Discrete Space Vector Model-Free Predictive Control Method for Three-Phase Three-Level Energy Storage Inverter

Abstract

The present invention discloses a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter, including: establishing a mathematical model in a stationary coordinate system according to the topology, performing current sampling to obtain 27 basic voltage vectors, and using the current gradient as the current gradient calculated based on parameters; generating 36 virtual voltage vectors and corresponding 36 current gradients according to the 27 basic voltage vectors for current prediction; establishing a current gradient equation, and updating the current gradients of the unapplied virtual voltage vectors according to the current gradient equation; calculating the predicted current, analyzing the current gradient, evaluating the cost function values of the 36 predicted currents obtained by the analysis, selecting the virtual voltage vector with the minimum cost function value as the optimal voltage vector, and applying it to the next control cycle; the present invention completely eliminates the stagnation phenomenon of current gradient update and the current spikes caused by the stagnation of current gradient update, reduces the current ripple and total harmonic distortion rate, and simultaneously greatly reduces the computational burden caused by current gradient update.

Description

Technical Field

[0001] The present invention relates to the technical field of three-phase three-level energy storage converters, and particularly to a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter. Background Art

[0002] With the development of new energy, the proportion of new energy technology in the entire energy system has increased rapidly. Due to the natural instability of wind power generation and photovoltaic power generation, supporting energy storage is required to completely replace traditional fossil energy. Therefore, energy storage technology is a key technology for the development of new energy, and has the functions of eliminating the peak-valley difference of electric power, realizing the smooth output of new energy such as photovoltaic and wind power, peak shaving and frequency modulation, and serving as backup energy, enabling new energy power generation to be smoothly connected to the power grid.

[0003] The energy storage system in the prior art is divided into two parts: an energy storage device composed of energy storage elements and an access system composed of power electronic devices. In the prior art, three-phase three-level energy storage converters have been widely used in the access system due to their own advantages. However, in the prior art, the control performance of the model-free control of three-phase three-level energy storage converters is affected by the current gradient update frequency. When only the current gradient under the action of the updated vector is measured through current in each control cycle, without considering the current gradients of other unused vectors, it will cause spikes in the output current and capacitor voltage of the three-phase three-level energy storage converter.

[0004] When the control method in the prior art uses only one basic vector in each cycle, the switching frequency is low and the current ripple is large; and when the vectors applied in adjacent control cycles are the same, it is easy to fail; resulting in a large current ripple, a high total harmonic distortion rate, a stagnation in current gradient update, and current spikes.

[0005] Therefore, there is a need for the emergence of a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter to solve the above problems existing in the model-free prediction of a three-phase three-level energy storage converter. Summary of the Invention

[0006] The object of the present invention is to provide a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter, which can completely eliminate the stagnation of current gradient update and the resulting current spikes in the model-free prediction of a three-phase three-level energy storage converter, reduce the ripple of the output current and the total harmonic distortion rate, and at the same time greatly reduce the computational burden caused by current gradient update.

[0007] To solve the above technical problems, the present invention provides a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter, including the following steps:

[0008] S1. Based on the three-phase three-level energy storage converter topology, establish its mathematical model in the stationary coordinate system, obtain 27 basic voltage vectors, perform current sampling, and use the measured current gradient as the current gradient of the three-phase three-level energy storage converter based on parameter calculation;

[0009] S2. Generate 36 virtual voltage vectors and their corresponding 36 current gradients according to the 27 basic voltage vectors, and perform current prediction based on the current gradient of the discrete space vector;

[0010] S3. Establish the current gradient equation relationship of the 36 virtual voltage vectors, and update the current gradient of the unapplied virtual voltage vectors according to the current gradient of the applied virtual voltage vectors;

[0011] S4. Calculate 36 predicted currents according to the current gradients corresponding to the 36 virtual voltage vectors, evaluate the value function values of the obtained 36 predicted currents, select the virtual voltage vector with the smallest corresponding value function value as the optimal voltage vector, and apply the optimal voltage vector to the next control cycle.

[0012] Optionally, the mathematical model of the three-phase three-level energy storage converter in the stationary coordinate system in S1 is:

[0013]

[0014] where L is the filter inductor, t is the time, and i αβ is the output current vector of the three-phase three-level energy storage converter in the stationary coordinate system; e αβ is the grid voltage vector in the stationary coordinate system, u αβ is the output voltage vector of the three-phase three-level energy storage converter in the stationary coordinate system, and R is the filter resistance.

[0015] Optionally, the 36 virtual voltage vectors in S2 are expressed as:

[0016] u siαβ = 0.5u mαβ + 0.5u nαβ

[0017] where u siαβ is the virtual voltage vector, u mαβ is the first basic voltage vector, and u nαβ is the second basic voltage vector.

[0018] Optionally, in S2,

[0019] the predicted current at time k + 1 is expressed as:

[0020] i αβ (k + 1)= iαβ (k) + Δi usiαβ (k)

[0021] The predicted current at time k+2 is expressed as:

[0022] i αβ (k + 2) = i αβ (k + 1) + Δi usiαβ (k + 1)

[0023] Wherein, i αβ (k) is the output current in the stationary coordinate system at time k, i αβ (k + 1) is the predicted current in the stationary coordinate system at time k+1, i αβ (k + 2) is the predicted current in the stationary coordinate system at time k+2, Δi usiαβ (k + 1) is the virtual voltage vector u applied at time k+1 siαβ (k + 1) current gradient in the stationary coordinate system, Δi usiαβ (k) is the virtual voltage vector u applied at time k siαβ (k) current gradient in the stationary coordinate system, u siαβ (k) is the virtual voltage vector applied at time k, u siαβ (k + 1) is the virtual voltage vector applied at time k+1

[0024] Optionally, in S3, when the coordinate components of the virtual voltage vector in the stationary coordinate system are equal, the corresponding current gradients are also equal, and the current gradient update formula is expressed as:

[0025] Δi usjαβ (k) = Δi usiαβ (k)

[0026] Wherein, u sjαβ (k) is the virtual voltage vector not applied at time k, Δi usjαβ (k) is the virtual voltage vector u not applied at time k sjαβ (k) current gradient in the stationary coordinate system, u siαβ (k) is the virtual voltage vector applied at time k, Δi usiαβ (k) is the current gradient of the virtual voltage vector applied at time k in the stationary coordinate system

[0027] When the coordinate components are equal, the current gradient of the unapplied virtual voltage vector u sjαβ (k) in the stationary coordinate system is equal to the current gradient of the virtual voltage vector applied at time k in the corresponding α stationary coordinate system and the corresponding β stationary coordinate system

[0028] Optionally, in S3, when the coordinate components of the virtual voltage vector in the stationary coordinate system are not equal, the current gradient update formula is expressed as:

[0029] Δi usjαβ (k) = (Δi usjαβ (k - 1) - Δi usiαβ (k - 1)) + Δi usiαβ (k)

[0030] where u sjαβ (k) is the virtual voltage vector not applied at time k, Δi usjαβ (k) is the current gradient of the virtual voltage vector u sjαβ (k) in the stationary coordinate system, u sjαβ (k - 1) is the virtual voltage vector not applied at time k - 1, Δi usjαβ (k - 1) is the current gradient of the virtual voltage vector u sjαβ (k - 1) in the stationary coordinate system, u siαβ (k - 1) is the virtual voltage vector applied at time k - 1, Δi usiαβ (k - 1) is the current gradient of the virtual voltage vector u siαβ (k - 1) in the stationary coordinate system, u siαβ (k) is the virtual voltage vector applied at time k, Δi usiαβ (k) is the current gradient of the virtual voltage vector applied at time k in the stationary coordinate system, k - 1 is the previous control time, and k is the current control time.

[0031] Optionally, in S4, the value function value G of the 36 predicted currents is expressed as:

[0032] G = (i refα - i α (k + 2)) 2 + (i refβ - i β (k + 2)) 2 .

[0033] where G is the value function, i refα and i refβ are both reference current vectors, and i α (k + 2) and i β (k + 2) are both predicted current vectors at time k + 2.

[0034] Compared with the prior art, the present invention has at least the following beneficial effects: Through the combination of virtual voltage vectors, two basic voltage vectors are applied in each control cycle, reducing the ripple of the output current and the total harmonic distortion rate, fixing the switching frequency, and improving the accuracy of the model-free prediction of the three-phase three-level energy storage converter; A sampling point is set at the end of each control cycle to measure the current gradient of the applied virtual voltage vector. By measuring and storing the current gradient under the action of the virtual voltage vector in the previous control cycle, the current gradient relationship of different virtual voltage vectors is established, realizing the real-time update of the current gradient of the unapplied virtual voltage vector in each control cycle, and completely eliminating the stagnation phenomenon of the current gradient update and the current spike caused by the stagnation of the current gradient update.

[0035] Furthermore, the discrete space vector model-free prediction control method for the three-phase three-level energy storage converter provided by the present application does not depend on any system parameters, still has good parameter robustness when the system parameters are mismatched, eliminates the dependence of the model-free control method in the prior art on system parameters, has good output current quality, further improves the output current quality compared with the prior art, and at the same time greatly reduces the computational burden caused by the current gradient update. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the control flow of an embodiment of the present invention;

[0037] Figure 2 It is a topology diagram of the three-phase three-level energy storage converter of an embodiment of the present invention;

[0038] Figure 3 It is a space vector diagram of the three-phase three-level energy storage converter of an embodiment of the present invention;

[0039] Figure 4 It is a discrete space vector diagram of the three-phase three-level energy storage converter of an embodiment of the present invention;

[0040] Figure 5 It is a discrete space vector current gradient diagram of the three-phase three-level energy storage converter of an embodiment of the present invention;

[0041] Figure 6 It is a discrete space vector α-axis current gradient relationship diagram of the three-phase three-level energy storage converter of an embodiment of the present invention;

[0042] Figure 7 It is a discrete space vector β-axis current gradient relationship diagram of the three-phase three-level energy storage converter of an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] The following will describe in more detail a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter of the present invention with reference to schematic diagrams. The preferred embodiments of the present invention are shown, and it should be understood that those skilled in the art can modify the present invention described herein while still achieving the advantageous effects of the present invention. Therefore, the following description should be understood as broad knowledge for those skilled in the art and not as a limitation on the present invention.

[0044] In the following paragraphs, reference is made to the attached Figure 1-7 The present invention will be described more specifically by way of example. The advantages and features of the present invention will be clearer according to the following description and the claims. It should be noted that the drawings are all in a very simplified form and use non-precise scales, only for the purpose of facilitating and clearly assisting in explaining the embodiments of the present invention.

[0045] In an existing model-free predictive control method for an energy storage converter, in order to increase the update frequency of the current gradient, an update frequency is set in the control. If a certain current gradient is not updated within 50 control cycles, the vector corresponding to this current gradient will be used in the next control cycle, and its current gradient will also be updated accordingly. However, this method will frequently use non-optimal vectors, thus reducing the current performance.

[0046] In another existing model-free predictive control method for an energy storage converter, the mathematical model is simplified, the current gradient relationship between two consecutive control cycles is established, and the current gradient of the unapplied vector is updated according to this relationship. However, when the vectors applied in two consecutive cycles are the same, this method will cause the prediction result to fail.

[0047] In another model-free predictive method in the prior art, a model-free predictive control method based on virtual vectors is proposed. Each virtual voltage vector is synthesized by two or three basic voltage vectors. In order to obtain the current gradient under the action of each basic voltage vector, sampling needs to be performed according to the action time of each vector within a control cycle, which increases the complexity of the control system. And only the current gradient of the applied vector is updated within each control cycle, resulting in a low update frequency of the current gradient.

[0048] Specifically, when performing model-free prediction in the prior art, two sampling points are set at the midpoint and the end of each control cycle. However, the increase in the number of sampling points will occupy more system memory and increase the complexity of system design.

[0049] In summary, the present invention proposes a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter. Only one sampling is set in each control cycle, and the current gradient can be updated in real time in each control cycle, completely eliminating the stagnation phenomenon of current gradient update and the current spikes caused by the stagnation of current gradient update, while significantly reducing the computational burden caused by current gradient update.

[0050] The present invention provides a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter. Please refer to Figure 1 as shown, which includes the following steps:

[0051] S1. According to the topology of the three-phase three-level energy storage converter, establish its mathematical model in the stationary coordinate system, obtain 27 basic voltage vectors, perform current sampling, and use the measured current gradient as the current gradient of the three-phase three-level energy storage converter based on parameter calculation.

[0052] S2. Generate 36 virtual voltage vectors and their corresponding 36 current gradients according to the 27 basic voltage vectors, and perform current prediction based on the current gradient of the discrete space vector.

[0053] S3. Establish the current gradient equation relationship of the 36 virtual voltage vectors, and update the current gradient of the unapplied virtual voltage vectors according to the current gradient of the applied virtual voltage vectors.

[0054] S4. Evaluate the cost function values of the 36 predicted currents according to the current gradients corresponding to the 36 virtual voltage vectors, select the virtual voltage vector with the smallest corresponding cost function value as the optimal voltage vector, and apply the optimal voltage vector to the next control cycle.

[0055] In an embodiment of the present invention, the discrete space vector model-free predictive control method for the three-phase three-level energy storage converter is carried out according to the following steps:

[0056] Step 1. According to the topology of the three-phase three-level energy storage converter, for the topology diagram, please refer to Figure 2 as shown, establish the mathematical model of the three-phase three-level energy storage converter in the stationary coordinate system, perform current sampling, and perform Clarke transformation on the sampled three-phase grid currents i abc to obtain i αβ :

[0057]

[0058] where, i α is the current in the α stationary coordinate system, i β is the current in the β stationary coordinate system,, i ais the current in the a-phase coordinate system of the three-phase coordinate system, i b is the current in the b-phase coordinate system of the three-phase coordinate system, i c is the current in the c-phase coordinate system of the three-phase coordinate system.

[0059] The mathematical model of the three-phase three-level energy storage converter in the stationary coordinate system is:

[0060]

[0061] Among them, L is the filter inductor, t is the time, i αβ is the output current of the three-phase three-level energy storage converter in the stationary coordinate system; e αβ is the grid voltage vector in the stationary coordinate system, u αβ is the output voltage vector of the three-phase three-level energy storage converter in the stationary coordinate system, and R is the filter resistance.

[0062] According to the forward Euler method, the predicted current at the (k + 1)th moment can be discretized as:

[0063]

[0064] Among them, i αβ (k + 1) is the predicted current in the stationary coordinate system at the (k + 1)th moment, i αβ (k) is the output current in the stationary coordinate system at the kth moment, T s is the control period, L is the filter inductor, e αβ (k) is the grid voltage vector in the stationary coordinate system at the kth moment, u αβ (k) is the basic voltage vector in the stationary coordinate system at the kth moment, R is the filter resistance, Δi αβ (k) is the current gradient of the basic voltage vector u αβ in the stationary coordinate system.

[0065] Specifically, since the predicted current of the model predictive control depends on the accuracy of the parameters of the three-phase three-level energy storage converter, the parameters of the three-phase three-level energy storage converter may be inaccurate due to measurement errors or may change due to changes in operating conditions. This parameter uncertainty will cause errors in the current gradient, thereby reducing the current performance. For the purpose of eliminating the influence of the model parameters on the predicted current, a model-free predictive control method is adopted.

[0066] Therefore, the current gradient can be obtained based on current measurement and expressed as:

[0067] Δi αβ (k - 1) = i αβ (k) - i αβ (k - 1)

[0068] Among them, Δiαβ (k - 1) is the current gradient of the voltage vector at the (k - 1)th moment in the stationary coordinate system, i αβ (k) is the output current in the stationary coordinate system at the kth moment, i αβ (k - 1) is the output current in the stationary coordinate system at the (k - 1)th moment.

[0069] In this embodiment, the current gradient corresponds one - to - one with the applied voltage vector and is stored in a look - up table for current prediction.

[0070] Therefore, the predicted current at the (k + 1)th moment can be expressed as:

[0071] i αβ (k + 1) = i αβ (k)+Δi αβ (k)

[0072] where, i αβ (k + 1) is the predicted current in the stationary coordinate system at the (k + 1)th moment, i αβ (k) is the output current in the stationary coordinate system at the kth moment, Δi αβ (k) is the current gradient of the voltage vector at the kth moment in the stationary coordinate system.

[0073] Step two, according to the volt - second balance principle, use the 27 basic voltage vectors to generate the 36 virtual voltage vectors, expressed as:

[0074] u siαβ = 0.5u mαβ + 0.5u nαβ

[0075] where, u siαβ is the virtual voltage vector, u mαβ is the first basic voltage vector, u nαβ is the second basic voltage vector.

[0076] The synthesis method of the 36 virtual voltage vectors is shown in the following table:

[0077]

[0078]

[0079] Furthermore, according to the synthesized 36 virtual voltage vectors, the corresponding 36 virtual voltage vector current gradients are generated accordingly. Please refer to Figure 4 shown.

[0080] According to the current gradients of the 36 virtual voltage vectors, current prediction is performed. Specifically, the predicted current at the (k + 1)th moment can be expressed as:

[0081] i αβ (k + 1) = i αβ (k) + Δi usiαβ (k)

[0082] The predicted current at time k + 2 can be expressed as:

[0083] i αβ (k + 2) = i αβ (k + 1) + Δi usiαβ (k + 1)

[0084] Wherein, i αβ (k + 1) is the predicted current in the stationary coordinate system at time (k + 1), i αβ (k) is the output current in the stationary coordinate system at time k, i αβ (k + 2) is the predicted current in the stationary coordinate system at time (k + 2), u siαβ (k) is the virtual voltage vector applied at time k, u siαβ (k + 1) is the virtual voltage vector applied at time k + 1, Δi usiαβ (k) is the current gradient of the virtual voltage vector u siαβ (k) at time k, Δi usiαβ (k + 1) is the current gradient of the virtual voltage vector u siαβ (k + 1) at time k + 1.

[0085] Step 3: Establish the relationship of the current gradient equations for different virtual vectors, and update the current gradient of the unapplied virtual vector according to the current gradient of the applied virtual vector.

[0086] Specifically, in this embodiment, sampling is set once in each control cycle to obtain the current gradient of the virtual voltage vector. By establishing the current gradient update equations for different virtual voltage vectors, the real-time update of the current gradients of other virtual voltage vectors is realized. Among them, the current gradient Δi usi (k) of the applied virtual voltage vector is expressed as:

[0087] Δi usiαβ (k) = X[e gαβ (k) - u siαβ (k) - Ri αβ (k)]

[0088] Wherein, u siαβ (k) is the virtual voltage vector applied at time k, Δi usiαβ (k) is the current gradient of the applied virtual voltage vector u siαβ (k) in the stationary coordinate system, X is a system parameter, e gαβ (k) is the grid voltage vector in the stationary coordinate system at time k, iαβ (k) is the output current in the stationary coordinate system at time k, and R is the filter resistance.

[0089] Similarly, the current gradient of the unapplied virtual voltage vector is expressed as:

[0090] Δi usjαβ (k) = X[e gαβ (k) - u sjαβ (k) - Ri αβ (k)]]

[0091] Among them, u sjαβ (k) is the unapplied virtual voltage vector at time k, and Δi usjαβ (k) is the current gradient of the unapplied virtual voltage vector u sjαβ (k) in the stationary coordinate system, e gαβ (k) is the grid voltage vector in the stationary coordinate system at time k, and i αβ (k) is the output current in the stationary coordinate system at time k, X is the system parameter, and R is the filter resistance.

[0092] Subtracting the above equations can obtain:

[0093] Δi usjαβ (k) - Δi usiαβ (k) = X(u siαβ (k) - u sjαβ (k))

[0094] Among them, Δi usjαβ (k) is the current gradient of the unapplied virtual voltage vector u sjαβ (k) in the stationary coordinate system, and Δi usiαβ (k) is the current gradient of the applied virtual voltage vector u siαβ (k) in the stationary coordinate system, u siαβ (k) is the applied virtual voltage vector at time k, and u sjαβ (k) is the unapplied virtual voltage vector at time k.

[0095] It can be seen from the above equation that when the coordinate components of the virtual voltage vectors u siαβ (k) and u sjαβ (k) in the stationary coordinate system are equal, the current gradient values corresponding to the virtual voltage vectors u siαβ (k) and u sjαβ (k) are also equal, so it can be obtained without complex calculations, effectively reducing the computational burden caused by the current gradient update.

[0096] The current gradient update formula is expressed as:

[0097] Δi usjαβ(k) = Δi usiαβ (k)

[0098] where Δi usjαβ (k) is the current gradient of the virtual voltage vector u siαβ (k) in the stationary coordinate system at time k, and Δi usiαβ (k) is the current gradient of the virtual voltage vector u siαβ (k) in the stationary coordinate system at time k.

[0099] When the coordinate components are equal, the current gradients corresponding to the applied virtual voltage vector u siαβ (k) and the unapplied virtual voltage vector u sjαβ (k) are equal in both the corresponding α stationary coordinate system and the corresponding β stationary coordinate system.

[0100] When the coordinate components of the virtual voltage vectors u siαβ (k) and u sjαβ (k) in the stationary coordinate system are not equal, the current gradient relationship of the virtual voltage vectors u siαβ (k - 1) and u sjαβ (k - 1) at time k - 1 can be obtained and expressed as

[0101] Δi usjαβ (k - 1) - Δi usiaβ (k - 1) = X(u siαβ (k - 1) - u sjαβ (k - 1))

[0102] To eliminate the parameter X, the current gradients can be jointly expressed as:

[0103]

[0104] Dividing the above equations can obtain:

[0105] (Δi usjαβ (k) - Δi usiαβ (k)) / (Δi usjαβ (k - 1) - Δi usiαβ (k - 1)) = (u siαβ (k) - u sjαβ (k)) / (u siαβ (k - 1) - u sjαβ (k - 1)) = U

[0106] Analyzing the above equation gives:

[0107] Δi usjαβ (k) = U(Δi usjαβ (k - 1) - Δi usiαβ (k - 1)) + Δiusiαβ (k)

[0108] When the coordinate components in the stationary coordinate system are not equal, it can be considered that U = 1. Therefore, the above formula is further expressed as:

[0109] Δi usjαβ (k)=(Δi usjαβ (k - 1)-Δi usiαβ (k - 1))+Δi usiαβ (k)

[0110] where, Δi usjαβ (k - 1) is the current gradient of the virtual voltage vector u sjαβ (k - 1) in the stationary coordinate system at time k - 1, Δi usiαβ (k - 1) is the current gradient of the virtual voltage vector u siαβ (k - 1) in the stationary coordinate system at time k - 1, u siαβ (k - 1) is the virtual voltage vector applied at time k - 1, u sjαβ (k - 1) is the virtual voltage vector not applied at time k - 1, Δi usjαβ (k) is the current gradient of the virtual voltage vector u sjαβ (k) not applied at time k in the stationary coordinate system, Δi usiαβ (k) is the current gradient of the virtual voltage vector u siαβ (k) applied at time k in the stationary coordinate system, u siαβ (k) is the virtual voltage vector applied at time k, u sjαβ (k) is the virtual voltage vector not applied at time k, X is the system parameter, and U is the vector ratio.

[0111] Through the above steps, the phenomenon of current gradient update stagnation is eliminated, and all current gradients can be updated within one control period.

[0112] Step 4: Evaluate the value function values of the 36 predicted currents obtained according to the current gradients corresponding to the 36 virtual voltage vectors, select the virtual voltage vector with the smallest corresponding value function value as the optimal voltage vector, and apply the optimal voltage vector to the next control period.

[0113] The value function values G of the 36 predicted currents are:

[0114] G=(i refα -i α (k + 2)) 2 +(i refβ -i β (k + 2)) 2 .

[0115] Among them, G is the value function, and i refα and i refβ is the reference current vector, i α (k + 2) and i β (k + 2) are both the predicted current vectors at the (k + 2)-th moment.

[0116] In this application, through the combination of virtual voltage vectors, two basic voltage vectors are applied in each control cycle, reducing the ripple of the output current and the total harmonic distortion rate, fixing the switching frequency, and improving the accuracy of the model-free prediction of the three-phase three-level energy storage converter; a sampling point is set at the end of each control cycle to measure the current gradient of the applied virtual voltage vector. By measuring and storing the current gradient under the action of the virtual voltage vector in the previous control cycle, the current gradient relationship of different virtual voltage vectors is established, realizing the real-time update of the current gradient of the unapplied virtual voltage vector in each control cycle, and completely eliminating the stagnation phenomenon of the current gradient update and the current spike caused by the stagnation of the current gradient update.

[0117] Furthermore, a discrete space vector model-free predictive control method for a three-phase three-level energy storage converter in this solution does not depend on any system parameters and still has good parameter robustness when the system parameters are mismatched, eliminating the dependence of the model-free control method in the prior art on system parameters, having good output current quality, and further improving the output current quality compared with the prior art.

[0118] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and its equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A discrete space vector model-free predictive control method for a three-phase three-level energy storage converter, characterized in that, It includes the following steps: S1. According to the three-phase three-level energy storage converter topology, establish its mathematical model in the stationary coordinate system, obtain 27 basic voltage vectors, perform current sampling, and use the measured current gradient as the current gradient of the three-phase three-level energy storage converter based on parameter calculation; S2. Generate 36 virtual voltage vectors and their corresponding 36 current gradients according to the 27 basic voltage vectors, and perform current prediction based on the current gradient of the discrete space vector; S3. Establish the relationship of the current gradient equation of the 36 virtual voltage vectors, and update the current gradient of the unapplied virtual voltage vectors according to the current gradient of the applied virtual voltage vectors; S4. Calculate 36 predicted currents according to the current gradients corresponding to the 36 virtual voltage vectors, evaluate the value function values of the obtained 36 predicted currents, select the virtual voltage vector with the smallest corresponding value function value as the optimal voltage vector, and apply the optimal voltage vector to the next control cycle; Among them, the mathematical model of the three-phase three-level energy storage converter in the stationary coordinate system in S1 is: ； Wherein, L is the filter inductor, t is the time, and i αβ is the output current vector of the three-phase three-level energy storage converter in the stationary coordinate system; e αβ is the grid voltage vector in the stationary coordinate system, u αβ is the output voltage vector of the three-phase three-level energy storage converter in the stationary coordinate system, and R is the filter resistance; In S2, the predicted current at the (k + 1)th moment is expressed as: ； The predicted current at the (k + 2)th moment is expressed as: ； where, i αβ (k) is the output current in the stationary coordinate system at time k, i αβ (k + 1) is the predicted current in the stationary coordinate system at time k + 1, i αβ (k + 2) is the predicted current in the stationary coordinate system at time k + 2, Δi usiαβ (k + 1) is the current gradient of the applied virtual voltage vector u siαβ (k + 1) in the stationary coordinate system, Δi usiαβ (k) is the current gradient of the applied virtual voltage vector u siαβ (k) in the stationary coordinate system, u siαβ (k) is the applied virtual voltage vector at time k, u siαβ (k + 1) is the applied virtual voltage vector at time k + 1.

2. The discrete space vector model-free predictive control method for a three-phase three-level energy storage converter according to claim 1, wherein The 36 virtual voltage vectors in S2 are expressed as: ； where, u siαβ is the virtual voltage vector, u mαβ is the first basic voltage vector, u nαβ is the second basic voltage vector.

3. The discrete space vector model-free predictive control method for a three-phase three-level energy storage converter according to claim 1, wherein In S3, when the coordinate components of the virtual voltage vector in the stationary coordinate system are equal, the corresponding current gradients are also equal, and the current gradient update formula is expressed as: ； Among them, u sjαβ (k) is the virtual voltage vector not applied at time k, and △i usjαβ (k) is the current gradient of the virtual voltage vector u sjαβ (k) in the stationary coordinate system at time k, and u siαβ (k) is the virtual voltage vector applied at time k, and △i usiαβ (k) is the current gradient of the virtual voltage vector applied at time k in the stationary coordinate system; When the coordinate components are equal, the unapplied virtual voltage vector u sjαβ (k) has the same current gradient in the stationary coordinate system as the applied virtual voltage vector at time k in both the corresponding α stationary coordinate system and the corresponding β stationary coordinate system.

4. The discrete space vector model-free predictive control method for a three-phase three-level energy storage converter according to claim 3, characterized in that, In S3, when the coordinate components of the virtual voltage vector in the stationary coordinate system are not equal, the current gradient update formula is expressed as: ； Among them, u sjαβ (k) is the virtual voltage vector not applied at time k, and △i usjαβ (k) is the current gradient of the virtual voltage vector u sjαβ (k) in the stationary coordinate system, and u sjαβ (k - 1) is the virtual voltage vector not applied at time k - 1, and △i usjαβ (k - 1) is the virtual voltage vector u sjαβ (k - 1) not applied, and the current gradient of u siαβ (k - 1) in the stationary coordinate system, and u usiαβ (k - 1) is the virtual voltage vector applied at time k - 1, and △i siαβ (k - 1) is the virtual voltage vector u siαβ (k - 1) applied, and the current gradient of u usiαβ (k) is the virtual voltage vector applied at time k, and △i (k) is the current gradient of the virtual voltage vector applied at time k in the stationary coordinate system. k - 1 is the previous control time, and k is the current control time.

5. The discrete space vector model-free predictive control method for a three-phase three-level energy storage converter according to claim 1, wherein In S4, the value function value G of the 36 predicted currents is expressed as: ； where G is the value function, i refα and i refβ are both reference current vectors, i α (k + 2) and i β (k + 2) are both predicted current vectors at time k + 2.

Images