Predictive Control Method for LCL Filtered Three-Phase Photovoltaic Grid-Connected Inverter System

Abstract

The application discloses a kind of prediction control methods of LCL filter type three-phase photovoltaic grid-connected inverter system, belong to the field of power electronics control. Including: using sampling information to establish the continuous domain state equation of LCL filter type three-phase photovoltaic grid-connected inverter system, the given value of inductance current and capacitor voltage is established by calculating steady-state value discrete prediction model;Resonance component is obtained by the difference between capacitor voltage prediction value and capacitor voltage given value, the prediction control instruction of integrated resonance suppression is calculated;Determine the optimal vector;Performance function of the tracking error of prediction control instruction synthesis is constructed, and the optimal vector acting time that can minimize performance function is calculated;Design out that symmetric 4 section type control pulse signal acts on inverter switch tube and carries out prediction control.This method can effectively suppress the resonance oscillation of LCL filter type three-phase photovoltaic grid-connected inverter system, realize fixed switching frequency, improve stability and steady-state performance.

Description

Technical Field

[0001] The invention relates to a prediction control method for an LCL filter type three-phase photovoltaic grid-connected inverter system, and belongs to the field of power electronics and electric power transmission. Background Art

[0002] With the increasing problems of energy depletion and environmental pollution, renewable energy power generation systems are experiencing rapid development, particularly distributed photovoltaic systems, whose installed capacity has increased annually. Photovoltaic renewable energy sources transmit generated electricity to the grid through grid-connected operation. Distributed photovoltaic grid-connected systems are often installed on rooftops or other supports, featuring small scale, simplicity, flexibility, and low line losses. As the key interface device between distributed photovoltaic power sources and the grid, the performance of photovoltaic grid-connected inverters directly impacts grid connection quality and grid stability. Because photovoltaic grid-connected inverters utilize power electronic components, their switching frequency is high during normal operation, injecting high-frequency harmonics into the grid, impacting power quality and operational stability. To achieve satisfactory high-order harmonic suppression, LCL filters are typically installed between the photovoltaic grid-connected inverter and the grid. However, the introduction of LCL filters not only increases the number of control steps but also causes resonance issues in the grid-connected current, seriously threatening the stability and reliability of photovoltaic grid-connected systems.

[0003] In recent years, model predictive control (MPC) methods have attracted widespread attention in the control of LCL-filtered three-phase photovoltaic grid-connected inverter systems due to their simple structure, excellent dynamic performance, and optimal control. Currently, to suppress LCL resonance, existing predictive control methods for photovoltaic grid-connected inverters utilize multi-objective finite set predictive model control. This requires simultaneous prediction of the inverter current, capacitor voltage, and grid current, and requires multiple iterations of the objective function to determine the optimal vector. Consequently, these methods face challenges such as multiple control parameters (two), difficulty in parameter tuning (trial and error based on empirical experience), and high computational complexity (due to multiple iterations), further complicating their implementation. Furthermore, existing MPC methods only apply a single vector within a control cycle, resulting in large steady-state harmonics and ripple, as well as unstable switching frequency, in LCL-filtered three-phase photovoltaic grid-connected inverter systems. This not only hinders the design of LCL filters but also significantly hinders the application of MPC in photovoltaic grid-connected inverters. Summary of the Invention

[0004] In view of the limitations of the existing technology, a predictive control method for an LCL filter type three-phase photovoltaic grid-connected inverter system is provided. This method extracts the resonant component by subtracting the capacitor voltage set value from the capacitor voltage predicted value, and feeds it back to the photovoltaic grid-connected inverter control instruction to obtain the photovoltaic grid-connected inverter control instruction with integrated resonance suppression, which effectively solves the resonance problem of the LCL filter type three-phase photovoltaic grid-connected inverter system, simplifies the implementation and enhances the stability of the LCL filter type three-phase photovoltaic grid-connected inverter system; this method only has a damping coefficient K r There is only one parameter to be designed, and the cyclic optimization process is eliminated, which greatly reduces the workload and complexity of parameter tuning; the symmetrical four-segment pulse generation of this method does not use a zero vector, so it can reduce the common-mode voltage and leakage current of the LCL filter-type three-phase photovoltaic grid-connected inverter system; this method achieves a fixed switching frequency, which is beneficial to the parameter design of the LCL filter.

[0005] To achieve the above technical objectives, the present invention provides a predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system, comprising the following steps:

[0006] Step 1: By sampling the inductor current, capacitor voltage, grid current and grid voltage of the LCL filter type three-phase photovoltaic grid-connected inverter system, the continuous domain state equation of the LCL filter type three-phase photovoltaic grid-connected inverter system in the dq coordinate system is established, and the inductor current given value i is calculated through the steady-state value of the state variable. fd,k * 、i fq,k * and capacitor voltage given value u cd,k * 、u cq,k * ;

[0007] Step 2: Based on the continuous-time domain state equation of the LCL filter type three-phase photovoltaic grid-connected inverter system, a discrete prediction model of the LCL filter type three-phase photovoltaic grid-connected inverter system is established using the forward Euler method;

[0008] Step 3: Based on the given values ​​of inductor current and capacitor voltage and the discrete prediction model of the LCL filter type three-phase photovoltaic grid-connected inverter system, the basic prediction control instruction v is calculated based on the principle that the inductor current prediction value tracks the given value of the inductor current within one sampling period. d,k * and v q,k * , using the resonance component obtained by the difference between the capacitor voltage prediction value and the capacitor voltage set value, calculate the prediction control instruction v for integrated resonance suppression rd,k * and v rq,k *, and integrate the predicted control instruction v of the resonance suppression rd,k * and v rq,k * Perform Park inverse transform to obtain the predictive control instruction v for integrated resonance suppression in the αβ coordinate system rα,k * and v rβ,k * ;

[0009] Step 4: Obtain the predicted control command v for integrated resonance suppression rα,k * and v rβ,k * The spatial position angle γ is used to determine the predictive control command v that can optimally synthesize the integrated resonance suppression. rα,k * and v rβ,k * The two optimal vectors v p,1 and v p,2 ;

[0010] Step 5: With the help of two optimal vectors and two optimal vector action times that can optimally synthesize the integrated resonance suppression predictive control command, construct the performance function of the predicted control command synthesis tracking error based on integrated resonance suppression, and use the Lagrange multiplier method to calculate the optimal vector action time T that can minimize the performance function. i ;

[0011] Step 6: Design a symmetrical four-segment control pulse signal generation strategy S and apply the strategy S to the inverter switch tube to achieve predictive control of the LCL filter type three-phase photovoltaic grid-connected inverter system.

[0012] Furthermore, the step 1 specifically includes: establishing a continuous time domain state equation of the sampling LCL filter type three-phase photovoltaic grid-connected inverter system in the dq coordinate system as follows:

[0013]

[0014] Where i fd 、i fq Indicates the dq axis inductor current, u cd 、u cq represents the dq axis capacitor voltage, i gd 、i gq Indicates the dq axis grid-connected current, v d 、v q Indicates the output voltage of the dq axis inverter, u gd 、u gq Indicates the dq axis grid voltage; L f and R fL represents the filter inductance and the inverse side of the parasitic resistance, L g and R g L represents the grid-side filter inductance and the grid-side parasitic resistance, C is the filter capacitance, and ω e is the fundamental angular frequency of the grid voltage;

[0015] Considering that the steady-state value of the LCL filter type three-phase photovoltaic grid-connected inverter system state variable is no longer changed and the steady-state value of the grid-connected current is equal to the grid-connected current reference value: i gd =i gd,k * , i gq =i gq,k * The given value of the LCL filter type three-phase photovoltaic grid-connected inverter system capacitor voltage u cd,k * , u cq,k * and the given value of the inductance current i fd,k * , i fq,k * are calculated as follows:

[0016]

[0017] In the formula, i gd,k * and i gq,k * are the given values of the dq-axis grid-connected current.

[0018] Further, the step 2 specifically comprises: setting the sampling period of the LCL filter type three-phase photovoltaic grid-connected inverter system as T s , discretizing the continuous time domain state equation (1) by using the forward Euler method, and establishing the prediction model of the inductance current and the capacitor voltage of the LCL filter type three-phase photovoltaic grid-connected inverter system as follows:

[0019]

[0020] In the formula, the subscripts "k" and "k+1" respectively represent the current k-step sampling time and the future k+1-step prediction value.

[0021] Further, the step 3 specifically comprises: based on the principle that the inductance current prediction value tracks the inductance current reference value in one sampling period, i.e., according to the given values of the inductance current and the capacitor voltage of the LCL filter type three-phase photovoltaic grid-connected inverter system and the prediction model of the inductance current and the capacitor voltage, i fd,k+1 =i fd,k * , i fq,k+1 =i fq,k *, calculate the basic predictive control instruction v of the LCL filter type three-phase photovoltaic grid-connected inverter system d,k * and v q,k * for:

[0022]

[0023] In order to suppress the de-resonant oscillation and improve the stability of the LCL filter type three-phase photovoltaic grid-connected inverter system, considering that the capacitor voltage prediction value contains both fundamental and resonant components, while the reference value only contains the fundamental component, the resonant component can be obtained by making a difference between the capacitor voltage prediction value and the capacitor voltage set value, and the resonant component is taken from the basic prediction control instruction v d,k * and v q,k * Subtract from the original, and get the predictive control instruction v with integrated resonance suppression rd,k * and v rq,k * :

[0024]

[0025] Where K r >0 is the adjustable resonance suppression gain, which is obtained through simulation or experimental trial;

[0026] Finally, through the Park inverse transform The predictive control command v for integrated resonance suppression in the αβ coordinate system is obtained rα,k * and v rβ,k * :

[0027]

[0028] Furthermore, the step 4 specifically includes: obtaining the prediction control instruction v of the integrated resonance suppression by the inverse tangent function rα,k * and v rβ,k * The spatial position angles are as follows:

[0029]

[0030] According to the spatial sector where γ is located, the predictive control instruction v that can optimally synthesize the integrated resonance suppression is determined. rα,k * and v rβ,k * The two optimal vectors v p,1 =[v pα,1 v pβ,1 ] and vp,2 =[v pα,2 v pβ,2 ].

[0031] Furthermore, step 5 specifically includes: constructing a performance function of the integrated resonance suppression predictive control instruction-based tracking error based on the two optimal vectors determined in step 4 that can optimally synthesize the integrated resonance suppression predictive control instruction in combination with the action time of the two optimal vectors as follows:

[0032] G=(T1e α,1 ) 2 +(T1e β,1 ) 2 +(T2e α,2 ) 2 +(T2e β,2 ) 2 (8),

[0033] Where T1 and T2 are the action times of the two optimal vectors, respectively, satisfying T1+T2=T s Constraints;

[0034] and are the tracking errors between the two optimal vectors and the predictive control instructions with integrated resonance suppression in the αβ coordinate system, respectively.

[0035] Furthermore, in order to solve T1+T2=T s Under the equality constraint, the optimal vector action time that can minimize the performance function G is constructed by using the Lagrange multiplier method. The following Lagrange function f is first constructed:

[0036] f(T1,T2,λ)=G+λ(T1+T2-T s ) (9),

[0037] Where λ is the Lagrangian operator;

[0038] Secondly, let the partial derivatives of the Lagrangian function f with respect to the time of the two optimal vectors be equal to 0, and we can obtain:

[0039]

[0040] Combine Equation (10) with the equality constraint T1+T2=T s Combined, we get the expression of the Lagrangian operator λ:

[0041]

[0042] Derived the action time T of the two optimal vectors i (i=1, 2) is:

[0043]

[0044] Furthermore, the step 6 specifically includes: in order to achieve a fixed switching frequency and a low common mode voltage, designing the following symmetrical 4-segment pulse pattern to generate the control pulse signal S:

[0045]

[0046] Finally, the pulse signal S is applied to the switch tube of the inverter to realize the predictive control of the LCL filter type three-phase photovoltaic grid-connected inverter system.

[0047] A computer device comprising: a processor, a memory, and a network interface;

[0048] The processor is connected to the memory and the network interface, wherein the network interface is used to provide network communication functions, the memory is used to store program code, and the processor is used to call the program code to execute the predictive control method of the LCL filter type three-phase photovoltaic grid-connected inverter system.

[0049] A computer-readable storage medium storing a computer program suitable for being loaded and executed by a processor to implement the predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system. Compared with the prior art, the present invention has the following advantages:

[0050] 1. The method provided by the present invention uses the capacitor voltage predicted value minus the capacitor voltage set value to extract the resonant component, which effectively solves the resonance problem of the LCL filter type three-phase photovoltaic grid-connected inverter system and enhances the stability of the LCL filter type three-phase photovoltaic grid-connected inverter system.

[0051] 2. The method provided by the present invention only includes one damping coefficient K r Design parameters, and there is no cyclic traversal optimization process, which greatly reduces the workload of parameter tuning and the complexity of control method implementation.

[0052] 3. The optimal vector action time calculation and pulse signal generation method used in the method provided by the present invention can effectively improve the steady-state performance of the LCL filter-type three-phase photovoltaic grid-connected inverter system, reduce the common-mode voltage / leakage current, and achieve a fixed switching frequency, which is beneficial to the design of the LCL filter. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a schematic block diagram of a predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system according to the present invention;

[0054] Figure 2 is a basic voltage space vector diagram in an embodiment of the present invention;

[0055] Figure 3 Schematic diagram of grid voltage and grid-connected current transient waveforms and grid-connected current harmonic spectrum in an embodiment of the present invention;

[0056] Figure 4 Schematic diagram of grid voltage and grid-connected current transient waveforms and grid-connected current harmonic spectrum in a traditional embodiment. DETAILED DESCRIPTION

[0057] The following drawings further illustrate the embodiments of the present invention:

[0058] like Figure 1 As shown, the present invention provides an LCL filter type three-phase photovoltaic grid-connected inverter system using a predictive control method thereof, wherein a DC bus voltage source is converted into an AC voltage square wave signal by a three-phase voltage source inverter, and then connected to the grid through an output LCL filter; and the three-phase inductor current, three-phase capacitor voltage, three-phase grid-connected current and three-phase grid voltage are sampled in sequence.

[0059] Figure 1 In, V dc Indicates DC bus voltage; L f and R f Indicates the filter inductance and parasitic resistance on the output side of the inverter, L g and R g Represents the grid side filter inductance and parasitic resistance, and C represents the filter capacitor. f,abc Indicates the three-phase filter inductor current; u c,abc Represents the three-phase filter capacitor voltage; i g,abc Indicates the grid-connected current; u g,abc Indicates the grid voltage; i g,k * is the given value of grid-connected current;

[0060] i g,k 、u g,k 、u c,k and i f,k They represent the grid-connected current, grid voltage, capacitor voltage and inductor current in the dq coordinate system at time k respectively; the sampled grid voltage is sent to the phase-locked loop to obtain the grid voltage phase θ; S represents the switching state of the inverter upper arm switch tube.

[0061] A predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system comprises the following steps:

[0062] Step 1: Establish the continuous time domain state equation of the system in the dq coordinate system as follows:

[0063]

[0064] Where i fd 、i fq Indicates the dq axis inductor current, u cd 、u cq represents the dq axis capacitor voltage, i gd 、i gq Indicates the dq axis grid-connected current, v d 、v q Indicates the output voltage of the dq axis inverter, u gd 、u gq Indicates the dq axis grid voltage; L f and R f Indicates the filter inductance and parasitic resistance on the inverter side, L g and R g is the grid-side filter inductor and grid-side parasitic resistance, C is the filter capacitor, ω e is the grid voltage fundamental angular frequency.

[0065] Considering that the steady-state value of the state variable of the LCL filter type three-phase photovoltaic grid-connected inverter system no longer changes, and the steady-state value of the grid-connected current is equal to the grid-connected current reference value: i gd =i gd,k * ,i gq =i gq,k * , the capacitor voltage given value u can be calculated cd,k * 、u cq,k * With the given value of inductor current i fd,k * 、i fq,k * for:

[0066]

[0067] Where i gd,k * and i gq,k * is the given value of the dq axis grid-connected current.

[0068] Step 2: In order to realize the predictive control of the inverter side current reference tracking, the sampling period of the LCL filter type three-phase photovoltaic grid-connected inverter system is set to T s , using the forward Euler method to discretize the continuous domain state equation in step 1, the prediction model of the inductor current and capacitor voltage of the LCL filter type three-phase photovoltaic grid-connected inverter system can be established as follows:

[0069]

[0070] Where the subscripts “k” and “k+1” represent the current k-th sampling moment and the predicted value in the k+1th step in the future, respectively.

[0071] Step 3: Based on the principle that the inductor current prediction value tracks the inductor current reference value within a sampling period, that is, according to the given values ​​of the inductor current and capacitor voltage and the prediction model of the inductor current and capacitor voltage, let i fd,k+1 =i fd,k * ,i fq,k+1 =i fq,k * , the basic predictive control instruction v of the LCL filter type three-phase photovoltaic grid-connected inverter system can be calculated d,k * and v q,k * :

[0072]

[0073] Then, in order to suppress the resonant oscillation and improve the stability of the LCL filter type three-phase photovoltaic grid-connected inverter system, considering that the capacitor voltage prediction value contains both fundamental and resonant components, while the reference value only contains the fundamental component, the resonant component can be obtained by making a difference between the capacitor voltage prediction value and the capacitor voltage given value, and the resonant component is taken from the basic prediction control instruction v d,k * and v q,k * Subtracting from the above, we can get the predictive control instruction v with integrated resonance suppression. rd,k * and v rq,k * :

[0074]

[0075] Where K r >0 is an adjustable resonance suppression gain, which can be obtained through simulation or experimental trial.

[0076] Finally, through the Park inverse transform The predictive control command v for integrated resonance suppression in the αβ coordinate system can be obtained rα,k * and v rβ,k * :

[0077]

[0078] Step 4: Obtain the predicted control command v for integrated resonance suppression using the inverse tangent function rα,k * and v rβ,k *The spatial position angles are as follows:

[0079]

[0080] like Figure 2 As shown, v αβ,1 to v αβ,6 is the basic voltage vector, and according to the spatial sector l where γ is located, the predictive control instruction v that can optimally synthesize the integrated resonance suppression is determined. rα,k * and v rβ,k * The two optimal vectors v p,1 =[v pα,1 v pβ,1 ] and v p,2 =[v pα,2 v pβ,2 ].

[0081] Step 5: Based on the two optimal vectors that can optimally synthesize the integrated resonance suppression predictive control command determined in step 4, the performance function of the integrated resonance suppression predictive control command synthesis tracking error is constructed by combining the action time of the two optimal vectors as follows:

[0082] G=(T1e α,1 ) 2 +(T1e β,1 ) 2 +(T2e α,2 ) 2 +(T2e β,2 ) 2 ,

[0083] Where T1 and T2 are the action times of the two optimal vectors, respectively, satisfying T1+T2=T s Constraints;

[0084] and are the tracking errors between the two optimal vectors and the predictive control instructions with integrated resonance suppression in the αβ coordinate system, respectively.

[0085] In order to solve T1+T2=T s Under the equality constraint, the optimal vector action time that can minimize the performance function G is constructed by using the Lagrange multiplier method. The following Lagrange function f is first constructed:

[0086] f(T1,T2,λ)=G+λ(T1+T2-T s ),

[0087] Where λ is the Lagrangian operator.

[0088] Secondly, let the partial derivatives of the Lagrangian function f with respect to the action time of the two optimal vectors be equal to 0, and we can get:

[0089]

[0090] Combine the above equation with the equality constraint T1+T2=T s By combining the above, we can get the expression of Lagrangian operator λ:

[0091]

[0092] The action time T of the two optimal vectors can be derived i (i=1, 2) is:

[0093]

[0094] Step 6: To achieve a fixed switching frequency and low common-mode voltage, design the following symmetrical 4-segment pulse pattern to generate the control pulse signal S:

[0095]

[0096] Finally, the pulse signal S is applied to the switch tube of the inverter to realize the predictive control of the LCL filter type three-phase photovoltaic grid-connected inverter system.

[0097] Each vector represents the switching state of the inverter upper arm switch tube. Figure 2 As shown in Figure 1 (1 for on, 0 for off), the switching state of the lower-arm switch is opposite to that of the upper-arm. Finally, the pulse signal is applied to the inverter switch to achieve predictive control of the LCL photovoltaic grid-connected inverter.

[0098] In order to verify the predictive control method of an LCL filter type three-phase photovoltaic grid-connected inverter system provided by the present invention, the method provided by the present invention is applied to the LCL filter type three-phase photovoltaic grid-connected inverter system, and the parameters are given in Table 1.

[0099] Table 1

[0100] Parameter Symbol Value DC bus voltage <![CDATA[V dc ]]> 600V Line-to-line RMS voltage <![CDATA[U g ]]> 220V Grid fundamental angular frequency <![CDATA[ω e ]]> 100π Grid side parasitic resistance <![CDATA[R g ]]> 0.2 Ω Grid side filter inductance <![CDATA[L g ]]> 2 mH Inverter side filter inductance <![CDATA[L f ]]> 1 mH Inverter side parasitic resistance <![CDATA[R f ]]> 0.2 Ω Filter capacitance C 10 μF Sampling frequency <![CDATA[f s ]]> 10 kHz

[0101] Figure 3 and Figure 4 Figures 1 and 2 are the transient waveforms of the grid voltage and grid current under the control method of the present invention and the traditional multi-objective finite set prediction model, and a schematic diagram of the grid current harmonic spectrum. The given value of the grid current suddenly changes from 10A to 30A at 0.1s and suddenly changes to 20A at 0.15s.

[0102] Grid voltage u under two control methods g and grid-connected current i gBoth can quickly follow the given reference amplitude when the grid-connected current suddenly changes, and both can complete the smooth transition of the grid-connected current without generating resonance, so the dynamic performance of the two control methods is comparable. Nevertheless, compared with the method of the present invention, the total harmonic distortion rate of the grid-connected current under the traditional control method is larger. On the contrary, the steady-state performance of the proposed control method is better, and it can be seen from the harmonic spectrum that the method of the present invention achieves a fixed switching frequency of 10kHz. Therefore, the method provided by the present invention can effectively improve the steady-state performance and operational stability of the LCL filter-type three-phase photovoltaic grid-connected inverter system.

Claims

1. A predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system, characterized in that: The steps include: Step 1: By sampling the inductor current, capacitor voltage, grid current and grid voltage of the LCL filter type three-phase photovoltaic grid-connected inverter system, the continuous time domain state equation of the LCL filter type three-phase photovoltaic grid-connected inverter system in the dq coordinate system is established, and the inductor current given value i is calculated through the steady-state value of the state variable. fd,k * 、i fq,k * and capacitor voltage given value u cd,k * 、u cq,k * ; Step 2: Based on the continuous-time domain state equation of the LCL filter type three-phase photovoltaic grid-connected inverter system, a discrete prediction model of the LCL filter type three-phase photovoltaic grid-connected inverter system is established using the forward Euler method; Step 3: Based on the given values ​​of inductor current and capacitor voltage and the discrete prediction model of the LCL filter type three-phase photovoltaic grid-connected inverter system, the basic prediction control instruction v is calculated based on the principle that the inductor current prediction value tracks the given value of the inductor current within one sampling period. d,k * and v q,k * , using the capacitor voltage resonance component obtained by making the difference between the capacitor voltage prediction value and the capacitor voltage given value, calculate the prediction control instruction v for integrated resonance suppression rd,k * and v rq,k * , and integrate the predicted control instruction v of the resonance suppression rd,k * and v rq,k * Perform Park inverse transform to obtain the predictive control instruction v for integrated resonance suppression in the αβ coordinate system rα,k * and v rβ,k * ; Step 4: Obtain the predicted control command v for integrated resonance suppression rα,k * and v rβ,k * The spatial position angle γ is used to determine the predictive control command v that can optimally synthesize the integrated resonance suppression. rα,k * and v rβ,k * The two optimal vectors v p,1 and v p,2 ; Step 5: With the help of the two optimal vectors and two optimal vector action times that can optimally synthesize the predictive control instructions with integrated resonance suppression, a performance function of the synthetic tracking error of the predictive control instructions based on integrated resonance suppression is constructed, and the optimal vector action time T that can minimize the performance function is calculated using the Lagrange multiplier method. i ; Step 6: Design a symmetrical four-segment control pulse signal generation strategy and apply it to the inverter switches to achieve predictive control of the LCL filter-type three-phase photovoltaic grid-connected inverter system; The step 2 specifically includes: assuming that the sampling period of the LCL filter type three-phase photovoltaic grid-connected inverter system is T s , the forward Euler method is used to discretize the continuous-time domain state equation, and the prediction model of the inductor current and capacitor voltage of the LCL filter type three-phase photovoltaic grid-connected inverter system is established as follows: Wherein, the subscripts "k" and "k+1" represent the current k-th sampling moment and the predicted value in the k+1th step in the future, respectively; In the formula, the state variable i fd 、i fq represents the dq axis inductor current, the state variable u cd 、u cq represents the dq axis capacitor voltage, the state variable i gd 、i gq Indicates the dq axis grid-connected current; v d 、v q Indicates the output voltage of the dq axis inverter; L f and R f represents the filter inductor and the parasitic resistance on the inverter side, C is the filter capacitor, ω e is the grid voltage fundamental angular frequency; The step 3 specifically includes: tracking the inductor current reference value within a sampling period based on the inductor current prediction value, that is, according to the inductor current, capacitor voltage given values ​​and the inductor current and capacitor voltage prediction models, Let i fd,k+1 =i fd,k * ,i fq,k+1 =i fq,k * , the basic predictive control instruction v can be calculated d,k * and v q,k * for: In order to suppress resonant oscillation and improve the stability of the LCL filter type three-phase photovoltaic grid-connected inverter system, considering that the capacitor voltage prediction value contains both fundamental and resonant components, while the reference value only contains the fundamental component, the resonant component can be obtained by subtracting the capacitor voltage prediction value from the capacitor voltage given value, and the resonant component is taken from the basic prediction control instruction v d,k * and v q,k * Subtract from the original, and get the predictive control instruction v with integrated resonance suppression rd,k * and v rq,k * : Where K r >0 is the adjustable resonance suppression gain, which is obtained through simulation or experimental trial; Finally, through the Park inverse transform The predictive control command v for integrated resonance suppression in the αβ coordinate system is obtained rα,k * and v rβ,k * :

2. The predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system according to claim 1, characterized in that: The step 1 specifically includes: The continuous time domain state equation of the LCL filter type three-phase photovoltaic grid-connected inverter system in the dq coordinate system is established as follows: In the formula, the state variable i fd 、i fq represents the dq axis inductor current, the state variable u cd 、u cq represents the dq axis capacitor voltage, the state variable i gd 、i gq Indicates the dq axis grid-connected current; v d 、v q Indicates the dq axis inverter output voltage, u gd 、u gq Indicates the dq axis grid voltage; L f and R f Indicates the filter inductance and parasitic resistance on the inverter side, L g and R g Indicates the grid-side filter inductance and grid-side parasitic resistance, C is the filter capacitor, ω e is the grid voltage fundamental angular frequency; Considering that the steady-state value of the state variable of the LCL filter type three-phase photovoltaic grid-connected inverter system no longer changes, and the steady-state value of the grid-connected current is equal to the grid-connected current reference value: i gd =i gd,k * ,i gq =i gq,k * , the capacitor voltage given value u of the LCL filter type three-phase photovoltaic grid-connected inverter system can be calculated cd,k * 、u cq,k * With the given value of inductor current i fd,k * 、i fq,k * They are: Where i gd,k * and i gq,k * is the given value of the dq-axis grid-connected current.

3. The predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system according to claim 2, characterized in that: The step 4 specifically includes: According to the prediction control command v obtained by the inverse tangent function for integrated resonance suppression rα,k * and v rβ,k * The spatial position angles are as follows: According to the spatial sector where γ is located, the predictive control instruction v that can optimally synthesize the integrated resonance suppression is determined. rα,k * and v rβ,k * The two optimal vectors v p,1 =[v pα,1 v pβ,1 ] and v p,2 =[v pα,2 v pβ,2 ].

4. The predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system according to claim 3, characterized in that: The step 5 specifically includes: Based on the two optimal vectors that can optimally synthesize the integrated resonance suppression predictive control command determined in step 4, the performance function of the integrated resonance suppression predictive control command synthesis tracking error is constructed by combining the action time of the two optimal vectors as follows: G=(T1e α,1 ) 2 +(T1e β,1 ) 2 +(T2e α,2 ) 2 +(T2e β,2 ) 2 (8), Where T1 and T2 are the action times of the two optimal vectors, respectively, satisfying T1+T2=T s Constraints; and are the tracking errors between the two optimal vectors and the integrated resonance suppression predictive control instructions in the αβ coordinate system.

5. The predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system according to claim 4, characterized in that: In order to solve T1+T2=T s Under the equality constraint, the optimal vector that can minimize the performance function G is used to construct the following Lagrangian function f using the Lagrangian multiplier method: f(T1,T2,λ)=G+λ(T1+T2-T s ) (9), Where λ is the Lagrangian operator; Secondly, let the partial derivatives of the Lagrangian function f with respect to the action time of the two optimal vectors be equal to 0, and we can obtain: Combine Equation (10) with the equality constraint T1+T2=T s Combined, we get the expression of the Lagrangian operator λ: Derived the action time T of the two optimal vectors i for:

6. The predictive control method for an LCL filter-type three-phase photovoltaic grid-connected inverter system according to claim 5, characterized in that: The step 6 specifically includes: In order to achieve a fixed switching frequency and low common-mode voltage, the following symmetrical 4-segment pulse pattern is designed to generate the control pulse signal S: Finally, the pulse signal S is applied to the switch tube of the inverter to realize the predictive control of the LCL filter type three-phase photovoltaic grid-connected inverter system.

7. A computer device, characterized in that: include: processor, memory, and network interface; The processor is connected to the memory and the network interface, wherein the network interface is used to provide network communication functions, the memory is used to store program code, and the processor is used to call the program code to execute the predictive control method of the LCL filter-type three-phase photovoltaic grid-connected inverter system according to any one of claims 1-6.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which is suitable for being loaded by a processor and executing the predictive control method of the LCL filter-type three-phase photovoltaic grid-connected inverter system according to any one of claims 1 to 6.

Images