Converter control method and power grid system

Abstract

The invention discloses a converter control method and a power grid system, and the method comprises the steps: carrying out the linear programming of sectors, defining two regular hexagons in each sector, carrying out the lineation of each sector, enabling a first straight line and a second straight line to extend along the diagonal line of the first regular hexagon through a small vector, and enabling the first straight line and the second straight line to be collinear with one side of the second regular hexagon; constructing a predictive control framework, and predicting a reference voltage vector by adopting an autoregression model; the angle and amplitude of a reference voltage vector are calculated, a target sector is determined, when the amplitude of the reference voltage vector is larger than 1 / 3 of the bus voltage, if the amplitude of the reference voltage vector is smaller than the voltage absolute value of any straight line, the middle vector closest to the corresponding straight line is selected as the optimal voltage vector, and otherwise, the large vector in the sector is selected as the optimal voltage vector; and generating a switching state of the converter based on the optimal voltage vector. According to the method, the dependence of the control method on converter system parameters can be eliminated, the candidate voltage vectors can be quickly screened out, and parameter-free robust predictive control is realized.

Description

Technical Field

[0001] The present application relates to the technical field of converter control, and particularly to a control method for a converter and a power grid system. Background Art

[0002] In the field of converter control, the development of microprocessors has promoted the research and application of model predictive control technology. Among them, finite set model predictive control has received key attention and application due to its advantages such as multi-objective optimization ability, superior dynamic performance, and steady-state performance.

[0003] However, in practical applications, the accuracy of the system model will have a great impact on the control performance of finite set model predictive control. With the change of working conditions and the accumulation of time, the parameters of the power converter will change, resulting in parameter mismatch problems. In addition, the unknown parameters of some power converters also limit their applications. Existing methods for solving parameter mismatch problems incur high computational costs and are difficult to run on low-power processors. Summary of the Invention

[0004] To solve the deficiencies of the prior art, the present application adopts the following technical solutions: In a first aspect, the present application provides a control method for a converter. The control method includes: Performing linear programming on sectors, equally dividing the complex plane of 27 voltage vectors into 6 sectors. In each sector, a first regular hexagon and a second regular hexagon with equal areas and sharing a common side are defined. Among them, the first regular hexagon is centered on the small vector of the corresponding sector, and the second regular hexagon is centered on the large vector of the corresponding sector. Drawing lines for each sector to obtain a first line and a second line of the corresponding sector. Both the first line and the second line extend along the diagonal of the first regular hexagon through the small vector of the corresponding sector and are collinear with one side of the second regular hexagon respectively, and obtaining the absolute values of the voltages corresponding to the first line and the second line of each sector; Constructing a data-driven predictive control framework. Under the predictive control framework, using the current and voltage output information of the converter system, predicting the reference voltage vector of the converter system by means of an autoregressive model; Calculating the angle of the reference voltage vector and determining the target sector where the reference voltage vector is located according to the angle; Calculating the amplitude of the reference voltage vector. When the amplitude of the reference voltage vector is greater than 1 / 3 of the bus voltage, comparing the amplitude of the reference voltage vector with the absolute values of the voltages of the first line and the second line of the target sector respectively. If the amplitude of the reference voltage vector is less than any one of the absolute values of the voltages, selecting the medium vector closest to the corresponding line as the optimal voltage vector; otherwise, selecting the large vector of the corresponding sector as the optimal voltage vector; Generate the switching state of the converter based on the selected optimal voltage vector.

[0005] In summary, a control method for a converter provided by this application is based on a vector optimization strategy of linear programming. It performs linear programming on the complex plane of voltage vectors to obtain the absolute values of voltages corresponding to the first straight line and the second straight line in each sector. When the amplitude of the reference voltage vector is greater than 1 / 3 of the bus voltage, it judges the magnitude relationship between the amplitude of the reference voltage vector and the absolute values of voltages corresponding to the first straight line and the second straight line, quickly screens out candidate voltage vectors, thereby avoiding traversing all switching states, improving the calculation efficiency, and reducing the calculation cost in vector optimization. Moreover, by constructing a data-driven predictive control framework and using an autoregressive model to predict the reference voltage vector, the dependence of the control method on the converter system parameters is eliminated, and parameter-free robust predictive control is achieved.

[0006] Furthermore, the control method further includes: The autoregressive model takes the reference voltage vector of the converter system as the prediction object. The reference voltage vector is obtained by the inner product of the regression coefficient matrix and the regression vector, where the regression vector includes the expected current information at the next moment, the current information at the current moment, and the historical current information with a dimension of and also includes the historical voltage information with a dimension of

[0007] Furthermore, the control method further includes: Use the NLMS algorithm to update the regression coefficients. The regression coefficients updated at the current moment are corrected from the regression coefficients at the previous moment. Among them, the correction term is determined by the adjustable learning rate and the current regression vector multiplied by the prediction error , and this product is controlled by the regularization factor composed of the inner product of the regression vectors.

[0008] Furthermore, the control method further includes: Use the autoregressive model to predict the reference voltage vector at the next moment to achieve delay compensation, and perform predictive control on the converter based on the predicted reference voltage vector at the next moment.

[0009] Furthermore, the control method further includes: Judge whether the amplitude of the reference voltage vector is less than 1 / 6 of the bus voltage. If so, select the zero vector as the optimal voltage vector.

[0010] Furthermore, the control method further includes: ​Determine whether the magnitude of the reference voltage vector is less than 1 / 3 of the bus voltage and greater than 1 / 6 of the bus voltage. If so, select the small vector corresponding to the target sector as the optimal voltage vector.

[0011] Further, when a small vector is selected as the optimal voltage vector, a cost function is used to screen the redundant small vectors, and the one that makes the result of the cost function closer to zero is selected as the optimal switching state. The cost function is obtained by multiplying the sign of the predicted neutral point voltage by the sign of the neutral point current.

[0012] Further, the control method further includes predicting the neutral point voltage of the converter system through the following steps: Design a neutral point voltage update law , and add the product of the neutral point voltage update law and the neutral point current at the current moment to the neutral point voltage at the current moment to obtain the predicted value of the neutral point voltage at the next moment .

[0013] Further, the control method further includes: The neutral point voltage update law is configured to satisfy: if the neutral point current at the previous moment is 0, the neutral point voltage update law is not updated; otherwise, it is obtained by adding the neutral point voltage update law at the previous moment and a correction term. The correction term of the voltage update law is determined by the neutral point voltage prediction error and the neutral point current.

[0014] In a second aspect, the present application further provides a power grid system, which includes a converter, and the power grid system applies the control method of the above converter. Description of the Drawings

[0015] Figure 1 is a flowchart of the steps of the control method of the converter provided by an embodiment of the present application; Figure 2 is a schematic diagram of the topology of a neutral point clamped three-level converter provided by an embodiment of the present application; Figure 3 is a schematic diagram of the spatial distribution of voltage vectors in the control method of the converter provided by an embodiment of the present application; Figure 4 is a schematic diagram of sector planning for the spatial distribution of voltage vectors in the control method of the converter provided by an embodiment of the present application; Figure 5 is a schematic diagram of drawing lines in one sector in the control method of the converter provided by an embodiment of the present application; Figure 6The control block diagram of the control method for the converter provided by an embodiment of the present application is applied to the converter system; Figure 7 The flowchart for finding the optimal voltage vector of the control method for the converter provided by an embodiment of the present application; Figure 8a The schematic diagram of the experimental results of the steady-state performance of the traditional FCS-MPC method provided by a comparative example of the present application under the condition of system parameter matching; Figure 8b The schematic diagram of the experimental results of the steady-state performance of the control method for the converter provided by an embodiment of the present application under the condition of system parameter matching; Figure 9a The schematic diagram of the experimental results of the steady-state performance of the traditional FCS-MPC method provided by a comparative example of the present application under the condition of system parameter mismatch; Figure 9b The schematic diagram of the experimental results of the steady-state performance of the control method for the converter provided by an embodiment of the present application under the condition of system parameter mismatch; Detailed implementation manners

[0016] The present application will be described in detail below in conjunction with the specific implementation manners shown in the accompanying drawings. However, these implementation manners do not limit the present application, and any structural, method, or functional transformation made by those of ordinary skill in the art based on these implementation manners is included in the protection scope of the present application.

[0017] In order to solve the deficiencies of the prior art, an embodiment of the present application provides a control method for a converter. Through a linear programming vector optimization strategy, the optimal voltage vector is quickly screened to realize the predictive control of the converter system. As Figure 1 shown, the control method includes the following steps: Step S101: Perform linear programming on the sectors, equally divide the complex plane of 27 voltage vectors into 6 sectors, define a first regular hexagon and a second regular hexagon with equal areas and a common side in each sector. Among them, the first regular hexagon is centered on the small vector of the corresponding sector, and the second regular hexagon is centered on the large vector of the corresponding sector. Draw lines for each sector to obtain the first line and the second line of the corresponding sector. The first line and the second line both pass through the small vector of the corresponding sector and extend along the diagonal of the first regular hexagon, and are respectively collinear with one side of the second regular hexagon, and obtain the absolute value of the voltage corresponding to the first line and the second line of each sector.

[0018] Step S102: Construct a data-driven predictive control framework. Under the predictive control framework, use the current and voltage output information of the converter system, and adopt an autoregressive model to predict the reference voltage vector of the converter system.

[0019] Step S103: Calculate the angle of the reference voltage vector, and determine the target sector where the reference voltage vector is located according to the angle.

[0020] Step S104: Calculate the amplitude of the reference voltage vector. When the amplitude of the reference voltage vector is greater than 1 / 3 of the bus voltage, compare the amplitude of the reference voltage vector with the absolute values of the first line voltage and the second line voltage of the target sector respectively. If the amplitude of the reference voltage vector is less than any of the voltage absolute values, select the medium vector closest to the corresponding line as the optimal voltage vector; otherwise, select the large vector of the corresponding sector as the optimal voltage vector.

[0021] Step S105: Generate the switching state of the converter based on the selected optimal voltage vector.

[0022] Exemplarily, taking the converter control method provided in the embodiment of the present application applied to a neutral-point clamped three-level converter as an example, the topological structure of the neutral-point clamped three-level converter is as Figure 2 shown. The neutral-point clamped three-level converter has a total of three bridge arms. According to the different switching states of the switching tubes, each bridge arm has three working states: 1, 0, and -1. The three bridge arms can form a total of 27 switching states. According to the different action characteristics of different voltage vectors, the 27 voltage vectors can be divided into four types of vectors: large vectors, medium vectors, small vectors, and zero vectors. Among them, each zero vector corresponds to three switching states, each small vector corresponds to two switching states, each medium vector corresponds to one switching state, and each large vector corresponds to one switching state.

[0023] Among the 27 voltage vectors, there are 19 effective voltage vectors and 8 redundant voltage vectors.

[0024] As Figure 3 shown, the 19 effective voltage vectors include: zero vectors V0 (-1, -1, -1), (0, 0, 0), (1, 1, 1); small vectors: V1 (1, 0, 0), (0, -1, -1), V2 (1, 1, 0), (0, 0, -1), V3 (0, 1, 0), (-1, 0, -1), V4 (0, 1, 1), (-1, 0, 0), V5 (0, 0, 1), (-1, -1, 0), V6 (1, 0, 1), (-1, -1, 0); medium vectors: V8 (1, 0, -1), V 10 (0, 1, -1), V 12 (-1, 1, 0), V 14 (-1, 0, 1), V 16 (0, -1, 1), V 18 (1, -1, 0); large vectors: V7 (1, -1, -1), V9 (1, 1, -1), V 11 (-1, 1, -1), V 13(-1, 1, 1), V 15 (-1, -1, 1), V 17 (1, -1, 1). The 8 redundant voltage vectors include 6 small vectors and 2 zero vectors.

[0025] As Figure 4 shown, sector planning is carried out on the spatial distribution of voltage vectors, dividing the complex plane of 27 voltage vectors into 6 equal sectors, and each sector corresponds to a spatial angle range of 60°. Two regular hexagons with equal areas and a common side are further defined within each sector. The first regular hexagon is centered on the small vector within the sector, and the second regular hexagon is centered on the large vector within the sector. From Figure 3 and Figure 4 it can be seen that the small vectors are distributed on the inner side of the sector close to the zero vector, the large vectors are distributed on the outer side of the sector far from the zero vector, and the first regular hexagon is closer to the zero vector than the second regular hexagon.

[0026] For the convenience of explanation, as Figure 4 shown, define the distance between the small vector and the zero vector as I, and I is equal to 1 / 3 times the bus voltage V dc . Taking the zero vector V0 as the center point and I as the radius, the area outside the radius I in the sector is called the outer vector circle, and the area inside the radius I in the sector is called the inner vector circle; moreover, the inner vector circle includes the zero vector circle and the small vector circle. Taking the zero vector V0 as the center point and (1 / 2)I as the radius, the area inside the radius (1 / 2)I in the inner vector circle is called the zero vector circle, and the area outside the radius (1 / 2)I in the inner vector circle is called the small vector circle.

[0027] Draw lines for each sector respectively to obtain the first straight line and the second straight line in the corresponding sector. The first straight line and the second straight line extend along the diagonals of the first regular hexagon respectively, and moreover, the first straight line and the second straight line extend to the second regular hexagon respectively. The first straight line and the second straight line are collinear with one side of the second regular hexagon respectively, and the intersection point of the first straight line and the second straight line coincides with the small vector in the sector. Optionally, lines can also be drawn along the diagonals of the second regular hexagon respectively to obtain the third straight line and the fourth straight line in the corresponding sector. The third straight line and the fourth straight line extend to the first regular hexagon respectively, and moreover, the third straight line and the fourth straight line are collinear with one side of the first regular hexagon respectively, and the intersection point of the third straight line and the fourth straight line coincides with the large vector in the sector. Further, in one embodiment, select one side of the first regular hexagon and the second regular hexagon that are collinear as the fifth straight line, and select one side of the first regular hexagon that is parallel to the fifth straight line as the sixth straight line.

[0028] Exemplarily, as Figure 5 shown, in the figure, V0 is the zero vector, V in is the small vector, V bo and Vup is the medium vector, V out is the large vector. Extend the lines along the diagonals of the first regular hexagon respectively to obtain the first straight line L up1 and the second straight line L bo1 . Further, extend the lines along the diagonals of the second regular hexagon respectively to obtain the third straight line L up2 and the fourth straight line L bo2 ; and select the common side line of the first regular hexagon and the second regular hexagon as the fifth straight line L out , and select a side line parallel to the fifth straight line L out in the first regular hexagon as the sixth straight line L in .

[0029] Exemplarily, in one embodiment, the numbers of all the straight lines in different sectors can be as shown in the following table:

[0030] Table 1 As an optional implementation manner, the voltage values corresponding to the straight lines after scribing in the sector can be expressed by the following formula: ; ; ; In the formula, is the angle of the voltage vector, represents the voltage value expression corresponding to the straight line in the rectangular coordinate system, and r represents the voltage value expression corresponding to the straight line in the polar coordinate system.

[0031] After completing the linear programming for each sector, based on the above voltage value expression formula of the straight line, substituting the numbers of the straight lines in Table 1 into the voltage value expression formula, the absolute values of the voltages corresponding to the first straight line and the second straight line in each sector can be obtained. When the reference voltage vector is located in a certain sector, by judging the magnitude relationship between the amplitude of the reference voltage vector and the absolute values of the voltages corresponding to the first straight line and the second straight line, the candidate voltage vectors can be quickly screened out, thus avoiding traversing all the switching states and improving the calculation efficiency.

[0032] After completing the linear programming of the sectors and obtaining the absolute values of the voltages corresponding to the first and second lines of each sector, in step S102, the current and voltage signals of the converter are collected in real time, and combined with the historical input and output data of the converter system, a data-driven predictive control framework is constructed. Based on the predictive control framework, using the current and voltage output information of the converter system, the model parameters are dynamically updated through an autoregressive model, so as to predict the reference voltage vector of the converter system at the next moment. For example, using the current value of the converter system at the current moment, historical current and voltage data, the regression coefficients are continuously optimized through an online learning algorithm, so that the predictive control framework can adapt to system parameter changes or external disturbances. Based on the autoregressive model and combined with the current and voltage output information of the converter system, the prediction of the reference voltage vector of the converter system at the next moment is realized, without the need to pre-obtain electrical parameters such as the inductor and resistor of the converter system, avoiding the dependence of the control system on the electrical parameters of the converter, and improving the robustness of the control system.

[0033] According to the predicted reference voltage vector, in step S103, calculate the angle of the reference voltage vector to determine the sector where the reference voltage vector is located. As an optional implementation manner, the angle calculation of the reference voltage vector can be expressed by the following formula: ; In the formula, represents the angle of the reference voltage vector, represents αβ the input voltage information in the two-phase stationary coordinate system.

[0034] Furthermore, determining the target sector where the reference voltage vector is located significantly reduces the computational complexity and realizes the rapid positioning of the reference voltage vector. As an optional implementation manner, the sector angle distribution can be expressed as follows: ; After determining the target sector where the reference voltage vector is located, in step S104, calculate the amplitude of the reference voltage vector, and compare the amplitude of the reference voltage vector with the bus voltage V dc If the amplitude of the reference voltage vector is greater than 1 / 3 of the bus voltage V dc, it indicates that the reference voltage vector is located outside the sector vector circle, and further compare the magnitude of the reference voltage vector with the absolute value of the first linear voltage and the absolute value of the second linear voltage of the target sector respectively. If the magnitude of the reference voltage vector is less than either the absolute value of the first linear voltage or the absolute value of the second linear voltage, it indicates that the reference voltage vector is close to the middle vector region. If the magnitude of the reference voltage vector is less than the absolute value of the first linear voltage, select the middle vector closest to the first straight line as the optimal voltage vector. If the magnitude of the reference voltage vector is less than the absolute value of the second linear voltage, select the middle vector closest to the second straight line as the optimal voltage vector. If the magnitude of the reference voltage vector is greater than both the absolute value of the first linear voltage and the absolute value of the second linear voltage, it indicates that the reference voltage vector is close to the large vector region, then select the large vector within the sector as the optimal voltage vector.

[0035] Exemplarily, in combination with Figure 3 and Figure 5 , taking the sector corresponding to Figure 5 as an example for sector 6, sector 6 takes the zero vector V0 as the origin and includes: the small vector V1, the middle vectors V8, V 18 and the large vector V7. If the magnitude of the reference voltage vector is less than the absolute value of the first linear voltage, select the vector V bo closest to the first straight line as the optimal voltage vector (i.e., the middle vector V 18 corresponding to sector 6); if the magnitude of the reference voltage vector is less than the absolute value of the second linear voltage, select the middle vector V up closest to the second straight line as the optimal voltage vector (i.e., the middle vector V8 corresponding to sector 6); if the magnitude of the reference voltage vector is greater than both the absolute value of the first linear voltage and the absolute value of the second linear voltage, select the large vector V out within the sector as the optimal voltage vector (i.e., the large vector V7 corresponding to sector 6).

[0036] After obtaining the optimal voltage vector, in step S105, based on the selected optimal voltage vector, generate the switching state corresponding to the converter. The generated switching state can act on the converter system through the drive circuit, thereby realizing the control of the converter system and completing the complete closed-loop from the reference voltage vector prediction to the converter output. Exemplarily, in a neutral-point clamped three-level converter, each arm has three switching states (1, 0, -1). When the optimal voltage vector is a middle vector or a large vector, convert the vector into the switching state of the corresponding arm and act on the converter system to realize the predictive control of the converter system.

[0037] To further illustrate the control method provided by the embodiments of the present application, the control method provided by the embodiments of the present application is applied to a converter system such as Figure 6As shown, the control system is based on the current and voltage output information of the converter system. Through the data-driven controller, linear programming vector screening is performed to predict the optimal voltage vector for the converter system at the current moment. By converting the optimal voltage vector into the switching signal for controlling the converter bridge arm, the predictive control of the converter system is realized.

[0038] According to the above description, a control method for a converter provided by an embodiment of the present application is based on a vector optimization strategy of linear programming. Linear programming is performed on the complex plane of the voltage vector to obtain the absolute values of the voltages corresponding to the first straight line and the second straight line in each sector. When the amplitude of the reference voltage vector is greater than 1 / 3 of the bus voltage, the magnitude relationship between the amplitude of the reference voltage vector and the absolute values of the voltages corresponding to the first straight line and the second straight line is judged, and the candidate voltage vectors are quickly screened out, thereby avoiding traversing all switching states, improving the calculation efficiency, and reducing the calculation cost in vector optimization; moreover, by constructing a data-driven predictive control framework and using an autoregressive model to predict the reference voltage vector, the dependence of the control method on the parameters of the converter system is eliminated, and parameter-free robust predictive control is realized.

[0039] As an optional implementation manner, in step S102, an autoregressive model is used to predict the reference voltage vector of the converter system. Exemplarily, taking a neutral-point clamped three-level converter as an example, the traditional model-based predictive control equation can be expressed as follows: ; In the formula, represents resistance, represents inductance, and respectively represent the system input voltage and output current in the

[0040] coordinate system. For the current control of a neutral-point clamped three-level converter, a matrix form of a controller based on an autoregressive model is set. The autoregressive model takes the reference voltage vector of the converter system as the prediction object, and the reference voltage vector of the converter system is obtained through the inner product of the regression coefficient matrix and the regression vector. In one embodiment, the regression vector includes the expected current information of the converter system at the next moment, the current information at the current moment, the historical current information with a dimension of and the historical voltage information with a dimension of .

[0041] As an optional implementation manner, the autoregressive model can be expressed by the following formula: ; In the formula, represents the voltage information of the axis at the kth moment, represents the transpose of the regression coefficient matrix, represents the regression vector, and the regression vector is expressed by the following formula: ; In the formula, represents the predicted output current information of the axis at time k, represents the output current information of the axis at time (k - 1), represents the input voltage information of the axis at time (k - 1), represents the historical current information with a dimension of represents the historical voltage information with a dimension of .

[0042] Furthermore, the regression coefficient can be expressed by the following formula: ; In the formula, and are constants to be predicted.

[0043] The regression coefficients in the autoregressive model are predicted through a criterion function, and the criterion function can be expressed by the following formula: ; In the formula, represents the voltage vector acting at time (k - 1), represents the predicted value of the voltage vector acting at time (k - 1). can be obtained by fitting through the following formula based on historical data: ; In the formula, represents the predicted value of the input voltage information of the axis at time (k - 1), represents the predicted value of the regression coefficient of the axis at time (k - 1).

[0044] Furthermore, the prediction error of the voltage vector can be expressed by the following formula: ; In the formula, represents the prediction error of the voltage vector of the

[0045] As an alternative implementation, the NLMS algorithm is used to update the regression coefficients in the autoregressive model. The regression coefficients updated at the current time are corrected from the regression coefficients at the previous time. Among them, the correction term is due to the adjustable learning rate and the current regression vector is determined by the product with the prediction error , and this product is controlled by a regularization factor composed of the inner product of the regression vectors.

[0046] In one embodiment, the updated regression coefficient can be expressed by the following formula: ; wherein, represents the regression vector, is an adjustable learning rate, is a very small positive number to ensure the effectiveness of the denominator, represents the regression coefficient updated at the current moment, represents the regression coefficient at the previous moment, represents the prediction error of the voltage vector on the

[0047] As an alternative implementation, considering the delay effect of the control system comprehensively, an autoregressive model is used to predict the reference voltage vector at the next moment to achieve delay compensation, and the converter is predicted and controlled based on the predicted reference voltage vector at the next moment. In one embodiment, the reference voltage vector at the next moment (k + 1) is obtained by multiplying the regression coefficient matrix at the next moment by the regression vector, and the autoregressive model can predict the reference voltage vector at the next moment through the following formula: ; wherein, represents the regression vector at the (k + 1)-th moment. The regression vector at the (k + 1)-th moment includes the expected current information at the (k + 2)-th moment, the current information at the (k + 1)-th moment, the current and voltage information at the current moment (the k-th moment), the historical current information with a dimension of and the historical voltage information with a dimension of . In one embodiment, the regression vector at the (k + 1)-th moment can be expressed by the following formula: ; wherein, represents the predicted output current information on the axis at the (k + 2)-th moment, represents the output current information on the axis at the (k + 1)-th moment, represents the input voltage information on the axis at the k-th moment, represents the historical current information with a dimension of , represents the historical voltage information with a dimension of .

[0048] Furthermore, the current information at the (k + 1)-th moment can be predicted from the information at the k-th moment. Specifically, in the case of digital delay, the reference voltage vector at the k-th moment is the actual voltage vector applied at the (k - 1)-th moment. The predicted current information at the (k + 1)-th moment generated under its action can be approximated as the actual current information at the (k + 1)-th moment. For the convenience of calculation, it is assumed that the regression coefficient at the (k + 1)-th moment remains unchanged. There is , in the regression vector to predict the current value at the (k + 1)-th moment , by shifting the voltage vector at the (k - 1)-th moment to the k-th moment, the current value can be obtained through the following formula : ; In the formula, represents the output current information of the axis at the (k + 1)-th moment, represents the output voltage information of the axis at the k-th moment, represents the first item of the coefficient matrix of the axis at the k-th moment, represents the historical current information with a dimension of represents the historical voltage information with a dimension of .

[0049] As an alternative implementation, as shown in Figure 7 , calculate the amplitude of the reference voltage vector, and determine whether the amplitude of the reference voltage vector is less than 1 / 6 of the bus voltage, that is, determine whether the amplitude of the reference voltage vector is less than (1 / 2)I. If the reference voltage vector is less than (1 / 2)I, it means that the reference voltage vector is within the sector zero vector circle. Select the zero vector as the optimal voltage vector to generate the switching state, and apply the switching state to the converter system to achieve the predictive control of the converter system.

[0050] As an alternative implementation, as shown in Figure 7 , if the amplitude of the reference voltage vector is greater than 1 / 6 of the bus voltage, further determine whether the amplitude of the reference voltage vector is less than 1 / 3 of the bus voltage and greater than 1 / 6 of the bus voltage. If the amplitude of the reference voltage vector is less than 1 / 3 of the bus voltage and greater than 1 / 6 of the bus voltage, it means that the reference voltage vector is within the small vector circle. Select the small vector in the sector where the reference voltage vector is located as the optimal voltage vector, and generate the switching state based on this optimal voltage vector. Apply the switching state to the converter system to achieve the predictive control of the converter system.

[0051] Furthermore, based on the neutral point voltage balance strategy, when a small vector is selected as the optimal voltage vector, the value function is used to screen the redundant small vectors, calculate the value function values of different redundant small vectors, select the one that makes the value function result closer to zero as the optimal switching state, and apply the switching state to the converter system to maintain the neutral point voltage balance of the converter system. In one embodiment, the value function is obtained by multiplying the sign of the predicted neutral point voltage by the sign of the neutral point current, and the value function can be expressed by the following formula: ; In the formula, represents the balance requirement for the neutral point voltage, is the sign function, represents the predicted value of the neutral point voltage, is the neutral point current.

[0052] It should be noted that the case where the amplitude of the reference voltage vector is greater than 1 / 3 of the bus voltage has been described above and will not be elaborated here.

[0053] As an alternative implementation, considering the time delay effect of the control system, an adaptive architecture is adopted to apply a time delay compensation strategy to achieve the prediction of the neutral point voltage of the converter system and maintain the neutral point voltage balance. The prediction of the neutral point voltage of the converter system specifically includes: designing a neutral point voltage update rate, sampling and obtaining the current and voltage information of the converter system at the current moment, and adding the product of the neutral point voltage update rate and the neutral point current at the current moment to the neutral point voltage at the current moment to obtain the predicted value of the neutral point voltage at the next moment.

[0054] In one embodiment, the neutral point voltage of the converter system can be predicted by the following formula: ; ; In the formula, represents the predicted value of the neutral point voltage at the k+1 moment, represents the neutral point voltage at the k moment, represents the neutral point current at the k moment, represents the three-switch expressions of the voltage vector used, represents the neutral point voltage update law.

[0055] Furthermore, the neutral point voltage update law is configured as follows: if the neutral point current at the previous moment (the k-1 moment) is 0, the neutral point voltage update rate is not updated; if the neutral point current at the previous moment is not 0, it is obtained by adding the neutral point voltage update rate at the previous moment and the correction term, where the correction term of the neutral point voltage update rate is determined by the neutral point voltage prediction error and the neutral point current.

[0056] As an alternative implementation, the neutral point voltage update rate can be expressed by the following formula: ; In the formula, represents the step factor; represents the predicted value of , and , represents the sampling time, represents the bus capacitor on the DC side; represents the neutral point current at the (k - 1)-th moment; represents the neutral point voltage at the k-th moment; represents the predicted value of the neutral point voltage at the k-th moment.

[0057] Substitute the expression of the neutral point voltage update law into the above calculation formula of the predicted value of the neutral point voltage . Based on the neutral point voltage and neutral point current at the current moment, predict the neutral point voltage of the converter system at the next moment, so as to maintain the balance of the neutral point voltage of the converter system.

[0058] As an alternative implementation, the control method provided in the embodiments of the present application further includes: based on the vector synthesis characteristic, within one control period, combine through the multi-vector synthesis method, screen multiple voltage vectors for vector synthesis, and then apply the switching state corresponding to the synthesized voltage vector to the converter system, so as to effectively improve the quality of the output current of the converter. Further, the autoregressive model architecture in the control method provided in the embodiments of the present application can be applied to various power electronic topologies, such as two-level converters, modular multilevel converters, permanent magnet synchronous motors, etc. The prediction equation based on the physical model can be rewritten as a prediction equation based on the autoregressive model and combined with an online neural network predictor to effectively improve the anti-interference ability and the ability to resist parameter changes of the control system.

[0059] To further illustrate a control method of a converter provided in the embodiments of the present application, the following experimental tests are carried out. By comparing the traditional FCS-MPC (Finite Control-Set Model Predictive Control) method and the control method provided in the embodiments of the present application, the effectiveness of the control method provided in the embodiments of the present application is verified. All the adopted parameters in the comparative experiment are shown in the following table:

[0060] Table 2 Under the condition of system parameter matching (inductance value L = 10 mH), the steady-state performance experimental results of the traditional FCS-MPC method are as follows Figure 8a shown. Under the control of the traditional FCS-MPC method, the total harmonic distortion (THD) of the output current of the converter is 2.17%; the steady-state performance experimental results of the control method provided by the embodiments of the present application are as follows Figure 8b shown. Under the control of this control method, the THD of the output current of the converter is 1.09%. Compared with the traditional control method, the relative reduction of THD of the control method provided by the embodiments of the present application reaches 49.8%. From Figure 8a and Figure 8b it can be concluded that under the condition of system parameter matching, the current quality output by the converter based on the control method provided by the embodiments of the present application is higher, and the control method provided by the embodiments of the present application has better harmonic suppression ability and better neutral point voltage balance effect compared with the traditional FCS-MPC method.

[0061] Under the condition of system parameter mismatch (inductance value L = 8 mH), the steady-state performance experimental results of the traditional FCS-MPC method are as follows Figure 9a shown. Under the control of the traditional FCS-MPC method, the THD of the output current of the converter is 3.43%; the steady-state performance experimental results of the control method provided by the embodiments of the present application are as follows Figure 9b shown. Under the control of this control method, the THD of the output current of the converter is 1.28%. Compared with the traditional control method, the relative reduction of THD of the control method provided by the embodiments of the present application reaches 62.7%. From Figure 9a and Figure 9b it can be concluded that under the condition of system parameter mismatch, the control effect of the traditional FCS-MPC method deteriorates significantly, and the control method provided by the embodiments of the present application can still maintain a relatively stable steady-state control effect.

[0062] Furthermore, under the same computing processor, the computing time of the control method provided by the embodiments of the present application is compared with the existing control methods for improving system robustness, and the results are shown in the following table:

[0063] Table 3 From the above comparison results, it can be concluded that the control method proposed in the embodiments of the present application can significantly reduce the computational burden and improve the computational efficiency of the control system compared with the existing ARX-based MFPC for improving system robustness. It should be noted that in the above experimental process, the off-grid operation mode of the converter was adopted for experimental verification. The output indicators of the converter grid-connected operation model are still physical quantities such as current and voltage, which are the same as the control objectives described above. Therefore, in the converter grid-connected operation mode, the control method provided in the embodiments of the present application can still predict the converter parameters and disturbances, and achieve the stable operation of the converter system without relying on system parameters.

[0064] According to the above description, a control method for a converter provided in an embodiment of the present application is based on a vector optimization strategy of linear programming. Linear programming is performed on the complex plane of the voltage vector to obtain the absolute values of the voltages corresponding to the first straight line and the second straight line in each sector. The magnitude relationship between the magnitude of the reference voltage vector and the bus voltage is judged. When the magnitude of the reference voltage vector is greater than 1 / 3 of the bus voltage, by judging the magnitude relationship between the magnitude of the reference voltage vector and the absolute values of the voltages corresponding to the first straight line and the second straight line, the optimal voltage vector is quickly screened and obtained; when the magnitude of the reference voltage vector is less than 1 / 6 of the bus voltage, the zero vector in the target sector is selected as the optimal voltage vector; when the magnitude of the reference voltage vector is greater than 1 / 6 of the bus voltage and less than 1 / 3 of the bus voltage, the small vector in the target sector is selected as the optimal voltage vector. Through the above optimal voltage vector screening method, traversing all switching states is avoided, the computational efficiency is improved, and the computational burden of the control method is greatly reduced; moreover, by constructing a data-driven predictive control framework and using an autoregressive model to predict the reference voltage vector, the dependence of the control method on the converter system parameters is eliminated, and parameter-free robust predictive control is achieved.

[0065] Furthermore, a control method for a converter provided in an embodiment of the present application comprehensively considers the need for delay compensation in the control system, designs the control system delay compensation, predicts the neutral point voltage, realizes the prediction of the neutral point voltage without increasing the algorithm complexity, maintains the neutral point voltage balance of the converter system, and ensures the stable operation of the converter system.

[0066] In the second aspect, an embodiment of the present application further provides a power grid system. The power grid system includes a converter. The power grid system applies the control method for the converter described above, can eliminate the dependence of the control system on the converter system parameters, achieve parameter-free robust predictive control, quickly screen out candidate voltage vectors, improve the computational efficiency, greatly reduce the computational burden of the control system, and ensure the stable operation of the converter system.

[0067] It will be understood that the term "exemplary" as used herein means "serving as an example, instance, or illustration". Any embodiment described as "exemplary" is not necessarily preferred or superior to other embodiments and / or does not exclude incorporating features of other embodiments. It should be understood that certain features of the present application described in the context of separate embodiments may also be provided in combination in a single embodiment. Conversely, the various features of the present application described in the context of a single embodiment may also be provided separately or in any suitable combination or as any other described embodiment of the present application.

[0068] In the description of the present application, unless otherwise specified, " / " means "or", for example, A / B may mean A or B. The "and / or" herein is merely a relationship describing the associated objects, indicating that three relationships may exist. For example, A and / or B may mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, "at least one" means one or more, and "a plurality" means two or more. The terms "first", "second", etc. do not limit the quantity and execution order, and the terms "first", "second", etc. do not necessarily limit to being different.

[0069] The above-disclosed are only the preferred embodiments of the present application, but they are not intended to limit the scope of the rights of the present application. Those of ordinary skill in the art can understand that within the spirit and scope of the present application and the appended claims, changes, modifications, substitutions, combinations, and simplifications should all be equivalent replacement methods and still fall within the scope covered by the invention.

Claims

1. A control method for an inverter, characterized in that The control method includes: Perform linear programming on the sectors, equally divide the complex plane of 27 voltage vectors into 6 sectors, define a first regular hexagon and a second regular hexagon with equal areas and a common side in each sector. Among them, the first regular hexagon is centered on the small vector of the corresponding sector, and the second regular hexagon is centered on the large vector of the corresponding sector. Draw lines for each sector to obtain the first line and the second line of the corresponding sector. Both the first line and the second line pass through the small vector of the corresponding sector and extend along the diagonal of the first regular hexagon, and are collinear with one side of the second regular hexagon respectively. Obtain the absolute value of the voltage corresponding to the first line and the second line of each sector. Construct a data-driven predictive control framework. Under this predictive control framework, use the current and voltage output information of the converter system, and use an autoregressive model to predict the reference voltage vector of the converter system. Calculate the angle of the reference voltage vector, and determine the target sector where the reference voltage vector is located according to the angle. Calculate the amplitude of the reference voltage vector. When the amplitude of the reference voltage vector is greater than 1 / 3 of the bus voltage, compare the amplitude of the reference voltage vector with the absolute value of the voltage of the first line and the absolute value of the voltage of the second line of the target sector respectively. If the amplitude of the reference voltage vector is less than any of the absolute values of the voltages, select the medium vector closest to the corresponding line as the optimal voltage vector. Otherwise, select the large vector of the corresponding sector as the optimal voltage vector. Generate the switching state of the converter based on the selected optimal voltage vector.

2. The control method of the converter according to claim 1, wherein The control method further includes: The autoregressive model uses the reference voltage vector of the converter system as the prediction object. The reference voltage vector is obtained by the inner product of the regression coefficient matrix and the regression vector, where the regression vector includes the expected current information at the next moment, the current information at the current moment, and the historical current information with a dimension of , and also includes the historical voltage information with a dimension of .

3. The control method of the converter according to claim 2, wherein The control method further includes updating the regression coefficients by using the NLMS algorithm, and the regression coefficients updated at the current moment are obtained by correcting the regression coefficients at the previous moment, where the correction term is determined by an adjustable learning rate and the current regression vector and the prediction error The product is determined, and the product is controlled by a regularization factor composed of the inner product of the regression vectors.

4. The control method of the converter according to claim 2, wherein The control method further includes: Use the autoregressive model to predict the reference voltage vector at the next moment to achieve delay compensation, and perform predictive control on the converter based on the predicted reference voltage vector at the next moment.

5. The control method of the converter according to claim 1, wherein The control method further includes: Judge whether the amplitude of the reference voltage vector is less than 1 / 6 of the bus voltage. If so, select the zero vector as the optimal voltage vector.

6. The control method of the converter according to claim 1, characterized in that The control method further includes: Judge whether the amplitude of the reference voltage vector is less than 1 / 3 of the bus voltage and greater than 1 / 6 of the bus voltage. If so, select the small vector of the corresponding target sector as the optimal voltage vector.

7. The control method of the converter according to claim 6, characterized in that When selecting the small vector as the optimal voltage vector, use a cost function to screen the redundant small vectors, and select the one that makes the result of the cost function closer to zero as the optimal switching state. The cost function is obtained by multiplying the sign of the predicted neutral point voltage by the sign of the neutral point current.

8. The control method of the converter according to claim 7, characterized in that, The control method further includes predicting the neutral point voltage of the converter system through the following steps: Design neutral point voltage update law , the neutral point voltage update law is multiplied by the neutral point current at the current moment , and then added to the neutral point voltage at the current moment to obtain the predicted value of the neutral point voltage at the next moment .

9. The control method of the converter according to claim 8, wherein The control method further includes: The neutral point voltage update law is configured to satisfy: if the neutral point current at the previous moment is 0, the neutral point voltage update law is not updated. Otherwise, it is obtained by adding the neutral point voltage update law at the previous moment and the correction term. The correction term of the voltage update law is determined by the neutral point voltage prediction error and the neutral point current.

10. A power grid system, the power grid system includes an inverter, characterized in that, The power grid system applies the control method of the converter according to any one of claims 1 to 9.

Images