Method and device for controlling three-phase four-leg LCL grid-connected converter

Abstract

The invention discloses a control method and device for a three-phase four-leg LCL grid-connected converter, and belongs to the technical field of grid-connected converter control. According to the method, the converter system prediction model and the resonance state extension observer of the converter in the static three-phase coordinate system are constructed through the operation variable data, and then the converter is observed through the resonance state extension observer. Grid-connected voltage observation data of the converter at the current time step and grid-connected voltage estimation data of the converter at the first prediction time step are obtained, and then prediction is carried out through a converter system prediction model to obtain power prediction data of a second prediction time step; and then calculating power reference prediction data of the converter in the second prediction time step and constructing a control objective function of the converter, solving the control objective function to obtain an estimated switching state of the converter in the first prediction time step, and further controlling the converter, thereby improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter.

Description

Technical Field

[0001] The present invention belongs to the technical field of grid-connected converter control, and particularly relates to a control method and device for a three-phase four-leg LCL grid-connected converter. Background Art

[0002] With the rapid development of new energy and distributed generation systems, the current demand for high-performance grid-connected converters is increasing day by day. In the current power system, a three-phase four-leg LCL grid-connected converter, which includes a four-leg topology and an inductor-capacitor-inductor (LCL) filter, compared with traditional three-phase three-wire converters and three-phase four-wire converters, can achieve efficient utilization of the DC bus voltage by controlling the neutral point voltage, and can provide good harmonic suppression ability through the LCL filter structure. At present, three-phase four-leg LCL grid-connected converters have been widely used in power systems.

[0003] Due to the inherent resonance characteristics of the three-phase four-leg LCL grid-connected converter, it is necessary to control the three-phase four-leg LCL grid-connected converter. The currently more commonly used solution is a control solution without a grid voltage sensor. For example, the three-phase four-leg LCL grid-connected converter is controlled by a single model predictive control method. However, this method has problems such as model parameter mismatch and control delay, resulting in a larger prediction deviation and poor tracking effect on grid power data. In addition, there is also a solution to control the converter through an extended state observer. However, this solution does not consider the frequency fluctuation of the converter under resonance characteristics or the performance under unbalanced conditions. Currently, the methods for controlling the converter all result in poor stability and dynamic performance of the converter. Therefore, there is an urgent need for a control method and device for a three-phase four-leg LCL grid-connected converter to solve the defects of the existing technology. Summary of the Invention

[0004] The present invention aims to provide a control method and device for a three-phase four-leg LCL grid-connected converter to solve the technical problem of poor stability and dynamic performance of the converter in the existing technology. Voltage observation and prediction are carried out through a converter system prediction model and a resonance state extended observer, realizing the control of the converter and improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter.

[0005] To solve the above technical problems, an embodiment of the present invention provides a control method for a three-phase four-leg LCL grid-connected converter, which is applicable to a converter of three-phase four-leg LCL and includes:

[0006] Obtain the operating variable data of the converter, and construct a converter system prediction model and a resonant state extended observer of the converter in the stationary three-phase coordinate system based on the operating variable data; observe the converter based on the resonant state extended observer to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step; input the grid-connected voltage observation data and the grid-connected voltage estimation data into the converter system prediction model to obtain the power prediction data of the converter at the second prediction time step; determine the power reference prediction data of the converter at the second prediction time step based on the grid-connected voltage observation data, the grid-connected voltage estimation data and the preset power reference data; obtain the real-time switching state of the converter at the current time step, and construct a control objective function of the converter based on the real-time switching state, the predicted power data and the predicted power reference data; solve the control objective function to obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state.

[0007] It can be understood that, compared with the prior art, the present invention constructs a converter system prediction model and a resonant state extended observer of the converter in a stationary three-phase coordinate system through the operating variable data of a three-phase four-leg LCL grid-connected converter. Then, the converter is observed through the resonant state extended observer to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step. Then, the power prediction data of the converter at the second prediction time step is predicted through the converter system prediction model. Then, the power reference prediction data of the converter at the second prediction time step is calculated through the grid-connected voltage observation data, the grid-connected voltage estimation data and the preset power reference data, and the control objective function of the converter is constructed through the real-time switch state, the predicted power data and the predicted power reference data. Then, the control objective function is solved to obtain the estimated switch state of the converter at the first prediction time step, and the converter is controlled based on the estimated switch state, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter. By constructing a converter system prediction model and a resonant state extended observer in the stationary coordinate system, the present invention avoids the computational complexity caused by introducing links such as rotational coordinate transformation, cross decoupling and positive and negative sequence component decomposition in the traditional model predictive control technology, reduces the computational complexity, and improves the dynamic performance of the converter and its ability to cope with unbalanced operating conditions; through the converter system prediction model and the resonant state extended observer, the present invention realizes the deep integration of the dynamic observation and multi-step prediction of the grid-connected voltage. The resonant state extended observer can accurately and directly observe the frequency fluctuation of the converter under resonant characteristics or the operating data under unbalanced conditions, that is, it can ensure high-precision tracking of the disturbance at a specific resonant frequency, thereby realizing the grid voltage observation in the three-phase stationary coordinate system; with the multi-step prediction of the current time step, the first prediction time step and the second prediction time step, a closed-loop feedback mechanism is formed. The double-time-step prediction mechanism (current time step observation + first prediction time step estimation) compensates for the inherent delay defect of digital control, effectively avoiding the prediction deviation increase caused by model parameter mismatch and control delay in the traditional model predictive control and the poor tracking effect on grid power data; by constructing and solving the control objective function, the control strategy of the converter can be dynamically adjusted, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, thereby realizing the independent control of the grid-connected power of each phase under the condition of grid asymmetry or frequency fluctuation.

[0008] Correspondingly, an embodiment of the present invention provides a control device for a three-phase four-leg LCL grid-connected converter, including: a converter modeling module, a resonant state extended observer observation module, a converter system prediction model prediction module, a power reference prediction data acquisition module, a control objective function construction module, and a converter control module; wherein, the converter modeling module is used to obtain the operating variable data of the converter, and construct a converter system prediction model and a resonant state extended observer of the converter in a stationary three-phase coordinate system based on the operating variable data; the resonant state extended observer observation module is used to observe the converter based on the resonant state extended observer to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step; the converter system prediction model prediction module is used to input the grid-connected voltage observation data and the grid-connected voltage estimation data into the converter system prediction model to obtain the power prediction data of the converter at the second prediction time step; the power reference prediction data acquisition module is used to determine the power reference prediction data of the converter at the second prediction time step based on the grid-connected voltage observation data, the grid-connected voltage estimation data, and the preset power reference data; the control objective function construction module is used to obtain the real-time switching state of the converter at the current time step, and construct a control objective function of the converter based on the real-time switching state, the predicted power data, and the predicted power reference data; the converter control module is used to solve the control objective function to obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state.

[0009] It can be understood that, compared with the prior art, the present device constructs a converter system prediction model and a resonance state extended observer of the converter in the stationary three-phase coordinate system through the operating variable data of the three-phase four-leg LCL grid-connected converter. Then, the converter is observed through the resonance state extended observer to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step. Then, the power prediction data of the converter at the second prediction time step is predicted through the converter system prediction model. Then, the power reference prediction data of the converter at the second prediction time step is calculated through the grid-connected voltage observation data, the grid-connected voltage estimation data and the preset power reference data, and the control objective function of the converter is constructed through the real-time switch state, the predicted power data and the predicted power reference data. Then, the control objective function is solved to obtain the estimated switch state of the converter at the first prediction time step, and the converter is controlled based on the estimated switch state, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter. By constructing a converter system prediction model and a resonance state extended observer in the stationary coordinate system, the present invention avoids the computational complexity caused by introducing links such as rotational coordinate transformation, cross decoupling and positive and negative sequence component decomposition in the traditional model predictive control technology, reduces the computational complexity, and improves the dynamic performance of the converter and its ability to cope with unbalanced operating conditions; through the converter system prediction model and the resonance state extended observer, the dynamic observation of the grid-connected voltage and the deep fusion of multi-step prediction are realized. The resonance state extended observer can accurately and directly observe the frequency fluctuation of the converter under the resonance characteristic or the operating data under unbalanced conditions, that is, it can ensure high-precision tracking of the disturbance at a specific resonance frequency, thus realizing the grid voltage observation in the three-phase stationary coordinate system; through the multi-step prediction of the current time step, the first prediction time step and the second prediction time step, a closed-loop feedback mechanism is formed, and the double-time step prediction mechanism (current time step observation + first prediction time step estimation) compensates for the inherent delay defect of digital control, effectively avoiding the prediction deviation caused by model parameter mismatch and control delay in the traditional model predictive control from becoming larger and the poor tracking effect on the grid power data; by constructing and solving the control objective function, the control strategy of the converter can be dynamically adjusted, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, thus realizing the independent control of the grid-connected power of each phase in the case of grid asymmetry or frequency fluctuation. Description of the Drawings

[0010] Figure 1 It is a flowchart of the steps of a control method for a three-phase four-leg LCL grid-connected converter provided by an embodiment of the present invention;

[0011] Figure 2 It is a schematic structural diagram of a three-phase four-leg LCL grid-connected converter provided by an embodiment of the present invention;

[0012] Figure 3 Block diagram of a model predictive control method without a grid voltage sensor provided by an embodiment of the present invention;

[0013] Figure 4 Waveform diagram of grid current and grid voltage under three-phase symmetrical grid conditions provided by an embodiment of the present invention;

[0014] Figure 5 Waveform diagram of grid current and grid voltage under three-phase asymmetrical grid conditions provided by an embodiment of the present invention;

[0015] Figure 6 Waveform diagram of grid current and grid voltage when the reference values of grid-connected currents of each phase are different under three-phase symmetrical grid conditions provided by an embodiment of the present invention;

[0016] Figure 7 Waveform diagram of grid current and grid voltage when the grid frequency suddenly changes under three-phase symmetrical grid conditions provided by an embodiment of the present invention;

[0017] Figure 8 Structural schematic diagram of a control device for a three-phase four-leg LCL grid-connected converter provided by an embodiment of the present invention. Detailed implementation manners

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] Embodiment 1

[0020] Please refer to Figure 1 , Figure 1 which is a flowchart of the steps of a control method for a three-phase four-leg LCL grid-connected converter provided by an embodiment of the present invention, Figure 2 which is a structural schematic diagram of a three-phase four-leg LCL grid-connected converter provided by an embodiment of the present invention. A control method for a three-phase four-leg LCL grid-connected converter in this embodiment is applied to a three-phase four-leg LCL grid-connected converter as shown in Figure 2 . First, a three-phase four-leg LCL grid-connected converter (hereinafter simply referred to as a converter) as shown in Figure 2 . In the figure, L1 is the filter inductor on the bridge arm side, L2 is the filter inductor on the grid side, C is the filter capacitor, L0 is the neutral line filter inductor, R1 and R2 respectively represent the equivalent series resistances on L1 and L2, V dc is the DC side voltage, which is represented by a DC voltage source in this embodiment; i 1a, i 1b , i 1c respectively represent the arm currents on the a, b, and c phases, i.e., the three-phase arm currents (i1), v Ca , v Cb , v Cc represent the filter capacitor voltage (i.e., v C ); i ga , i gb , i gc represent the grid-connected currents corresponding to the a, b, and c phases (i g ), v ga , v gb , v gc represent the grid-connected voltages corresponding to the a, b, and c phases (v g ), v af , v bf , v cf respectively represent the voltages of the midpoints of the a, b, and c phase arms relative to the midpoint of the fourth arm, i.e., the three-phase arm voltages; Based on the converter shown in Figure 2 , according to Kirchhoff's voltage law and Kirchhoff's current law, the circuit equation of the converter can be obtained as follows:

[0021]

[0022] It should be clear that the three-phase four-leg LCL grid-connected converter shown in this embodiment Figure 2 , and the corresponding circuit equation, belong to the commonly used converter applications in this field, so the structural connection of this converter will not be described in detail here. Specifically, as shown in Figure 1 , a control method for a three-phase four-leg LCL grid-connected converter provided in this embodiment includes steps S101 to S106.

[0023] Step S101: Obtain the operating variable data of the converter, and construct a converter system prediction model and a resonant state extended observer of the converter in the stationary three-phase coordinate system based on the operating variable data.

[0024] In this embodiment, obtaining the operating variable data of the converter and constructing a converter system prediction model and a resonant state extended observer of the converter in a stationary three-phase coordinate system based on the operating variable data specifically includes: obtaining the operating variable data of the converter, where the operating variable data includes: three-phase bridge arm side current data, filter capacitor voltage data, grid-connected current data, three-phase bridge arm side voltage data, and grid-connected voltage data; constructing a state space equation of the converter according to the three-phase bridge arm side current data, filter capacitor voltage data, grid-connected current data, three-phase bridge arm side voltage data, and grid-connected voltage data, and discretizing the state space equation to determine a converter system prediction model of the converter in a stationary three-phase coordinate system; constructing an observation expression of the grid-connected voltage data and the power frequency signal data in the converter, and constructing a resonant state extended observer of the converter in a stationary three-phase coordinate system based on the observation expression and a preset error feedback gain matrix.

[0025] In this embodiment, through the acquisition of multi-dimensional operating variable data and the construction of a discretized state space equation, a complete physical model basis is provided for the control of the converter, enabling the converter system prediction model to accurately reflect the dynamic coupling characteristics and grid interaction behavior of the three-phase four-leg LCL grid-connected converter. The resonant state extended observer can solve the problem of observation failure of traditional observers due to frequency offset or harmonic interference under complex grid conditions, fully considering the frequency fluctuation of the converter under resonant characteristics or the performance under unbalanced conditions. Through the converter system prediction model and the resonant state extended observer, the deep fusion of dynamic observation and multi-step prediction of the grid-connected voltage is realized, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, thereby achieving independent control of the grid-connected power of each phase under grid asymmetry or frequency fluctuation conditions.

[0026] In this embodiment, the constructing a state space equation of the converter according to the three-phase bridge arm side current data, filter capacitor voltage data, grid-connected current data, three-phase bridge arm side voltage data, and grid-connected voltage data, and discretizing the state space equation to determine a converter system prediction model of the converter in a stationary three-phase coordinate system specifically includes: constructing a state variable matrix of the converter according to the grid-connected current data, filter capacitor voltage data, and three-phase bridge arm side current data; constructing an input voltage vector matrix of the converter according to the three-phase bridge arm side voltage data; constructing a grid-connected voltage vector matrix of the converter according to the grid-connected voltage data; constructing a state space equation of the converter according to the state variable matrix, the input voltage vector matrix, and the grid-connected voltage vector matrix; and discretizing the state space equation according to a preset zero-order hold method to obtain a converter system prediction model of the converter in a stationary three-phase coordinate system.

[0027] In an alternative embodiment, as Figure 2 shown, a state variable matrix X of the converter is constructed according to grid-connected current data, filter capacitor voltage data, and three-phase bridge arm side current data, X = [i 1a i 1b i 1c i ga i gb i gc v Ca v Cb v Cc T , and an input voltage vector matrix u of the converter is constructed according to three-phase bridge arm side voltage data, u = [v af v bf v cf T ; a grid-connected voltage vector matrix v g of the converter is constructed according to grid-connected voltage data, v ga = [v gb v gc T ; then, a state space equation of the converter is constructed according to the state variable matrix, the input voltage vector matrix, and the grid-connected voltage vector matrix In Equation (4), A is a system matrix, which is used to represent how the state itself changes over time, B is an input matrix, which is used to represent how an external input affects the state of the system, and C is an output matrix, which is used to represent how an external disturbance affects the state of the system. Specifically, A, B, and C are as follows:

[0028]

[0029] In Equations (5) and (6), both g and f are used to represent electrical characteristics and the interactions between various parts, specifically Then, the state space equation is discretized according to the preset zero-order hold method to obtain a discretized state space equation, which is specifically: X(k + 1) = A d X(k) + B d u(k) + C d v g (k)(9); In Formulas (9) and (10), A d is a discrete-time state transition matrix, which is used to represent the state transition of the converter at discrete time steps, B d is a discrete-time input matrix, which is used to represent the influence of the control input on the state of the converter system at discrete time steps, C d is a discrete-time output matrix, which is used to represent how the system state affects the output of the converter system at discrete time steps; T s ​​​$T_s$ is the sampling time of the discrete system, which is used to represent the discrete time step, and $\tau$ is the integration variable, which is used to calculate the variable of the discretization matrix; based on formulas (9) and (10), the prediction model of the converter system in the stationary three-phase coordinate system of the converter is further derived as follows:

[0030] In this embodiment, by decomposing the system state into a state variable matrix, an input voltage vector matrix, and a grid-connected voltage vector matrix, the prediction model of the converter system can accurately reflect the dynamic coupling characteristics and grid interaction behavior of the three-phase four-leg LCL grid-connected converter; using the preset zero-order hold method to discretize the state space equation can better approximate the continuous behavior of the actual converter, making the obtained system prediction model closer to the actual operating conditions, thereby improving the accuracy of prediction and the precision of control; at the same time, the combination with the resonant state extended observer realizes the deep integration of the dynamic observation of the grid-connected voltage and multi-step prediction, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, and thus realizing the independent control of the grid-connected power of each phase under the condition of grid asymmetry or frequency fluctuation.

[0031] In this embodiment, constructing the observation expression of the grid-connected voltage data and the power frequency signal data in the converter, and constructing the resonant state extended observer of the converter in the stationary three-phase coordinate system based on the observation expression and the preset error feedback gain matrix specifically includes: constructing the observation expression of the grid-connected voltage data and the power frequency signal data, and constructing the converter state observation matrix and the resonant angular frequency observation matrix based on the observation expression; constructing the inductance observation matrix according to the inductance data of the converter; constructing the observation expression of the converter in the stationary three-phase coordinate system according to the observation matrix, the resonant angular frequency observation matrix, the inductance observation matrix, and the preset error feedback gain matrix; discretizing the observation expression and performing pole placement to obtain the resonant state extended observer of the converter in the stationary three-phase coordinate system.

[0032] Since the control is carried out in the stationary three-phase coordinate system of the converter, that is, the grid voltage shows a power frequency sine signal in the stationary three-phase coordinate system, and the traditional extended state observer can only ensure the asymptotic convergence estimation of a constant or slowly changing disturbance, so this optional embodiment designs a resonant state extended observer (Resonant Extended State Observer, RESO) for observation, which can ensure high-precision tracking of the disturbance at a specific resonant frequency. As Figure 2 shown in the converter, the grid-connected current $i$ g satisfies the relational expression shown in formula (12), that is Based on this, the observation expressions of the grid-connected voltage data and the power frequency signal data are constructed, as shown in Formulas (13) and (14).

[0033]

[0034] In Formulas (13) and (14), F p is the constant perturbation component, F s is the power frequency sine component, is the derivative of the grid-connected voltage v g , is the derivative of F s ; then, based on the observation expressions (i.e., Formulas (13) and (14)), the converter state observation matrix X ob and the resonant angular frequency observation matrix A ob are constructed. According to the inductor data of the converter, the inductor observation matrix B ob is constructed. Then, combined with the preset error feedback gain matrix L and the filter capacitor voltage, the observation expression of the converter in the stationary three-phase coordinate system is constructed Specifically:

[0035]

[0036] v c =[v Ca v Cb v Cc T (17);

[0037]

[0038] L = [β1 β2 β3 β4] (20); H = [1 0 0 0] T (21);

[0039] In Formulas (15) to (21), X ob represents the state variable of the converter, represents the estimated value of the state variable X ob of the converter; represents 's derivative; is the derivative of F s , is the second derivative of F s , ω r = 2πf0 is the resonant angular frequency of the resonant state extended observer, set to the power grid power frequency; L is the error feedback gain matrix, used to adjust the error feedback of the resonant state extended observer, β1 is the gain coefficient related to the grid-connected current (i g ), β2 is the gain coefficient related to F, β3 is related to​ The relevant gain coefficient, and β4 is related to the relevant gain coefficient. β1, β2, β3, and β4 are all used to adjust the influence of the state estimation error on the system state estimation. Then, the observation expression is discretized, specifically into a recursive form, which is:

[0040]

[0041] In formulas (22) to (24), represents the estimated value of the state variable of the converter at time k + 1; X ob (k) represents the state variable of the converter at time k; represents the estimated value of the state variable of the converter at time k; v C (k) represents the filter capacitor voltage of the converter at time k; φ ob is the discrete-time input matrix, used to represent the influence of the control input on the system state of the converter, Γ ob is the discrete-time input matrix, used to represent the influence of the control input on the system state of the converter, L d is the discrete-time gain matrix, used to adjust the feedback gain of the system, β 1d is the gain coefficient related to the grid-connected current (i g ); β 2d is the gain coefficient related to F; β 3d is related to the relevant gain coefficient; β 4d is related to the relevant gain coefficient; β 1d , β 2d , β 3d , β 4d are all used to adjust the influence of the discrete-time gain on the system state estimation. Formula (24) is used to represent the error recurrence relation. After discretization, it is necessary to perform pole placement on the matrix φ ob -L d H to ensure the asymptotic convergence of the resonance state extended observer error tracking. In this embodiment, all poles are placed at where ω ob represents the grid angular frequency, which corresponds to the bandwidth of the resonance state extended observer. On this basis, the resonance state extended observer also satisfies the characteristic polynomial of formula (25) as follows; In formula (25), I is the identity matrix, used to maintain the dimension matching of matrix operations in the characteristic equation, z is a complex variable, used to represent the complex plane variable of the pole position. After determining the value of ω ob and expanding and calculating formula (25), the discrete-time gain matrix L can be solved by making the coefficients of z on both sides of formula (25) equal.d .

[0042] In this embodiment, by constructing an observation expression of the power frequency signal and the grid-connected voltage and combining a dynamic update mechanism of the inductance parameter and the resonant angular frequency, the active observation and suppression of the resonant characteristics of the LCL filter are realized. Through the dynamic adjustment of the resonant angular frequency observation matrix, the grid frequency change is tracked in real time, and the problem of observation failure caused by frequency deviation or harmonic interference of the traditional observer under complex grid conditions can be solved. The performance of the converter under frequency fluctuations or unbalanced conditions under resonant characteristics is fully considered. Through discretization processing and pole placement, the dynamic response characteristics of the observer are optimized, so that it can converge quickly within the digital control period, and at the same time, high-frequency noise interference is suppressed, so that the resonant state extended observer can ensure high-precision tracking of the disturbance at a specific resonant frequency, thereby realizing the grid voltage observation in the three-phase stationary coordinate system, and further improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter.

[0043] Step S102: Based on the resonant state extended observer, observe the converter to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step.

[0044] In this embodiment, the observing the converter based on the resonant state extended observer to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step specifically includes: observing the power frequency sine signal of the converter based on the resonant state extended observer to obtain the power frequency observation data of the converter at the current time step; converting the power frequency observation data based on the observation expression to obtain the grid-connected voltage observation data of the converter at the current time step; synchronizing the grid-connected voltage observation data through a phase-locked loop to obtain the grid angular frequency observation value; obtaining the real-time grid-connected current data and the real-time filter capacitor voltage data of the converter at the current time step, and updating the converter state observation matrix of the resonant state extended observer based on the real-time grid-connected current data and the power frequency observation data; updating the resonant angular frequency observation matrix of the resonant state extended observer based on the power frequency observation data; solving the discrete-time gain matrix of the resonant state extended observer according to the grid angular frequency observation value; obtaining the grid-connected voltage estimation data of the converter at the first prediction time step based on the updated resonant state extended observer, the solution result of the discrete-time gain matrix, and the real-time filter capacitor voltage data.

[0045] In an optional embodiment, the power frequency sine signal of the converter is observed based on the resonant state extended observer to obtain the power frequency observation data of the converter at the current time step, and then the power frequency observation data is converted in combination with the observation expressions (i.e., formulas (13) and (14)) to obtain the grid-connected voltage observation data of the converter at the current time step. After that, the grid-connected voltage observation data is synchronized by a Phase-Locked Loop (PLL) to obtain the observed value of the grid angular frequency. After that, the real-time grid-connected current data i g (k) and the real-time filter capacitor voltage data v C (k) of the converter at the current time step (time step k) are obtained; then, based on the real-time grid-connected current data i g (k) and the power frequency observation data, the converter state observation matrix X of the resonant state extended observer is updated. ob Based on the power frequency observation data, the resonant angular frequency observation matrix A of the resonant state extended observer is updated. ob After that, the observed value of the grid angular frequency is substituted into formula (25), and the coefficients on both sides of the parameter z are made to correspond equally one by one to solve for the discrete-time gain matrix L d . Then, the updated resonant state extended observer (i.e., the updated converter state observation matrix X ob and the resonant angular frequency observation matrix A ob ), the solution result of the discrete-time gain matrix L d , and the real-time filter capacitor voltage data v C (k) are substituted into formulas (15) to (24) for mathematical operations to obtain the grid-connected voltage estimation data of the converter at the first prediction time step (set to time step k + 1 in this embodiment) (including the grid-connected voltage estimated value and the derivative of the grid-connected voltage estimated value ).

[0046] In this embodiment, through the phase-locked loop synchronization and the dynamic update of the resonant angular frequency, the real-time tracking of the grid frequency and the adaptive adjustment of the parameters of the resonant state extended observer are realized, avoiding the asymptotic convergence estimation of the traditional extended state observer that can only guarantee the perturbation of constant or slow change, so that the resonant state extended observer can guarantee the high-precision tracking of the perturbation at a specific resonant frequency; the resonant state extended observer obtains the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step, forming a closed-loop feedback mechanism, and using the double-time step prediction mechanism to compensate for the inherent delay defect of digital control, effectively avoiding the prediction deviation caused by the mismatch of model parameters and control delay in the traditional model predictive control and the poor tracking effect on the grid power data, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, thereby realizing the independent control of the grid-connected power of each phase in the case of grid asymmetry or frequency fluctuation.

[0047] Step S103: Input the grid-connected voltage observation data and the grid-connected voltage estimation data into the converter system prediction model to obtain the power prediction data of the converter at the second prediction time step.

[0048] In an optional embodiment, the grid-connected voltage observation data and the grid-connected voltage estimation data (i.e., the grid-connected voltage estimation value ) are respectively substituted into v g (k) and v g (k + 1) in formula (11) to obtain the power prediction data of the converter at the second prediction time step. Specifically, in step S102, the real-time grid-connected current data i g (k), the real-time filter capacitor voltage data v C (k), the grid-connected voltage observation data and the grid-connected voltage estimation data (i.e., the grid-connected voltage estimation value ) have been obtained. Then, the real-time current i1(k) of the three-phase bridge arm side is collected through the sensor, and combined with formulas (1), (2), (3), and (9), u(k) and u(k + 1) can be calculated. Finally, based on formula (11), the power prediction data (X(k + 2)) of the converter at the second prediction time step (i.e., at the moment of k + 2) can be solved.

[0049] Step S104: Determine the power reference prediction data of the converter at the second prediction time step based on the grid-connected voltage observation data, the grid-connected voltage estimation data, and the preset power reference data.

[0050] In this embodiment, determining the power reference prediction data of the converter at the second prediction time step based on the grid-connected voltage observation data, the grid-connected voltage estimation data, and the preset power reference data specifically includes: performing phase-locked loop synchronization on the grid-connected voltage observation data to obtain the observed value of the grid voltage phase angle; the preset power reference data includes: the reference value of the grid-connected current; calculating the predicted value of the grid-connected current reference of the converter at the second prediction time step according to the observed value of the grid voltage phase angle and the reference value of the grid-connected current; calculating the predicted value of the filter capacitor voltage reference and the predicted value of the three-phase bridge arm side current reference of the converter at the second prediction time step according to the preset Lagrange extrapolation method in combination with the grid-connected voltage observation data, the grid-connected voltage estimation data, and the predicted value of the grid-connected current reference; determining the power reference prediction data of the converter at the second prediction time step according to the predicted value of the grid-connected current reference, the predicted value of the filter capacitor voltage reference, and the predicted value of the three-phase bridge arm side current reference.

[0051] In this embodiment, the Lagrange extrapolation method is used to fuse the grid-connected voltage observation data and the grid-connected voltage estimation data, realizing the dynamic prediction of the power reference value at the second prediction time step. By combining the grid-connected voltage observation data, the grid-connected voltage estimation data, and the predicted value of the grid-connected current reference, the future state of the converter can be predicted more accurately, thereby improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, and thus realizing the independent control of the grid-connected power of each phase under the condition of grid asymmetry or frequency fluctuation.

[0052] In this embodiment, calculating the predicted value of the filter capacitor voltage reference and the predicted value of the three-phase bridge arm side current reference of the converter at the second prediction time step according to the preset Lagrange extrapolation method in combination with the grid-connected voltage observation data, the grid-connected voltage estimation data, and the predicted value of the grid-connected current reference specifically includes: calculating the predicted value of the grid-connected voltage and the derivative of the predicted value of the grid-connected voltage of the converter at the second prediction time step according to the preset Lagrange extrapolation method in combination with the grid-connected voltage observation data and the grid-connected voltage estimation data; calculating the predicted value of the filter capacitor voltage reference of the converter at the second prediction time step based on the predicted value of the grid-connected voltage and the predicted value of the grid-connected current reference; calculating the predicted value of the three-phase bridge arm side current reference of the converter at the second prediction time step based on the derivative of the predicted value of the grid-connected voltage and the predicted value of the grid-connected current reference.

[0053] In this embodiment, by introducing the derivative calculation of the grid-connected voltage prediction value, not only can the instantaneous change trend of the converter system state be captured, but also the deep law of the energy exchange of the LCL filter can be explored. By calculating the predicted reference value of the filter capacitor voltage at the second prediction time step of the converter, the coupling relationship between the grid-connected voltage and the grid-connected current can be considered, and the resonance risk of the converter can be predicted in advance with the predicted reference value of the three-phase bridge arm side current, so as to provide a prediction reference for the control of the three-phase four-leg LCL grid-connected converter, improve the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, and thus achieve independent control of the grid-connected power of each phase under the condition of grid asymmetry or frequency fluctuation.

[0054] In an optional embodiment, the grid-connected voltage observation data is synchronized by a phase-locked loop to obtain the observed value of the grid voltage phase angle After that, the grid-connected current reference value is obtained After that, on the premise of the in-place power factor, the observed value of the grid voltage phase angle is substituted into formula (26) to generate the real-time value of the grid-connected current reference at the current time step And the observed value of the grid voltage phase angle and the observed value of the grid angular frequency are substituted into formula (27) to deduce the predicted reference value of the grid-connected current at the second prediction time step (at time k + 2)

[0055]

[0056] In formulas (26) and (27), cosθ i The θ in (k) i (k) is the grid voltage phase angle, ω is the grid angular frequency; then, according to the preset Lagrange extrapolation method, combining the grid-connected voltage observation data and the grid-connected voltage estimation data, the predicted value of the grid-connected voltage and the derivative of the predicted value of the grid-connected voltage of the converter at the second prediction time step are calculated, and the calculation process is shown in formula (28). After that, based on the predicted value of the grid-connected voltage and the predicted reference value of the grid-connected current, the predicted reference value of the filter capacitor voltage of the converter at the second prediction time step is calculated Based on the derivative of the predicted value of the grid-connected voltage and the predicted reference value of the grid-connected current, the predicted reference value of the three-phase bridge arm side current of the converter at the second prediction time step is calculated The predicted reference value of the filter capacitor voltage and the predicted reference value of the three-phase bridge arm side current The calculation process is shown in formula (29);

[0057]

[0059] In Formulas (28) and (29), is the predicted grid-connected voltage value of the i-th phase of the converter at the second prediction time step (k + 2), is the derivative of the predicted grid-connected voltage value of the i-th phase of the converter at the second prediction time step; and are respectively the grid-connected voltage values of the i-th phase of the converter at (k - 1) and (k - 2), and are respectively and the corresponding derivatives.

[0060] Step S105: Obtain the real-time switching state of the converter at the current time step, and construct a control objective function of the converter based on the real-time switching state, predicted power data, and predicted power reference data.

[0061] In this embodiment, obtaining the real-time switching state of the converter at the current time step and constructing a control objective function of the converter based on the real-time switching state, predicted power data, and predicted power reference data specifically include: obtaining the real-time switching state of the converter at the current time step, and constructing a switching action loss term in combination with a preset switching action weight coefficient; the predicted power data includes: predicted grid-connected current value, predicted three-phase bridge arm side current value, and predicted filter capacitor voltage value; constructing a grid-connected current steady-state tracking accuracy term according to the predicted grid-connected current value, predicted grid-connected current reference value, and preset grid-connected current steady-state weight coefficient; constructing a three-phase bridge arm side current change term according to the predicted three-phase bridge arm side current value and predicted three-phase bridge arm side current reference value; constructing a converter resonance suppression term according to the predicted filter capacitor voltage value, predicted filter capacitor voltage reference value, and preset resonance suppression weight coefficient; constructing the control objective function of the converter according to the switching action loss term, grid-connected current steady-state tracking accuracy term, three-phase bridge arm side current change term, and converter resonance suppression term.

[0062] In an alternative embodiment, obtain the real-time switching state S i (k) of the converter at the current time step (k), and construct a switching action loss term λ S in combination with a preset switching action weight coefficient λ S ∑ i＝a,b,c,f (S i (k + 1) - S i (k)) 2 ; the predicted power data includes: predicted grid-connected current value i gi (k + 2), predicted three-phase bridge arm side current value i 1i (k + 2), and predicted filter capacitor voltage value v Ci (k + 2); then according to the predicted grid-connected current value igi (k + 2), the predicted value of the grid-connected current reference and the preset steady-state weight coefficient λ of the grid-connected current I Construct the steady-state tracking accuracy term of the grid-connected current After that, according to the predicted value i of the three-phase bridge arm side current 1i (k + 2) and the predicted value of the three-phase bridge arm side current reference Construct the three-phase bridge arm side current change term After that, according to the predicted value v of the filter capacitor voltage Ci (k + 2), the predicted value of the filter capacitor voltage reference and the preset resonance suppression weight coefficient λ V Construct the converter resonance suppression term Thus, construct the control objective function J of the converter, which is specifically shown in formula (30);

[0063]

[0064] In formula (30), increasing λ I can improve the steady-state tracking accuracy of the grid-connected current. Increasing λ V can improve the resonance suppression effect. Increasing λ S can effectively reduce the switching action and reduce the switching loss.

[0065] In this embodiment, by constructing the control objective function, the comprehensive optimization of switching loss, steady-state accuracy, dynamic response and resonance suppression is realized, to track the reference value, reduce the switching action to reduce the loss and suppress the resonance characteristics of the LCL system, improve the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, and thus realize the independent control of the grid-connected power of each phase under the condition of grid asymmetry or frequency fluctuation.

[0066] Step S106: Solve the control objective function to obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state.

[0067] In this embodiment, the solving of the control objective function to obtain the estimated switching state of the converter at the first prediction time step and controlling the converter based on the estimated switching state specifically includes:

[0068] Taking the minimization of the control objective function as the solution target, solve the control objective function according to the preset finite control set model predictive method to obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state.

[0069] In an optional embodiment, the three-phase four-leg LCL grid-connected converter has 16 switching states. For the finite control set model predictive control method (FCS-MPC), one of the switching states is used as the control input of the converter in each control cycle. The control input calculated at the k-th moment is loaded into the register and executed at the k+1-th moment. That is, there is a control delay when using the finite control set model predictive method. To compensate for this delay, it is necessary to predict the system state at the k+2-th moment at the k-th moment (i.e., the power prediction data of the converter at the second prediction time step in step S103), and accordingly calculate the switching state at the k+1-th moment (i.e., the estimated switching state of the converter at the first prediction time step). For each switching state S(k+1) = [S a (k+1)S b (k+1)S c (k+1)S f (k+1)] T , the corresponding input voltage vector matrix can be obtained as: u(k+1) = [v af (k+1)v bf (k+1)v cf (k+1)] T = V dc [S a (k+1)-S f (k+1)S b (k+1)-S f (k+1)S c (k+1)-S f (k+1)] T (31); Therefore, by calculating the switching state at the k+1-th moment, the input voltage vector matrix at the k+1-th moment can be directly calculated, and then the control of the converter can be realized.

[0070] In this embodiment, through the real-time solution of the objective function by the finite control set model predictive method, the computational complexity can be greatly reduced while ensuring the control accuracy. By restricting the switching state combination within the physically realizable range through the finite control set, the problem of increased delay caused by the traditional continuous control set method's over-reliance on the modulation strategy is avoided; by constructing and solving the control objective function, the control strategy of the converter can be dynamically adjusted, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, and thus realizing the independent control of the grid-connected power of each phase in the case of grid asymmetry or frequency fluctuation.

[0071] In this embodiment, based on the operating variable data of a three-phase four-leg LCL grid-connected converter, a converter system prediction model and a resonant state extended observer in the stationary three-phase coordinate system are constructed. Then, the converter is observed through the resonant state extended observer to obtain the grid-connected voltage observation data at the current time step and the grid-connected voltage estimation data at the first prediction time step. After that, the power prediction data of the converter at the second prediction time step is obtained through prediction by the converter system prediction model. Then, the power reference prediction data of the converter at the second prediction time step is calculated through the grid-connected voltage observation data, the grid-connected voltage estimation data, and the preset power reference data, and the control objective function of the converter is constructed through the real-time switch state, the predicted power data, and the predicted power reference data. Then, the control objective function is solved to obtain the estimated switch state of the converter at the first prediction time step, and the converter is controlled based on the estimated switch state, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter. In this embodiment, by constructing the converter system prediction model and the resonant state extended observer in the stationary coordinate system, the computational complexity caused by introducing links such as rotational coordinate transformation, cross decoupling, and positive and negative sequence component decomposition in the traditional model predictive control technology is avoided, the computational complexity is reduced, and the dynamic performance of the converter and its ability to handle unbalanced operating conditions are improved; through the converter system prediction model and the resonant state extended observer, the deep integration of the dynamic observation and multi-step prediction of the grid-connected voltage is realized. The resonant state extended observer can accurately and directly observe the frequency fluctuations of the converter under resonant characteristics or the operating data under unbalanced conditions, that is, it can ensure high-precision tracking of disturbances at specific resonant frequencies, thereby realizing the grid voltage observation in the three-phase stationary coordinate system; with the multi-step prediction at the current time step, the first prediction time step, and the second prediction time step, a closed-loop feedback mechanism is formed, and the dual-time-step prediction mechanism (current time step observation + first prediction time step estimation) is used to compensate for the inherent delay defect of digital control, effectively avoiding the prediction deviation increase caused by model parameter mismatch and control delay in the traditional model predictive control and the poor tracking effect on grid power data; by constructing and solving the control objective function, the control strategy of the converter can be dynamically adjusted, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, thereby realizing the independent control of the grid-connected power of each phase under grid asymmetry or frequency fluctuation conditions.

[0072] Embodiment 2

[0073] Please refer to Figure 3 , Figure 3 which is a block diagram of a model predictive control method without a grid voltage sensor provided by an embodiment of the present invention. This model predictive control method without a grid voltage sensor is applied to a three-phase four-leg LCL grid-connected converter as shown in Figure 2 First, the real-time grid-connected current data i g(k), real-time data v of the filter capacitor voltage C (k), real-time current i1(k) on the three-phase bridge arm side; the estimated grid voltage value at time k+1 is obtained through the resonant extended state observer and the grid voltage observation data Grid voltage observation data The observed value of the grid angular frequency is obtained through PLL synchronization and the observed value of the grid voltage phase angle (i.e., ), through the observed value of the grid angular frequency periodically recalculate and update the resonant extended state observer gain, and then perform reference value generation to obtain the power reference prediction data of the converter at the second prediction time step ( Figure 3 in its equivalent implementation example one's grid-connected current prediction value i gi (k+2), three-phase bridge arm side current prediction value i 1i (k+2) and filter capacitor voltage prediction value ); then through the prediction model and delay compensation (i.e., step S103 of Embodiment 1) to obtain the power prediction data at time k+2 ( Figure 3 in i1(k+2), i g (k+2), v C (k+2), its equivalent implementation example one's grid-connected current prediction value i gi (k+2), three-phase bridge arm side current prediction value i 1i (k+2) and filter capacitor voltage prediction value v Ci (k+2)), and finally obtain the estimated switch state S(k+1) at time k+1 by minimizing the value function (i.e., the control objective function constructed in step S105 of Embodiment 1).

[0074] Based on the specific steps described in Embodiment 1, a simulation model is constructed to verify the effectiveness of the proposed control method. Specifically, Figure 4 is the waveform diagram of the grid current and grid voltage under the three-phase symmetric grid conditions provided by the embodiment of the present invention; Figure 5 is the waveform diagram of the grid current and grid voltage under the three-phase asymmetric grid conditions provided by the embodiment of the present invention; Figure 6 is the waveform diagram of the grid current and grid voltage when the grid-connected current reference values of each phase are different under the three-phase symmetric grid conditions provided by the embodiment of the present invention; Figure 7 is the waveform diagram of the grid current and grid voltage when the grid frequency suddenly changes under the three-phase symmetric grid conditions provided by the embodiment of the present invention; Figures 4 to 7 The grid voltage in Figure 4As shown, the reference values of the three-phase grid-connected currents are the same under the condition of a three-phase symmetrical grid, and step from 70 A to 100 A, and reach a new steady state 2.6 ms after the reference value changes, indicating that the method of this embodiment has good dynamic performance. As Figure 5 shown Figure 5 Specifically, it is the waveforms of the grid-connected current and the grid voltage under the condition of a three-phase asymmetrical grid (a 20% drop in the grid voltage of phase a). The waveforms are stable, indicating that the three-phase four-leg LCL grid-connected converter has stability and dynamic performance under asymmetrical grid conditions. As Figure 6 shown, specifically, the reference values of the grid-connected currents of each phase are different under the condition of a three-phase symmetrical grid when the waveforms of the grid-connected current and the grid voltage are stable, indicating that the three-phase four-leg LCL grid-connected converter still has stability and dynamic performance when the reference values of the grid-connected currents of each phase are different under the condition of a three-phase symmetrical grid. As Figure 7 shown, specifically, under the condition of a three-phase symmetrical grid, when the grid frequency suddenly changes from 50 Hz to 52 Hz, the waveforms of the grid-connected current and the grid voltage can still maintain the stability of the waveforms of the grid-connected current and the grid voltage under frequency fluctuation, verifying the effectiveness of the control method of the three-phase four-leg LCL grid-connected converter proposed in this embodiment.

[0075] Embodiment 3

[0076] Please refer to Figure 8 , Figure 8 which is a schematic structural diagram of a control device for a three-phase four-leg LCL grid-connected converter provided by an embodiment of the present invention, including: a converter modeling module 201, a resonant state extended observer observation module 202, a converter system prediction model prediction module 203, a power reference prediction data acquisition module 204, a control objective function construction module 205, and a converter control module 206.

[0077] Among them, the converter modeling module 201 is used to obtain the operating variable data of the converter, and construct a converter system prediction model and a resonant state extended observer of the converter in a stationary three-phase coordinate system based on the operating variable data.

[0078] In this embodiment, the converter modeling module 201 includes: a converter modeling unit; the converter modeling unit is configured to obtain the operating variable data of the converter, where the operating variable data includes: three-phase bridge arm side current data, filter capacitor voltage data, grid-connected current data, three-phase bridge arm side voltage data, and grid-connected voltage data; construct a state space equation of the converter according to the three-phase bridge arm side current data, filter capacitor voltage data, grid-connected current data, three-phase bridge arm side voltage data, and grid-connected voltage data, and discretize the state space equation to determine a converter system prediction model of the converter in the stationary three-phase coordinate system; construct an observation expression of the grid-connected voltage data and the power frequency signal data in the converter, and construct a resonant state extended observer of the converter in the stationary three-phase coordinate system based on the observation expression and a preset error feedback gain matrix. In this embodiment, the converter modeling unit includes: a converter system prediction model construction sub-unit; the converter system prediction model construction sub-unit is configured to construct a state variable matrix of the converter according to the grid-connected current data, filter capacitor voltage data, and three-phase bridge arm side current data; construct an input voltage vector matrix of the converter according to the three-phase bridge arm side voltage data; construct a grid-connected voltage vector matrix of the converter according to the grid-connected voltage data; construct a state space equation of the converter according to the state variable matrix, input voltage vector matrix, and grid-connected voltage vector matrix; discretize the state space equation according to a preset zero-order hold method to obtain a converter system prediction model of the converter in the stationary three-phase coordinate system. In this embodiment, the converter modeling unit includes: a resonant state extended observer construction sub-unit; the resonant state extended observer construction sub-unit is configured to construct an observation expression of the grid-connected voltage data and the power frequency signal data, and construct a converter state observation matrix and a resonant angular frequency observation matrix based on the observation expression; construct an inductor observation matrix according to the inductor data of the converter; construct an observation expression of the converter in the stationary three-phase coordinate system according to the observation matrix, resonant angular frequency observation matrix, inductor observation matrix, and preset error feedback gain matrix; perform discretization processing and pole placement on the observation expression to obtain a resonant state extended observer of the converter in the stationary three-phase coordinate system.

[0079] The resonant state extended observer observation module 202 is configured to observe the converter based on the resonant state extended observer, and obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step. In this embodiment, the resonant state extended observer observation module 202 includes: a resonant state extended observer observation unit; the resonant state extended observer observation unit is configured to observe the power frequency sine signal of the converter based on the resonant state extended observer, and obtain the power frequency observation data of the converter at the current time step; convert the power frequency observation data based on the observation expression to obtain the grid-connected voltage observation data of the converter at the current time step; perform phase-locked loop synchronization on the grid-connected voltage observation data to obtain the grid angular frequency observation value; obtain the real-time grid-connected current data and the real-time filter capacitor voltage data of the converter at the current time step, and update the converter state observation matrix of the resonant state extended observer based on the real-time grid-connected current data and the power frequency observation data; update the resonant angular frequency observation matrix of the resonant state extended observer based on the power frequency observation data; solve the discrete-time gain matrix of the resonant state extended observer according to the grid angular frequency observation value; and obtain the grid-connected voltage estimation data of the converter at the first prediction time step based on the updated resonant state extended observer, the solution result of the discrete-time gain matrix, and the real-time filter capacitor voltage data.

[0080] The converter system prediction model prediction module 203 is configured to input the grid-connected voltage observation data and the grid-connected voltage estimation data into the converter system prediction model, and obtain the power prediction data of the converter at the second prediction time step.

[0081] The power reference prediction data acquisition module 204 is configured to determine the power reference prediction data of the converter at the second prediction time step based on the grid-connected voltage observation data, the grid-connected voltage estimation data, and the preset power reference data. In this embodiment, the power reference prediction data acquisition module 204 includes: a power reference prediction data acquisition unit; the power reference prediction data acquisition unit is configured to perform phase-locked loop synchronization on the grid-connected voltage observation data to obtain the observed value of the grid voltage phase angle; the preset power reference data includes: the reference value of the grid-connected current; calculate the predicted value of the grid-connected current reference of the converter at the second prediction time step according to the observed value of the grid voltage phase angle and the reference value of the grid-connected current; according to the preset Lagrange extrapolation method, combine the grid-connected voltage observation data, the grid-connected voltage estimation data, and the predicted value of the grid-connected current reference to calculate the predicted value of the filter capacitor voltage reference and the predicted value of the three-phase bridge arm side current reference of the converter at the second prediction time step; determine the power reference prediction data of the converter at the second prediction time step according to the predicted value of the grid-connected current reference, the predicted value of the filter capacitor voltage reference, and the predicted value of the three-phase bridge arm side current reference. In this embodiment, the power reference prediction data acquisition unit includes: a power reference prediction data acquisition subunit; the power reference prediction data acquisition subunit is configured to calculate the predicted value of the grid-connected voltage and the derivative of the predicted value of the grid-connected voltage of the converter at the second prediction time step according to the preset Lagrange extrapolation method, in combination with the grid-connected voltage observation data and the grid-connected voltage estimation data; calculate the predicted value of the filter capacitor voltage reference of the converter at the second prediction time step based on the predicted value of the grid-connected voltage and the predicted value of the grid-connected current reference; calculate the predicted value of the three-phase bridge arm side current reference of the converter at the second prediction time step based on the derivative of the predicted value of the grid-connected voltage and the predicted value of the grid-connected current reference.

[0082] The control objective function construction module 205 is configured to obtain the real-time switching state of the converter at the current time step, and construct the control objective function of the converter based on the real-time switching state, predicted power data, and predicted power reference data. In this embodiment, the control objective function construction module 205 includes: a control objective function construction unit; the control objective function construction unit is configured to obtain the real-time switching state of the converter at the current time step, and construct a switching action loss term by combining a preset switching action weight coefficient; the predicted power data includes: a predicted value of grid-connected current, a predicted value of three-phase bridge arm side current, and a predicted value of filter capacitor voltage; construct a grid-connected current steady-state tracking accuracy term according to the predicted value of grid-connected current, the predicted reference value of grid-connected current, and a preset grid-connected current steady-state weight coefficient; construct a three-phase bridge arm side current change term according to the predicted value of three-phase bridge arm side current and the predicted reference value of three-phase bridge arm side current; construct a converter resonance suppression term according to the predicted value of filter capacitor voltage, the predicted reference value of filter capacitor voltage, and a preset resonance suppression weight coefficient; construct the control objective function of the converter according to the switching action loss term, the grid-connected current steady-state tracking accuracy term, the three-phase bridge arm side current change term, and the converter resonance suppression term.

[0083] The converter control module 206 is configured to solve the control objective function to obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state. In this embodiment, the converter control module 206 includes: a converter control unit; the converter control unit is configured to take the minimization of the control objective function as the solution target, solve the control objective function according to a preset finite control set model prediction method, obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state.

[0084] In summary, in the embodiments of the present invention, based on the operating variable data of a three-phase four-leg LCL grid-connected converter, a converter system prediction model and a resonant state extended observer in the stationary three-phase coordinate system are constructed for the converter. Then, the converter is observed through the resonant state extended observer to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step. Subsequently, the power prediction data of the converter at the second prediction time step is predicted through the converter system prediction model. Then, the power reference prediction data of the converter at the second prediction time step is calculated through the grid-connected voltage observation data, the grid-connected voltage estimation data, and the preset power reference data. In addition, a control objective function of the converter is constructed through the real-time switch state, the predicted power data, and the predicted power reference data. Then, the control objective function is solved to obtain the estimated switch state of the converter at the first prediction time step, and the converter is controlled based on the estimated switch state, thereby improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter. In the embodiments of the present invention, by constructing a converter system prediction model and a resonant state extended observer in the stationary coordinate system, the calculation complexity caused by introducing links such as rotational coordinate transformation, cross decoupling, and positive and negative sequence component decomposition in the traditional model predictive control technology is avoided, the calculation complexity is reduced, and the dynamic performance of the converter and the ability to cope with unbalanced operating conditions are improved. Through the converter system prediction model and the resonant state extended observer, the deep integration of the dynamic observation and multi-step prediction of the grid-connected voltage is realized. The resonant state extended observer can accurately and directly observe the frequency fluctuation of the converter under resonant characteristics or the operating data under unbalanced conditions, that is, it can ensure high-precision tracking of the disturbance at a specific resonant frequency, thereby realizing the grid voltage observation in the three-phase stationary coordinate system. Through the multi-step prediction at the current time step, the first prediction time step, and the second prediction time step, a closed-loop feedback mechanism is formed. The double-time-step prediction mechanism (current time step observation + first prediction time step estimation) compensates for the inherent delay defect of digital control, effectively avoiding the prediction deviation increase caused by model parameter mismatch and control delay in the traditional model predictive control and the poor tracking effect on grid power data. By constructing and solving the control objective function, the control strategy of the converter can be dynamically adjusted, improving the stability and dynamic performance of the three-phase four-leg LCL grid-connected converter, thereby realizing the independent control of the grid-connected power of each phase under the condition of grid asymmetry or frequency fluctuation.

[0085] In the specific embodiments described above, the purpose, technical solutions, and beneficial effects of the present invention are further described in detail. It should be understood that the above description is only for the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. It is particularly pointed out that for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A control method for a three-phase four-leg LCL grid-connected converter, characterized in that A converter applicable to a three-phase four-leg LCL, comprising: Obtain the operating variable data of the converter, and construct a converter system prediction model and a resonant state extended observer of the converter in the stationary three-phase coordinate system based on the operating variable data; Observe the converter based on the resonant state extended observer to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step; Input the grid-connected voltage observation data and the grid-connected voltage estimation data into the converter system prediction model to obtain the power prediction data of the converter at the second prediction time step; Determine the power reference prediction data of the converter at the second prediction time step based on the grid-connected voltage observation data, the grid-connected voltage estimation data and the preset power reference data; Obtain the real-time switching state of the converter at the current time step, and construct a control objective function of the converter based on the real-time switching state, the predicted power data and the predicted power reference data; Solve the control objective function to obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state.

2. The control method of a three-phase four-leg LCL grid-connected converter according to claim 1, characterized in that, The obtaining the operating variable data of the converter, and constructing a converter system prediction model and a resonant state extended observer of the converter in the stationary three-phase coordinate system based on the operating variable data specifically includes: Obtain the operating variable data of the converter, wherein the operating variable data includes: three-phase bridge arm side current data, filter capacitor voltage data, grid-connected current data, three-phase bridge arm side voltage data and grid-connected voltage data; Construct a state space equation of the converter according to the three-phase bridge arm side current data, the filter capacitor voltage data, the grid-connected current data, the three-phase bridge arm side voltage data and the grid-connected voltage data, and discretize the state space equation to determine a converter system prediction model of the converter in the stationary three-phase coordinate system; Construct an observation expression of the grid-connected voltage data and the power frequency signal data in the converter, and construct a resonant state extended observer of the converter in the stationary three-phase coordinate system based on the observation expression and a preset error feedback gain matrix.

3. The control method of a three-phase four-leg LCL grid-connected converter according to claim 2, characterized in that, The constructing a state space equation of the converter according to the three-phase bridge arm side current data, the filter capacitor voltage data, the grid-connected current data, the three-phase bridge arm side voltage data and the grid-connected voltage data, and discretizing the state space equation to determine a converter system prediction model of the converter in the stationary three-phase coordinate system specifically includes: Construct a state variable matrix of the converter according to the grid-connected current data, the filter capacitor voltage data and the three-phase bridge arm side current data; Construct an input voltage vector matrix of the converter according to the three-phase bridge arm side voltage data; Construct a grid-connected voltage vector matrix of the converter according to the grid-connected voltage data; Construct a state space equation of the converter according to the state variable matrix, the input voltage vector matrix and the grid-connected voltage vector matrix; Discretize the state space equation according to the preset zero-order hold method to obtain the converter system prediction model of the converter in the stationary three-phase coordinate system.

4. The control method of a three-phase four-leg LCL grid-connected converter according to claim 2 or 3, characterized in that, Construct the observation expressions of the grid-connected voltage data and the power frequency signal data in the converter, and construct the resonant state extended observer of the converter in the stationary three-phase coordinate system based on the observation expressions and the preset error feedback gain matrix, specifically including: Construct the observation expressions of the grid-connected voltage data and the power frequency signal data, and construct the converter state observation matrix and the resonant angular frequency observation matrix based on the observation expressions; Construct the inductor observation matrix according to the inductor data of the converter; Construct the observation expressions of the converter in the stationary three-phase coordinate system according to the observation matrix, the resonant angular frequency observation matrix, the inductor observation matrix and the preset error feedback gain matrix; Perform discretization processing and pole placement on the observation expressions to obtain the resonant state extended observer of the converter in the stationary three-phase coordinate system.

5. The control method of a three-phase four-leg LCL grid-connected converter according to claim 4, characterized in that, Perform observation on the converter based on the resonant state extended observer to obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step, specifically including: Perform observation on the power frequency sine signal of the converter based on the resonant state extended observer to obtain the power frequency observation data of the converter at the current time step; Convert the power frequency observation data based on the observation expressions to obtain the grid-connected voltage observation data of the converter at the current time step; Synchronize the grid-connected voltage observation data through a phase-locked loop to obtain the grid angular frequency observation value; Obtain the real-time grid-connected current data and the real-time filter capacitor voltage data of the converter at the current time step, and update the converter state observation matrix of the resonant state extended observer based on the real-time grid-connected current data and the power frequency observation data; Update the resonant angular frequency observation matrix of the resonant state extended observer based on the power frequency observation data; Solve the discrete-time gain matrix of the resonant state extended observer according to the grid angular frequency observation value; Obtain the grid-connected voltage estimation data of the converter at the first prediction time step based on the updated resonant state extended observer, the solution result of the discrete-time gain matrix and the real-time filter capacitor voltage data.

6. The control method of a three-phase four-leg LCL grid-connected converter according to claim 1, characterized in that, Determine the power reference prediction data of the converter at the second prediction time step based on the grid-connected voltage observation data, the grid-connected voltage estimation data and the preset power reference data, specifically including: Synchronize the grid-connected voltage observation data through a phase-locked loop to obtain the grid voltage phase angle observation value; The preset power reference data includes: the grid-connected current reference value; Calculate the grid-connected current reference prediction value of the converter at the second prediction time step according to the grid voltage phase angle observation value and the grid-connected current reference value; Calculate the filter capacitor voltage reference prediction value and the three-phase bridge arm side current reference prediction value of the converter at the second prediction time step according to the preset Lagrange extrapolation method, in combination with the grid-connected voltage observation data, the grid-connected voltage estimation data and the grid-connected current reference prediction value. Determine the predicted power reference data of the converter at the second prediction time step according to the predicted grid-connected current reference value, the predicted filter capacitor voltage reference value, and the predicted three-phase bridge arm side current reference value.

7. The control method of a three-phase four-leg LCL grid-connected converter according to claim 6, characterized in that, According to the preset Lagrangian extrapolation method, calculate the predicted filter capacitor voltage reference value and the predicted three-phase bridge arm side current reference value of the converter at the second prediction time step by combining the grid-connected voltage observation data, the grid-connected voltage estimation data, and the predicted grid-connected current reference value. Specifically, it includes: According to the preset Lagrangian extrapolation method, calculate the predicted grid-connected voltage value and the derivative of the predicted grid-connected voltage value of the converter at the second prediction time step by combining the grid-connected voltage observation data and the grid-connected voltage estimation data. Based on the predicted grid-connected voltage value and the predicted grid-connected current reference value, calculate the predicted filter capacitor voltage reference value of the converter at the second prediction time step. Based on the derivative of the predicted grid-connected voltage value and the predicted grid-connected current reference value, calculate the predicted three-phase bridge arm side current reference value of the converter at the second prediction time step.

8. The control method of a three-phase four-leg LCL grid-connected converter according to claim 1, wherein, Obtain the real-time switching state of the converter at the current time step, and construct the control objective function of the converter based on the real-time switching state, the predicted power data, and the predicted power reference data. Specifically, it includes: Obtain the real-time switching state of the converter at the current time step, and construct a switching action loss term by combining a preset switching action weight coefficient. The predicted power data includes: the predicted grid-connected current value, the predicted three-phase bridge arm side current value, and the predicted filter capacitor voltage value. Construct a grid-connected current steady-state tracking accuracy term according to the predicted grid-connected current value, the predicted grid-connected current reference value, and a preset grid-connected current steady-state weight coefficient. Construct a three-phase bridge arm side current change term according to the predicted three-phase bridge arm side current value and the predicted three-phase bridge arm side current reference value. Construct a converter resonance suppression term according to the predicted filter capacitor voltage value, the predicted filter capacitor voltage reference value, and a preset resonance suppression weight coefficient. Construct the control objective function of the converter according to the switching action loss term, the grid-connected current steady-state tracking accuracy term, the three-phase bridge arm side current change term, and the converter resonance suppression term.

9. The control method of a three-phase four-leg LCL grid-connected converter according to claim 1 or 8, characterized in that, Solve the control objective function to obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state. Specifically, it includes: Taking the minimization of the control objective function as the solution target, solve the control objective function according to the preset finite control set model predictive method to obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state.

10. A control device for a three-phase four-leg LCL grid-connected converter, characterized in that, It includes: A converter modeling module, a resonant state extended observer observation module, a converter system prediction model prediction module, a power reference prediction data acquisition module, a control objective function construction module, and a converter control module; Among them, the converter modeling module is used to obtain the operating variable data of the converter, and construct a converter system prediction model and a resonant state extended observer of the converter in the stationary three-phase coordinate system based on the operating variable data. The resonant state extended observer observation module is used to observe the converter based on the resonant state extended observer, and obtain the grid-connected voltage observation data of the converter at the current time step and the grid-connected voltage estimation data at the first prediction time step; The converter system prediction model prediction module is used to input the grid-connected voltage observation data and the grid-connected voltage estimation data into the converter system prediction model, and obtain the power prediction data of the converter at the second prediction time step; The power reference prediction data acquisition module is used to determine the power reference prediction data of the converter at the second prediction time step based on the grid-connected voltage observation data, the grid-connected voltage estimation data and the preset power reference data; The control objective function construction module is used to obtain the real-time switching state of the converter at the current time step, and construct the control objective function of the converter based on the real-time switching state, the predicted power data and the predicted power reference data; The converter control module is used to solve the control objective function, obtain the estimated switching state of the converter at the first prediction time step, and control the converter based on the estimated switching state.

Images