Model-free predictive control method and system for T-type three-level energy storage device

Abstract

The present application provides a model-free predictive control method and system for a T-type three-level energy storage device, the method comprising: splitting the reference voltage vector output by the energy storage device into a first voltage vector component and a second voltage vector component by a split transformation method; obtaining the non-repeating first voltage vector component and the second voltage vector component as the standard voltage vector component; obtaining three applied voltage vectors of the energy storage device and splitting them into applied voltage vector components; calculating the current gradient corresponding to the applied voltage vector component by a current gradient calculation method, and recording the current gradient as the current gradient corresponding to the standard voltage vector component that is the same as the applied voltage vector component. In the method, the voltage vector output by the T-type three-level energy storage device is split into voltage vector components, and only the current gradient corresponding to the non-repeating voltage vector component is calculated, thereby reducing the amount of calculation and improving the output power quality and dynamic response speed of the energy storage device.

Description

Technical Field

[0001] The present application relates to the field of new energy technology, and in particular to a model-free predictive control method and system for a T-type three-level energy storage device. Background Art

[0002] Energy storage devices can convert direct current into alternating current, which can be used in renewable energy, electric vehicles, computer equipment, industrial production and other fields. However, the output of energy storage devices fluctuates greatly, and control strategies are needed to reduce current ripple and total harmonic distortion.

[0003] As a control strategy, the model predictive control method has the advantages of high-speed dynamic response, multi-object processing capability and simple implementation. However, since the model predictive control method is based on the actual physical model of the energy storage device, when the model parameters change, such as temperature rise leading to increased loop impedance and inductor saturation, the control accuracy will be reduced. As another control strategy, the model-free predictive control method does not need to be based on the actual physical model, but calculates the current gradient based on the voltage vector to achieve current prediction. Therefore, the impact of model parameters on the control strategy can be reduced.

[0004] However, the model-free predictive control method needs to calculate the current gradient based on the applied voltage vector, which increases the amount of calculation for energy storage devices with multiple voltage vectors. In addition, the current gradient cannot be calculated based on the unapplied voltage vector, which leads to current control lag, increases current ripple and current total harmonic distortion, and reduces the output power quality and dynamic response speed of the energy storage device. Summary of the invention

[0005] The present application provides a model-free predictive control method and system for a T-type three-level energy storage device to improve the output power quality and dynamic response speed of the energy storage device.

[0006] In a first aspect, the present application provides a model-free predictive control method for a T-type three-level energy storage device, comprising:

[0007] Splitting the reference voltage vector output by the T-type three-level energy storage device into a first voltage vector component and a second voltage vector component by a split transformation method;

[0008] Acquire a non-repeating first voltage vector component in the first voltage vector component and a non-repeating second voltage vector component in the second voltage vector component as standard voltage vector components;

[0009] Acquire a first applied voltage vector, a second applied voltage vector and a third applied voltage vector of the T-type three-level energy storage device within a control period;

[0010] Splitting the first applied voltage vector, the second applied voltage vector and the third applied voltage vector into applied voltage vector components by a split transformation method; wherein the standard voltage vector component includes the applied voltage vector component;

[0011] The current gradient corresponding to the applied voltage vector component is calculated by a current gradient calculation method, and the current gradient is recorded as the current gradient corresponding to the standard voltage vector component which is the same as the applied voltage vector component.

[0012] In the method, the voltage vector output by the T-type three-level energy storage device is split into voltage vector components, and only the current gradients corresponding to the non-repetitive voltage vector components are calculated, thereby reducing the amount of calculation and improving the output power quality and dynamic response speed of the energy storage device.

[0013] In an implementation manner of the first aspect, recording the current gradient as a current gradient corresponding to a standard voltage vector component that is the same as the applied voltage vector component includes:

[0014] A current gradient lookup table is created, the current gradient lookup table including a standard voltage vector component, a current gradient at a first moment corresponding to the standard voltage vector component, and a current gradient at a second moment corresponding to the standard voltage vector component, wherein the first moment is a previous predicted time point of the second moment.

[0015] In the method, the standard voltage vector component, the current gradient at the first moment and the current gradient at the second moment are filled into the current gradient lookup table, so that the current gradient lookup table can be quickly updated and the current prediction accuracy can be improved.

[0016] In an implementation manner of the first aspect, the current gradient at the first moment corresponding to the standard voltage vector component is zero.

[0017] In the method, the current gradient of the T-type three-level energy storage device is zero during initialization, and zero is filled into the current gradient lookup table as the current gradient at the first moment corresponding to the standard voltage vector component, so as to facilitate the subsequent calculation of the predicted current value.

[0018] In an implementation manner of the first aspect, the method further includes:

[0019] Subtracting the current gradient corresponding to the unapplied voltage vector component at the first moment from the current gradient corresponding to the applied voltage vector component at the first moment to obtain an intermediate value of the current gradient;

[0020] Adding the current gradient corresponding to the applied voltage vector component at the second moment to the intermediate value of the current gradient to obtain the current gradient corresponding to the unapplied voltage vector component at the second moment;

[0021] The current gradient corresponding to the non-acting voltage vector component at the second moment is recorded as the current gradient at the second moment corresponding to the standard voltage vector component that is the same as the non-acting voltage vector component.

[0022] In the method, when there is an unavailable unacted voltage vector in the T-type three-level energy storage device, the current gradient corresponding to the unacted voltage vector at the second moment can be calculated by using the current gradients corresponding to the applied voltage vector components at the first moment and the second moment, and the current gradient corresponding to the unacted voltage vector at the first moment, so as to obtain the current gradients corresponding to all standard voltage vector components, thereby improving the real-time performance of the current gradient update and further improving the dynamic response of the energy storage device.

[0023] In an implementation manner of the first aspect, the method further includes:

[0024] Dividing the reference voltage vector into a zero vector, a small vector, a medium vector and a large vector, wherein the number of the zero vectors is a first number, the number of the small vectors is a second number, the number of the medium vectors is a third number, and the number of the large vectors is a fourth number;

[0025] The first number, the second number, the third number and the fourth number are related to the magnitude value of the reference voltage vector; the third number is equal to the fourth number, the third number or the fourth number is less than the second number, and the third number or the fourth number is greater than the first number;

[0026] The reference voltage vector is divided into a first vector group according to a first division standard, wherein the first division standard is to take a large vector, a small vector and a zero vector having the smallest amplitude difference with a middle vector and the middle vector as one first vector group.

[0027] In an implementation manner of the first aspect, the method further includes:

[0028] Obtaining a first inductor current value component and a second inductor current value component at a second moment corresponding to the center vector;

[0029] Searching in the current gradient lookup table a first current gradient at a second moment corresponding to a first standard voltage vector component of a mid vector and a second current gradient at a second moment corresponding to a second standard voltage vector component;

[0030] The first inductor current value component is added to the first current gradient to obtain a first predicted current value at a third moment corresponding to the first standard voltage vector component of the center vector; the second inductor current value component is added to the second current gradient to obtain a second predicted current value at a third moment corresponding to the second standard voltage vector component of the center vector, wherein the third moment is a subsequent predicted time point of the second moment;

[0031] Performing a first operation on a first reference current value at a third moment corresponding to a first standard voltage vector component of the center vector and a first predicted current value to obtain a first intermediate value; performing a first operation on a second reference current value at a third moment corresponding to a second standard voltage vector component of the center vector and a second predicted current value to obtain a second intermediate value;

[0032] Adding the first intermediate value and the second intermediate value to obtain a cost function corresponding to the mid vector;

[0033] The first vector group where the middle vector with the minimum cost function is located is taken as the selected first vector group.

[0034] In the method, the reference voltage vector of the T-type three-level energy storage device is divided into a third number of first vector groups according to the amplitude value, each first vector group has a middle vector, the cost function of each middle vector is calculated by a cost function calculation method, and the first vector group where the middle vector with the smallest cost function is located is selected as the first vector group, thereby improving the rate of generating the drive control signal.

[0035] In an implementation manner of the first aspect, the method further includes:

[0036] Calculating the cost functions corresponding to the large vector, the small vector and the zero vector in the selected first vector group to obtain the cost functions of all the reference voltage vectors in the selected first vector group;

[0037] The first vector group is divided into second vector groups according to a second division criterion, wherein the second division criterion is to select a reference voltage vector with the smallest amplitude value difference in the first vector group as one second vector group;

[0038] The second vector group with the smallest cost function is determined as the optimal vector group.

[0039] In the method, the cost function of the reference voltage vector in each second vector group in the selected first vector group is calculated, and the second vector group with the smallest sum of cost functions is determined as the optimal vector group, thereby improving the control accuracy of the drive control signal.

[0040] In an implementation manner of the first aspect, the method further includes:

[0041] Obtaining a cost function corresponding to a reference voltage vector in an optimal vector group;

[0042] A second operation is performed on the cost function and the control period to obtain the action time of the reference voltage vector in the optimal vector group.

[0043] In the method, after the optimal vector group is determined, the action time of the reference voltage vector in the optimal vector group is calculated by a second operation to generate a driving control signal.

[0044] In an implementation manner of the first aspect, the method further includes:

[0045] Acquire a first capacitor voltage and a second capacitor voltage of a T-type three-level energy storage device;

[0046] Determine a vector sequence of the optimal vector group according to the first capacitor voltage and the second capacitor voltage, wherein when the first capacitor voltage is greater than the second capacitor voltage, the vector sequence of the optimal vector group is the first voltage vector, the second voltage vector, the third voltage vector, the second voltage vector, and the first voltage vector; when the first capacitor voltage is less than the second capacitor voltage, the vector sequence of the optimal vector group is the third voltage vector, the fourth voltage vector, the fifth voltage vector, the fourth voltage vector, and the third voltage vector;

[0047] The driving control signal of the T-type three-level energy storage device is determined according to the reference voltage vector of the optimal vector group, the action time of the reference voltage vector and the vector sequence.

[0048] In a second aspect, the present application provides a model-free predictive control system for a T-type three-level energy storage device, comprising:

[0049] An energy storage module is configured to obtain an inductor current value, an applied voltage vector, a first capacitor voltage, and a second capacitor voltage;

[0050] A vector splitting module configured to split the voltage vector into a first voltage vector component and a second voltage vector component;

[0051] a current gradient calculation module, configured to calculate a current gradient based on the first voltage vector component and the second voltage vector component;

[0052] An optimal vector group determination module is configured to determine a cost function according to the inductor current value, the current gradient and the reference current value, and determine an optimal vector group according to the cost function;

[0053] The vector sequence determination module determines the vector sequence of the optimal vector group according to the magnitude relationship between the first capacitor voltage and the second capacitor voltage;

[0054] The drive control signal generating module is configured to calculate the action time according to the cost function, and determine the drive control signal of the T-type three-level energy storage device according to the reference voltage vector of the optimal vector group, the action time of the reference voltage vector and the vector sequence.

[0055] It can be seen from the above technical solutions that the present application provides a model-free predictive control method and system for a T-type three-level energy storage device, the method comprising: splitting the reference voltage vector output by the energy storage device into a first voltage vector component and a second voltage vector component by a split transformation method; obtaining the non-repeating first voltage vector component and the second voltage vector component as the standard voltage vector component; obtaining three applied voltage vectors of the energy storage device and splitting them into applied voltage vector components; calculating the current gradient corresponding to the applied voltage vector component by a current gradient calculation method, and recording the current gradient as the current gradient corresponding to the standard voltage vector component that is the same as the applied voltage vector component. In the method, the voltage vector output by the T-type three-level energy storage device is split into voltage vector components, and only the current gradient corresponding to the non-repeating voltage vector component is calculated, thereby reducing the amount of calculation and improving the output power quality and dynamic response speed of the energy storage device. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0057] Figure 1 This is a schematic diagram of the structure of a T-type three-level topology circuit as an energy storage device described in an embodiment of the present application;

[0058] Figure 2 This is a flow chart of the model-free predictive control method described in an embodiment of the present application;

[0059] Figure 3 This is a flow chart of a method for calculating a current gradient corresponding to an unacted voltage vector described in an embodiment of the present application;

[0060] Figure 4 This is a flow chart of the first vector group division method described in an embodiment of the present application;

[0061] Figure 5 A voltage vector distribution diagram of a T-type three-level topology circuit according to an embodiment of the present application;

[0062] Figure 6 A flow chart of a method for determining a selected first vector group according to an embodiment of the present application;

[0063] Figure 7 This is a flow chart of a method for determining an optimal vector group according to an embodiment of the present application;

[0064] Figure 8 1 is an overall block diagram of the model-free predictive control method for a T-type three-level energy storage device described in an embodiment of the present application. DETAILED DESCRIPTION

[0065] The following embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following embodiments do not represent all implementations consistent with the present application. They are only examples of systems and methods consistent with some aspects of the present application as detailed in the claims.

[0066] It should be noted that the brief description of terms in this application is only for the convenience of understanding the embodiments described below, and is not intended to limit the embodiments of this application. Unless otherwise specified, these terms should be understood according to their ordinary and common meanings.

[0067] The terms "first", "second", "third", etc. in the specification and claims of this application and the above drawings are used to distinguish similar or similar objects or entities, and do not necessarily mean to limit a specific order or sequence, unless otherwise noted. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances.

[0068] The main function of an energy storage device is to convert direct current (DC) into alternating current (AC) to adapt to various power needs and application scenarios. In the field of renewable energy, such as solar and wind power generation, energy storage devices can store these intermittent energy sources and provide stable power output when needed. For electric vehicles, energy storage devices can store electrical energy for use when the vehicle is driving. In addition, in computer equipment and industrial production, energy storage devices can provide a stable power supply for equipment to ensure the normal operation of the equipment. However, the output characteristics of energy storage devices will be accompanied by fluctuations, affecting the stability of the power system and the operating efficiency of the equipment. Therefore, a control strategy is needed to reduce the current ripple and the total harmonic distortion rate of the current to ensure that the energy storage device can operate safely and efficiently.

[0069] In some embodiments, the model predictive control method can be used as a control strategy for the energy storage device. The model predictive control method calculates the corresponding current gradient based on the applied voltage vector and creates a current gradient lookup table to achieve current predictive control and ensure a stable supply of electricity. The model predictive control method has multi-object processing capabilities and can handle the control requirements of multiple energy storage devices or power equipment at the same time, improving the overall performance of the control system. However, the model predictive control method has certain limitations. Since it is based on the actual physical model of the energy storage device, when the model parameters change, such as when the temperature rises, causing the loop impedance to increase and the inductor to saturate, its control accuracy may be affected.

[0070] In order to overcome these limitations, in some embodiments, a model-free predictive control method can be used. The model-free predictive control method does not rely on the actual physical model, but calculates the current gradient based on the voltage vector to achieve current prediction. This method can reduce the impact of model parameters on the control strategy and improve the robustness and adaptability of the control system.

[0071] However, the model-free predictive control method needs to calculate the current gradient based on the applied voltage vector, which is too computationally intensive for energy storage devices with multiple voltage vectors. In addition, since the current gradient cannot be calculated based on the unapplied voltage vector, the current control has hysteresis, which increases the current ripple and total harmonic distortion rate, and reduces the output power quality and dynamic response speed of the energy storage device.

[0072] The present application embodiment provides a model-free predictive control method for a T-type three-level energy storage device, which can improve the output power quality and dynamic response speed of the energy storage device. Figure 1 The T-type three-level topology circuit in the embodiment is used as the main body of the energy storage device, and the technical solution of the present application is introduced in detail.

[0073] like Figure 2 As shown, the model-free predictive control method specifically includes the following steps:

[0074] Step S201 : splitting a reference voltage vector output by a T-type three-level energy storage device into a first voltage vector component and a second voltage vector component by a split transformation method.

[0075] In some embodiments, Figure 1 As shown, the T-type three-level topology circuit consists of three-phase bridge arms a, b, and c, and each phase bridge arm consists of S x1 , S x2 , S x3 and S x4 It consists of four switch tubes. Among them, x=a, b, c, S x1 , S x3 Switching states are complementary, S x2 and S x4 The switch states are complementary. The first capacitor C 1 The voltage is V p , the second capacitor C 2 The voltage is V n , when V p and V n When the equilibrium state is reached, the output voltage of each phase of the topological circuit is as shown in formula (1):

[0076]

[0077] Among them, V dc is the DC side voltage, Sx is the switch state of the bridge arm, including -1, 0, and 1. The four switch tube states and output voltages corresponding to each bridge arm state are shown in the following table, where x = a, b, c:

[0078] <![CDATA[S x ]]> <![CDATA[S x1 ]]> <![CDATA[S x2 ]]> <![CDATA[S x3 ]]> <![CDATA[S x4 ]]> <![CDATA[V xo ]]> 1(P) 1 1 0 0 <![CDATA[V dc / 2]]> 0(O) 0 1 1 0 0 -1(N) 0 0 1 1 <![CDATA[-V dc / 2]]>

[0079] It can be seen that each phase of the T-type three-level topology circuit has three switching states, so there are 27 switching states for the three phases, corresponding to 27 reference voltage vectors V 0 -V 26 .

[0080] In some embodiments, the split transformation method is Clark transformation. Clark transformation can transform a three-phase current vector or a three-phase voltage vector into two equivalent two-phase current vectors or two-phase voltage vectors.

[0081] For example, in a T-type three-level topology circuit, 27 reference voltage vectors V 0 -V 26 The Clark transformation method is used to split the voltage vector into 27 components V xα and 27 second voltage vector components V xβ , where x=0, 1, ..., 26.

[0082] Step S202 : acquiring non-repeating first voltage vector components in the first voltage vector components and non-repeating second voltage vector components in the second voltage vector components as standard voltage vector components.

[0083] In some embodiments, 27 reference voltage vectors V 0 -V 26 The 27 first voltage vector components and the 27 second voltage vector components obtained by the Clark transform method all have repeated values. In order to reduce the amount of calculation and improve the working efficiency of the T-type three-level energy storage device, it is only necessary to obtain non-repeating first voltage vector components and non-repeating second voltage vector components for calculation.

[0084] Exemplarily, the first voltage vector components obtained by splitting 27 reference voltage vectors of the T-type three-level topology circuit include And so on 9 non-repeating first voltage vector components.

[0085] The second voltage vector components obtained by splitting the 27 reference voltage vectors of the T-type three-level topology circuit include And so on for 5 non-repeating second voltage vector components.

[0086] Step S203 , obtaining a first applied voltage vector, a second applied voltage vector and a third applied voltage vector of the T-type three-level energy storage device within a control period.

[0087] In some embodiments, the current gradient may be calculated by acquiring the applied voltage vector in the current control cycle, and then the predicted current value of the next control cycle may be predicted based on the current gradient.

[0088] For example, in the current control cycle, the first applied voltage vector V at time k-1 is obtained. 1 (k-1), the second applied voltage vector V 2 (k-1) and the third applied voltage vector V 3 (k-1).

[0089] Step S204: split the first applied voltage vector, the second applied voltage vector and the third applied voltage vector into applied voltage vector components by using a split transformation method.

[0090] Exemplarily, the first applied voltage vector V 1 (k-1), the second applied voltage vector V 2 (k-1) and the third applied voltage vector V 3 (k-1) is split into 6 applied voltage vector components V by Clark transformation 1α (k-1), V 1β (k-1), V 2α (k-1), V 2β (k-1), V 3α (k-1) and V 3β (k-1).

[0091] It can be understood that the 6 applied voltage vector components obtained by the above splitting are values ​​among the 14 standard voltage vector components.

[0092] Step S205 , calculating the current gradient corresponding to the applied voltage vector component by a current gradient calculation method, and recording the current gradient as the current gradient corresponding to the standard voltage vector component that is the same as the applied voltage vector component.

[0093] In some embodiments, the output voltage V o Calculate the current gradient and output voltage V o The calculation method of is shown in formula (2):

[0094]

[0095] Among them, v ao 、v bo and v coThey represent the output voltage vector of each phase in the T-type three-level topology circuit, and γ represents a phase lag of 120 degrees.

[0096] In some embodiments, before calculating the current gradient, the filter inductor circuit equation in the T-type three-level topology circuit is first calculated, as shown in equation (3):

[0097]

[0098] Among them, i αβ are the α and β components of the inductor current; L f for Figure 1 The filter inductance value shown; V αβ are the α and β components of the applied voltage vector, V oαβ are the α and β components of the output voltage.

[0099] Secondly, the forward Euler method is used to discretize equation (3), and the result is shown in equation (4):

[0100]

[0101] Where k represents the current sampling time; k-1 represents the previous sampling time; i xαβ (k) represents the α and β components of the inductor current value at time k; i xαβ (k-1) represents the α and β components of the inductor current value at time k-1; V xαβ (k-1) represents the α and β components of the x-th applied voltage vector at the previous sampling moment, x = 1, 2, ..., 27; V oαβ (k-1) represents the α and β components of the output voltage at the previous sampling moment; Δi xαβ (k-1) represents the current gradient from time k-1 to time k corresponding to the α and β components of the x-th applied voltage vector; T s Represents the sampling period, that is, the control period.

[0102] Since the filter inductance value L is included in formula (4) f , which causes the model equation to be affected by the change of the filter inductance value. Therefore, in order to avoid the influence of the change of the filter inductance value on the accuracy of the current gradient calculation, it is assumed that the first applied voltage vector used at time k-1 is V m , the second applied voltage vector is V n 、The third applied voltage vector is V z , then we have formula (5):

[0103]

[0104] Among them, t m The first applied voltage vector is V mThe action time; t n The second applied voltage vector is V n The action time; t z The third applied voltage vector is V z The action time of Δi mαβ (k-1) is the α and β components of the inductor current value corresponding to the α and β components of the first applied voltage vector at time k-1; Δi nαβ (k-1) is the α and β components of the inductor current value corresponding to the α and β components of the second applied voltage vector at time k-1; Δi zαβ (k-1) are the α and β components of the inductor current value corresponding to the α and β components of the third applied voltage vector at time k-1.

[0105] Combining equations (4) and (5), we can get the relationship between the three applied voltage vectors and the current gradient at time k-1, as shown in equation (6):

[0106]

[0107] Similarly, the relationship between the three applied voltage vectors and the current gradient at time k-2 can be obtained as shown in equation (7):

[0108]

[0109] Among them, k-2 is the previous sampling time of k-1.

[0110] It can be understood that in the T-type three-level topology circuit, the filter inductance value L at adjacent moments f They can be considered equal, so by dividing equation (6) and equation (7), we can get equation (8):

[0111]

[0112] Finally, by combining equations (5) and (8), we can obtain the first applied voltage vector V at time k-1: m , the second applied voltage vector V n and the third applied voltage vector V z The current gradient Δi corresponding to the respective α and β components mαβ (k-1), Δi nαβ (k-1), Δi zαβ (k-1), as shown in formula (9):

[0113]

[0114] By respectively substituting 9 non-repeating first voltage vector components and 5 non-repeating second voltage vector components into the above formula for calculation, 14 current gradients at time k-1 can be obtained.

[0115] For example, if the first applied voltage vector V sampled at time k-1 is m The first voltage vector component obtained by Clark transformation The second voltage vector component The second applied voltage vector V n The first voltage vector component obtained by Clark transformation The second voltage vector component The third applied voltage vector V z The first voltage vector component obtained by Clark transformation The second voltage vector component Then V mα 、V mβ 、V nα 、V nβ 、V zα 、V zβ The six voltage vector components are all substituted into equation (9) to calculate the current gradients corresponding to the six voltage vector components at time k-1. The calculated current gradients are recorded as standard voltage vector components and The corresponding current gradient.

[0116] It can be understood that when the two voltage vector components in the denominator of coefficients A and B in equation (8) are equal, the corresponding current gradients are also equal. Therefore, there is no need to repeatedly calculate the current gradients corresponding to the two voltage vector components, thereby achieving the purpose of reducing the amount of calculation.

[0117] In some embodiments, to improve the current prediction accuracy and efficiency, a current gradient lookup table may be created. The current gradient lookup table includes a standard voltage vector component, a current gradient at a first moment corresponding to the standard voltage vector component, and a current gradient at a second moment corresponding to the standard voltage vector component, wherein the first moment is a previous prediction time point of the second moment.

[0118] Exemplarily, the T-type three-level topology circuit includes 14 non-repetitive standard voltage vector components, which are 9 α components: And 5 beta components: The first moment is k-2, and the second moment is k-1. The current gradient lookup table is as follows:

[0119]

[0120] The current gradients corresponding to the various standard voltage vector components calculated by equation (9) are respectively filled into the corresponding positions in Table 2 to obtain an updated current gradient lookup table.

[0121] It should be noted that the current gradient lookup table shown in Table 2 includes the current gradients at two moments k-1 and k-2. According to formula (9), it is necessary to know the current gradient at moment k-2 in the process of calculating the current gradient at moment k-1. However, the current gradient of the T-type three-level energy storage device at the initial moment is known to be zero. Therefore, it is necessary to update the current gradient starting from the initial state of the T-type three-level energy storage device.

[0122] In some embodiments, in order to reduce control delay, the current gradient lookup table needs to be updated in each control cycle. However, there is still an unacted voltage vector in the control cycle, and the current gradient corresponding to the unacted voltage vector cannot be calculated using formula (9). Therefore, the current gradient corresponding to the unacted voltage vector at the current moment can be calculated based on the current gradient corresponding to the applied voltage vector at the current moment and the current gradient corresponding to the unacted voltage vector at the previous moment. Figure 3 As shown, the current gradient calculation method corresponding to the unacted voltage vector includes:

[0123] Step S301: subtract the current gradient corresponding to the unapplied voltage vector component at the first moment from the current gradient corresponding to the applied voltage vector component at the first moment to obtain an intermediate value of the current gradient.

[0124] For example, if the unacted voltage vector component is V yαβ , then according to the current gradient lookup table, we know that V yαβ The current gradient corresponding to time k-2 is Δi yαβ (k-2), and the first applied voltage vector component V mαβ The current gradient corresponding to time k-2 is Δi mαβ (k-2), subtract the current gradient corresponding to the unapplied voltage vector component at time k-2 from the current gradient corresponding to the first applied voltage vector component at time k-2 to obtain the current gradient intermediate value Δi yαβ (k-2)-Δi mαβ (k-2).

[0125] Step S302: Add the current gradient corresponding to the applied voltage vector component at the second moment to the intermediate value of the current gradient to obtain the current gradient corresponding to the unapplied voltage vector component at the second moment.

[0126] Exemplarily, the first applied voltage vector component V mαβ The current gradient corresponding to time k-1 is Δi mαβ (k-1), add the current gradient corresponding to the first applied voltage vector component at time k-1 to the intermediate value of the current gradient to obtain the unapplied voltage vector component V yαβ The current gradient corresponding to the time k-1 is shown in equation (10):

[0127] Δi yαβ (k-1) = Δi yαβ (k-2)-Δi mαβ (k-2)+Δi mαβ (k-1), (10)

[0128] Among them, y≠m,n,z.

[0129] Step S303: Record the current gradient corresponding to the unacted voltage vector component at the second moment as the current gradient at the second moment corresponding to the standard voltage vector component that is the same as the unacted voltage vector component.

[0130] Unapplied voltage vector component V yαβ is the value of the standard voltage vector component, so the unacted voltage vector component V yαβ The current gradient Δi corresponding to the second moment yαβ (k-1) is the vector component V yαβ The same standard voltage vector component corresponds to the current gradient at the second moment.

[0131] For example, the unacted voltage vector component At the second moment, the corresponding current gradient is Δi yαβ (k-1), then the standard voltage vector component At the second moment, the corresponding current gradient is also Δi yαβ (k-1).

[0132] In some embodiments, the current gradients of adjacent control cycles are similar, so the current gradient at time k can be approximated by the current gradient at time k-1, as shown in formula (11):

[0133] Δi jαβ (k) = Δi jαβ (k-1) j = 0…26, (11)

[0134] Therefore, the inductor current value at the next moment can be predicted.

[0135] In some embodiments, in order to realize the control of the T-type three-level energy storage device, it is necessary to select 3 optimal voltage vectors from 27 reference voltage vectors as vector information for generating a driving signal. First, the 27 reference voltage vectors need to be divided into 6 first vector groups. Figure 4 As shown, the following steps are included:

[0136] Step S401, dividing the reference voltage vector into zero vector, small vector, medium vector and large vector, wherein the number of zero vectors is a first number, the number of small vectors is a second number, the number of medium vectors is a third number, and the number of large vectors is a fourth number.

[0137] The first number, the second number, the third number and the fourth number are related to the amplitude value of the reference voltage vector: the reference voltage vector has four different amplitude values, namely the minimum amplitude value, the second minimum amplitude value, the second maximum amplitude value and the maximum amplitude value. There are 3 reference voltage vectors with the minimum amplitude value, which are determined as zero vectors, and the first number is 3; there are 12 reference voltage vectors with the second minimum amplitude value, which are determined as small vectors, and the second number is 12; there are 6 reference voltage vectors with the second maximum amplitude value, which are determined as medium vectors, and the third number is 6; there are 6 reference voltage vectors with the maximum amplitude value, which are determined as large vectors, and the fourth number is 6.

[0138] Exemplarily, the 27 reference voltage vectors of the T-type three-level topology circuit are V 0 , V 1 , …, V 26 The 27 reference voltage vectors are divided into 3 zero vectors, 12 small vectors, 6 medium vectors and 6 large vectors. The specific division method is shown in the following table:

[0139]

[0140] Step S402: divide the reference voltage vector into a first vector group according to a first division standard, wherein the first division standard is to take a large vector, a small vector and a zero vector having the smallest difference in amplitude value with a middle vector and the middle vector as one first vector group.

[0141] Since the 27 reference voltage vectors include 6 middle vectors, the middle vector can be used as the division reference vector to be divided into 6 first vector groups.

[0142] For example, Figure 5 As shown in the figure, each center vector and all the reference voltage vectors at the five surrounding points are taken as a first vector group, which are first vector groups I (V 16 、V 17 、V 9 、V 2 、V 0 、V 7 、V 14 、V 8 、V 1 、V 15 ), the first vector group II (y 18 、V 17 、V 19 、V0 、V 7 、V 14 、V 9 、V 2 、V 10 、V 3 ), the first vector group III (V 19 、V 20 、V 21 、V 11 、V 0 、V 7 、V 14 、V 4 、V 10 、V 3 ), the first vector group IV (V 21 、V 22 、V 23 、V 12 、V 0 、V 7 、V 14 、V 5 、V 4 、V 11 ), the first vector group V(V 23 、V 24 、V 25 、V 12 、V 0 、V 7 、V 14 、V 5 、V 6 、V 13 ) and the first vector group VI (V 15 、V 25 、V 26 、V 6 、V 0 、V 7 、V 14 、V 13 、V 1 、V 8 ).

[0143] In some embodiments, after the reference voltage vector is divided into 6 first vector groups, a first vector group that meets the standard is selected from the 6 first vector groups as a selected first vector group for subsequent selection of the optimal voltage vector. Figure 6 As shown, the step of determining the first vector group includes:

[0144] Step S601, obtaining a first inductor current value component and a second inductor current value component at a second moment corresponding to a center vector.

[0145] For example, the mid vector V 16 For example, the vector V 16The corresponding first inductor current value component at the second moment is i 16α (k), the vector V 16 The corresponding second inductor current value component at the second moment is i 16β (k).

[0146] Step S602 , searching in a current gradient lookup table the first current gradient at the second moment corresponding to the first standard voltage vector component of the center vector and the second current gradient at the second moment corresponding to the second standard voltage vector component.

[0147] For example, the mid vector V 16 For example, find the vector V 16 The first current gradient at the second moment corresponding to the first standard voltage vector component is Δi 16α (k), the vector V 16 The second current gradient at the second moment corresponding to the second standard voltage vector component is Δi 16β (k).

[0148] Step S603, adding the first inductor current value component to the first current gradient to obtain a first predicted current value at a third moment corresponding to the first standard voltage vector component of the center vector; adding the second inductor current value component to the second current gradient to obtain a second predicted current value at a third moment corresponding to the second standard voltage vector component of the center vector, wherein the third moment is a predicted time point after the second moment.

[0149] For example, the mid vector V 16 Take the first inductor current value component u as an example. 16α (k) and the first current gradient Δi 16α (k) add to obtain the first predicted current value The second inductor current value component i 16β (k) and the second current gradient Δi 16β (k) add to obtain the second predicted current value

[0150] Step S604, performing a first operation on the first reference current value at the third moment corresponding to the first standard voltage vector component of the center vector and the first predicted current value to obtain a first intermediate value; performing a first operation on the second reference current value at the third moment corresponding to the second standard voltage vector component of the center vector and the second predicted current value to obtain a second intermediate value; adding the first intermediate value and the second intermediate value to obtain a cost function corresponding to the center vector.

[0151] For example, the mid vector V 16 For example, the first reference current value at the third moment With the first predicted current value Subtract to get the first difference, then square the first difference to get the first intermediate value The second reference current value at the third moment With the second predicted current value Subtract to get the second difference, then square the second difference to get the second intermediate value Then add the first intermediate value and the second intermediate value to get the midpoint vector V 16 The corresponding cost function is shown in formula (12):

[0152]

[0153] Where x = 16, 18, 20, 22, 24, 26, in the above vector V 16 In the example embodiment, x=16.

[0154] Step S605: The first vector group in which the middle vector with the minimum cost function belongs is selected as the first vector group.

[0155] According to formula (12), the cost functions corresponding to the six mid vectors are calculated in turn, and the first vector group where the mid vector with the smallest cost function is located is the selected first vector group. 16 The cost function is the smallest, then the vector V 16 The first vector group I is used as the selected first vector group.

[0156] In some embodiments, after the selected first vector group is obtained, a second vector group is selected from the selected first vector group as the optimal vector group, and three reference voltage vectors in the optimal vector group are used as optimal voltage vectors to generate a driving signal to control the T-type three-level energy storage device. Figure 7 As shown, the steps of selecting the optimal vector group include:

[0157] Step S701 : Calculate the cost functions corresponding to the large vectors, small vectors and zero vector in the selected first vector group to obtain the cost functions of all reference voltage vectors in the selected first vector group.

[0158] For example, if the vector y 16 The first vector group where is the selected first vector group, then the mid vector V is calculated according to formula (12) 16 The cost functions corresponding to all reference voltage vectors in the first vector group include: V 16 、V 17 、V 9 、V 2 、V 0 、V 7 、V 14 、V 8、V 1 、V 15 The cost function of .

[0159] Step S702: divide the selected first vector group into second vector groups according to a second division criterion, wherein the second division criterion is to select a reference voltage vector with the smallest amplitude value difference in the selected first vector group as one second vector group.

[0160] Depend on Figure 5 As shown, the amplitude difference of the reference voltage vectors at every three adjacent points is the smallest, so the reference voltage vectors at every three adjacent points are taken as one second vector group.

[0161] For example, any three adjacent points in the first vector group I form four second vector groups, namely, the second vector groups S 1 (V 9 、V 2 、V 0 、V 7 、V 14 ), the second vector group S 2 (V 1 、V 8 、V 15 、V 16 ), the second vector group S 3 (V 9 、V 2 、V 1 、V 8 、V 16 ), the second vector group S 4 (V 9 、V 2 、V 17 、V 16 ).

[0162] Step S703: Determine the second vector group with the minimum cost function as the optimal vector group.

[0163] because Figure 5 The reference voltage vector at each point in the is equal, so when calculating the cost function, only one reference voltage vector at the same point needs to be calculated.

[0164] For example, taking the first vector group I as an example, the second vector group S is calculated according to formula (12) 1 Medium V 7 、V 8 、V 9 The cost function, the second vector group S 2 Medium V 18 、V 15 、V 16 The cost function, the second vector group S 3 Medium V8 、V 9 、V 16 The cost function and the second vector group S 4 Medium V 9 、V 16 、V 17 By comparing the calculation results of the cost function, the second vector group S with the smallest cost function is selected. 1 As the optimal vector group.

[0165] In some embodiments, after determining the optimal vector group, it is also necessary to obtain the switching sequence of the three optimal voltage vectors in the optimal vector group. The switching sequence needs to be based on the first capacitor C 1 The voltage V p and the second capacitor C 2 The voltage V n The size relationship between them (i.e., NP condition) is determined as follows:

[0166] At the first capacitor voltage V p Greater than the second capacitor voltage V n When the switching sequence of the three optimal voltage vectors is the first voltage vector, the second voltage vector, the third voltage vector, the second voltage vector, and the first voltage vector; when the first capacitor voltage V p Less than the second capacitor voltage V n When , the switching sequence of the three optimal voltage vectors is the third voltage vector, the fourth voltage vector, the fifth voltage vector, the fourth voltage vector, and the third voltage vector.

[0167] Exemplarily, taking the first vector group I as an example, the corresponding relationship between the NP condition and the switch sequence is shown in the following table:

[0168]

[0169] In some embodiments, in order to obtain the driving signal, it is also necessary to calculate the action time corresponding to the optimal voltage vector. The driving signal of the T-type three-level energy storage device can be obtained by PWM modulation of the optimal vector group, the switching sequence and the time used. The method of calculating the action time includes:

[0170] According to formula (12), the cost functions corresponding to the three optimal voltage vectors in the optimal vector group are calculated, and the cost functions corresponding to the three optimal voltage vectors are combined to obtain formula (13):

[0171]

[0172] Among them, G m , G n , G z Represent the optimal voltage vector V m 、V n、V z The corresponding cost function.

[0173] The cost functions and control periods corresponding to the three optimal voltage vectors are jointly calculated to obtain the action times of the three optimal voltage vectors, as shown in formula (14):

[0174]

[0175] Among them, t m ,t n ,t z Represent the optimal voltage vector V m 、V n 、V z action time.

[0176] For example, if the second vector group S 1 is the optimal vector group, and the three optimal vector voltages are V 7 、V 8 、V 9 , then the V 7 、V 8 、V 9 The corresponding action time t 7 ,t 8 ,t 9 for:

[0177]

[0178] like Figure 8 As shown, the overall process of the model-free predictive control method for a T-type three-level energy storage device described in the embodiment of the present application includes:

[0179] S801: Acquire the inductor current value, applied voltage vector, first capacitor voltage, and second capacitor voltage output by the T-type three-level topology circuit in the current control cycle.

[0180] S802: Using Clark transformation to split each voltage vector into a first voltage vector component and a second voltage vector component.

[0181] S803: Calculate the current gradient according to the first voltage vector component and the second voltage vector component, and update the current gradient lookup table in the current control cycle.

[0182] S804: Determine a cost function according to the inductor current value, the current gradient and the reference current value, and determine an optimal vector group according to the cost function.

[0183] S805: Determine a vector sequence of the optimal vector group according to the magnitude relationship between the first capacitor voltage and the second capacitor voltage.

[0184] S806: Calculate the action time according to the cost function, and determine the driving signal of the T-type three-level energy storage device according to the reference voltage vector of the optimal vector group, the action time of the reference voltage vector and the vector sequence, so as to realize model-free predictive control.

[0185] Corresponding to the embodiment of the model-free predictive control method of the aforementioned T-type three-level energy storage device, some embodiments of the present application also provide a model-free predictive control system for a T-type three-level energy storage device. The system includes: an energy storage module, a vector splitting module, a current gradient calculation module, an optimal vector group determination module, a vector sequence determination module, and a drive control signal generation module.

[0186] The energy storage module is configured to obtain an inductor current value, an applied voltage vector, a first capacitor voltage, and a second capacitor voltage.

[0187] The vector splitting module is configured to split the voltage vector into a first voltage vector component and a second voltage vector component.

[0188] The current gradient calculation module is configured to calculate the current gradient according to the first voltage vector component and the second voltage vector component.

[0189] The optimal vector group determination module is configured to: determine a cost function according to the inductor current value, the current gradient and the reference current value, and determine the optimal vector group according to the cost function.

[0190] The vector sequence determination module is configured to determine the vector sequence of the optimal vector group according to the magnitude relationship between the first capacitor voltage and the second capacitor voltage.

[0191] The drive control signal generating module is configured to: calculate the action time according to the cost function, and determine the drive control signal of the T-type three-level energy storage device according to the reference voltage vector of the optimal vector group, the action time of the reference voltage vector and the vector sequence.

[0192] It can be seen from the above technical solutions that the present application provides a model-free predictive control method and system for a T-type three-level energy storage device, the method comprising: splitting the reference voltage vector output by the energy storage device into a first voltage vector component and a second voltage vector component by a split transformation method; obtaining the non-repeating first voltage vector component and the second voltage vector component as the standard voltage vector component; obtaining three applied voltage vectors of the energy storage device and splitting them into applied voltage vector components; calculating the current gradient corresponding to the applied voltage vector component by a current gradient calculation method, and recording the current gradient as the current gradient corresponding to the standard voltage vector component that is the same as the applied voltage vector component. In the method, the voltage vector output by the T-type three-level energy storage device is split into voltage vector components, and only the current gradient corresponding to the non-repeating voltage vector component is calculated, thereby reducing the amount of calculation and improving the output power quality and dynamic response speed of the energy storage device.

[0193] Similar parts between the embodiments provided in this application can be referenced to each other. The specific implementation methods provided above are only a few examples under the general concept of this application and do not constitute a limitation on the protection scope of this application. For those skilled in the art, any other implementation methods expanded based on the scheme of this application without creative work belong to the protection scope of this application.

Claims

1. A model-free predictive control method for a T-type three-level energy storage device, characterized in that: include: Splitting the reference voltage vector output by the T-type three-level energy storage device into a first voltage vector component and a second voltage vector component by a split transformation method, wherein the split transformation method is a Clark transformation, and the Clark transformation converts the voltage vector into two equivalent two-phase voltage vectors; Acquire a non-repeating first voltage vector component in the first voltage vector component and a non-repeating second voltage vector component in the second voltage vector component as standard voltage vector components; Acquire a first applied voltage vector, a second applied voltage vector and a third applied voltage vector of the T-type three-level energy storage device within a control period; Splitting the first applied voltage vector, the second applied voltage vector and the third applied voltage vector into applied voltage vector components by the split transformation method; wherein the standard voltage vector component includes the applied voltage vector component; The current gradient corresponding to the applied voltage vector component is calculated by a current gradient calculation method, and the current gradient is recorded as the current gradient corresponding to the standard voltage vector component which is the same as the applied voltage vector component.

2. The model-free predictive control method for a T-type three-level energy storage device according to claim 1, characterized in that: The step of recording the current gradient as a current gradient corresponding to the standard voltage vector component which is the same as the applied voltage vector component comprises: A current gradient lookup table is created, wherein the current gradient lookup table includes the standard voltage vector component, the current gradient at a first moment corresponding to the standard voltage vector component, and the current gradient at a second moment corresponding to the standard voltage vector component, wherein the first moment is a previous predicted time point of the second moment.

3. The model-free predictive control method for a T-type three-level energy storage device according to claim 2, characterized in that: The current gradient at the first moment corresponding to the standard voltage vector component is zero.

4. The model-free predictive control method for a T-type three-level energy storage device according to claim 2, characterized in that: The method further comprises: subtracting the current gradient corresponding to the unapplied voltage vector component at the first moment from the current gradient corresponding to the applied voltage vector component at the first moment to obtain an intermediate value of the current gradient; adding the current gradient corresponding to the applied voltage vector component at the second moment to the intermediate value of the current gradient to obtain the current gradient corresponding to the unapplied voltage vector component at the second moment; The current gradient corresponding to the non-acting voltage vector component at the second moment is recorded as the current gradient at the second moment corresponding to the standard voltage vector component which is the same as the non-acting voltage vector component.

5. The model-free predictive control method for a T-type three-level energy storage device according to claim 2, characterized in that: The method further comprises: Dividing the reference voltage vector into a zero vector, a small vector, a medium vector and a large vector, wherein the number of the zero vectors is a first number, the number of the small vectors is a second number, the number of the medium vectors is a third number, and the number of the large vectors is a fourth number; The first number, the second number, the third number and the fourth number are related to the amplitude value of the reference voltage vector; the third number is equal to the fourth number, the third number or the fourth number is smaller than the second number, and the third number or the fourth number is larger than the first number; The reference voltage vector is divided into a first vector group according to a first division standard, wherein the first division standard is to take a large vector, a small vector and a zero vector having the smallest amplitude difference with the middle vector and the middle vector as one first vector group.

6. The model-free predictive control method for a T-type three-level energy storage device according to claim 5, characterized in that: The method further comprises: Obtaining a first inductor current value component and a second inductor current value component at a second moment corresponding to the midpoint vector; Searching in the current gradient lookup table a first current gradient at a second moment corresponding to the first standard voltage vector component of the middle vector and a second current gradient at a second moment corresponding to the second standard voltage vector component; The first inductor current value component is added to the first current gradient to obtain a first predicted current value at a third moment corresponding to a first standard voltage vector component of the center vector; the second inductor current value component is added to the second current gradient to obtain a second predicted current value at a third moment corresponding to a second standard voltage vector component of the center vector, wherein the third moment is a predicted time point after the second moment; A first operation is performed on a first reference current value at a third moment corresponding to a first standard voltage vector component of the mid vector and the first predicted current value to obtain a first intermediate value; a first operation is performed on a second reference current value at a third moment corresponding to a second standard voltage vector component of the mid vector and the second predicted current value to obtain a second intermediate value; wherein the first operation is to perform a difference operation on two current values ​​to obtain a current difference value, and then perform a square operation on the current difference; Adding the first intermediate value and the second intermediate value to obtain a cost function corresponding to the midpoint vector; The first vector group where the middle vector with the minimum cost function belongs is taken as the selected first vector group.

7. The model-free predictive control method for a T-type three-level energy storage device according to claim 6, characterized in that: The method further comprises: Calculating the cost functions corresponding to the large vector, the small vector and the zero vector in the selected first vector group to obtain the cost functions of all reference voltage vectors in the selected first vector group; dividing the selected first vector group into second vector groups according to a second division criterion, wherein the second division criterion is to use the reference voltage vector with the smallest amplitude value difference in the selected first vector group as one second vector group; The second vector group with the smallest cost function is determined as the optimal vector group.

8. The model-free predictive control method for a T-type three-level energy storage device according to claim 7, characterized in that: The method further comprises: Obtaining a cost function corresponding to a reference voltage vector in the optimal vector group; A second operation is performed on the cost function and the control period to obtain the action time of the reference voltage vector in the optimal vector group, wherein the second operation is as follows: Among them, t m ,t n ,t z are the action time of the reference voltage vector in the optimal vector group, G n , G m , G z are the cost functions corresponding to the reference voltage vectors in the optimal vector group, T s is the control period.

9. The model-free predictive control method for a T-type three-level energy storage device according to claim 8, characterized in that: The method further comprises: Acquire a first capacitor voltage and a second capacitor voltage of the T-type three-level energy storage device; Determine a vector sequence of the optimal vector group according to the first capacitor voltage and the second capacitor voltage, wherein when the first capacitor voltage is greater than the second capacitor voltage, the vector sequence of the optimal vector group is the first voltage vector, the second voltage vector, the third voltage vector, the second voltage vector, and the first voltage vector; when the first capacitor voltage is less than the second capacitor voltage, the vector sequence of the optimal vector group is the third voltage vector, the fourth voltage vector, the fifth voltage vector, the fourth voltage vector, and the third voltage vector; The driving control signal of the T-type three-level energy storage device is determined according to the reference voltage vector of the optimal vector group, the action time of the reference voltage vector and the vector sequence.

10. A model-free predictive control system for a T-type three-level energy storage device, used to execute the model-free predictive control method for a T-type three-level energy storage device according to any one of claims 1 to 9, characterized in that: include: The energy storage module is configured to: obtain an inductor current value, an applied voltage vector, a first capacitor voltage, and a second capacitor voltage; A vector splitting module is configured to: split the voltage vector into a first voltage vector component and a second voltage vector component; A current gradient calculation module, configured to: calculate a current gradient according to the first voltage vector component and the second voltage vector component; An optimal vector group determination module is configured to: determine a cost function according to the inductor current value, the current gradient and the reference current value, and determine an optimal vector group according to the cost function; A vector sequence determination module is configured to: determine a vector sequence of the optimal vector group according to a magnitude relationship between the first capacitor voltage and the second capacitor voltage; The drive control signal generating module is configured to: calculate the action time according to the cost function, and determine the drive control signal of the T-type three-level energy storage device according to the reference voltage vector of the optimal vector group, the action time of the reference voltage vector and the vector sequence.

Images