A modular multilevel converter disturbance model predictive control method and system

Abstract

This invention discloses a model predictive control method and system for modular multilevel converters (MMC), belonging to the field of power electronic converter control. A hyperlocal predictive model incorporating grid-side power and circulating current is established based on the MMC topology, and lumped disturbances are estimated in real time using an extended state observer. To address the decreased observation accuracy caused by fixed inductance parameters, the inductance parameter channels are reconstructed, and the equivalent inductance estimate is calculated online using observation errors and input / output data. The updated inductance value is fed back to the observer and controller to correct the gain matrix and prediction equations, achieving rolling optimization based on real-time equivalent parameters. Traversing the local finite control set, with power point tracking and circulating current suppression as multiple objectives, a sequential optimization strategy is employed to select the optimal number of sub-modules to be engaged in the bridge arm and apply them to the next cycle. This invention improves the robustness of traditional methods to parameter mismatch through an online inductance update mechanism, enhancing steady-state accuracy, dynamic response, and operational reliability.

Description

Technical Field

[0001] This invention belongs to the field of power electronic converter control, specifically relating to a modular multilevel converter disturbance rejection model predictive control method. Background Technology

[0002] Modular multilevel converters (MCs) have become core equipment in fields such as flexible DC transmission, solid-state transformers, and medium- and high-voltage motor drives due to their modular structure, low harmonic content, and high voltage levels. Model predictive control (MPC) is widely used in MMC control due to its advantages such as fast dynamic response and ease of handling multi-objective constraints. Among them, MPC based on extended state observers effectively improves the robustness of MPC by estimating the unmodeled dynamics of the system and external disturbances as new state variables.

[0003] However, the disturbance rejection performance of this framework is highly dependent on the accuracy of the system model. Traditional methods use fixed inductance parameters (AC equivalent inductance and bridge arm inductance) to construct the system model and observer gain. In actual operation, the inductance value of the MMC changes significantly due to factors such as temperature rise, magnetic saturation, and aging, leading to a mismatch between the model and the actual system. This mismatch weakens the accuracy of the extended state observer's disturbance estimation, thereby reducing the controller's performance in regulating power and circulating current, and in severe cases, even causing system instability, significantly reducing the original disturbance rejection capability. Existing countermeasures often require injecting additional disturbance signals or relying on offline measurements, making it difficult to maintain the controller's disturbance rejection performance without increasing hardware costs. Summary of the Invention

[0004] The technical problem to be solved by this invention is: addressing the issue that traditional MMC model predictive control based on extended state observers relies on fixed inductor parameters, which is prone to reduced accuracy of controlled variable observation and total disturbance estimation due to parameter mismatch, resulting in inaccurate model prediction. This invention provides a modular multilevel converter disturbance rejection model predictive control method that introduces an online inductor update mechanism into the model predictive control framework based on extended state observers. This enables the controller to correct the prediction model in real time based on operating data, thereby maintaining stable disturbance rejection capability under parameter changing conditions.

[0005] To solve the above technical problems, the present invention adopts the following technical solution:

[0006] First, this invention proposes a modular multilevel converter disturbance rejection model predictive control method, comprising the following steps:

[0007] Step S1: Collect the operating variables of MMC, including AC side voltage and current, DC side voltage, bridge arm current, submodule capacitor voltage and bridge arm engagement status of the previous control cycle, and calculate the controlled variable vector Y(k), input regression quantity Φ(k) and known dynamic term c(k) based on these variables.

[0008] Step S2: Based on the MMC model, establish a hyperlocal model of the MMC that includes equivalent inductance and lumped perturbations. ,in, The input gain is determined by the equivalent inductance, and D(k) is the lumped disturbance. An extended state observer is constructed based on the hyperlocal model to simultaneously estimate the controlled variable Y(k) and the total disturbance D(k).

[0009] Step S3: Estimate the equivalent inductance value online based on the state change, observation error, and input regression of the MMC extended state observer. ;

[0010] Step S4: Update the input gain of the extended state observer based on the estimated MMC equivalent inductance value. And the predictive model of the predictive controller;

[0011] Step S5, based on the updated equivalent inductance value of MMC Total disturbance observations The system continuously predicts the state of controlled variables in future time steps within a local candidate control set. Then, it employs a hierarchical sequential optimization strategy combining primary objective power tracking and secondary objective circulation suppression to select the optimal number of sub-modules for the bridge arm. And apply it to the next control cycle.

[0012] As a further optimization of the model predictive control method for disturbance rejection in modular multilevel converters, this invention expands the state change, observation error, and input regression of the state observer. Directly estimate the equivalent input gain This leads to the estimated value of the equivalent inductance. Compared to the extended state observer model predictive control method that uses fixed inductance parameters, this invention allows the update of inductance parameters to directly affect the input gain of the hyperlocal model, the extended state observer, and the prediction equation of the model predictive controller, thereby effectively suppressing the impact of inductance parameter mismatch on the accuracy of total disturbance estimation and the prediction results of candidate control quantities.

[0013] Meanwhile, the present invention proposes an electronic system comprising a memory, a processor, and program instructions stored in the memory that can be executed by the processor to implement the various steps of the control method proposed in the present invention.

[0014] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0015] First, this invention maintains consistency between the hyperlocal model and the actual system by updating the equivalent inductance parameters online in real time, effectively suppressing the negative impact of inductance mismatch on control performance and significantly improving the robustness of system parameters. Simultaneously, the extended state observer can estimate and compensate for lumped disturbances including unmodeled dynamics, parameter errors, and external disturbances, further enhancing the anti-interference capability of the control system.

[0016] Second, the present invention adopts a local candidate control set and a hierarchical sequential optimization strategy, which avoids the traditional model predictive control from performing a full ergonomic search on the global candidate set, significantly reducing the amount of real-time computation, while ensuring multi-objective coordinated control of power tracking and circulating current suppression.

[0017] Third, the inductor update mechanism proposed in this invention, in conjunction with disturbance observation, enables the MMC to operate stably and efficiently under complex conditions such as inductor saturation and temperature fluctuations, significantly improving the long-term reliability of the system. Furthermore, this solution requires no additional sensors, has simple control logic, is easy to implement in engineering, and can be flexibly extended to power electronic systems with different parameters, power levels, and topologies, demonstrating significant engineering application value. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the MMC circuit structure according to an embodiment of the present invention.

[0020] Figure 2 This is a schematic diagram of the half-bridge submodule structure according to an embodiment of the present invention.

[0021] Figure 3 This is a flowchart of a modular multilevel converter anti-disturbance model predictive control method according to an embodiment of the present invention.

[0022] Figure 4 This is a comparison chart showing the effects of a modular multilevel converter anti-disturbance model predictive control method according to an embodiment of the present invention. Detailed Implementation

[0023] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0024] Example 1: The overall control block diagram of the modular multilevel converter disturbance rejection model predictive control method proposed in this invention is as follows: Figure 3 As shown. This method consists of two parts: online updating of equivalent inductance and finite control set model predictive control. The MMC structure is as follows. Figure 1 As shown, the MMC includes phases A, B, and C. Each phase includes an upper bridge arm and a lower bridge arm. Each bridge arm is composed of several half-bridge sub-modules, one bridge arm inductor, and one bridge arm resistor cascaded together. The sub-module structure is as follows: Figure 2 As shown. Assume the total number of submodules in each bridge arm is N, and the bridge arm voltage can be adjusted by the number of submodules engaged. The control cycle is T. s The following steps are executed during the kth control cycle.

[0025] The active power P(k) and reactive power Q(k) on the AC side of the MMC can be obtained from the AC side voltage and current, and their expressions are as follows:

[0026]

[0027] In the formula, e α (k) and e β (k) represents the AC side voltage component in the αβ coordinate system, i α (k) and i β (k) represents the component of the AC side current in the αβ coordinate system.

[0028] The three-phase circulating current i can be obtained from the currents in the upper and lower bridge arms. cir,x (k), whose expression is:

[0029]

[0030] In the formula, i xu (k) and i xl (k) represents the upper arm current and lower arm current of phase x, respectively.

[0031] Construct the vector of controlled variables in S1 using equations (1) and (2):

[0032]

[0033] Here, Y(k) includes both grid-side power variables and bridge arm circulating current variables, representing the main control objective of MMC model predictive control.

[0034] For phase x, the equivalent voltages of the upper and lower bridge arms are expressed as follows:

[0035]

[0036] In the formula, M xu (k-1), M xl (k-1) represents the number of sub-modules engaged in the upper and lower bridge arms of phase x in the previous control cycle. , These represent the average capacitor voltages of the upper and lower bridge arm submodules for phase x, respectively, during the current control cycle.

[0037] The MMC phase voltage can be obtained from the upper and lower bridge arm voltages:

[0038]

[0039] definition:

[0040]

[0041] The power channel input regression is constructed from the dynamic power relationship of the MMC AC side. Definition:

[0042]

[0043]

[0044] The power channel input regression is:

[0045]

[0046] In the formula, R e This is the equivalent resistance value.

[0047] The input regression value of the circulation channel is expressed as:

[0048]

[0049] In the formula, U dc (k) is the DC side voltage, R arm This is the resistance of the bridge arm.

[0050] From equations (9) and (10), the MMC input regression value in S1 can be obtained:

[0051]

[0052] The known dynamic terms in MMC in S1 are:

[0053]

[0054] In the formula, Let be the AC side angular frequency. The first two terms in equation (12) represent the angular frequency coupling terms in the power dynamic equation, and the last three terms represent the known dynamic terms in the circulating channel that are not extracted separately.

[0055] Based on the MMC electrical relationship, the combined effect of the AC side filter inductance, bridge arm inductance, and their parameter changes on the dynamics of the controlled variable is uniformly equivalent to inductance L. e Thus, a hyperlocal model of the MMC in S2, including equivalent inductance and lumped perturbation, is established:

[0056]

[0057] In the formula, D(k) is the lumped disturbance, which represents the model error caused by unmodeled dynamics, external disturbances, and inductor parameter mismatch; The input gain, determined by the equivalent inductance, is expressed as:

[0058]

[0059] Construct the MMC extended state observer in S2 based on equation (13). Let the observed value of the controlled variable be... The total disturbance observation value is The observation error is:

[0060]

[0061] The discrete update model of the extended state observer is:

[0062]

[0063] In the formula, B1 and B2 are the observation error feedback gain matrices. Through equation (16), the extended state observer synchronously obtains the observed values ​​of the controlled variables and the total disturbance in each control cycle.

[0064] To enable the extended state observer and model predictive controller to follow the changes in inductor parameters, a residual vector is constructed in the k-th control cycle based on the state change of the extended state observer, the observation error, and the input regression. Let:

[0065]

[0066] The residual vector is then:

[0067]

[0068] Based on the residual vector obtained from equation (18), the equivalent input gain estimate is:

[0069]

[0070] In the formula, ε is a positive number to prevent the denominator from being zero. From equation (19), the estimated value of the MMC equivalent inductance in S3 can be obtained:

[0071]

[0072] For the j-th candidate control variable in the local candidate control set, its bridge arm input submodule number vector is:

[0073]

[0074] The local candidate control set is constructed based on the number of sub-modules engaged in the bridge arm in the previous control cycle.

[0075]

[0076] In the formula, ΔM is the search radius.

[0077] for Any candidate control quantity Recalculate the bridge arm equivalent voltage and input regression quantity of MMC in S4 corresponding to the candidate control quantity according to equations (4)-(12). and known dynamic terms And use the updated equivalent input gain and total perturbation observations to predict the controlled variables at the next time step:

[0078]

[0079] In the formula, This represents the predicted value of the controlled variable at the next time step corresponding to the j-th candidate control variable.

[0080] To evaluate the candidate control variables for MMC in S5, we first construct the main objective cost function with grid-side power tracking as the primary objective:

[0081]

[0082] In the formula, and These are the reference values ​​for active power and reactive power at the next time step, respectively. and These are the predicted active power and reactive power values ​​corresponding to the j-th candidate control variable, respectively.

[0083] Let the power error boundary be When the following conditions are met:

[0084]

[0085] The power point tracking error has not yet entered the acceptable range. At this time, priority should be given to ensuring fast power point tracking. Select:

[0086]

[0087] When the following conditions are met:

[0088]

[0089] This indicates the existence of candidate control variables that meet the power error requirements. Therefore, a subset of candidate control variables is constructed:

[0090]

[0091] Then, using circulation suppression as the secondary objective in the candidate control subset, a secondary objective cost function is constructed:

[0092]

[0093] In the formula, This serves as a reference value for the x-phase circulation at the next moment. Let x be the predicted value of the circulating current in phase x corresponding to the j-th candidate control variable.

[0094] In candidate control subset Select the candidate control variable that minimizes the secondary objective cost function:

[0095]

[0096] The optimal number of submodules to be deployed in the MMC arm in S5 is finally obtained:

[0097]

[0098] The result obtained from equation (31) Apply to the next control cycle.

[0099] Through equations (1)-(31), this embodiment unifies MMC running variable sampling, controlled variable construction, input regression construction, hyperlocal model establishment, extended state observer design, online update of equivalent inductance, model predictive control, and hierarchical sequential optimization under the same control framework. This method estimates the equivalent inductance L online. e This information is then fed back to the extended state observer and model predictive controller, thereby reducing the impact of inductor parameter mismatch on the total disturbance estimation and candidate control quantity prediction results, and improving the power tracking performance, circulating current suppression performance and operational reliability of the MMC under parameter mismatch.

[0100] Figure 4 This is a simulation comparison of the proposed method and a traditional control method under inductor parameter mismatch conditions. In this rectification case, each bridge arm submodule has 10 modules, the AC side rated voltage is 1200V, and at t=1.5s, the output DC voltage drops from 3300V to 2800V. The controller's AC side filter inductor L... f Decreased to 0.5L in 0.5s f It decreased to 0.2L in 1 second. f It decreased to 0.1L in 1.5 seconds. f It rises to 1.4L in 2.5 seconds. f Controller bridge arm inductance L arm Reduced to 0.5L arm The proposed method simulates inductor mismatch. During inductor mismatch, both grid-side current and power can be well tracked, and the arm current and circulating current suppression effects are superior to traditional methods. Furthermore, the estimated inductance value remains consistent with the actual circuit value. Therefore, the modular multilevel converter disturbance rejection model predictive control method proposed in this invention can estimate inductance online, improving system performance.

[0101] Example 2: This application also discloses an electronic system, which includes a memory, a processor, and program instructions stored in the memory that can be executed by the processor to implement the various steps of the control method proposed in this invention.

[0102] It should be noted that the electronic system can use terminal devices such as desktop computers, laptops, or cloud servers. Furthermore, terminal devices include, but are not limited to, processors and memory. For example, terminal devices can also include input / output devices, network access devices, and buses.

[0103] Furthermore, the processor can be a central processing unit (CPU). Of course, depending on the actual use, other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be used. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it in this regard.

[0104] Furthermore, the memory can be an internal storage unit of the terminal device, such as the hard disk or RAM of the terminal device, or an external storage device of the terminal device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) equipped on the terminal device. In addition, the memory can also be a combination of the internal storage unit and the external storage device of the terminal device. The memory is used to store computer programs and other programs and data required by the terminal device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.

[0105] Furthermore, through this electronic system, any one of the methods described in the above embodiments can be stored in the memory of the electronic system and loaded and executed on the processor of the terminal device for convenient use.

[0106] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0107] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A modular multilevel converter disturbance rejection model predictive control method, characterized in that, Includes the following steps: A hyperlocal prediction model including grid-side power and circulating current is established based on the MMC topology, and the lumped disturbance is estimated in real time through an extended state observer; Based on the state changes, observation errors, and input regressions of the MMC extended state observer, the equivalent inductance value is estimated online. Then update the input gain of the extended state observer and the prediction model of the model prediction controller; Based on the updated equivalent inductance value of MMC Total disturbance observations The system continuously predicts the state of controlled variables in future time steps within a local candidate control set. Then, it employs a hierarchical sequential optimization strategy combining primary objective power tracking and secondary objective circulation suppression to select the optimal number of sub-modules for the bridge arm. And apply it to the next control cycle.

2. The method according to claim 1, characterized in that, The process involves establishing a hyperlocal prediction model based on the MMC topology, including grid-side power and circulating current, and estimating lumped disturbances in real time using an extended state observer. Specifically: The operating variables of the MMC are collected, including AC side voltage and current, DC side voltage, bridge arm current, submodule capacitor voltage and bridge arm engagement status of the previous control cycle, and the controlled variable vector Y(k), input regression quantity Φ(k) and known dynamic term c(k) are calculated accordingly. Based on the MMC model, a hyperlocal model of MMC including equivalent inductance and lumped perturbation is established: ， in, Let D(k) be the input gain determined by the equivalent inductance, and D(k) be the lumped disturbance. An extended state observer is constructed based on the hyperlocal model to simultaneously estimate Y(k) and D(k); Equivalent input gain From the equivalent inductance L e (k) Determine: .

3. The method according to claim 2, characterized in that, The controlled variable vector Y(k) is: ， Where P(k) is the active power on the AC side of the MMC, Q(k) is the reactive power on the AC side of the MMC, and i cir,a (k), i cir,b (k), i cir,c (k) represent the circulation of phases A, B, and C, respectively; For phase x, x = a, b, c, the equivalent voltages of the upper and lower bridge arms are respectively: and , in, , These represent the capacitor voltages of the upper and lower bridge arm submodules for phase x, respectively, during the current control cycle. , These are the average values ​​of the capacitor voltages of the upper and lower bridge arm submodules for phase x, respectively, during the current control cycle. , These represent the number of sub-modules for the upper and lower bridge arms of phase x, respectively. The AC phase voltage is: .

4. The method according to claim 3, characterized in that, The input regression parameters are: ; , These are the input regression values ​​for the power channel and the input regression value for the circulating current channel, respectively. The power channel input regression is: ， in, , , For AC phase voltage, R e e is the equivalent resistance value. α and e β Indicates the AC side grid voltage; The input regression value of the circulation channel is: ， Among them, U dc For DC voltage, R arm For the bridge arm resistance, The dynamic term is known to be: ,in It is the AC angular frequency.

5. The method according to claim 4, characterized in that, The discrete model of the controlled variables is: ， ， ， Among them, T s For MMC control cycle, (k) represents the observed values ​​of the controlled variable. Let e(k) be the total disturbance observation value, and e(k) be the observation error. and The observation error feedback gain matrix is ​​obtained from the observer pole configuration.

6. The method according to claim 5, characterized in that, The residual vector of the extended state observer is: ， in, = B1 / T s , = B2 / T s , The equivalent input gain estimate obtained from the residual vector is: , Where ε is a positive number to prevent the denominator from being zero, the equivalent inductance value is: .

7. The method according to claim 6, characterized in that, Input gain updated to The prediction model is updated as follows: , in, Let be the predicted value of the controlled variable at the next time step corresponding to the j-th candidate control variable. and These are the input regression quantity and the known dynamic quantity corresponding to the j-th candidate control quantity, respectively.

8. The method according to claim 7, characterized in that, The local candidate control set is: ， ， ， Where ΔM is the search radius. , These represent the number of upper and lower bridge arm submodules deployed for the j-th candidate control quantity, respectively, for any candidate control quantity in the local candidate control set. Calculate the corresponding bridge arm equivalent voltage and input regression quantity Φ. j (k) and known dynamic quantity c j (k), and predict the future state of the controlled variable based on the updated equivalent inductance value and total disturbance observation.

9. The method according to claim 8, characterized in that, The sequential optimization strategy is as follows: First, calculate the primary objective cost function: ， like Then choose to make The smallest candidate control variable. like Then construct a candidate control subset. and in Calculate the secondary objective cost function , choose to The minimum candidate control variable is used to ultimately select the optimal number of bridge arm sub-modules. And apply it to the next control cycle. in, This is the active power reference value. This is the reactive power reference value. As the power error boundary, This is a reference value for circulation.

10. An electronic system comprising a memory, a processor, and program instructions stored in the memory and executable by the processor, characterized in that, The processor executes the program instructions to implement the steps of the method according to any one of claims 1-9.

Images