Predictive control method and system for network-forming inverter

Abstract

The invention discloses a predictive control method and system for a network-building type inverter, and belongs to the technical field of power electronics, and the method comprises the steps: building an output voltage prediction model according to the output voltage data and historical control input signals of the network-building type inverter at the current moment; obtaining a current output voltage prediction value through the output voltage prediction model; obtaining a reference voltage vector of the inverter at the next moment according to a preset optimization criterion function and the current output voltage predicted value, and obtaining an optimal control input signal at the next moment; and according to the optimal control input signal at the next moment, regulating and controlling a switch element of the grid-forming inverter. Therefore, by implementing the method and the device, the problems that a traditional output voltage prediction model is easy to generate model mismatch for an actual inverter, and the reliability of a prediction result is greatly reduced when the data of the current sensor is changed due to the fact that a traditional grid-forming inverter model excessively depends on the data of the current sensor in the prior art can be solved.

Description

Technical Field

[0001] This application belongs to the field of power electronics technology, and particularly relates to a predictive control method and system for a grid-forming inverter. Background Art

[0002] In recent years, with the continuous development of new energy power generation technologies such as photovoltaic and wind power, great attention has been paid to vigorously developing distributed new energy power generation systems to improve the utilization rate of new energy. Among them, the grid-forming inverter, due to its voltage source characteristics and the characteristics of fast response and high controllability, plays a key role in ensuring the stable voltage and frequency support of the system, and has become an indispensable part of the safe and stable operation of distributed new energy power generation systems. The grid-forming inverter mainly designs its optimal switching element based on the predicted output state to achieve the purpose of stabilizing the voltage of the power system. Therefore, how to accurately predict the output state of the inverter to more precisely control the inverter is worthy of research.

[0003] However, the predictive control performance of the traditional grid-forming inverter model depends severely on the accuracy of the prediction model. When there is a model mismatch between the prediction model and the actual inverter system, the output voltage of the inverter will be severely distorted, and the output performance of the inverter will deteriorate sharply. At the same time, the predictive control of the traditional grid-connected inverter model needs to collect information such as the filter capacitor voltage, inverter-side current, and output current simultaneously. More data input will make the operation reliability of the inverter worse. Once there is an error in a certain data, the voltage control effect of the inverter will be reduced. Summary of the Invention

[0004] This application proposes a predictive control method and system for a grid-forming inverter, which constructs a corresponding output voltage prediction model only using the input and output data of the grid-forming inverter, and can obtain the optimal control signal of the inverter switching element accurately, so that the voltage output by the inverter can effectively maintain the stable operation of the power system at the next moment.

[0005] The first aspect of this application provides a predictive control method for a grid-forming inverter, and the method includes:

[0006] Construct an output voltage prediction model according to the output voltage data of the grid-forming inverter collected at the current moment and the historical control input signal;

[0007] Input the output voltage prediction value of the previous moment into the output voltage prediction model and perform calculation delay compensation to obtain the current output voltage prediction value;

[0008] According to the preset optimization criterion function and the current output voltage prediction value, combined with the corresponding relationship between the control input signal and the output voltage of the grid-forming inverter, obtain the reference voltage vector of the grid-forming inverter at the next moment;

[0009] Screen the reference voltage vector based on a preset midpoint balance principle to obtain the optimal control input signal for the next moment;

[0010] Regulate the switching elements of the network-forming inverter according to the optimal control input signal for the next moment.

[0011] The above solution can construct an output voltage prediction model only through the output voltage of the network-forming inverter at the current moment and the historical input and output data, without relying on any system parameters, completely eliminating the problem of model mismatch, and based on the data at the current moment, it can more ensure that the prediction result of the model conforms to the actual situation. Moreover, inputting the output voltage prediction value of the previous moment into the output voltage prediction model can obtain the accurate current output voltage prediction value, fundamentally eliminating the dependence of the prediction model on the filtered current and output current sensors, and preventing the prediction result from being distorted due to incorrect input data, improving the operation reliability of the inverter. In addition, considering the influence of the control input signal of the network-forming inverter on the output voltage, aiming at the stability of the DC-side voltage of the inverter, the switching state of the inverter that can make the power system operate stably is obtained in reverse, and the inverter is regulated accordingly, effectively maintaining the stable operation of the power system.

[0012] In a possible implementation method of the first aspect, according to the collected output voltage data of the network-forming inverter at the current moment and the historical control input signals, construct an output voltage full-format prediction model, specifically:

[0013] Collect the output voltage data of the network-forming inverter at the current moment, and the historical control input signals of the switching elements of the network-forming inverter;

[0014] Construct an inverter data matrix according to the output voltage data at the current moment and the historical control input signals;

[0015] Construct an output voltage prediction model based on the inverter data matrix and a preset pseudo-partial derivative matrix;

[0016] Among them, the pseudo-partial derivative matrix is used to describe the relationship between the output voltage of the network-forming inverter and the historical input and output data of the network-forming inverter.

[0017] The above solution can construct the corresponding output voltage full-format prediction model only through the collected output voltage data of the network-forming inverter at the current moment and the relevant historical input data, making the prediction result of the model more matched with the actual operation of the inverter. Moreover, considering the relationship between the output voltage of the network-forming inverter and the historical input and output data of the network-forming inverter, a pseudo-partial derivative matrix is established, greatly reducing the risk of model mismatch.

[0018] In a possible implementation method of the first aspect, the output voltage prediction model is specifically:

[0019] v(k + 1)=v(k)+Ψ f (k)ΔH(k);

[0020] In the formula, v(k + 1) is the output voltage of the grid-forming inverter at the (k + 1)-th moment, v(k) is the output voltage of the grid-forming inverter at the k-th moment, ψ f (k) is the pseudo partial derivative matrix, and H(k) is the inverter data matrix.

[0021] In a possible implementation method of the first aspect, according to the preset optimization criterion function and the current output voltage prediction value, combined with the corresponding relationship between the control input signal and the output voltage of the grid-forming inverter, the reference voltage vector of the grid-forming inverter at the next moment is obtained, specifically:

[0022] Construct an output reference curve based on the historical output voltage data of the grid-forming inverter;

[0023] Based on the corresponding relationship between the control input signal and the output voltage of the grid-forming inverter, construct the optimization criterion function through the output reference curve;

[0024] With the goal of minimizing the optimization criterion function, calculate the reference voltage vector at the next moment by the least squares method based on the current output voltage prediction value.

[0025] In the above solution, because it is necessary to evaluate the influence of the control input signal of the inverter on the output voltage magnitude, an output reference curve is constructed based on the historical output voltage data of the grid-forming inverter to obtain the relationship between the control input signal and the output voltage, providing data support for subsequent obtaining the optimal output voltage by adjusting the switch state. Then, the optimization criterion function is constructed according to the output reference curve, and the reference voltage vector at the next moment is solved by the least squares method. The optimal output voltage exists in this vector, providing data support for subsequent data screening.

[0026] In a possible implementation method of the first aspect, the reference voltage vector at the next moment is specifically:

[0027]

[0028] In the formula, u * (k + 1) is the reference voltage vector of the grid-forming inverter at the next moment, Ξ(k + 1) is the response matrix of the least squares method, Θ(k + 1) is the response matrix of the least squares method, I is the identity matrix, Υ is the weight factor, A sub-module for the estimated value of the preset pseudo partial derivative matrix at time k+1, F(k+1) is an intermediate calculation variable related to the pseudo partial derivative sub-module, data matrix extraction data, and reference output voltage at time k+1, and L(k+1) is an intermediate calculation variable related to the pseudo partial derivative matrix sub-module, inverter data matrix, and reference output voltage at time k+1.

[0029] In a possible implementation method of the first aspect, the reference voltage vector is screened based on a preset midpoint balance principle to obtain the optimal control input signal at the next moment, specifically:

[0030] Compare the magnitudes of the capacitor voltages on the DC side of the grid-forming inverter, and screen the collected candidate voltage vectors based on the capacitor voltage comparison result to obtain the screened voltage vectors;

[0031] Calculate the distance between the reference voltage vector and the screened voltage vector, and use the screened voltage vector corresponding to the minimum distance as the optimal voltage vector;

[0032] Determine the optimal control input signal at the next moment according to the switching state of the grid-forming inverter corresponding to the optimal voltage vector.

[0033] The above scheme determines whether the output voltage at the next moment needs to increase or decrease the capacitor voltage to maintain voltage balance by comparing the magnitudes of the capacitor voltages on the DC side of the inverter, and screens out the appropriate screened voltage vectors accordingly. Since the above process has considered the midpoint balance, the process of calculating the optimal voltage vector can be directly converted into the process of calculating the distance between the reference voltage vector and the screened voltage vector, greatly reducing the computational complexity.

[0034] In a possible implementation method of the first aspect, the candidate voltage vectors are specifically:

[0035] Collect all the switching states of the grid-forming inverter, determine the power vectors corresponding to each switching state, and use the voltage vectors as the candidate voltage vectors.

[0036] In the above scheme, since the magnitude of the output voltage is related to the switching state, and each switching state has a corresponding control input signal, and each control input signal corresponds to a voltage vector, the optimal output voltage can be obtained by selecting an appropriate voltage vector.

[0037] In a possible implementation method of the first aspect, calculating the distance between the reference voltage vector and the screened voltage vector is specifically:

[0038] Calculate the distance between the reference voltage vector and the screened voltage vector, and the specific calculation formula is:

[0039]

[0040] where J is the distance between the reference voltage vector and the screened voltage vector, u jα (k + 1)u jβ (k + 1) is the screened voltage vector, u * α (k + 1)u * β (k + 1) is the reference voltage vector.

[0041] In a possible implementation method of the first aspect, the switching elements of the network-forming inverter are regulated according to the optimal control input signal at the next moment, specifically:

[0042] Set the optimal switching state of the switching elements according to the optimal control input signal at the next moment;

[0043] Based on the optimal switching state, regulate the output voltage of the network-forming inverter at the next moment to stabilize the voltage of the power system.

[0044] The second aspect of the present application provides a predictive control system for a network-forming inverter, and the system includes: a model construction module, an output voltage prediction module, a reference voltage vector calculation module, a data screening module, and an inverter switch control module;

[0045] Among them, the model construction module is used to construct an output voltage prediction model according to the output voltage data and historical control input signals of the network-forming inverter collected at the current moment;

[0046] The output voltage prediction module is used to input the output voltage prediction value at the previous moment into the output voltage prediction model and perform calculation delay compensation to obtain the current output voltage prediction value;

[0047] The reference voltage vector calculation module is used to obtain the reference voltage vector of the network-forming inverter at the next moment according to a preset optimization criterion function and the current output voltage prediction value, in combination with the corresponding relationship between the control input signal and the output voltage of the network-forming inverter;

[0048] The data screening module is used to screen the reference voltage vector based on a preset midpoint balance principle to obtain the optimal control input signal at the next moment;

[0049] The inverter switch control module is used to regulate the switching elements of the network-forming inverter according to the optimal control input signal at the next moment. Description of the Drawings

[0050] To more clearly illustrate the technical solutions of the present application, the accompanying drawings required for implementation will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0051] Figure 1 is a specific flowchart of a predictive control method for a grid-forming inverter provided by an embodiment of the present application;

[0052] Figure 2 is a space vector diagram of an inverter for a predictive control method of a grid-forming inverter provided by an embodiment of the present application;

[0053] Figure 3 is a schematic diagram of an output voltage prediction model based on a three-level grid-forming inverter for a predictive control method of a grid-forming inverter provided by an embodiment of the present application;

[0054] Figure 4 is a comparison diagram of output waveforms for a predictive control method of a grid-forming inverter provided by an embodiment of the present application;

[0055] Figure 5 is a comparison diagram of output voltage waveforms for a predictive control method of a grid-forming inverter provided by an embodiment of the present application;

[0056] Figure 6 is a specific structural diagram of a predictive control system for a grid-forming inverter provided by an embodiment of the present application. Specific Embodiments

[0057] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present application belong to the scope of protection of the present application.

[0058] It should be understood that the step numbers used in the text are only for convenience of description and are not intended to limit the order of execution of the steps.

[0059] First Embodiment

[0060] The grid-forming inverter adjusts the output voltage by setting the states of the switching elements, and then regulates the line voltage of the transmission line where it is located. Therefore, how to accurately predict which switching state the switching element should be in at the next moment to timely adjust the line voltage is the main solution to achieve fast control of the inverter. However, when the existing output voltage prediction model mismatches the actual inverter situation, the output voltage of the inverter will be severely distorted, which will affect the regulation effect of the line voltage. Moreover, the traditional output voltage prediction model highly relies on the data collected by the current sensor. When these collected data are incorrect, it is easy to cause the prediction result to be seriously distorted, resulting in a decrease in the reliability of the inverter. Therefore, how to reduce the risk of model mismatch and get rid of the dependence on the current sensor data is the main research direction of the embodiments of this application.

[0061] As Figure 1 shown, Figure 1 This application provides a specific flowchart of a predictive control method for a grid-forming inverter in one embodiment. To solve the problems in the prior art that the traditional output voltage prediction model is prone to model mismatch with the actual inverter, and the traditional grid-forming inverter model highly depends on the current sensor data, resulting in a significant reduction in the reliability of the prediction result when the current sensor data changes. The predictive control method of the grid-forming inverter in this embodiment includes steps S1 to S5, which are described in detail as follows:

[0062] Step S1: Construct an output voltage prediction model based on the output voltage data and historical control input signals of the grid-forming inverter collected at the current moment.

[0063] In the embodiments of this application, an output voltage prediction model of the inverter is established only through the input and output data of the grid-forming inverter without other parameters, which greatly reduces the risk of model mismatch.

[0064] Among them, the grid-forming inverter is an inverter that adopts the grid-forming grid connection technology, which allows distributed energy sources such as solar photovoltaic, wind energy, and energy storage systems to form a local power grid in an autonomous and coordinated manner through the inverter. This enables distributed energy resources to operate independently without a traditional central power grid and to be connected to the main power grid when needed. The grid-forming inverter is the basis of the grid-forming grid connection technology. It can connect the DC power sources of wind power, photovoltaic, and energy storage to the AC power grid and can perform bidirectional power flow control in the power grid. At the same time, it also has grid support functions such as voltage and frequency control, as well as low voltage ride-through capabilities to ensure stable operation during grid faults.

[0065] For constructing an output voltage prediction model of a grid-forming inverter, first collect the output voltage data of the grid-forming inverter at the current moment, and at the same time collect the historical control input signals of the grid-forming inverter, and construct an inverter data matrix based on these. Among them, the control input signal is used to set the switching state of the grid-forming inverter, and the output voltage of the inverter is adjusted by adjusting the switching elements.

[0066] The inverter data matrix has the following specific expression:

[0067]

[0068] In the formula, H(k) is the inverter data matrix, Δv(k) and Δu(k) are the output voltage change amount of the grid-forming inverter at the current moment and the control input signal change amount of the grid-forming inverter at the previous moment respectively; k is the current moment, L v is the output dynamic linearization length constant, L u is the input dynamic linearization length constant.

[0069] Then, based on the inverter data matrix and a preset pseudo-partial derivative matrix, construct an output voltage prediction model, and the specific expression is:

[0070] v(k + 1) = v(k) + Ψ f (k)ΔH(k);

[0071] In the formula, v(k + 1) is the output voltage of the grid-forming inverter at the k + 1 moment, v(k) is the output voltage of the grid-forming inverter at the k moment, ψ f (k) is the pseudo-partial derivative matrix, and H(k) is the inverter data matrix.

[0072] Furthermore, for the pseudo-partial derivative matrix, it is actually used to describe the relationship between the output state of the inverter and the control input signal. Since the true value of the pseudo-partial derivative matrix cannot be directly calculated, an approximate value needs to be estimated through an estimation algorithm, and the estimation result is approximately equal to the true value of the pseudo-partial derivative matrix. In addition, based on the pseudo-partial derivative matrix and the inverter data matrix, the output voltage prediction value of the grid-forming inverter can be calculated, and the current pseudo-partial derivative value can be estimated based on the inverter data matrix and the previous moment's pseudo-partial derivative.

[0073] Since the output voltage prediction value obtained through the pseudo-partial derivative matrix and the inverter data matrix is based on the dynamic linearization theory, the output voltage prediction model is a full-format dynamic linearization model.

[0074] Among them, the specific expression of the pseudo-partial derivative matrix is:

[0075]

[0076] Where, Ψ i (k), i = 1, …, L v +L u represents the pseudo partial derivative matrix sub-module.

[0077] For the estimation formula and reset algorithm of the pseudo partial derivative matrix, they are respectively:

[0078]

[0079] Where, is the estimated value of the preset pseudo partial derivative matrix at time k, η is the step size factor of the estimation formula, η ∈ (0, 2), μ is the weight factor, μ > 0.

[0080]

[0081] Where, are all the estimated values of the estimation matrix sub-module , ε1 and ε2 are both normal constants, α >= 1 and satisfy ε1 > ε2(2α + 1).

[0082] Step S2, input the predicted value of the output voltage at the previous moment into the output voltage prediction model and perform calculation delay compensation to obtain the predicted value of the current output voltage.

[0083] Considering that there is a certain calculation delay in calculating the predicted value of the current output voltage by the digital controller, in order to reduce the error caused by the calculation delay, in the embodiment of the present application, a two-step prediction method is adopted to bring the control input signal of the network-forming inverter at the previous moment into the output voltage prediction model to obtain the predicted value of the current output voltage of the network-forming inverter at the current moment. Among them, the two-step prediction method is to predict the values at time k + 1 and time k + 2 using the previous data, that is, to predict the values of the next two steps.

[0084] The predicted value of the current output voltage, the specific expression is:

[0085]

[0086] Where, v(k + 1)| u(k) is the predicted value of the output voltage at time k + 1, v(k) is the measured value of the output voltage at time k, is the estimated value of the preset pseudo partial derivative matrix at time k, H(k)| u(k) is the inverter data matrix at time k.

[0087] Step S3, according to the preset optimization criterion function and the predicted value of the current output voltage, combined with the corresponding relationship between the control input signal and the output voltage of the network-forming inverter, obtain the reference voltage vector of the network-forming inverter at the next moment.

[0088] Since the control input signal directly affects the switching state of the network-forming inverter, and the switching state change will cause the magnitude of the output voltage of the network-forming inverter to change. For example, the control input signal can modulate the duty cycle of the switching element of the network-forming inverter, and the duty cycle is related to the conduction duration of the switching element in one cycle. The duty cycle has a linear relationship with the magnitude of the output voltage of the network-forming inverter. When the duty cycle increases, the output voltage of the network-forming inverter increases. Therefore, there is a certain correspondence between the control input signal and the output voltage of the network-forming inverter.

[0089] To study the correspondence between the control input signal and the output voltage of the network-forming inverter, in the embodiments of the present application, an output reference curve is first constructed according to the historical output voltage data of the network-forming inverter, and this curve is generally a power frequency sine waveform curve.

[0090] Then, based on the output reference curve, considering the influence of different control input signals on the output voltage of the network-forming inverter, an optimization criterion function is established, and the specific expression is:

[0091]

[0092] In the formula, J1 is the optimization criterion function, J v is the penalty term of the output voltage of the network-forming inverter, J der is the differential penalty term of the output voltage of the network-forming inverter, v * (k + 2) is the output reference curve, Δv * (k + 2) is the change amount of the output reference curve at the previous moment, λ1 is the tracking weight factor of the output voltage differential term, T s is the control period of the network-forming inverter.

[0093] In the embodiments of the present application, not only the influence of the control input signal of the inverter on the output voltage is evaluated through the optimization criterion function, but also the influence of the control input signal on the output voltage and its differential term of the network-forming inverter is considered. Therefore, taking the three-level inverter as an example, one of the 19 candidate switching states is selected as the corresponding inverter control input signal in each control period. Therefore, through the optimization criterion function here, the influence of each switching state on the system output can be evaluated, and then the optimal candidate vector can be selected.

[0094] Then, based on the current output voltage prediction value and the estimation formula of the pseudo-partial derivative matrix, the estimated value of the pseudo-partial derivative matrix at the next moment is calculated At the same time, the optimization criterion function is expressed by the intermediate variables of the pseudo-partial derivative matrix and the inverter data matrix, and the following formula is obtained:

[0095]

[0096] In the formula, J1 is the optimization criterion function, is the weight factor, F(k + 1) is the intermediate calculation variable related to the pseudo - derivative sub - module, data matrix extraction data, and reference output voltage at the (k + 1) - th moment, E(k + 1) is the intermediate calculation variable related to the pseudo - derivative matrix sub - module and inverter data matrix extraction data at the (k + 1) - th moment, and L(k + 1) is the intermediate calculation variable related to the pseudo - derivative matrix sub - module, inverter data matrix, and reference output voltage at the (k + 1) - th moment.

[0097] Then, aiming at minimizing the optimization criterion function, the reference voltage vector at the next moment is calculated by the least - squares method based on the predicted value of the current output voltage. Among them, each control input signal corresponds to a voltage vector. By comparing the reference voltage vector with the voltage vector, the optimal switching state that can achieve the best line voltage stability effect can be selected.

[0098] Specifically, first define the error term according to the output reference curve, and the specific expression is:

[0099]

[0100] In the formula, ξ is the error term, Ξ(k + 1) is the response matrix of the least - squares method, and Θ(k + 1) is the parameter matrix of the least - squares method.

[0101] In this way, the least - squares method for minimizing the optimization criterion function is obtained:

[0102]

[0103] Then, by solving the extreme value of the above equation, the reference voltage vector at the next moment can be obtained, and its expression is:

[0104]

[0105] In the formula, u * (k + 1) is the reference voltage vector of the grid - forming inverter at the next moment, Ξ(k + 1) is the response matrix of the least - squares method, Θ(k + 1) is the parameter matrix of the least - squares method, I is a 2×2 identity matrix, Υ is the weight factor, is the sub - module of the estimated value of the preset pseudo - derivative matrix at the (k + 1) - th moment, F(k + 1) is the intermediate calculation variable related to the pseudo - derivative sub - module, data matrix extraction data, and reference output voltage at the (k + 1) - th moment, and L(k + 1) is the intermediate calculation variable related to the pseudo - derivative matrix sub - module, inverter data matrix, and reference output voltage at the (k + 1) - th moment.

[0106] Step S4: Screen the reference voltage vector based on a preset midpoint balance principle to obtain the optimal control input signal for the next moment.

[0107] In the embodiment of the present application, by comparing the magnitudes of the capacitor voltages on the DC side of the grid-forming inverter, the candidate voltage vectors collected are screened based on the capacitor voltage comparison result to obtain the screened voltage vectors. Among them, all the switching states of the grid-forming inverter are collected to determine the power vectors corresponding to each switching state, and the voltage vectors are used as the candidate voltage vectors.

[0108] Among them, the midpoint balance mainly means that the control input signal of the inverter will cause the offset and fluctuation of the midpoint potential. The offset of the midpoint potential will cause uneven voltage sharing of the two DC-side capacitors. One of the capacitors may be damaged due to excessive voltage, and at the same time, it will also cause distortion of the inverter output voltage. Therefore, when solving the optimal control input signal, the influence of the signal on the midpoint potential needs to be considered to reduce the risk of midpoint imbalance in the inverter.

[0109] Exemplarily, Figure 2 Show the SVPWM space vector diagram taking a three-level grid-forming inverter as an example. SVPWM is the abbreviation of Space Vector Pulse Width Modulation, which is a specific switching mode composed of six switching elements of a three-level grid-forming inverter, making the output voltage waveform as close as possible to an ideal sine waveform. The green dots in the figure are large vectors, the blue dots are medium vectors, the orange dots are small vectors, and the red dots are zero vectors, each representing the magnitude of the candidate voltage vectors; there are 12 small vectors that affect the midpoint balance in the figure, namely 6 P-type vectors and 6 N-type vectors. Each P-type small vector and N-type small vector at the same spatial position in the figure correspond to the same spatial vector but different switching states. Among them, the P-type small vector (switching state) causes the upper capacitor voltage V p to decrease, and the N-type small vector causes the upper capacitor voltage V p to increase. Compare the magnitudes of the upper and lower capacitor voltages V p 、V n on the current DC side of the three-level grid-forming inverter, and eliminate some candidate voltage vectors that cause midpoint imbalance. If V p < V n , it means that the upper capacitor voltage should be ensured not to decrease to achieve midpoint balance. Therefore, 6 P-type small vectors (V P1 、V P2 、V P3 、V P4 、V P5 、V P6 ) need to be eliminated, and only N-type small vectors (V N1 、V N2 、V N3 、V N4 、V N5 、VN6 );Conversely, if V p > V n , then the N-type small vectors are eliminated and the P-type small vectors are retained. At the same time, two zero vectors V 0P and V 0N are also eliminated from the candidate voltage vectors to avoid the grid-forming inverter from outputting a large common-mode voltage.

[0110] Since the neutral point balance has been considered during the process of screening the candidate voltage vectors, and the space vectors do not change during the elimination process, the optimality of the inverter output voltage will not be affected either. Therefore, the process of calculating the optimal voltage vector can be transformed into the process of calculating the distance between the reference voltage vector and the screened voltage vector. The specific expression is:

[0111]

[0112] In the formula, J is the distance between the reference voltage vector and the screened voltage vector, u jα (k + 1)u jβ (k + 1) is the screened voltage vector, u * α (k + 1)u * β (k + 1) is the reference voltage vector.

[0113] Then, the screened voltage vector corresponding to the minimum distance is used as the optimal voltage vector and applied to the switching tube control of the grid-forming inverter in the next control period.

[0114] Specifically, since the embodiment of the present application is finite set predictive control and only one switching state is applied in each control period, the switching state of the grid-forming inverter corresponding to the obtained optimal voltage vector can be used to determine the optimal control input signal at the next moment to adjust the switching state of the inverter, and further adjust the output voltage of the inverter.

[0115] Figure 3 The schematic diagram of the output voltage prediction model based on the three-level grid-forming inverter is provided. As shown in the figure, in the embodiment of the present application, candidate voltage vectors are obtained by sampling the DC-side capacitor C of the inverter, and pseudo-derivative estimation, delay compensation, reference voltage calculation, etc. are performed during the prediction process of the output voltage.

[0116] Step S5, regulate the switching elements of the grid-forming inverter according to the optimal control input signal at the next moment.

[0117] Set the optimal switching state of the switching elements through the calculated optimal control input signal at the next moment. Based on the optimal switching state, regulate the output voltage of the grid-forming inverter at the next moment to stabilize the voltage of the power system.

[0118] To better demonstrate the technical effects of the embodiments of the present application, Figure 4 show the comparison of the output voltage waveforms between traditional model predictive control under accurate model parameters and the three-level grid-forming inverter of the embodiments of the present application, that is, the comparison of the output voltage waveforms when no model mismatch occurs. Figure 4 In the left figure (a) of, it is the output voltage fluctuation of traditional model predictive control, and in the right figure (b) is the comparison of the output voltage waveforms of the embodiments of the present application. It can be seen that when no model mismatch occurs, the control output performances of the traditional output voltage prediction model and the output voltage prediction model provided by the embodiments of the present application are similar, indicating that the output voltage prediction model provided by the embodiments of the present application can effectively predict the system output state and achieve accurate tracking of the output voltage.

[0119] Figure 5 Then it shows the changes in the output voltage waveforms of the traditional output voltage prediction model and the output voltage prediction model provided by the embodiments of the present application in the case of model mismatch. Figure (a) is the output voltage waveforms of the traditional output voltage prediction model before and after model mismatch. The left side of Figure (a) is the output voltage waveform before model mismatch, and the right side of Figure (a) is the output voltage waveform after model mismatch. It can be seen that the output voltage waveform is greatly affected by model mismatch because it depends on multiple system parameters, resulting in the output voltage waveform being very sensitive to system parameter changes. Therefore, when model mismatch occurs, the performance of the traditional output voltage prediction model drops significantly, and the reliability of its prediction results is greatly reduced; Figure (a) is the output voltage waveforms of the output voltage prediction model of the embodiments of the present application before and after model mismatch. The left side of Figure (a) is the output voltage waveform before model mismatch, and the right side of Figure (a) is the output voltage waveform after model mismatch. It can be seen that the output voltage waveform is not greatly affected. This is because the output voltage prediction model of the embodiments of the present application does not depend on any system parameters. Therefore, the output voltage waveform is not sensitive to system parameter changes, avoiding the performance degradation caused by model mismatch.

[0120] Implementing the embodiments of the present application has the following beneficial effects:

[0121] In the embodiment of the present application, an output voltage prediction model can be constructed only through the output voltage of the network-forming inverter at the current moment and the historical input and output data, without relying on any system parameters, completely eliminating the problem of model mismatch, and based on the data at the current moment, it can more ensure that the prediction result of the model conforms to the actual situation. Moreover, by inputting the predicted value of the output voltage at the previous moment into the output voltage prediction model, an accurate predicted value of the current output voltage can be obtained, fundamentally eliminating the dependence of the prediction model on the filtering current and output current sensors, and preventing the prediction result from being distorted due to incorrect input data, thereby improving the operation reliability of the inverter. In addition, considering the influence of the control input signal of the network-forming inverter on the output voltage, the switching state of the inverter that can enable the stable operation of the power system is obtained in reverse with the goal of stabilizing the DC-side voltage of the inverter, and the inverter is regulated accordingly to effectively maintain the stable operation of the power system.

[0122] Second Embodiment

[0123] Further, in order to implement the prediction control system of the network-forming inverter corresponding to the above method embodiment to achieve the corresponding functions and technical effects, Figure 6 A structural diagram of a prediction control system for a network-forming inverter is provided. For the sake of convenience of description, only the parts related to this embodiment are shown. The prediction control system for a network-forming inverter provided by the embodiment of the present application includes:

[0124] A model construction module 201, configured to construct an output voltage prediction model according to the output voltage data of the network-forming inverter collected at the current moment and the historical control input signals.

[0125] In the embodiment of the present application, the output voltage data of the network-forming inverter at the current moment and the historical control input signals of the switching elements of the network-forming inverter are collected;

[0126] An inverter data matrix is constructed according to the output voltage data at the current moment and the historical control input signals;

[0127] An output voltage prediction model is constructed based on the inverter data matrix and a preset pseudo partial derivative matrix;

[0128] Wherein, the pseudo partial derivative matrix is used to describe the relationship between the output voltage of the network-forming inverter and the historical input and output data of the network-forming inverter.

[0129] An output voltage prediction module 202, configured to input the predicted value of the output voltage at the previous moment into the output voltage prediction model and perform calculation delay compensation to obtain the predicted value of the current output voltage.

[0130] Considering that there is a certain calculation delay in the digital controller when calculating the predicted value of the current output voltage, in order to reduce the error caused by the calculation delay, in the embodiment of the present application, a two-step prediction method is adopted to bring the control input signal of the grid-forming inverter at the previous moment into the output voltage prediction model to obtain the predicted value of the current output voltage of the grid-forming inverter at the current moment. Among them, the two-step prediction method is to predict the values at the k+1 and k+2 moments by using the previous data, that is, to predict the values of the next two steps.

[0131] The specific expression of the predicted value of the current output voltage is:

[0132]

[0133] In the formula, v(k+1)| u(k) is the predicted value of the output voltage at the k+1 moment, v(k) is the measured value of the output voltage at the k moment, is the estimated value of the pseudo partial derivative matrix preset at the k moment, H(k)| u(k) is the inverter data matrix at the k moment.

[0134] The reference voltage vector calculation module 203 is used to obtain the reference voltage vector of the grid-forming inverter at the next moment according to the preset optimization criterion function and the predicted value of the current output voltage, considering the influence of the control input signal of the grid-forming inverter on the output voltage.

[0135] In the embodiment of the present application, an output reference curve is constructed according to the historical output voltage data of the grid-forming inverter;

[0136] Based on the influence of different control input signals on the output voltage of the grid-forming inverter, the optimization criterion function is constructed through the output reference curve;

[0137] With the goal of minimizing the optimization criterion function, the reference voltage vector at the next moment is calculated by the least square method based on the predicted value of the current output voltage.

[0138] The data screening module 204 is used to screen the reference voltage vector based on the preset midpoint balance principle to obtain the optimal control input signal at the next moment.

[0139] In the embodiment of the present application, the magnitudes of the capacitor voltages on the DC side of the grid-forming inverter are compared, and the candidate voltage vectors collected are screened based on the comparison result of the capacitor voltages with the goal of ensuring that the grid-forming inverter can achieve midpoint balance to obtain the screened voltage vectors;

[0140] Calculate the distance between the reference voltage vector and the screened voltage vector, and use the screened voltage vector corresponding to the minimum distance as the optimal voltage vector;

[0141] Determine the optimal control input signal at the next moment according to the switching state of the network-forming inverter corresponding to the optimal voltage vector.

[0142] The inverter switch control module 205 is configured to regulate the switching elements of the network-forming inverter according to the optimal control input signal at the next moment.

[0143] In the embodiment of the present application, the optimal switching state of the switching element is set by the calculated optimal control input signal at the next moment, and based on the optimal switching state, the output voltage of the network-forming inverter at the next moment is regulated to stabilize the voltage of the power system.

[0144] In some embodiments, the model construction module 201 is specifically:

[0145] An output voltage prediction model of the inverter is established only through the input and output data of the network-forming inverter without other parameters, which greatly reduces the risk of model mismatch.

[0146] Among them, the network-forming inverter is an inverter adopting network-forming grid connection technology, which can allow distributed energy sources such as solar photovoltaic, wind energy, energy storage systems, etc. to form a local power grid through the inverter in an autonomous and coordinated manner, so that in the absence of a traditional central power grid, distributed energy resources can operate independently and at the same time be connected to the main power grid for grid connection operation when needed. The network-forming inverter is the basis of the network-forming grid connection technology. It can connect the DC power sources of wind power, photovoltaic and energy storage to the AC power grid, and can perform bidirectional power flow control in the power grid. At the same time, it also has grid support functions such as voltage and frequency control, and low voltage ride-through ability to ensure stable operation during power grid faults.

[0147] For constructing the output voltage prediction model of the network-forming inverter, first collect the output voltage data of the network-forming inverter at the current moment, and at the same time collect the historical control input signals of the network-forming inverter, and construct an inverter data matrix therefrom. Among them, the control input signal is used to set the switching state of the network-forming inverter, and the output voltage of the inverter is adjusted by adjusting the switching element.

[0148] The specific expression of the inverter data matrix is:

[0149]

[0150] In the formula, H(k) is the inverter data matrix, Δv(k) and Δu(k) are respectively the output voltage change amount of the network-forming inverter at the current moment and the control input signal change amount of the network-forming inverter at the previous moment; k is the current moment, L v is the output dynamic linearization length constant, L uIt is the input dynamic linearization length constant.

[0151] Then, based on the inverter data matrix and the preset pseudo partial derivative matrix, an output voltage prediction model is constructed, and the specific expression is:

[0152] v(k + 1) = v(k) + Ψ f (k)ΔH(k);

[0153] In the formula, v(k + 1) is the output voltage of the grid-forming inverter at time k + 1, v(k) is the output voltage of the grid-forming inverter at time k, ψ f (k) is the pseudo partial derivative matrix, and H(k) is the inverter data matrix.

[0154] Furthermore, for the pseudo partial derivative matrix, it is actually used to describe the relationship between the output state of the inverter and the control input signal. Since the true value of the pseudo partial derivative matrix cannot be directly calculated, an approximate value needs to be estimated through an estimation algorithm, and the estimation result is approximately equal to the true value of the pseudo partial derivative matrix. In addition, based on the pseudo partial derivative matrix and the inverter data matrix, the predicted value of the output voltage of the grid-forming inverter can be calculated, and the current pseudo partial derivative value can be estimated based on the inverter data matrix and the pseudo partial derivative at the previous moment.

[0155] Since the predicted value of the output voltage obtained through the pseudo partial derivative matrix and the inverter data matrix is based on the dynamic linearization theory, the output voltage prediction model is a full-format dynamic linearization model.

[0156] Among them, the specific expression of the pseudo partial derivative matrix is:

[0157]

[0158] For the estimation formula and reset algorithm of the pseudo partial derivative matrix, they are respectively:

[0159]

[0160] In the formula, is the estimated value of the preset pseudo partial derivative matrix at time k, η is the step factor of the estimation formula, η ∈ (0, 2), μ is the weight factor, μ > 0;

[0161]

[0162] In the formula, are all the estimated values of the estimation matrix sub-module , ε1 and ε2 are both normal constants, α >= 1 and satisfy ε1 > ε2(2α + 1).

[0163] In some embodiments, the reference voltage vector calculation module 203 is specifically:

[0164] Since the control input signal directly affects the switching state of the network-forming inverter, and the switching state transition causes the magnitude of the output voltage of the network-forming inverter to change. For example, the control input signal can modulate the duty cycle of the switching element of the network-forming inverter, and the duty cycle is related to the conduction duration of the switching element in one cycle. The duty cycle has a linear relationship with the magnitude of the output voltage of the network-forming inverter. When the duty cycle increases, the output voltage of the network-forming inverter increases. Therefore, there is a certain correspondence between the control input signal and the output voltage of the network-forming inverter.

[0165] To study the correspondence between the control input signal and the output voltage of the network-forming inverter, in the embodiment of the present application, an output reference curve is first constructed according to the historical output voltage data of the network-forming inverter, and this curve is generally a power frequency sine waveform curve.

[0166] Then, based on the output reference curve, considering the influence of different control input signals on the output voltage of the network-forming inverter, an optimization criterion function is established, and the specific expression is:

[0167]

[0168] In the formula, J1 is the optimization criterion function, J v is the penalty term of the output voltage of the network-forming inverter, J der is the differential penalty term of the output voltage of the network-forming inverter, v * (k + 2) is the output reference curve, Δv * (k + 2) is the change amount of the output reference curve at the previous moment, λ1 is the tracking weight factor of the output voltage differential term, T s is the control period of the network-forming inverter.

[0169] In the embodiment of the present application, not only the influence of the control input signal of the inverter on the output voltage is evaluated through the optimization criterion function, but also the influence of the control input signal on the output voltage and its differential term of the network-forming inverter is considered. Therefore, taking the three-level inverter as an example, one is selected from 19 candidate switching states as the corresponding inverter control input signal in each control period. Therefore, through the optimization criterion function here, the influence of each switching state on the system output can be evaluated, and then the optimal candidate vector can be selected.

[0170] Then, based on the current output voltage prediction value and the estimation formula of the pseudo-partial derivative matrix, the estimated value of the pseudo-partial derivative matrix at the next moment is calculated At the same time, the optimization criterion function is expressed by the intermediate variables of the pseudo-partial derivative matrix and the inverter data matrix, and the following formula is obtained:

[0171]

[0172] In the formula, J1 is the optimization criterion function, is the weight factor, F(k + 1) is an intermediate calculation variable related to the pseudo - derivative sub - module, data matrix extraction data, and reference output voltage at the (k + 1) - th moment, E(k + 1) is an intermediate calculation variable related to the pseudo - derivative matrix sub - module and inverter data matrix extraction data at the (k + 1) - th moment, and L(k + 1) is an intermediate calculation variable related to the pseudo - derivative matrix sub - module, inverter data matrix, and reference output voltage at the (k + 1) - th moment.

[0173] Then, aiming at minimizing the optimization criterion function, the reference voltage vector at the next moment is calculated by the least - squares method based on the predicted value of the current output voltage. Among them, each control input signal corresponds to a voltage vector. By comparing the reference voltage vector with the voltage vector, the optimal switching state that can achieve the best line voltage stability effect can be selected.

[0174] Specifically, first define the error term according to the output reference curve, and the specific expression is:

[0175]

[0176] In the formula, ξ is the error term, Ξ(k + 1) is the response matrix of the least - squares method, and Θ(k + 1) is the parameter matrix of the least - squares method.

[0177] In this way, the least - squares method for minimizing the optimization criterion function is obtained:

[0178]

[0179] Then, by solving the extreme value of the above - mentioned equation, the reference voltage vector at the next moment can be obtained, and its expression is:

[0180]

[0181] In the formula, u * (k + 1) is the reference voltage vector of the grid - forming inverter at the next moment, Ξ(k + 1) is, Θ(k + 1) is, I is a 2×2 identity matrix, Υ is the weight factor, is the sub - module of the estimated value of the preset pseudo - derivative matrix at the (k + 1) - th moment, F(k + 1) is an intermediate calculation variable related to the pseudo - derivative sub - module, data matrix extraction data, and reference output voltage at the (k + 1) - th moment, and L(k + 1) is an intermediate calculation variable related to the pseudo - derivative matrix sub - module, inverter data matrix, and reference output voltage at the (k + 1) - th moment.

[0182] In some embodiments, the data screening module 204 is specifically:

[0183] By comparing the magnitudes of the capacitor voltages on the DC side of the grid-forming inverter, with the goal of ensuring that the grid-forming inverter can achieve midpoint balance, candidate voltage vectors collected are screened based on the comparison results of the capacitor voltages to obtain the screened voltage vectors. Among them, all the switching states of the grid-forming inverter are collected, and the power vectors corresponding to each of the switching states are determined, and the voltage vectors are used as the candidate voltage vectors.

[0184] Among them, the midpoint balance mainly means that the control input signal of the inverter will cause the offset and fluctuation of the midpoint potential. The offset of the midpoint potential will cause uneven voltage sharing of the two DC-side capacitors. One of the capacitors may be damaged due to excessive voltage, and at the same time, it will also cause distortion of the inverter output voltage. Therefore, when solving the optimal control input signal, the influence of the signal on the midpoint potential needs to be considered to reduce the risk of midpoint imbalance in the inverter.

[0185] Because midpoint balance has been considered during the process of screening the candidate voltage vectors, and the space vectors do not change during the elimination process, it will not affect the optimality of the inverter output voltage. Therefore, the process of calculating the optimal voltage vector can be transformed into the process of calculating the distance between the reference voltage vector and the screened voltage vector. The specific expression is:

[0186]

[0187] In the formula, J is the distance between the reference voltage vector and the screened voltage vector, u jα (k + 1)u jβ (k + 1) is the screened voltage vector, u * α (k + 1)u * β (k + 1) is the reference voltage vector.

[0188] Then, the screened voltage vector corresponding to the minimum distance is used as the optimal voltage vector and applied to the switching tube control of the grid-forming inverter in the next control cycle.

[0189] Specifically, because the embodiment of the present application is finite set predictive control, and only one switching state is applied in each control cycle, the switching state of the grid-forming inverter corresponding to the obtained optimal voltage vector can be used to determine the optimal control input signal at the next moment to adjust the switching state of the inverter, and further adjust the output voltage of the inverter.

[0190] Implementing the embodiment of the present application has the following beneficial effects:

[0191] In the embodiment of the present application, an output voltage prediction model can be constructed only by using the output voltage of the network-forming inverter at the current moment and the historical input and output data, without relying on any system parameters, completely eliminating the problem of model mismatch, and the prediction result of the model can better conform to the actual situation based on the data at the current moment. Moreover, by inputting the predicted value of the output voltage at the previous moment into the output voltage prediction model, an accurate predicted value of the current output voltage can be obtained, fundamentally eliminating the dependence of the prediction model on the filter current and output current sensors, and preventing the prediction result from being distorted due to incorrect input data, thus improving the operation reliability of the inverter. In addition, considering the influence of the control input signal of the network-forming inverter on the output voltage, the inverter switching state that can enable the stable operation of the power system is obtained in reverse with the goal of stabilizing the DC-side voltage of the inverter, and the inverter is regulated accordingly to effectively maintain the stable operation of the power system.

[0192] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present application. It should be understood that the above are only specific embodiments of the present application and are not used to limit the protection scope of the present application. In particular, for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A predictive control method for a grid-connected inverter, characterized in that: include: An output voltage prediction model is constructed based on the collected output voltage data of the grid-connected inverter at the current moment and the historical control input signal; Inputting the output voltage prediction value at the previous moment into the output voltage prediction model and performing calculation delay compensation to obtain the current output voltage prediction value; According to the preset optimization criterion function and the current output voltage prediction value, combined with the corresponding relationship between the control input signal and the output voltage of the grid-type inverter, a reference voltage vector of the grid-type inverter at the next moment is obtained; The reference voltage vector is screened based on a preset midpoint balance principle to obtain an optimal control input signal at the next moment; The switching elements of the grid-connected inverter are regulated according to the optimal control input signal at the next moment.

2. The predictive control method for a grid-connected inverter according to claim 1, characterized in that: The output voltage full-format prediction model is constructed based on the collected output voltage data of the grid-connected inverter at the current moment and the historical control input signal, specifically: Collecting output voltage data of the grid-forming inverter at the current moment and historical control input signals of the switching elements of the grid-forming inverter; Constructing an inverter data matrix according to the output voltage data at the current moment and the historical control input signal; Based on the inverter data matrix and the preset pseudo partial derivative matrix, an output voltage prediction model is constructed; The pseudo partial derivative matrix is ​​used to describe the relationship between the output voltage of the grid-type inverter and the historical input and output data of the grid-type inverter.

3. The predictive control method for a grid-connected inverter according to claim 2, characterized in that: The output voltage prediction model is specifically: v(k+1)=v(k)+Ψ f (k)ΔH(k); Where v(k+1) is the output voltage of the grid-connected inverter at time k+1, v(k) is the output voltage of the grid-connected inverter at time k, ψ f (k) is the pseudo partial derivative matrix, and H(k) is the inverter data matrix.

4. The predictive control method for a grid-connected inverter according to claim 1, characterized in that: According to the preset optimization criterion function and the current output voltage prediction value, combined with the corresponding relationship between the control input signal and the output voltage of the grid-type inverter, the reference voltage vector of the grid-type inverter at the next moment is obtained, which is specifically: According to the historical output voltage data of the grid-connected inverter, an output reference curve is constructed; Based on the corresponding relationship between the control input signal and the output voltage of the grid-connected inverter, the optimization criterion function is constructed through the output reference curve; With the goal of minimizing the optimization criterion function, the reference voltage vector at the next moment is calculated by the least square method based on the current output voltage prediction value.

5. The predictive control method for a grid-connected inverter according to claim 4, characterized in that: The reference voltage vector at the next moment is specifically: In the formula, u * (k+1) is the reference voltage vector of the grid-type inverter at the next moment, Ξ(k+1) is the response matrix of the least squares method, Θ(k+1) is the parameter matrix of the least squares method, I is the unit matrix, Υ is the weight factor, It is a submodule for estimating the pseudo partial derivative matrix preset at time k+1, F(k+1) is an intermediate calculation variable related to the pseudo partial derivative submodule, data matrix extracted data and reference output voltage at time k+1, and L(k+1) is an intermediate calculation variable related to the pseudo partial derivative matrix submodule, inverter data matrix and reference output voltage at time k+1.

6. The predictive control method for a grid-connected inverter according to claim 1, characterized in that: The reference voltage vector is screened based on the preset midpoint balance principle to obtain the optimal control input signal at the next moment, specifically: Comparing the capacitor voltages on the DC side of the grid-type inverter, and screening the collected candidate voltage vectors based on the capacitor voltage comparison results to obtain a screened voltage vector; Calculating the distance between the reference voltage vector and the screening voltage vector, and taking the screening voltage vector corresponding to the minimum distance as the optimal voltage vector; The optimal control input signal at the next moment is determined according to the switching state of the grid-connected inverter corresponding to the optimal voltage vector.

7. The predictive control method for a grid-connected inverter according to claim 6, characterized in that: The candidate voltage vector is specifically: All switch states of the grid-type inverter are collected to determine the quantity vector corresponding to each switch state, and the voltage vector is used as the candidate voltage vector.

8. The predictive control method for a grid-connected inverter according to claim 6, characterized in that: The calculating of the distance between the reference voltage vector and the screening voltage vector is specifically as follows: The distance between the reference voltage vector and the screening voltage vector is calculated, and the specific calculation formula is: Where, J is the distance between the reference voltage vector and the screening voltage vector, u jα (k+1)u jβ (k+1) is the screening voltage vector, u * α (k+1)u * β (k+1) is the reference voltage vector.

9. The predictive control method for a grid-connected inverter according to any one of claims 1 to 8, characterized in that: The step of regulating the switch element of the grid-type inverter according to the optimal control input signal at the next moment is specifically as follows: Setting the optimal switching state of the switch element according to the optimal control input signal at the next moment; Based on the optimal switching state, the output voltage of the grid-connected inverter at the next moment is regulated to stabilize the voltage of the power system.

10. A predictive control system for a grid-connected inverter, characterized in that: include: Model building module, output voltage prediction module, reference voltage vector calculation module, data screening module and inverter switch control module; The model building module is used to build an output voltage prediction model based on the collected output voltage data of the grid-connected inverter at the current moment and the historical control input signal; The output voltage prediction module is used to input the output voltage prediction value at the previous moment into the output voltage prediction model and perform calculation delay compensation to obtain the current output voltage prediction value; The reference voltage vector calculation module is used to obtain the reference voltage vector of the grid-type inverter at the next moment according to the preset optimization criterion function and the current output voltage prediction value, combined with the corresponding relationship between the control input signal and the output voltage of the grid-type inverter; The data screening module is used to screen the reference voltage vector based on a preset midpoint balance principle to obtain an optimal control input signal at the next moment; The inverter switch control module is used to regulate the switch elements of the grid-connected inverter according to the optimal control input signal at the next moment.

Images