A novel finite control set model predictive control method for grid-connected inverters

Abstract

The present invention discloses a novel finite control set model predictive control method applied to a grid-connected inverter, comprising: selecting a candidate inverter output voltage vector at the current moment according to the system operation state; screening out an optimal output voltage vector according to the selected candidate inverter output voltage vector, and activating the optimal output voltage vector in the next cycle; storing the switch sequence corresponding to the obtained optimal output voltage vector in a memory, and updating and iterating the stored data with the power frequency cycle as a cycle. By applying the embodiments of the present invention, the number of candidate vectors can be fundamentally reduced without increasing the sampling frequency and without damaging the power quality, the calculation complexity of the algorithm can be reduced, the calculation time of the grid-connected inverter control algorithm can be greatly reduced, and the work efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical automation and new energy grid-connected power electronic conversion equipment, and in particular to a novel finite control set model predictive control method applied to a grid-connected inverter. Background Art

[0002] In recent years, with the rapid development of renewable energy, distributed power generation systems have received extensive attention. As an important component connecting DC power supply and power grid, the control performance of grid-connected inverter directly determines the power generation quality of the system. In order to obtain high-quality electric energy, in addition to the traditional linear control strategy, many nonlinear control strategies such as sliding mode control, passive control, model predictive control, etc. are widely used in grid-connected inverter control. Among them, finite control set model predictive control has attracted widespread attention due to its advantages such as no need for modulator, fast dynamic response, simple operation, and ability to achieve multi-objective constraints.

[0003] However, traditional grid-connected inverters need to traverse all possible switch combinations, which places a heavy computational burden on the system. For example, for a three-phase two-level grid-connected inverter, the traditional finite control set model predictive control needs to traverse eight possible switch combinations in each sampling period, and as the number of grid-connected inverter levels increases, the computational burden of the traditional finite control set model predictive control will increase exponentially, greatly increasing the computational complexity of the control system. Summary of the invention

[0004] The purpose of the present invention is to provide a new finite control set model predictive control method applied to grid-connected inverters, aiming to solve the problem of heavy computational burden of traditional finite control set model predictive control methods, reduce system computational complexity, and optimize the control performance of grid-connected inverters.

[0005] In order to achieve the above object, the present invention provides a novel finite control set model predictive control method applied to a grid-connected inverter, comprising:

[0006] According to the system operation status, select the inverter output voltage candidate vector at the current moment;

[0007] Screening out an optimal output voltage vector according to the selected inverter output voltage candidate vector, and activating the optimal output voltage vector in the next cycle;

[0008] The switch sequence corresponding to the obtained optimal output voltage vector is stored in a memory, and the stored data is updated and iterated with the power frequency cycle as a period.

[0009] In one implementation, the step of selecting a candidate inverter output voltage vector at the current moment according to the system operating state is:

[0010] Determine whether the system operation time reaches the set value, if not, select all possible inverter output voltage vectors as candidate vectors;

[0011] If so, based on the historical data of the optimal vector sequence, the inverter output voltage vector adjacent to the vector in spatial position and the vector itself are selected as candidate vectors.

[0012] In one implementation, the step of screening out the optimal output voltage vector according to the selected inverter output voltage candidate vector is:

[0013] According to the state variables and grid voltage sampled at the current time k, and using the delay compensation algorithm, a prediction model for time k+2 is constructed, and the reference values ​​of the three state variables of the inverter side current, grid current and capacitor voltage of the grid-connected inverter at time k+2 are calculated;

[0014] Substitute the selected candidate voltage vectors into the prediction model respectively to obtain the predicted values ​​of the inverter side current, capacitor voltage and grid current at time k+2 corresponding to each candidate voltage vector;

[0015] Construct a cost function, substitute the predicted values ​​and reference values ​​of the inverter side current, capacitor voltage and grid current at time k+2, and use the inverter output voltage vector that minimizes the cost function value as the optimal output voltage vector of the grid-connected inverter, and activate the selected inverter optimal output voltage vector within time k+1.

[0016] In one implementation, the prediction model is:

[0017]

[0018]

[0019]

[0020] Among them, v i (k) is the output voltage vector of the grid-connected inverter at time k, L2 is the grid-side inductance in the inverter, C is the filter capacitor in the inverter, v iαβ Output voltage of the grid-connected inverter 、 u cαβ is the filter capacitor voltage, i 1αβ is the inverter side current, i gαβ is the grid side current, v gαβ is the grid voltage, L1 is the inverter side inductance in the inverter, T s is the system sampling time.

[0021] In one implementation, the cost function is:

[0022]

[0023] Among them, λ i1 , uc , ig Respectively represent the inverter side current i1 and capacitor voltage u c and grid current i g Weight factor for the error term, ε i1 , ε ig , ε uc They represent the inverter side current i1 and the grid current i at time k+2 respectively. g , capacitor voltage u c The error between the reference value and the predicted value. The subscripts “α” and “β” represent the real and imaginary components of the error term; the superscript “2” represents the square value of the error.

[0024] A novel finite control set model predictive control method for grid-connected inverters provided by an embodiment of the present invention is applied. The present invention takes the historical data of the optimal voltage vector into consideration in the selection of the optimal vector at the future moment. The optimal vector sequence data of the previous base frequency cycle is stored through a memory. Moreover, based on the stored historical data of the optimal vector, the optimal voltage vector of the previous base frequency cycle and its adjacent vectors are selected as a candidate vector set in the current base frequency cycle. Based on this, without increasing the sampling frequency and without damaging the power quality, the number of candidate vectors can be fundamentally reduced, the computational complexity of the algorithm can be reduced, the computational time of the grid-connected inverter control algorithm can be greatly reduced, and the work efficiency can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The present invention is a flow chart of a novel finite control set model predictive control method applied to a grid-connected inverter.

[0026] Figure 2 It is a circuit structure diagram of a grid-connected inverter provided by an embodiment of the present invention.

[0027] Figure 3 This is the spatial distribution of the voltage vector of the three-phase two-level grid-connected inverter provided by the embodiment of the present invention.

[0028] Figure 4 It is a steady-state experimental waveform of the proposed algorithm under strong power grid conditions provided by an embodiment of the present invention.

[0029] Figure 5 It is an experimental waveform of the algorithm proposed in the embodiment of the present invention when the grid voltage changes in steps.

[0030] Figure 6 This is the experimental result of the algorithm proposed in the embodiment of the present invention when the grid current changes in steps.

[0031] Figure 7 It is an experimental waveform of grid voltage and grid injection current under the condition of unbalanced grid voltage containing background harmonics provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention.

[0033] like Figure 1 The present invention provides a novel finite control set model predictive control method applied to a grid-connected inverter, comprising:

[0034] S110, selecting a candidate inverter output voltage vector at the current moment according to the system operation state.

[0035] It should be noted that the grid-connected inverter used in the embodiment of the present invention is an LCL type three-phase two-level grid-connected inverter. Figure 2 As shown, in the LCL type three-phase two-level grid-connected inverter used, the DC power supply output voltage is U dc ; The state of the semiconductor switch device is represented by S x , where the switching device S x and Complementary; the LCL filter comprises three inverter-side inductors L1 connected to the grid-connected inverter, a grid-side inductor L2 connected to the grid, and a filter capacitor C connected in parallel between the first inductor and the second inductor. Figure 2 In, v ix Represents the output voltage of each phase bridge arm of the three-phase two-level circuit, i 1x Represents the three-phase current flowing through the inductor on the inverter side, i gx Indicates the grid current, i.e. the three-phase current flowing through the grid-side inductance, u cx is the voltage across the filter capacitor, v gx is the three-phase grid voltage, where x = a, b, c.

[0036] Each switch device of the three-phase two-level grid-connected inverter has two switching states: x Closed, switch Disconnect, then switch S x The state is represented as "1", corresponding to the switch Therefore, there are 8 switching combinations for the grid-connected inverter, and the corresponding spatial positions of the inverter output voltage vector are as follows: Figure 3 shown.

[0037] Firstly, the present invention samples three state variables of the inverter side current, the grid current and the capacitor voltage and the grid voltage at the current moment.

[0038] The specific steps are: collect the three-phase inverter side current i in the grid-connected inverter 1a 、i 1b 、i 1c , capacitor voltage u ca 、u cb 、u cc , grid injection current i ga 、i gb 、i gc , and the grid voltage v ga 、v gb 、v gc , and then use Clark transformation to transform the coordinates of the collected electrical physical quantities to obtain the inverter side current i at time k in the stationary coordinate system 1α 、i 1β , capacitor voltage u cα 、u cβ , grid injection current i gα 、i gβ , and the grid voltage v gα 、v gβ . Where "α" is the real part of the variable, and "β" is the imaginary part of the variable.

[0039] Among them, the Clark transformation matrix is:

[0040]

[0041] The present invention sets the candidate voltage vector selection algorithm switching time to t s , judge whether the system running time reaches the set value; when the system running time reaches the set value t s , according to the optimal vector sequence information x stored at the previous moment opt , select the voltage vector adjacent to the optimal vector space position and itself as candidate vectors, as shown in Table 1.

[0042] When the system running time does not reach the set value t s ,Will Figure 3 All possible voltage vectors in are taken as candidate vectors.

[0043] Table 1 Candidate voltage vector selection

[0044]

[0045] S120, screening out an optimal output voltage vector according to the selected inverter output voltage candidate vector, and activating the optimal output voltage vector in the next cycle.

[0046] It is understandable that the delay compensation algorithm is used to calculate the reference values ​​of the three state variables of the inverter side current, the grid current and the capacitor voltage at the time k+2 of the grid-connected inverter according to the state variables and the grid voltage obtained by the above sampling.

[0047] Specifically, according to the grid voltage in the stationary coordinate system at time k, the grid voltage in the stationary coordinate system at time k+2 is calculated. The grid voltage v in the stationary coordinate system at time k+2 is gαβ The calculation formula for (k+2) is:

[0048]

[0049] Among them, V g represents the grid voltage; k and k+2 represent the kth moment and the k+2th moment respectively; Ts represents the system sampling period; ω represents the grid angular frequency; j represents the imaginary unit; the subscripts “α” and “β” represent the real and imaginary parts of the corresponding complex vectors.

[0050] Based on the given grid injection current reference value at time k, the calculation formula for the grid current reference value in the stationary coordinate system at time k+2 is derived as follows:

[0051]

[0052] Among them, i g represents the grid current, the superscript “*” represents the reference value corresponding to the variable; the subscripts “α” and “β” represent the real and imaginary parts of the corresponding complex vector; k, k-1, k-2, and k+2 represent the sampling moments, respectively.

[0053] It should be noted that the reference value of the grid current in the stationary coordinate system at time k+2 is For is the real part, is the complex vector of the imaginary part.

[0054] Further, according to the reference value of the grid current in the stationary coordinate system at time k+2, the calculation formula for calculating the reference value of the inverter side current and the reference value of the capacitor voltage in the stationary coordinate system at time k+2 is:

[0055]

[0056]

[0057] Among them, L2 is the grid-side inductance in the inverter, C is the filter capacitor in the inverter, and i * 1α 、i * 1β are the components of the inverter side current reference value, are the components of the capacitor voltage reference value of the inverter; v gα is the real component of the grid voltage vector, v gβ is the imaginary component of the grid voltage vector, ω is the grid angular frequency, is the real component of the grid current reference, is the imaginary component of the grid current reference, is the real component of the capacitor voltage reference, is the imaginary component of the capacitor voltage reference.

[0058] Constructing a prediction model based on the above sampled state variables and grid voltage;

[0059] The prediction model is:

[0060]

[0061]

[0062]

[0063] Among them, v i (k) is the output voltage vector of the grid-connected inverter at time k. L2 is the grid-side inductance in the inverter, C is the filter capacitor in the inverter, v iαβ is the grid-connected inverter output voltage, u cαβ is the filter capacitor voltage, i 1αβ is the inverter side current, i gαβ is the grid side current, v gαβ is the grid voltage, L1 is the inverter side inductance in the inverter, T s is the system sampling time.

[0064] It should be noted that the selected candidate vectors of grid-connected inverter output voltage are substituted into the constructed prediction model respectively to obtain the predicted values ​​of the three state variables of the inverter side current, grid current and capacitor voltage of the grid-connected inverter at time k+2 corresponding to each output voltage vector.

[0065] Specifically, the voltage vector components in the grid-connected inverter output voltage candidate vector set screened in step S110 are substituted into the above-constructed prediction model formula. The predicted values ​​of the converter side current, grid current and capacitor voltage of the grid-connected inverter at time k+2 are calculated. Each voltage vector corresponds to a set of predicted values.

[0066] Construct a cost function, and substitute the predicted values ​​and reference values ​​of the three state variables of the grid-connected inverter at time k+2. The output voltage vector corresponding to the predicted values ​​of the three state transformations that minimize the cost function value is used as the optimal output voltage vector of the grid-connected inverter and activated in the next cycle.

[0067] The cost function is:

[0068]

[0069] Among them, λ i1 , uc , ig Respectively represent the inverter side current i1, capacitor voltage u c and grid current i g The weight factor of the error term is used to adjust the priority of the corresponding control target; ε i1 , ε ig , ε uc They represent the inverter side current i1 and the grid current i at time k+2 respectively. g , capacitor voltage u c The error between the reference value and the predicted value. The subscripts “α” and “β” represent the real and imaginary components of the error term; the superscript “2” represents the square value of the error.

[0070] The present invention determines the corresponding optimal driving signal according to the screened optimal output voltage vector, thereby controlling the on and off of the switching device in the grid-connected inverter, and activating it in the next control cycle, thereby outputting the optimal driving signal, achieving the purpose of reducing the amount of calculation.

[0071] S130, storing the obtained optimal output voltage vector sequence in a memory, and updating and iterating the stored data with the power frequency cycle as a period.

[0072] It should be noted that in the FCS-MPC algorithm, there are eight possible output voltage vectors, corresponding to eight switching sequences, which form a switching table. In each sampling period, by calculating the cost function values ​​of the eight output voltage vectors, the output voltage vector that minimizes the cost function is selected as the optimal output voltage vector, and its corresponding switching sequence (i.e., the optimal output voltage vector sequence) is applied to the next moment. This sequence is the switching sequence corresponding to the optimal output voltage vector in the switching table.

[0073] Specifically, a memory is constructed to store the optimal vector sequence information of the previous power frequency cycle and use it as a selection reference for the candidate vector of the next power frequency cycle; the stored data is updated and iterated with one power frequency cycle as a period. It can be understood that the present invention stores the inverter optimal output voltage vector sequence information obtained in S120 for selection of the candidate vector of the next period.

[0074] According to the above steps, the grid-connected converter can work stably. In order to fully illustrate the implementation effect of the present invention, a comparative experiment was carried out on a three-phase 3kw prototype. The parameters and control parameters of the experimental device refer to Table 2.

[0075] Table 2 System parameters

[0076]

[0077] Figure 4 The steady-state grid voltage and grid injection current waveforms of the proposed algorithm under strong grid conditions are given. It can be found that the grid injection current frequency and phase obtained based on the proposed algorithm are consistent with the grid voltage, and the current waveform is an ideal sine curve, with a THD of only 1.28%, which is much lower than the IEEE harmonic standard (total harmonic distortion cannot exceed 5%).

[0078] Figure 5 and Figure 6 The experimental waveforms of the proposed algorithm when the grid voltage and grid current reference step changes are given respectively. It can be seen from the figure that, using the proposed algorithm, when the voltage effective value changes from 110V to 82.5V, the grid injection current can still track the reference value well; and when the grid injection current reference value changes from 6.43A to 12.86A, the grid injection current can quickly track the reference value within 1ms. Therefore, the dynamic performance of the proposed scheme is verified.

[0079] Figure 7 The performance of the proposed algorithm in suppressing grid harmonics and unbalanced disturbances is verified. In the experiment, the grid background voltage is set to have third, fifth, seventh, and ninth harmonics. The corresponding amplitudes are set to 7%, 5%, 3% and 3% of the fundamental voltage amplitude. At the same time, the B phase voltage amplitude is reduced by 25% to simulate the unbalanced grid voltage. It can be found from the figure that under the unbalanced grid voltage condition with background harmonics, the grid injection current can still quickly track the reference value, and the THD meets the IEEE harmonic standard.

[0080] It should be noted that the present invention is applicable to various topological structures. In the embodiments of the present invention, only the LCL topology is taken as an example, which is not a limitation of the present invention.

[0081] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A novel finite control set model predictive control method applied to grid-connected inverters, characterized in that: include: According to the system operation status, select the inverter output voltage candidate vector at the current moment; Screening out an optimal output voltage vector according to the selected inverter output voltage candidate vector, and activating the optimal output voltage vector in the next cycle; The switching sequence corresponding to the obtained optimal output voltage vector is stored in a memory, and the stored data is updated and iterated with the power frequency cycle as a period; The step of screening out the optimal output voltage vector according to the selected inverter output voltage candidate vector is: According to the state variables and grid voltage sampled at the current time k, and using the delay compensation algorithm, a prediction model for time k+2 is constructed, and the reference values ​​of the three state variables of the inverter side current, grid current and capacitor voltage of the grid-connected inverter at time k+2 are calculated; Substitute the selected candidate voltage vectors into the prediction model respectively to obtain the predicted values ​​of the inverter side current, capacitor voltage and grid current at time k+2 corresponding to each candidate voltage vector; Construct a cost function, substitute the predicted values ​​and reference values ​​of the inverter side current, capacitor voltage and grid current at time k+2, take the inverter output voltage vector that minimizes the cost function value as the optimal output voltage vector of the grid-connected inverter, and activate the selected inverter optimal output voltage vector at time k+1; The prediction model is: Among them, v i (k) is the output voltage vector of the grid-connected inverter at time k, L2 is the grid-side inductance in the inverter, C is the filter capacitor in the inverter, v iαβ is the grid-connected inverter output voltage, u cαβ is the filter capacitor voltage, i 1αβ is the inverter side current, i gαβ is the grid side current, v gαβ is the grid voltage, L1 is the inverter side inductance in the inverter, T s is the system sampling time.

2. The novel finite control set model predictive control method for grid-connected inverter according to claim 1, characterized in that: The step of selecting the inverter output voltage candidate vector at the current moment according to the system operation state is: Determine whether the system operation time reaches the set value, if not, select all possible inverter output voltage vectors as candidate vectors; If so, based on the historical data of the optimal vector sequence, the inverter output voltage vector adjacent to the vector in spatial position and the vector itself are selected as candidate vectors.

3. The novel finite control set model predictive control method for grid-connected inverter according to claim 1, characterized in that: The cost function is: Among them, λ i1 , uc , ig Respectively represent the inverter side current i1 and capacitor voltage u c and grid current i g Weight factor for the error term, ε i1 , ε i g , ε uc They represent the inverter side current i1 and the grid current i at time k+2 respectively. g , capacitor voltage u c The error between the reference value and the predicted value; the subscripts "α" and "β" represent the real and imaginary components of the error term.

Images