or load, offering advantages such as improved output voltage waveform quality, reduced filter size, and lower common-mode voltage between inverter phases, making them suitable for distributed battery energy storage systems. To control the phase and frequency of the inverter's grid-connected current and improve current quality, domestic and international researchers have employed pulse width modulation (PWM), vector control, and model predictive control (MMC). Among these, MMC offers advantages such as simple design, multi-objective control, effective containment and limitation of nonlinear components, and faster dynamic response, making it widely used in power electronic converters. Finite set MMC is particularly suitable for controlling cascaded H-bridge inverters. Finite set MMC calculates the cost function of each existing voltage vector combination within a sampling period and selects the switching state of the smallest vector combination to apply to the inverter's switching transistors, without requiring a modulator.

[0003] Due to differences in battery manufacturing processes and operating environments, as well as issues such as overcharging and over-discharging during operation, the capacity and state of charge (SOC) of the batteries in each cascaded H-bridge cell are inconsistent, thus affecting the utilization rate of the energy storage system. For cascaded H-bridge cells with energy storage systems, it is necessary to ensure SOC balance among the three phases. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] To address the shortcomings of existing technologies, this invention provides a method for charge balancing of interphase batteries in a three-phase cascaded H-bridge based on model predictive control. This method has advantages such as stopping SoC balancing once the error of the SoC is less than a certain given value, thus achieving the minimum common-mode voltage. It also solves the problem of inconsistent capacity and state of charge of batteries in each cascaded H-bridge unit.

[0006] (II) Technical Solution

[0007] To achieve the goal of stopping SoC equalization once the error of the existing SoC falls below a certain given value, thus minimizing the common-mode voltage, this invention provides the following technical solution:

[0008] A method for charge balancing between phase-to-phase batteries in a three-phase cascaded H-bridge based on model predictive control, characterized by the following steps:

[0009] S1, SoC in the sorted battery

[0010] L is the inductance value of the grid-connected filter, R is the equivalent internal resistance of the filter inductor, and v dc For each H-bridge unit, v a v b v c The inverter outputs three-phase voltage, v ga v gb v gc For the three-phase power grid voltage, i a 、、i b i c This refers to the three-phase grid-connected current.

[0011] S2, SoC in the balanced battery

[0012] The SoC in the battery is balanced by adjusting the action vector and data in the battery module.

[0013] Preferably, in step S1, the mathematical model for the grid-connected inverter in the αβ coordinate system is as follows:

[0014]

[0015] Among them, i α and i β These represent the values ​​of the three-phase grid-connected current on the α-axis and β-axis, respectively. gα and v gβ These represent the values ​​of the three-phase grid voltage on the α and β axes, respectively. α and v β These represent the values ​​of the three-phase voltage output by the inverter on the α-axis and β-axis, respectively.

[0016] Preferably, in step S1, the learned model shown in equation (1) is discretized using forward Euler method to obtain:

[0017]

[0018] Where T is the sampling time, representing the interval between sampling periods k and k+1. When T is sufficiently small, equation (2) will be approximately equal to equation (1), i α (k+1) and i β(k+1) represent the predicted values ​​of the grid-connected current along the α and β axes at time k+1, respectively. Different voltage vectors will produce different predicted values, and there will be varying degrees of error between the predicted values ​​and the reference current. Therefore, in order for the current to track the reference current, a cost function needs to be defined:

[0019]

[0020]

[0021] in, and This is the predicted current reference value at time k+1. When the sampling time T is sufficiently small, it can be...

[0022] Preferably, in step S1, the prediction error is evaluated by the cost function (3), and after calculating all cost functions, the voltage state corresponding to the smallest cost function is selected by formula (4) and applied to the inverter system (1).

[0023] Preferably, in step S1, in the αβ coordinate system, there exist redundant voltage vectors that have the same effect on current tracking but different common-mode voltages; that is, they are different voltage vectors in the three stationary coordinate systems, but the vectors are the same in the α-β system. For example... Figure 2 As shown. The minimum common-mode voltage vector is There are redundant voltage vectors corresponding to the same Relationship:

[0024]

[0025] Among them, T 3s-2s For Clark transform, λ is the coefficient between the redundant voltage vector and the minimum common-mode voltage vector, v dc The DC-side battery voltage for each H-bridge unit.

[0026] Preferably, in step S2, in order to achieve inter-phase SoC equalization, the following needs to be defined:

[0027]

[0028] ΔSOC y =SOC-SOC y ,(y=a,b,c) (7)

[0029] Where SoC refers to the average SoC of the three-phase bridge circuit, SOC y Let ΔSOC be the average SoC value of phase y. y This represents the difference between the three-phase balanced SoC and the y-th phase average SoC.

[0030] Preferably, in step S2, a reward function is defined to achieve SoC equalization between battery phases:

[0031]

[0032] Where k is the sampling period at time k. By selecting different λ values, power redistribution can be achieved in a three-phase cascaded H-bridge, realizing inter-phase SoC equalization. For example, when ΔSOC a >0, i a When (k)>0, in order to allocate more power to phase a, λ>0 should be satisfied; when ΔSOC a <0, i a When (k) > 0, to ensure that phase a receives less power, λ > 0 should be satisfied. This method similarly applies to the SoC equalization of phases b and c. Different λ values ​​will produce different reward function results; the maximum value g is chosen. y The voltage vector corresponding to (k) is applied in the inverter to achieve the fastest inter-phase SoC equalization, i.e.

[0033] {v′ a , v′ b , v′ c} = arg(max(g y (k))) (9).

[0034] Preferably, in step S2, in order to maintain a smaller common-mode voltage after inter-phase SoC equalization, the sum of errors e between inter-phase SoCs is defined. soc (k) is as follows:

[0035]

[0036] when e soc When (k) > ε, the system will execute equations (8) and (9). When the battery phase-to-phase SoC is balanced to a certain extent, that is, when e is satisfied... soc If (k) < ε, the system no longer executes equations (8) and (9), but directly selects the voltage vector combination with the minimum common-mode voltage, thereby achieving the minimum common-mode voltage in the balanced state. The expression for the common-mode voltage is:

[0037]

[0038] (III) Beneficial Effects

[0039] Compared with existing technologies, this invention provides a method for charge balancing of interphase batteries in a three-phase cascaded H-bridge based on model predictive control, which has the following advantages:

[0040] 1. The three-phase cascaded H-bridge phase-to-phase battery charge balancing method based on model predictive control controls the grid-connected current through model predictive control. Without affecting the current performance, a reward function is introduced to select redundant voltage vectors to balance the phase-to-phase common-mode voltage (SoC). After the SoC error is less than a certain given value, SoC balancing is stopped to achieve the minimum common-mode voltage. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the three-phase cascaded H-bridge grid-connected inverter structure of the present invention;

[0042] Figure 2 This is a schematic diagram illustrating the distribution of different numbers of voltage redundancy vectors in the αβ coordinate system according to the present invention;

[0043] Figure 3 This is a waveform diagram of phase-to-phase SoC equalization in this invention;

[0044] Figure 4 This is a schematic diagram of the three-phase power grid voltage and three-phase power grid current of the present invention;

[0045] Figure 5 This is a schematic diagram of the voltage of the three-phase inverter of the present invention;

[0046] Figure 6 This is a schematic diagram of the common-mode voltage of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] A method for charge balancing between phase-to-phase batteries in a three-phase cascaded H-bridge based on model predictive control, characterized by the following steps:

[0049] S1, SoC in the sorted battery

[0050] L is the inductance value of the grid-connected filter, R is the equivalent internal resistance of the filter inductor, and v dc For each H-bridge unit, v a v b v c The inverter outputs three-phase voltage, v ga v gb v gc For the three-phase power grid voltage, i a 、、i b i c This refers to the three-phase grid-connected current.

[0051] S2, SoC in the balanced battery

[0052] The SoC in the battery is balanced by adjusting the action vector and data in the battery module.

[0053] In the αβ coordinate system, the mathematical model for a grid-connected inverter is as follows:

[0054]

[0055] Among them, i α and i β These represent the values ​​of the three-phase grid-connected current on the α-axis and β-axis, respectively. gα and v gβ These represent the values ​​of the three-phase grid voltage on the α and β axes, respectively. α and v β Let α and β be the values ​​of the three-phase voltage output from the inverter, respectively. By performing forward Euler discretization on the learned model shown in equation (1), we can obtain:

[0056]

[0057] Where T is the sampling time, representing the interval between sampling periods k and k+1. When T is sufficiently small, equation (2) will be approximately equal to equation (1), i α (k+1) and i β (k+1) represent the predicted values ​​of the grid-connected current along the α and β axes at time k+1, respectively. Different voltage vectors will produce different predicted values, and there will be varying degrees of error between the predicted values ​​and the reference current. Therefore, in order for the current to track the reference current, a cost function needs to be defined:

[0058]

[0059] {v a v β}=arg(min(g(k+1))) (4)

[0060] in, and This is the predicted current reference value at time k+1. When the sampling time T is sufficiently small, it can be... The prediction error is evaluated by the cost function (3). After calculating all cost functions, the voltage state corresponding to the minimum cost function is selected by equation (4) and applied to the inverter system (1). In the αβ coordinate system, there are redundant voltage vectors that have the same impact on current tracking but have different common-mode voltages. That is, they are different voltage vectors in the three stationary coordinate systems, but the vectors are the same in the α-β system. Figure 2As shown. The minimum common-mode voltage vector is There are redundant voltage vectors corresponding to the same Relationship:

[0061]

[0062] Among them, T 3s-2s For Clark transform, λ is the coefficient between the redundant voltage vector and the minimum common-mode voltage vector, v dc The DC-side battery voltage for each H-bridge unit.

[0063] To achieve phase-to-phase SoC equalization, the following needs to be defined:

[0064]

[0065] ΔSOC y =SOC-SOC y ,(y=a,b,c) (7)

[0066] Where SoC refers to the average SoC of the three-phase bridge circuit, SOC y Let ΔSOC be the average SoC value of phase y. y This represents the difference between the three-phase balanced SoC and the y-th phase average SoC.

[0067] Battery phase-to-phase SoC equalization is achieved by defining a reward function:

[0068]

[0069] Where k is the sampling period at time k. By selecting different λ values, power redistribution can be achieved in a three-phase cascaded H-bridge, realizing inter-phase SoC equalization. For example, when ΔSOC a >0, i a When (k)>0, in order to allocate more power to phase a, λ>0 should be satisfied; when ΔSOC a <0, i a When (k) > 0, to ensure that phase a receives less power, λ > 0 should be satisfied. This method similarly applies to the SoC equalization of phases b and c. Different λ values ​​will produce different reward function results; the maximum value g is chosen. y The voltage vector corresponding to (k) is applied in the inverter to achieve the fastest inter-phase SoC equalization, i.e.

[0070] {v′ a , v′ b , v′ c} = arg(max(g y (k))) (9)

[0071] To maintain a smaller common-mode voltage after inter-phase SoC equalization, the sum of errors between inter-phase SoCs, e, is defined. soc (k) is as follows:

[0072]

[0073] when e soc When (k) > ε, the system will execute equations (8) and (9). When the battery phase-to-phase SoC is balanced to a certain extent, that is, when e is satisfied... soc If (k) < ε, the system no longer executes equations (8) and (9), but directly selects the voltage vector combination with the minimum common-mode voltage, thereby achieving the minimum common-mode voltage in the balanced state. The expression for the common-mode voltage is:

[0074]

[0075] A simulation circuit model of a 7-level cascaded H-bridge grid-connected inverter based on model predictive control was built in Matlab / Simulink simulation software, and the simulation results were recorded. The DC-side voltage vo of each H-bridge unit of the inverter was recorded. dc The value is 150V, and the mains voltage is v. g The values ​​are set to 220V, the fundamental frequency to 50Hz, the sampling time T for model predictive control to 0.1ms, the grid-connected filter inductor L to 1mH, and the equivalent resistance R of the grid-connected filter inductor to 0.1Ω. The SOC is set... a =70.1%, SOC b =70%, SOC c =69.9%, ε=0.01.

[0076] like Figure 3 As shown, the three-phase bridge circuit operates in charge / discharge switching mode. After using the proposed method, the phase-to-phase SoCs reached equilibrium at 2.52 seconds. This method does not affect the quality of the grid-connected current, maintaining a total harmonic distortion of 2.24% before and after equilibrium, and ensuring that the grid voltage and grid current maintain a power factor close to -1 (when the battery is in charging mode). The waveform is shown below. Figure 4 As shown. After the phase-to-phase SoC is balanced, the inverter's three-phase voltage changes from an asymmetrical and dissimilar waveform to a symmetrical and similar waveform, as shown. Figure 5 As shown, the common-mode voltage drops from a maximum of 432V to 65V, as... Figure 6 As shown.

[0077] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for inter-phase battery charge equalization for three-phase cascaded H-bridge based on model predictive control, characterized in that, Includes the following steps: S1, Sorting the SoC value in the battery L is the grid-connected filter inductance value, R is the equivalent internal resistance of the filter inductance, Vbat is the battery voltage value of each H-bridge unit, Vinv is the inverter output three-phase voltage, Vgrid is the three-phase grid voltage, Igrid is the three-phase grid-connected current; S2, SoC in the balanced battery By adjusting the action vector and data in the battery module, the SoC in the battery is brought into balance; In step S1, the mathematical model for the grid-connected inverter in the αβ coordinate system is as follows: in, and These represent the values ​​of the three-phase grid-connected current on the α-axis and β-axis, respectively. and These represent the values ​​of the three-phase grid voltage on the α-axis and β-axis, respectively. and These represent the values ​​of the three-phase voltage output by the inverter on the α-axis and β-axis, respectively. In step S1, the learned model shown in equation (1) is discretized using forward Euler method, resulting in: Where T is the sampling time, representing the interval between sampling periods k and k+1. When T is sufficiently small, equation (2) will be approximately equal to equation (1). and Let be the predicted values ​​of the grid-connected current along the α and β axes at time k+1, respectively. Different voltage vectors will produce different predicted values, and there will be varying degrees of error between the predicted values ​​and the reference current. Therefore, in order for the current to track the reference current, a cost function needs to be defined: in, and This is the predicted current reference value at time k+1. When the sampling time T is sufficiently small, it can be... ; In step S1, the prediction error is evaluated by the cost function (3), and after calculating all cost functions, the voltage state corresponding to the smallest cost function is selected by equation (4) and applied to the inverter system (1). In step S2, a reward function is defined to achieve battery phase SoC equalization: Where k is the sampling period at time k, which is determined by selecting different... This value enables power redistribution of a three-phase cascaded H-bridge, achieving inter-phase SoC equalization. In order to allocate more power to phase a, the following should be satisfied: ;when In order to allocate less power to phase a, the following should be satisfied: This method is similarly applicable to SoC equalization of phases b and c, for different Different reward function results will be generated; choose the maximum value. The corresponding voltage vector is applied in the inverter to achieve the fastest inter-phase SoC equalization, i.e. 。 2. The method for charge balancing of interphase batteries in a three-phase cascaded H-bridge based on model predictive control according to claim 1, characterized in that, In step S1, under the αβ coordinate system, there exist redundant voltage vectors that have the same effect on current tracking but different common-mode voltages. That is, they are different voltage vectors under the three stationary coordinate systems, but the vectors are the same under the α-β system. The minimum common-mode voltage vector is... There are redundant voltage vectors corresponding to the same Relationship: in, For Clark transform, , The coefficients between the redundant voltage vector and the minimum common-mode voltage vector are... The DC-side battery voltage for each H-bridge unit.

3. The method for charge balancing of interphase batteries in a three-phase cascaded H-bridge based on model predictive control according to claim 1, characterized in that, In step S2, in order to achieve inter-phase SoC equalization, the following needs to be defined: Where SoC is the average SoC of the three-phase bridge circuit. Let y be the average SoC value of the y-th phase. This represents the difference between the three-phase balanced SoC and the y-th phase average SoC.

4. The method for charge balancing of interphase batteries in a three-phase cascaded H-bridge based on model predictive control according to claim 1, characterized in that, In step S2, in order to maintain a smaller common-mode voltage after inter-phase SoC equalization, the sum of errors between inter-phase SoCs is defined. as follows: when When the system executes equations (8) and (9), the battery phase SoC is balanced to a certain extent, thus satisfying the condition. The system no longer executes equations (8) and (9), but directly selects the voltage vector combination with the minimum common-mode voltage, thereby achieving the minimum common-mode voltage under balanced conditions. The expression for the common-mode voltage is: 。

Images