A predictive control method and control system for a three-level converter

Abstract

The application discloses a predictive control method for a three-level converter, comprising the following steps: obtaining a reference vector based on the deadbeat prediction of the output current of the three-level converter, and establishing a target value function by using the Euclidean distance between the reference vector and an output vector; calculating a basic vector related to the reference vector to form a simplified control set; obtaining an optimized virtual vector and an optimized dead-zone time corresponding to each basic vector in the simplified control set according to the relative position relationship between each basic vector and the corresponding dead-zone voltage vector; obtaining an optimal virtual vector by using the target value function according to each basic vector and the corresponding optimized virtual vector, so as to control the three-level converter by using the optimized dead-zone time corresponding to the optimal virtual vector. The application reduces the calculation amount of the traditional FCS-MPC model prediction method, simplifies the control set, reduces the output current ripple and lowers the total harmonic distortion of the current, thereby improving the control performance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-level converter control, and particularly relates to a predictive control method and a control system for a three-level converter. BACKGROUND

[0002] With the continuous development of software and hardware technology, more control strategies have been applied to power electronic converters. In recent years, model predictive control has become a research hotspot. Among them, finite control set model predictive control (FCS-MPC) has attracted much attention due to its simple control structure and good control performance, and has become an effective control scheme for power electronic converters.

[0003] During the operation of an electronic power converter, in order to ensure smooth completion of switching state switching, a dead-time control process needs to be added during the switching process. However, the dead-time reduces the accuracy of the model, increases the distortion and ripple of the converter output voltage and current, and affects the control performance. Therefore, the FCS-MPC needs to consider the dead-time problem when applied in the converter.

[0004] Most existing FCS-MPC control strategies adopt two methods, one is compensation or elimination to reduce the influence of dead-time, and the other is to regard the output of the dead-time process as a dead-time voltage vector and use "variable dead-time" for double vector control. However, when the above methods are applied in a three-level converter, the virtual vector obtained by synthesizing the optimal basic vector and the corresponding dead-time voltage vector may not be the optimal solution. In addition, there are 27 basic vectors in a three-level converter, and if the dead-time is optimized one by one, 27 iterations are needed, resulting in heavy calculation. SUMMARY

[0005] The present application aims to provide a new method for reducing the calculation amount of the traditional FCS-MPC model predictive method, simplifying the control set, reducing the output current ripple and reducing the total harmonic distortion of the current, in order to control the three-level converter and improve the control performance of the traditional predictive model.

[0006] To solve the above technical problems, the embodiment of the present application provides a predictive control method for a three-level converter, comprising: obtaining a reference vector based on a deadbeat prediction of an output current of the three-level converter, and establishing a target value function by using a Euclidean distance between the reference vector and an output vector; calculating basic vectors related to the reference vector to form a simplified control set; obtaining an optimized virtual vector and an optimized dead zone time corresponding to each basic vector in the simplified control set according to a relative positional relationship between each basic vector in the simplified control set and a corresponding dead zone voltage vector; and obtaining an optimal virtual vector by using the target value function according to each basic vector and the corresponding optimized virtual vector, so as to control the three-level converter by using the optimized dead zone time corresponding to the optimal virtual vector.

[0007] Preferably, in the process of constructing the target value function, the following steps are included: obtaining a value function of a traditional FCS-MPC strategy, denoted as an original value function; and equivalently converting the Euclidean distance between the reference vector and the output vector into a deviation term formula between the output current prediction and a given current in the original value function to form the target value function.

[0008] Preferably, in the step of calculating the basic vectors related to the reference vector to form the simplified control set, the following steps are included: selecting three vectors with the closest Euclidean distance to the reference vector in a space vector plane, and determining redundant vectors of the three closest vectors to form the simplified control set composed of a plurality of basic vectors related to the reference vector, wherein the number of the basic vectors in the simplified control set is at most 5.

[0009] Preferably, the three closest vectors are sorted, the three closest vectors are converted into basic vectors to form a first type of basic vectors according to a difference between a maximum closest vector and a minimum closest vector, and a plurality of redundant vectors of the three closest vectors are obtained to be used as a second type of basic vectors; and the simplified control set is generated according to the first type of basic vectors and the second type of basic vectors.

[0010] Preferably, in the step of obtaining the optimized virtual vector and the optimized dead zone time corresponding to each basic vector according to the relative positional relationship between each basic vector in the simplified control set and the corresponding dead zone voltage vector, the following steps are included: judging whether there is a dead zone process; synthesizing each basic vector with the corresponding dead zone voltage vector to obtain an optimized virtual vector with the minimum Euclidean distance to the reference vector, and further obtaining the optimized dead zone time of each optimized virtual vector; and directly taking each basic vector without the dead zone process as the corresponding optimized virtual vector, and setting the optimized dead zone time of each basic vector without the dead zone process to zero.

[0011] Preferably, the each basic vector is compared with its corresponding dead-zone voltage vector, and the optimization virtual vector and the optimization dead-zone time corresponding to the current basic vector are calculated according to the comparison result, wherein if the same, the current basic vector is directly taken as the corresponding optimization virtual vector, and the minimum value of the action time of the current basic vector and the corresponding dead-zone voltage vector is taken as the corresponding optimization dead-zone time.

[0012] Preferably, if different, the optimization virtual vector and the optimization dead-zone time corresponding to the current basic vector are analyzed according to the following table:

[0013]

[0014]

[0015] wherein i represents the serial number of the basic vector with the dead-zone process, r i represents the distance between the terminal point of the i th basic vector and the terminal point of the dead-zone voltage vector of the i th basic vector, l i represents the distance between the projection on the line or the extension line between the terminal point of the i th basic vector and the terminal point of the dead-zone voltage vector of the i th basic vector and the terminal point of the dead-zone voltage vector of the i th basic vector, t min represents the minimum value of the terminal point of the i th basic vector and the action time of the dead-zone voltage vector of the i th basic vector, T s represents the unit control period, v i represents the i th basic vector, v id the dead-zone voltage vector of the i th basic vector.

[0016] Preferably, in the step of obtaining the optimal virtual vector according to the each basic vector and its corresponding optimization virtual vector by using the target value function, the following steps are included: substituting the optimization virtual vector corresponding to each basic vector into the Euclidean distance item in the target function, and substituting each basic vector and its corresponding dead-zone voltage vector and optimization dead-zone time into the direct-current side capacitor voltage balance control item in the target value function, to calculate the target evaluation value of each basic vector; selecting the optimal virtual vector and the corresponding optimal dead-zone time.

[0017] Preferably, the three nearest vectors are the three vectors corresponding to the vertices of the triangle in which the reference vector is located, wherein when the reference vector is outside the control area, the reference vector is converted into a vector with the same phase angle and located on the boundary of the hexagon.

[0018] In another aspect, the embodiments of the present application also provide a control system for a three-level converter, the system comprising: a three-level converter; and a processor connected to a main circuit of the three-level converter, wherein the processor is configured to implement the predictive control method as described above to control the three-level converter.

[0019] Compared with the prior art, one or more embodiments of the above solution can have the following advantages or beneficial effects:

[0020] The present application provides a predictive control method and a control system for a three-level converter. The control strategy adopted by the method and system has the following advantages or beneficial effects:

[0021] (1) Optimize the value function and effectively reduce the iteration calculation amount: the Euclidean distance between the basic vector and the reference vector is used to replace the current prediction deviation value in the traditional value function, thereby avoiding complex current prediction model calculation;

[0022] (2) Simplify the control set and significantly reduce the optimization space: the reference vector is obtained according to the deadbeat prediction current, and a general algebraic method is proposed to obtain the nearest three basic vectors and their redundant vectors of the reference vector, so as to construct a simplified control set;

[0023] (3) Consider the dead zone effect and construct a virtual vector control set: the dead zone voltage vector corresponding to the basic vector in the simplified control set is analyzed one by one; and the dead zone time is optimized, and the two are combined to obtain a virtual vector with the minimum Euclidean distance from the reference vector, thereby constructing a virtual vector control set;

[0024] (4) Realize double vector control and improve control performance: the optimal virtual vector is selected based on the value function, and the corresponding control waveform is output, so that the control performance is improved, and at the same time the number of switching times is basically unchanged.

[0025] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and achieved by the structure particularly pointed out in the specification, claims and drawings. BRIEF DESCRIPTION OF DRAWINGS

[0026] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, which together with the embodiments of the present application, serve to explain the present application, and do not constitute a limitation of the present application. In the drawings:

[0027] Figure 1 A flowchart of the predictive control method for a three-level converter according to the embodiments of the present application.

[0028] Figure 2A schematic diagram of reference vector positions in the space vector plane in the predictive control method for a three-level converter according to an embodiment of the present application.

[0029] Figure 3 A schematic diagram of the relationship between reference vector positions and optimal basic vectors in the predictive control method for a three-level converter according to an embodiment of the present application.

[0030] Figure 4 A schematic diagram of the positions of basic vectors, dead-zone voltage vectors and reference vectors in the predictive control method for a three-level converter according to an embodiment of the present application.

[0031] Figure 5 A schematic diagram of the relative positions of vectors in the predictive control method for a three-level converter according to an embodiment of the present application.

[0032] Figure 6 A schematic diagram of the framework structure of a control system for a three-level converter according to an embodiment of the present application. DETAILED DESCRIPTION

[0033] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and embodiments, so that how the present application applies technical means to solve technical problems and achieves technical effects can be fully understood and implemented. It should be noted that, as long as there is no conflict, each embodiment in the present application and each feature in each embodiment can be combined with each other, and the technical solutions formed thereby are all within the protection scope of the present application.

[0034] In addition, the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Moreover, although a logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown.

[0035] To solve the problems in the background art, an application embodiment proposes a three-level converter model predictive control strategy considering dead-zone voltage vectors. Specifically, it includes the following steps: obtaining a reference vector based on a dead-beat prediction, constructing a new value function; using relevant basic vectors and their redundant vectors to construct a simplified control set; synthesizing a virtual vector from the basic vectors in the simplified control set and the corresponding dead-zone voltage vectors, optimizing the dead-zone time one by one to obtain the virtual vector closest to the reference vector, and constructing a virtual vector control set; selecting the optimal virtual vector based on the Euclidean distance value function, and outputting the corresponding control waveform to realize the variable dead-zone time model predictive control. The present application can effectively reduce the complexity of iterative calculation, significantly reduce the optimization space, and compared with the traditional model predictive control, the control strategy can reduce the output current ripple, reduce the current total harmonic distortion by about 40%-50%, and significantly improve the control performance of the converter.

[0036] The topologies of three-level converters mainly include diode neutral point clamped, flying capacitor clamped and T-type neutral point clamped, etc. The application mode of finite control set model predictive control (FCS-MPC) strategy in the above topologies is referred to Figure 6 . The terminal voltage of the AC side is divided into three categories, and there are 3 3 27 kinds of switching states, and the space vector diagram composed of the corresponding 27 voltage vectors is shown in Figure 2 The switching function is defined as:

[0037]

[0038] There is a control cycle delay in the digital controller, that is, the control quantity calculated in the kth cycle (current control cycle) is actually controlled in the k+1th cycle. In order to compensate for the adverse effects of one beat delay, in the actual predictive control, the output value at k+2 time should be predicted in combination with the control quantity in the kth cycle, and then the voltage vector with the minimum cost function is selected to complete the evaluation and optimization of the optimal voltage vector and the control of the three-level converter.

[0039] According to the traditional FCS-MPC, the predicted values of the a and β axis currents at k+2 time are calculated. The value function of the traditional FCS-MPC is:

[0040]

[0041] Wherein, l represents the weight factor; represents the given value of the α and β axis currents at k+2 time, respectively; represents the predicted value at k+2 time, respectively; ΔU p (k+2) represents the predicted value of the voltage difference of the upper and lower capacitors on the DC side.

[0042] According to Kirchhoff's voltage law and Clark transformation, the mathematical model of the AC side in the αβ coordinate system can be obtained:

[0043]

[0044] Wherein, u α , u β represent the components of the output voltage in the α and β axes, respectively; i α , i β represent the components of the output current in the α and β axes, respectively; L represents the inductance of the AC side; R represents the load.

[0045] According to the forward Euler method, formula (2) is discretized. Considering the delay compensation, the estimated value of the phase current at k+1 time is:

[0046]

[0047] wherein i α (k) and i β (k) are obtained by Clark transformation of the output current sample value at the kth moment, respectively; and represents the estimation value of the output current at the k+1th moment; u α (k) and u β (k) represents the control quantity of the kth control period, and the voltage vector has been obtained in the previous period; T s represents a unit control period.

[0048] At the k+2th moment, the predicted values of the output currents of the α and β axes are:

[0049]

[0050] wherein u a (k+1) and u β (k+1) represent the control quantity of the k+1th control period.

[0051] Figure 1 is a step diagram of the predictive control method for a three-level converter according to an embodiment of the present application. The specific flow steps of the predictive control method for a three-level converter according to an embodiment of the present application (hereinafter referred to as “the predictive control method”) will be described below with reference to Figure 1 , the specific flow steps of the predictive control method for a three-level converter according to an embodiment of the present application (hereinafter referred to as “the predictive control method”) will be described below with reference to

[0052] Step S110 obtains the reference vector based on the deadbeat prediction of the output current of the three-level converter, and uses the Euclidean distance between the reference vector and the output vector, and further combines the direct current side capacitor voltage balance control to establish a target value function.

[0053] Reference Figure 6 After sampling the output current of the three-level converter, the three-phase output current is first Clark transformed and then delay compensation processed, so as to perform deadbeat prediction on the estimation value of the output current at the next control period (the k+1th moment) to obtain the reference vector v * .

[0054] Further, in the process of constructing the target value function, the value function of the traditional FCS-MPC strategy is first obtained, denoted as the original value function, and then the Euclidean distance between the reference vector and the output vector is equivalent to the deviation term between the output current prediction and the given current in the original value function, forming a new target value function.

[0055] Specifically, it is assumed that the reference vector v *Under the action of the system output current in a sampling period can achieve the given value of no difference tracking. At the same time, let the reference vector v * The coordinates of the above expression (4) are The current prediction value obtained by substituting the above expression (4) is the same as the given value, that is:

[0056]

[0057] The expression (4), (5) is substituted into the original value function expression (1), and the deviation of the current prediction item formula part is obtained:

[0058]

[0059] Wherein, d represents the Euclidean distance between the basic vector (output vector) and the reference vector in the aβ coordinate system.

[0060] Further, from expression (6), the Euclidean distance d of the basic vector and the reference vector is equivalent to the deviation between the corresponding current prediction value and the given value. Therefore, the above original value function expression (1) can be rewritten as a new target value function, and the target value function is expressed as follows:

[0061] g=d 2 +λΔU p (k+2) 2 (7)

[0062] Compared with the traditional FCS-MPC strategy, since the reference vector needs to be calculated only once, the calculation amount is greatly reduced.

[0063] After the construction of the target value function is completed, enter step S120. Referring to Figure 1 , step S120 calculates the basic vector related to the reference vector obtained in step S110 to form a simplified control set. In step S120, the three vectors with the closest Euclidean distance to the reference vector in the space vector plane are selected, and the redundant vectors of the three selected nearest vectors are determined, thereby forming a simplified control set composed of a plurality of basic vectors related to the reference vector. Among them, the number of basic vectors in the simplified control set is at most 5.

[0064] Furthermore, in the process of generating the simplified control set, firstly, the three vectors in the space vector plane with the closest Euclidean distance to the reference vector are selected and sorted from largest to smallest; then, based on the difference between the largest and smallest nearest vectors, the three nearest vectors are converted into basic vectors to form a first type of basic vector set, and several redundant vectors related to the three nearest vectors are obtained at the same time, so that the redundant vectors are used as a second type of basic vector set; finally, the simplified control set is generated based on the first type of basic vector (set) and the second type of basic vector (set).

[0065] The above expression (7) shows that in the space vector plane, for any fundamental vector, the effect of current control is equivalent to its Euclidean distance from the reference vector. In traditional FCS-MPC current control, compared with other indicators such as capacitor voltage balance, current control should account for the largest proportion in the value function. Therefore, the fundamental vector closer to the reference vector is more likely to become the optimal fundamental vector.

[0066] Figure 2 This is a schematic diagram of the reference vector in the space vector plane in the predictive control method for a three-level converter according to an embodiment of this application. Figure 2 As shown, when determining the three closest vectors to the reference vector, the three vectors corresponding to the vertices of the triangle containing the reference vector are taken as the three closest vectors. Specifically, when the reference vector is outside the control area, the current reference vector is transformed into a vector with the same phase angle and located on the hexagonal boundary.

[0067] Equation (6) shows that in the space vector plane, for any basic vector, the current control effect is equivalent to its Euclidean distance from the reference vector. In FCS-MPC current control, compared to other indicators such as capacitor voltage balance, current control should account for the largest proportion in the value function. Therefore, the basic vector closer to the reference vector is more likely to be the optimal basic vector. Based on the above analysis, after obtaining the reference vector based on the deadbeat predicted current, only the three basic vectors with the smallest Euclidean distance from the reference vector in the space vector plane and their redundant vectors are selected to construct a simplified control set, i.e., the basic vectors corresponding to the vertices of the small triangle where the reference vector is located. When the reference vector is outside the control region, it is transformed into a vector with the same phase angle and located on the hexagonal boundary. In one embodiment, using... Figure 2 The reference vector v in * For example, the simplified control set is: (1,0,0), (0,-1,-1), (1,0,-1), (1,-1,-1). The calculation method for the simplified control set is as follows:

[0068] Step S1201 (not shown), as follows Figure 2As shown, any vector can be represented by two non-collinear unit vectors (1, 0, 0) and (0, 1, 0). In the aβ coordinate system, the vector (1, 0, 0) corresponds to the vector (u dc / 3, 0), and the vector (0, 1, 0) corresponds to the vector (0, v Therefore, in the aβ coordinate system, any vector can be synthesized by the vector (u dc / 3, 0) and the vector (0, v .

[0069] Step 1202 (not shown), the three nearest vectors of the reference vector v * are derived. Among them, for the reference vector v * is expressed as follows:

[0070]

[0071] where u dc represents the DC bus voltage, represents a given value of the a-axis output voltage, represents a given value of the β-axis output voltage. According to expression (8), the three nearest vectors v1, v2, v3 of v * are obtained; according to the volt-second balance formula, the action times of the three nearest vectors are t1, t2, and t3, respectively, and the following relationship exists:

[0072]

[0073] In the abc coordinate system, expression (8) is converted to:

[0074]

[0075] where

[0076] 1) If e ≥ f, expression (10) is converted to the form of expression (9), that is:

[0077]

[0078] In this case, the three nearest vectors are (x, y, 0), (x + 1, y, 0), and (x + 1, y + 1, 0).

[0079] 2) If e < f, expression (10) is converted to:

[0080]

[0081] In this case, the three nearest vectors are (x, y, 0), (x, y + 1, 0), and (x + 1, y + 1, 0).

[0082] According to the above derivation, the reference vector v * The three nearest vectors v1, v2, v3 of the reference vector v

[0083] Table 1 nearest three vectors

[0084] e ≥ f e < f <![CDATA[v2]]> (x, y, 0) (x, y, 0) (x, y, 0) <![CDATA[v2]]> (x + 1, y, 0) (x, y + 1, 0) <![CDATA[v3]]> (x + 1, y + 1, 0) (x + 1, y + 1, 0)

[0085] Wherein, the parameters x, y, e, f are respectively:

[0086]

[0087] For any reference vector, after calculating the parameters x, y, e and f, the three nearest vectors can be determined without other complex operations;

[0088] Step 1203 (not shown in the figure), determine the redundant vectors of the three nearest vectors.

[0089] Since the obtained nearest vectors are not necessarily basic vectors, and the redundant vectors are not considered, further processing is needed. For the nearest vectors obtained in Table 1, first sort the elements of the three nearest vectors, and let the maximum value be max and the minimum value be min. Specifically, if the difference between the maximum value and the minimum value is 2, it represents a large vector or a medium vector, and there is no redundant vector, and after subtracting min+1 from the three nearest vector elements, the three nearest vectors can be converted into basic vectors; if the difference between the maximum value and the minimum value is 1, it represents a small vector, and after subtracting min+1 or min from the three nearest vector elements, two redundant basic vectors can be obtained; if the difference between the maximum value and the minimum value is 0, it represents a zero vector, and since the zero vector has no effect on the midpoint potential, only (0, 0, 0) is used as a redundant vector.

[0090] Reference Figure 2 For any reference vector, based on the different number of redundant vectors, the number of basic vectors in the simplified control set is not more than 5. Compared with the original FCS-MPC control strategy using the original control set containing 27 voltage vectors, the simplified control set used in the embodiment of the application significantly reduces the optimization space and reduces the number of iterations in a single control period.

[0091] Reference Figure 1 After generating the simplified control set, step S130 is entered. Step S130 obtains the optimized virtual vector and the optimized dead-time corresponding to each basic vector according to the relative position relationship between each basic vector in the simplified control set constructed in step S120 and the dead-zone voltage vector corresponding to each basic vector.

[0092] According to expressions (6) and (7), if only current control is considered, the optimal basic vector selected by the traditional strategy is the basic vector with the smallest Euclidean distance from the reference vector. Figure 3 This is a schematic diagram illustrating the relationship between the reference vector position and the optimal basic vector in the predictive control method for a three-level converter according to an embodiment of this application. Figure 2 Taking the shaded triangle in the image as an example, as Figure 3 As shown, when the reference vector is in region ① of the small triangle, the nearest fundamental vector is V1; similarly, region ② corresponds to V2; and region ③ corresponds to V3. During steady-state operation, the maximum Euclidean distance between the output vector and the reference vector is a constant. In expression (6), the maximum value of the difference between the output current and the reference current corresponding to the optimal basic vector is also a constant, independent of the magnitude of the reference current. When a multi-vector control strategy is adopted, the maximum value of the Euclidean distance between the output vector and the reference vector can be reduced, and the ripple amplitude of the output current will also decrease accordingly. However, existing multi-vector strategies use multiple basic vectors, which leads to an increase in the number of switching operations. To address this, this embodiment of the invention uses a dead-time voltage vector to implement a dual-vector strategy, which not only does not increase the number of switching operations but also provides a method for calculating the optimal dead time to optimize the fixed dead time in traditional control strategies.

[0093] Figure 4 This diagram illustrates the positions of the basic vector, dead-zone voltage vector, and reference vector in the predictive control method for a three-level converter according to an embodiment of this application. Analyzing the impact of the dead-zone voltage vector on the output steady-state performance reveals that two vectors exist in each control cycle: the basic vector v in the control set. i The dead zone voltage vector v corresponding to this basic vector. id Because of the fundamental vector v i and dead zone voltage vector v id The total duration of action is a fixed value T. s Therefore, according to the volt-second balance formula, the virtual vector resulting from the synthesis of the two vectors lies on the line connecting their endpoints. Thus, the corresponding dead-zone voltage vector can be determined based on each fundamental vector, further referencing... Figure 4 Through the basic vector v i Dead zone voltage vector v id and the reference vector v * The positional relationship allows for analysis of the impact of the dead-time process on output performance, as shown in Table 2.

[0094] Table 2. Impact of Dead Zone on Output Performance

[0095]

[0096] As can be seen from Table 2, by optimizing the dead time, vi and v id It is possible to synthesize a result closer to v * Virtual vector.

[0097] Further, in step S130, the optimized virtual vector and optimized dead time corresponding to each basic vector in the simplified control set are obtained according to the following steps. Specifically, first, it is determined whether a dead time process exists; then, each basic vector with a dead time process is synthesized with the corresponding dead time voltage vector to obtain the optimized virtual vector with the smallest Euclidean distance from the reference vector, thereby obtaining the optimized virtual vector v corresponding to each basic vector. io At the same time, the optimization dead time t of each optimized virtual vector is obtained. id Finally, the basic vector of the process without dead zone is directly used as its corresponding optimized virtual vector, and the optimized dead time of the basic vector of the process without dead zone is set to zero.

[0098] Furthermore, for each basic vector exhibiting a dead-zone process, each basic vector is compared with its corresponding dead-zone voltage vector. Based on the comparison results, the optimized virtual vector and optimized dead-zone time corresponding to the current basic vector are calculated. Specifically, if the current basic vector and its corresponding dead-zone voltage vector are identical, the current basic vector is directly used as its corresponding optimized virtual vector, and the minimum of the action times of the current basic vector and its corresponding dead-zone voltage vector is taken as the optimized dead-zone time corresponding to the current basic vector.

[0099] Figure 5 This is a schematic diagram of the relative positional relationship of vectors in the predictive control method for a three-level converter according to an embodiment of this application. If the current basic vector is different from its corresponding dead-zone voltage vector, the optimized virtual vector v corresponding to the current basic vector needs to be adjusted according to the following table. io And optimize dead time t id Analysis:

[0100] Conditions Optimizing virtual vector v io ]] Optimizing dead time t id ]]> i i r i t min T s ]]> t min / T s *v i +(1-t min / T s )*v id ]]> [CAT s -t min ]]> l i / r i ≥1-t min / T s ]]> (1-t min / T s )*v i +t min / T s *v id ]]> <![CDATA[t min ]]> Otherwise i i / r i * v i + (1 - l i / r i ) * v id ]]> T s (1-l i / r i )]]>

[0101] Where i represents the index of the basic vector of the dead-zone process, r i This represents the distance between the endpoint of the i-th fundamental vector and the endpoint of the dead-zone voltage vector of the i-th fundamental vector (reference). Figure 5 ), l i The projection onto the line connecting the endpoint of the i-th fundamental vector and the endpoint of the dead-zone voltage vector of the i-th fundamental vector, or the distance between this projection and the endpoint of the dead-zone voltage vector of the i-th fundamental vector (see reference). Figure 5 ), t min T represents the minimum time of action of the endpoint of the i-th fundamental vector and the dead zone voltage vector of the i-th fundamental vector.s denotes the unit control period, v i denotes the i-th basic vector, v id denotes the dead-zone voltage vector of the i-th basic vector.

[0102] Specifically, for any basic vector in the control set, if there is a dead-zone process, by reasonably setting the dead-zone time, v i and v id are synthesized to obtain the virtual vector v io with the minimum Euclidean distance to the reference vector; if there is no dead-zone process, equivalent to t id is 0, v io is v i Therefore, by optimizing the dead-zone time, the control set is converted into a new virtual vector control set;

[0103] When there is a dead-zone process, according to the relationship between v i and v id , the solution of v io can be divided into two cases (considering that the action time of v i and v id both has a minimum value t min limitation):

[0104] 1) v i and v id are the same: although there is a dead-zone, it has no effect on the output, the dead-zone time t id is t min , v io is v i ,

[0105] 2) v i and v id are different: the positional relationship of v * , v i and v id can be divided into two categories:

[0106] (a) the projection of v * on the line connecting v i and v id ;

[0107] (b) the projection of v * on the extension line of the line connecting v i and v id ;

[0108] Let the distance between the endpoints of v i and v id be r i ; the projection of v * on the line or extension line connecting v i and v id and vid The distance between the endpoints is l i ; when v id belongs to the control set, the position relationship of the three is Figure 5 (a), at this time l i ≤ r i ; when v id does not belong to the control set, v id cannot be closer to v i than v * , the position relationship of the three is Figure 5 (b), at this time l i > r i ; according to the relationship between r i and l i , the analytical solution of v io and the corresponding t id is as follows:

[0109] When l i / r i ≤ t min / T s , v io is t min / T s *v i +(1-t min / T s )*v id , t id is T s -t min ;

[0110] When l i / r i ≥ 1-t min / T s , v io is (1-t min / T s )*v i +t min / T s *v id , t id is t min ;

[0111] In other cases, v io is l i / r i *v i +(1-l i / r i )*v id , t id is T s (1-l i / r i ).

[0112] Therefore, the embodiment of the application simplifies the control set into a new virtual vector control set by calculating the optimized virtual vector and the optimized dead-time corresponding to each basic vector, and then enters step S140.

[0113] As shown in step S140, the optimal virtual vector is obtained by using the target value function according to each basic vector and the corresponding optimized virtual vector, and then the three-level converter to be controlled is controlled by using the optimized dead-time corresponding to the optimal virtual vector. Figure 1

[0114] In step S140, first, the optimized virtual vector corresponding to each basic vector is substituted into the Euclidean distance item in the target function, and each basic vector and the corresponding dead-zone voltage vector and the optimized dead-time are substituted into the DC side capacitor voltage balance control item in the target value function, and the target evaluation value of each basic vector is calculated; then, according to the minimum value of the target evaluation value of each basic vector, the optimal virtual vector and the optimal dead-time corresponding to the optimal virtual vector are selected.

[0115] Specifically, for the above virtual vector control set, the optimal virtual vector is selected by the target value function based on the Euclidean distance in combination with the DC side capacitor voltage balance control, and the control waveform corresponding to the optimal virtual vector is output.

[0116] The optimal virtual vector is selected from the virtual vector control set by the target value function expression (7), wherein the analytical expression of d is:

[0117] d=||v * -v io ||2

[0118] At the same time, the DC side capacitor voltage balance control item ΔU p (k+2) in the target value function for realizing the DC side capacitor voltage balance control is calculated by using the following expression:

[0119]

[0120] Wherein, i o1 (k+1) and i o2 (k+1) respectively represent the midpoint current value calculated according to the control amount of the dead-zone voltage vector and the basic voltage vector in the k+1 control period.

[0121] Based on the above predictive control method, the embodiment of the application further provides a control system for a three-level converter.

[0122] Figure 6 ​It is a schematic diagram of the framework structure of the control system for the three-level converter according to the embodiments of the present application. As shown in Figure 6 the control system according to the embodiments of the present application comprises a three-level converter and a processor. The processor is connected to the main circuit of the three-level converter. The processor is used to implement the predictive control method as described above to control the three-level converter.

[0123] The present application discloses a predictive control method and a control system for a three-level converter. The control strategy adopted by the method and the system has the following advantages or beneficial effects:

[0124] (1) Optimizing the value function and effectively reducing the iterative calculation amount: the Euclidean distance between the basic vector and the reference vector is adopted to replace the current prediction deviation value in the traditional value function, thereby avoiding the complex current prediction model calculation;

[0125] (2) Simplifying the control set and significantly reducing the optimization space: the reference vector is obtained according to the deadbeat prediction current, and a general algebraic method is proposed to obtain the nearest three basic vectors and their redundant vectors of the reference vector, so as to construct a simplified control set;

[0126] (3) Considering the dead zone effect and constructing a virtual vector control set: the dead zone voltage vector corresponding to the basic vector in the simplified control set is analyzed one by one; and the dead zone time is optimized, and the virtual vector with the minimum Euclidean distance from the reference vector is obtained by combining the two, thereby constructing a virtual vector control set;

[0127] (4) Realizing double vector control and improving control performance: the optimal virtual vector is selected based on the value function, and the corresponding control waveform is output, so that the control performance is improved, and at the same time the number of switching times is basically unchanged.

[0128] The above description is only the preferred specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be limited by the protection scope of the claims.

[0129] It should be understood that the disclosed embodiments of the present application are not limited to the specific structures, processing steps or materials disclosed herein, but should extend to equivalent alternatives of these features understood by those skilled in the relevant art. It should also be understood that the terms used herein are only for the purpose of describing the specific embodiments and do not mean limitation.

[0130] Reference in the specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. The appearances of the phrase "in one embodiment" or "an embodiment" in various places in the specification are not necessarily all referring to the same embodiment.

[0131] Although the present application has been described with reference to the above embodiments, the contents described are merely adopted embodiments for facilitating the understanding of the present application, and are not intended to limit the present application. Any modification and change in the form and details of the embodiments can be made by any person skilled in the art without departing from the spirit and scope of the present application, and the patent protection scope of the present application shall be subject to the scope defined by the appended claims.

Claims

1. A predictive control method for a three-level converter, characterized in that, The method comprises the following steps: a reference vector is obtained based on a zero-error prediction of an output current of a three-level converter, and a target value function is established by using a Euclidean distance between the reference vector and an output vector; basic vectors related to the reference vector are calculated to form a simplified control set; an optimized virtual vector and an optimized dead-time corresponding to each basic vector in the simplified control set are obtained according to a relative positional relationship between each basic vector and a corresponding dead-zone voltage vector; an optimal virtual vector is obtained by using the target value function according to each basic vector and the corresponding optimized virtual vector, so that the three-level converter is controlled by using the optimized dead-time corresponding to the optimal virtual vector.

2. The predictive control method of claim 1, wherein, In the process of establishing the target value function, the following steps are included: a value function of a traditional FCS-MPC strategy is obtained, which is denoted as an original value function; the Euclidean distance between the reference vector and the output vector is equivalent to a deviation term of the original value function between an output current prediction and a given current, so that the target value function is formed.

3. The predictive control method according to claim 1 or 2, characterized in that, In the step of calculating the basic vectors related to the reference vector to form the simplified control set, the following steps are included: three vectors closest to the reference vector in a space vector plane are selected, and redundant vectors of the three closest vectors are determined to form the simplified control set composed of a plurality of basic vectors related to the reference vector, wherein the number of the basic vectors in the simplified control set is at most 5.

4. The predictive control method according to claim 3, wherein the three closest vectors are sorted; the three closest vectors are converted into basic vectors to form a first type of basic vectors according to a difference between a largest closest vector and a smallest closest vector, and a plurality of redundant vectors of the three closest vectors are obtained to serve as a second type of basic vectors; the simplified control set is generated according to the first type of basic vectors and the second type of basic vectors.

5. The predictive control method according to any one of claims 1 to 4, characterized in that, In the step of obtaining the optimized virtual vector and the optimized dead-time corresponding to each basic vector according to a relative positional relationship between each basic vector and a corresponding dead-zone voltage vector in the simplified control set, the following steps are included: it is judged whether there is a dead-zone process; each basic vector with the corresponding dead-zone voltage vector is synthesized to obtain an optimized virtual vector with a minimum Euclidean distance from the reference vector, and further to obtain an optimized dead-time of each optimized virtual vector; a basic vector without a dead-zone process is directly used as a corresponding optimized virtual vector, and an optimized dead-time of the basic vector without the dead-zone process is set to zero.

6. The predictive control method of claim 5, wherein, The optimized virtual vector and the optimized dead-time corresponding to the current basic vector are calculated according to a comparison result, wherein if the same, the current basic vector is directly used as the corresponding optimized virtual vector, and a minimum value of an action time of the current basic vector and the corresponding dead-zone voltage vector is used as the corresponding optimized dead-time.

7. The predictive control method of claim 6, wherein, if different, the optimized virtual vector and the optimized dead-time corresponding to the current basic vector are analyzed according to the following table: wherein, i a serial number of a basic vector indicating presence of a dead-time process, a terminal point of the i th basic vector, i a distance between the terminal point of the th basic vector and a terminal point of a dead-time voltage vector of the i th basic vector, i a projection on a line or an extension line between the terminal point of the i th basic vector and the terminal point of the dead-time voltage vector of the th basic vector, i a minimum value of an action time of the i th basic vector and an action time of the dead-time voltage vector of the th basic vector, a unit control period, i a th basic vector, i a dead-time voltage vector of the th basic vector.

8. The predictive control method according to any one of claims 1 to 7, characterized in that, In the step of obtaining the optimal virtual vector by using the target value function according to each basic vector and the corresponding optimal virtual vector, the step comprises: substituting the optimal virtual vector corresponding to each basic vector into the Euclidean distance term in the target value function, and substituting each basic vector, the corresponding dead-zone voltage vector and the optimal dead-zone time into the direct-current side capacitor voltage balance control term in the target value function to calculate the target evaluation value of each basic vector; selecting the optimal virtual vector and the corresponding optimal dead-zone time.

9. The predictive control method according to claim 3 or 4, characterized in that, the three nearest vectors are three vectors corresponding to the vertices of a triangle in which the reference vector is located, wherein when the reference vector is outside the control region, the reference vector is converted into a vector with the same phase angle and located on the boundary of the hexagon.

10. A control system for a three-level converter, characterized in that The system comprises: a three-level converter; and a processor connected to a main circuit of the three-level converter, wherein the processor is configured to implement the predictive control method according to any one of claims 1 to 9 to control the three-level converter.

Images