Control methods, devices, electronic equipment, and storage media for suppressing voltage fluctuations

By introducing a dual radial basis function neural network control model into the MMC-HVDC transmission system and dynamically adjusting the control vector, the stability problem of the PI controller under voltage fluctuations and load changes is solved, achieving higher control accuracy and response speed, and enhancing the robustness and stability of the system.

CN123121780APending Publication Date: 2026-11-17EAST CHINA BRANCH OF STATE GRID CORP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611268149.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-20
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

When the AC side voltage fluctuates at the sending end or the AC side load changes at the receiving end, the PI controller of the MMC-HVDC transmission system cannot effectively control the situation, resulting in excessive overshoot, and even causing the DC bus voltage to exceed the limit and the transmission system to oscillate and become unstable.

Method used

A dual radial basis function neural network control model is adopted. By constructing the target power balance equation between the AC side of the sending end and the rectifier, and combining the main neural network and the disturbance observation neural network, the final control vector is dynamically adjusted. The model fits the changes in transmission system parameters and external disturbances in real time, and uses time-varying error boundary constraints to track errors, thereby improving control accuracy and response speed.

Benefits of technology

It effectively suppressed voltage oscillations, enhanced the robustness and stability of the power transmission system under complex operating conditions, reduced DC bus voltage overshoot, and shortened convergence time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123121780A_ABST
    Figure CN123121780A_ABST
Patent Text Reader

Abstract

This application provides a control method, apparatus, electronic device, and storage medium for suppressing voltage fluctuations, applied to a rectifier in a power transmission system. The method includes: acquiring and determining a state vector based on first output data and first input data; determining a tracking error based on the state vector, and determining an error transformation gain based on the tracking error and a first function; determining a first input vector based on the state vector, tracking error, and rate of change of the state vector, and determining a first update weight based on a main neural network, the first input vector, gain data, and a first historical weight; determining a second input vector based on the tracking error and a first output current, determining a second update weight based on a disturbance observation neural network, the second input vector, and a second historical weight, and determining a disturbance vector based on a first auxiliary state variable, observer gain, and the second update weight; and outputting a final control vector based on the first input vector, tracking error, gain data, first update weight, and disturbance vector.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of high voltage direct current transmission control technology, and in particular to a control method, device, electronic device and storage medium for suppressing voltage fluctuations. Background Technology

[0002] With the escalating global energy crisis and heightened environmental awareness, renewable energy power generation technologies, represented by wind and solar power, have developed rapidly. Since renewable energy sources are often located in remote areas far from load centers, high-voltage direct current (HVDC) transmission technology has become a key means of achieving long-distance, large-capacity power transmission. In power transmission systems, flexible direct current transmission technology based on modular multilevel converters (MMC) (MMC-HVDC) is commonly used.

[0003] In related technologies, the operating characteristics of MMC-HVDC transmission systems are affected by multiple factors. These factors may change gradually over time or may change abruptly, thus requiring real-time measurement. However, the classic proportional-integral (PI) controller cannot control the MMC-HVDC transmission system based on the real-time status of these factors. When the voltage output on the AC side of the sending end of the transmission system fluctuates drastically or the load on the AC side of the receiving end changes abruptly, the MMC-HVDC transmission system under the control of the PI controller may experience excessive overshoot, or even cause the DC bus voltage to exceed the limit, leading to transmission system protection shutdown or oscillation instability. Summary of the Invention

[0004] In view of this, this application provides a control method, device, electronic device and storage medium for suppressing voltage fluctuations, which effectively suppresses voltage oscillations in the power transmission system caused by fluctuations on the side of the new energy power generation base or changes in the load on the AC side of the receiving end, thereby enhancing the robustness and stability of the power transmission system under complex operating conditions.

[0005] In a first aspect, embodiments of this application provide a control method for suppressing voltage fluctuations, applied to a rectifier in a power transmission system. The power transmission system further includes a sending-end AC side, which is electrically connected to the rectifier. The control method for suppressing voltage fluctuations includes: Acquire the first output data of the AC side at the sending end and the first input data of the rectifier, determine the target power balance equation between the AC side at the sending end and the rectifier based on the first output data and the first input data, and determine the state vector based on the target power balance equation and the first output data; The tracking error is determined based on the state vector and the pre-configured target parameters, and the error transformation gain is determined based on the tracking error and the pre-configured first function. Based on the state vector, tracking error, and the rate of change of the state vector determined according to the state vector, a first input vector is determined; a first basis function vector is determined based on the pre-configured main neural network; and a first update weight is determined based on the first basis function vector, the first input vector, the gain data, and the obtained first historical weight. The second input vector is determined based on the tracking error and the first output data. The second basis function vector is determined based on the pre-configured perturbation observation neural network. The second update weight is determined based on the second input vector, the second basis function vector, and the obtained second historical weight. The perturbation vector is determined based on the pre-determined first auxiliary state variable, the observer gain, and the second update weight. The final control vector is output based on the first input vector, the tracking error, the gain data, the first update weight, and the disturbance vector.

[0006] Secondly, embodiments of this application provide a control device for suppressing voltage fluctuations, comprising: A rectifier used in a power transmission system, the power transmission system further including a sending-end AC side, the sending-end AC side being electrically connected to the rectifier, and the control device for suppressing voltage fluctuations comprising: The data processing module is used to acquire the first output data of the AC side of the sending end and the first input data of the rectifier, determine the target power balance equation between the AC side of the sending end and the rectifier based on the first output data and the first input data, and determine the state vector based on the target power balance equation and the first output data. A gain determination module is used to determine the tracking error based on the state vector and pre-configured target parameters, and to determine the error transformation gain based on the tracking error and a pre-configured first function. The variable fitting module is used to determine a first input vector based on the state vector, the tracking error, and the rate of change of the state vector determined according to the state vector; determine a first basis function vector based on a pre-configured main neural network; and determine a first update weight based on the first basis function vector, the first input vector, the gain data, and the acquired first historical weight. The perturbation fitting module determines a second input vector based on the tracking error and the first output data, determines a second basis function vector based on a pre-configured perturbation observation neural network, determines a second update weight based on the second input vector, the second basis function vector, and the obtained second historical weight, and determines a perturbation vector based on a pre-determined first auxiliary state variable, observer gain, and the second update weight. The vector output module is used to output a final control vector based on the first input vector, the tracking error, the gain data, the first update weight, and the disturbance vector.

[0007] Thirdly, embodiments of this application provide an electronic device including a processor and a memory, the memory storing a program or instructions that can run on the processor, the program or instructions implementing the steps of the method as described in the first aspect when executed by the processor.

[0008] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method as described in the first aspect.

[0009] Fifthly, embodiments of this application provide a chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the method as described in the first aspect.

[0010] In a sixth aspect, embodiments of this application provide a computer program product stored in a storage medium, which is executed by at least one processor to implement the method as described in the first aspect.

[0011] Thus, in the voltage fluctuation suppression control method provided in this application embodiment, the controller constructs a target power balance equation between the sending-end AC side and the rectifier and introduces a dual radial basis function neural network control model to achieve real-time fitting of complex influencing factors such as transmission system parameter changes, unmodeled dynamics, and external disturbances. The controller dynamically adjusts the final control vector based on the state vector, tracking error, and adaptive backstepping composite control law, while simultaneously constraining the tracking error using time-varying error boundaries. This improves the control accuracy and response speed of the DC voltage channel and the instantaneous reactive power channel, effectively suppressing voltage oscillations in the transmission system caused by fluctuations on the new energy power generation base side or changes in load on the receiving-end AC side, thereby enhancing the robustness and stability of the transmission system under complex operating conditions.

[0012] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0013] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a control method for suppressing voltage fluctuations according to an embodiment of this application is shown; Figure 2 One of the schematic diagrams of the power transmission system structure according to an embodiment of this application is shown; Figure 3 A second schematic diagram of the power transmission system structure according to an embodiment of this application is shown; Figure 4 One of the comparison diagrams shows the control effect of the embodiment of this application and the PI controller when the power transmission system parameters remain unchanged; Figure 5 The second diagram shows a comparison of the control effects of the embodiments of this application and the PI controller when the power transmission system parameters remain unchanged. Figure 6 One of the comparison diagrams shows the control effects of the embodiments of this application and the PI controller when the parameters of the power transmission system change; Figure 7 The second diagram shows a comparison of the control effects of the embodiments of this application and the PI controller when the parameters of the power transmission system change. Figure 8 A structural block diagram of a control device for suppressing voltage fluctuations according to an embodiment of this application is shown; Figure 9 A structural block diagram of an electronic device according to an embodiment of this application is shown. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0015] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0016] The control method for suppressing voltage fluctuations provided in this application will be described in detail below with reference to the accompanying drawings and through specific embodiments and application scenarios. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0017] This application provides a control method for suppressing voltage fluctuations. This method is applied to a rectifier in a power transmission system, which also includes a sending-end AC side electrically connected to the rectifier. The execution entity of this control method can be a controller or control chip for controlling the rectifier. The following explanation uses a controller as the execution entity to illustrate the voltage fluctuation suppression control method provided in this application. Figure 1 As shown, the control methods for suppressing voltage fluctuations include: Step 101: The controller acquires the first output data of the sending-end AC side and the first input data of the input rectifier, and determines the target power balance equation between the sending-end AC side and the rectifier based on the first output data and the first input data, and determines the state vector based on the target power balance equation and the first output data.

[0018] like Figure 2As shown, in a power transmission system, electrical energy is transmitted from the sending-end AC side to the receiving-end AC side. A rectifier and an inverter are installed between the sending-end and receiving-end AC sides. The rectifier can be an MMC rectifier, and correspondingly, the inverter can also be an MMC inverter. The sending-end AC side can be a power plant or a new energy power generation base. The following description uses a new energy power generation base as the main example of the sending-end AC side to introduce relevant embodiments. The rectifier is located near the new energy power generation base to convert electrical energy from AC to DC, thereby transmitting the electrical energy in the form of DC. The receiving-end AC side can be a receiving-end power grid or a load center. The following description uses the receiving-end power grid as the main example of the receiving-end AC side to introduce relevant embodiments. The inverter is located near the receiving-end power grid or load center to convert the transmitted DC back into AC and supply it to the receiving-end power grid or load center. Because there may be power equipment such as step-up transformers, AC filters, converter reactors, and circuit breakers between the new energy power generation base and the rectifier, the actual electrical energy output from the new energy power generation base will not be exactly the same as the electrical energy actually flowing into the rectifier. The controller acquires the first output data of the electrical energy output from the sending-end AC side and the first input data of the electrical energy flowing into the rectifier. The first output data includes the first output voltage and the first output current, and the first input data includes the first input voltage. Then, the controller performs rotational phase transformation on the first output data and the first input data respectively. Based on the first output data and the first input data after rotational phase transformation, the controller determines the target power balance equation between the sending-end AC side and the rectifier. Finally, the controller determines the state vector based on the target power balance equation and the first output data.

[0019] Step 102: The controller determines the tracking error based on the state vector and the pre-configured target parameters, and determines the error transformation gain based on the tracking error and the pre-configured first function.

[0020] The state vector can include two components. Specifically, the two components can be the square of the DC voltage output by the rectifier and the instantaneous reactive power flowing into the rectifier, respectively. The target parameters include a first target parameter and a second target parameter. The first target parameter is the preset square of the DC voltage output by the rectifier, and the second target parameter is the preset instantaneous reactive power flowing into the rectifier. Therefore, the control signal output by the controller includes a DC voltage channel and an instantaneous reactive power channel. The following description uses only the DC voltage channel as an example to illustrate the embodiments of this application.

[0021] Tracking error can be determined by the first formula, which is:

[0022] in, Represents tracking error. This represents the time when the rectifier experiences a disturbance. represent The state vector at time t, represent The objective parameters of the state vector at time t.

[0023] The controller determines the error change gain with respect to the DC voltage based on the square of the DC voltage in the state vector and the first objective parameter in the objective parameters.

[0024] Step 103: The controller determines the first input vector based on the state vector, tracking error, and the rate of change of the state vector determined according to the state vector; determines the first basis function vector based on the pre-configured main neural network; and determines the first update weight based on the first basis function vector, the first input vector, gain data, and the acquired first historical weight.

[0025] For example, the factors affecting the rectifier output voltage can be expressed in a first-order nonlinear affine state-space form, which is:

[0026] in, The variable represents the rate of change of the state vector, and u represents the control vector. and This represents an unknown nonlinear function, used to replace the influence of transmission system parameters, transmission system operating conditions, and unmodeled dynamics. and Cannot be obtained directly; This represents the disturbance vector, i.e., the change that occurs in the external environment. It cannot be directly determined either. The main neural network is used to learn the inverse dynamics model of the power transmission system, and the input vector of the main neural network is the first input vector. The neural network control model utilizes the main neural network to... and Estimate and use perturbation observation neural networks to perform An estimate is made to determine the rate of change of the state vector.

[0027] The controller is pre-configured with a neural network control model. This model allows the controller to fit factors affecting the rectifier's output voltage, enabling it to control the output voltage based on real-time acquired data. The controller transposes the matrix consisting of the state vector, tracking error, and the rate of change of the state vector determined from the state vector to determine the first input vector. The first input vector is:

[0028] in, Represents the first input vector. Represents the state vector. Represents the rate of change of the state vector. This represents the tracking error.

[0029] Then, the controller determines the first basis function vector based on the basic parameters of the main neural network in the pre-configured neural network control model. Based on the first basis function vector, the first input vector, the gain data, and the obtained first historical weights, the controller determines the first update weights.

[0030] Step 104: The controller determines the second input vector based on the tracking error and the first output data, determines the second basis function vector based on the pre-configured perturbation observation neural network, determines the second update weight based on the second input vector, the second basis function vector and the obtained second historical weight, and determines the perturbation vector based on the pre-determined first auxiliary state variable, the observer gain and the second update weight.

[0031] The disturbance observation neural network is used to handle rapid time-varying disturbances such as sudden changes in external load. The input vector of the disturbance observation neural network is the second input vector. In the disturbance observation neural network within the neural network control model, the controller transposes the matrix formed by the tracking error and the first output current from the first output data, and uses this transpose as the second input vector. The second input vector is:

[0032] in, Represents the second input vector. represents the tracking error, and u represents the previous control vector output by the controller.

[0033] The controller determines the second basis function vector based on the basic parameters of the perturbation observation neural network, and then determines the second update weight based on the second input vector, the second basis function vector, the obtained second historical weight, the pre-determined first auxiliary state variable and the observer gain.

[0034] To determine the perturbation vector, an auxiliary state variable is introduced into the perturbation observation neural network. The controller substitutes the pre-configured observer gain, the first auxiliary state variable, the second update weight, and the second input vector into the pre-configured first dynamic equation to obtain the perturbation vector and the second auxiliary state vector. It is worth noting that the second auxiliary state variable obtained by the controller in this calculation will be used as the first auxiliary state variable in the next calculation, and resubmitted into the first dynamic equation to obtain a new second auxiliary state variable and perturbation vector. Specifically, when the controller first calculates the perturbation vector and the second auxiliary state variable, the first auxiliary state vector is pre-configured in the controller. The first dynamic equation is:

[0035] in, Represents the perturbation vector, as mentioned above. , Represents the first auxiliary state variable. The value represents the pre-configured observer gain, and x represents the state vector. Represents the second auxiliary state variable. Represents the transpose of the second update weight. Represents the second basis function vector. represents the second input vector, and u represents the previous control vector output by the controller.

[0036] Preferably, to facilitate the implementation of steps 103 and 104, a neural network control model with a main neural network and a disturbance observation neural network can be pre-configured in the controller. This neural network control model is a dual radial basis function (RBF) neural network control model.

[0037] Step 105: The controller outputs the final control vector based on the first input vector, tracking error, gain data, first update weights, and disturbance vector.

[0038] The controller substitutes the first input vector, tracking error, gain data, first update weights, and disturbance vector determined according to the second update weights into the pre-configured adaptive backstepping composite control law to obtain the final control vector. The adaptive backstepping composite control law is as follows:

[0039] in, This represents the final control vector, i.e., the updated control vector; This represents the linear feedback term that provides damping; Represents tracking error; Represents pre-configured reference trajectory data; Represents the transpose of the first updated weight; Represents the first basis function vector; Represents the first input vector. Represents the compensation term of the main neural network; This represents the disturbance estimate, which can be determined based on the second basis function vector, the state vector, the control vector, and relevant data pre-configured in the controller. This represents the time-varying error boundary, which can be determined based on the tracking error and data pre-configured in the controller. This represents the error transformation gain.

[0040] Thus, in the voltage fluctuation suppression control method provided in this application embodiment, the controller constructs a target power balance equation between the sending-end AC side and the rectifier and introduces a dual radial basis function neural network control model to achieve real-time fitting of complex influencing factors such as transmission system parameter changes, unmodeled dynamics, and external disturbances. The controller dynamically adjusts the final control vector based on the state vector, tracking error, and adaptive backstepping composite control law, while simultaneously constraining the tracking error using time-varying error boundaries. This improves the control accuracy and response speed of the DC voltage channel and the instantaneous reactive power channel, effectively suppressing voltage oscillations in the transmission system caused by fluctuations on the new energy power generation base side or changes in load on the receiving-end AC side, thereby enhancing the robustness and stability of the transmission system under complex operating conditions.

[0041] In some embodiments, step 102 can be implemented by steps 1021 to 1022: Step 1021: The controller determines the time-varying error boundary based on the pre-configured initial error limit, the pre-configured expected steady-state error tolerance, the pre-configured convergence rate factor, and the time when the rectifier is disturbed.

[0042] Specifically, the controller is pre-configured with a boundary function for calculating the time-varying error boundary, and the boundary function is:

[0043] in, Represents the time-varying error boundary. This represents the time when the rectifier experiences a disturbance. This represents the pre-configured initial error limit. This represents the pre-configured expected steady-state error tolerance. represents the natural constant, approximately equal to 2.71828, and is the base of the natural logarithm. This represents the pre-configured convergence rate factor.

[0044] Initial error limit With tracking error The relationship between them satisfies: .

[0045] Step 1022: The controller limits the range of tracking error based on the time-varying error boundary, and determines the error transformation gain based on the tracking error and the time-varying error boundary.

[0046] The relationship between the tracking error and the time-varying error boundary is set in the controller to limit the error range of the DC voltage output by the rectifier. With time-varying error boundary Relationship satisfaction .

[0047] The controller determines the error transformation gain based on the tracking error and the time-varying error boundary using a pre-configured second formula. The second formula is:

[0048] in, Represents the error transformation gain. Represents tracking error. Represents the time-varying error boundary. Represents a time variable.

[0049] When tracking error Approaching the time-varying error boundary When the denominator in the second formula approaches zero, the error transformation gain will increase sharply. Introducing a time-varying error boundary can generate enhanced feedback when the tracking error approaches the time-varying error boundary, preventing the tracking error from going out of bounds.

[0050] In the above embodiments, the time-varying error boundary is determined by the boundary function. The error constraint range can be adaptively adjusted according to the disturbance occurrence time, initial error limit, expected steady-state error tolerance and convergence rate factor. The error transformation gain is used to limit the allowable boundary of the tracking error in real time. This effectively solves the problem of error exceeding the limit caused by sudden disturbances or parameter changes in traditional fixed boundary control. It ensures that the tracking error of rectifier output voltage and instantaneous reactive power is always within the preset convergence trajectory, improves the ability to suppress voltage fluctuations and the steady-state accuracy of the system during transient processes, and enhances the adaptive adjustment capability and robustness of the transmission system to disturbances such as new energy power generation fluctuations and load changes.

[0051] In some embodiments, the first output data includes a first output voltage and a first output current, and the first input data includes a second output voltage. Step 101 includes: Step 1011: The controller acquires the first output data of the AC side of the sending end and the first input data of the input rectifier. Based on the first output data and the pre-configured first coordinate system, it determines the current component and the first voltage component of the first output data on the corresponding coordinate axis, and the second voltage component of the first input data on the corresponding coordinate axis.

[0052] The controller can convert the AC output from the sending-end AC side, which is a sinusoidal AC quantity in a three-phase stationary coordinate system, into a DC quantity in a dq rotating coordinate system. Specifically, the first output data includes the current data and the first voltage data output from the sending-end AC side, and the first input data includes the second voltage data of the rectifier near the sending-end AC side. The current data can be decomposed into current components in the dq rotating coordinate system, including a d-axis current component and a q-axis current component; the first voltage data can be decomposed into a first voltage component in the dq rotating coordinate system, including a d-axis first voltage component and a q-axis first voltage component; the second voltage data can be decomposed into a second voltage component in the dq rotating coordinate system, including a d-axis second voltage component and a q-axis second voltage component.

[0053] like Figure 3 As shown, each arm of the rectifier converter station consists of multiple identical sub-modules and arm inductors connected in series. Figure 3 Each phase of the MMC consists of two bridge arms, upper and lower. Each bridge arm is composed of N identical half-bridge sub-modules connected in series, along with the bridge arm inductor. and bridge arm resistance The three-phase AC terminals are connected to the sending-end AC system via a converter transformer. Since this application mainly studies the external characteristics and power dynamics of the AC side, the submodule-level switching process can be equivalent to a continuously controllable voltage source, and the bridge arm resistance, inductance, and converter transformer leakage reactance can be uniformly converted to the AC side, thereby obtaining the AC side equivalent circuit used in the following embodiments.

[0054] Therefore, the equivalent circuit on the side of the converter station connected to the AC side of the sending end can be considered as a controllable voltage source connected to the AC grid through an equivalent impedance. According to Kirchhoff's voltage law, in the synchronously rotating dq coordinate system, the dynamic equation of the AC loop in the rectifier can be expressed as:

[0055] in, Represents equivalent inductance. Represents the equivalent resistance. Represents the d-axis current component. Represents the q-axis current component. This represents the voltage vector angular frequency in the transmission system pre-configured in the controller. Represents the second voltage component along the d-axis. Represents the second voltage component along the q-axis. Represents the first voltage component along the d-axis. This represents the first voltage component along the q-axis.

[0056] Step 1012: The controller determines the first-order tracking element based on the current component, the first voltage component, and the second voltage component, and determines the instantaneous active power and instantaneous reactive power of the rectifier based on the current component and the first voltage component.

[0057] The controller determines the first-order tracking element based on the equivalent inductance and equivalent resistance determined by the current component, the first voltage component, and the second voltage component, and using a pre-configured third formula. The third formula is:

[0058] in, Represents the first-order tracking stage. Represents equivalent inductance. Represents the equivalent resistance.

[0059] The controller determines the instantaneous active and reactive power at the rectifier input based on the d-axis current component, q-axis current component, d-axis first voltage component, and q-axis first voltage component, using pre-configured fourth and fifth formulas. The fourth and fifth formulas are as follows:

[0060] in, Represents the instantaneous active power at the rectifier input. Represents the instantaneous reactive power at the rectifier input. Represents the second voltage component along the d-axis. Represents the second voltage component along the q-axis. Represents the d-axis current component. This represents the q-axis current component.

[0061] In the dq rotating coordinate system, let Then the instantaneous reactive power is: Therefore, it can be obtained from the reactive power command. ,in, This is the specified reactive power reference value.

[0062] When the inner PI closed loop is approximated as a first-order tracking element, the sixth formula can be obtained. The sixth formula is the dynamic current response equation on the q-axis obtained after approximating the inner PI current control closed loop as a first-order inertial element. It represents the actual q-axis current. For current reference value The first-order tracking process. The sixth formula is:

[0063] in, Represents the first-order tracking stage. Represents the q-axis current component. Represents a time variable. The reference value representing the q-axis current component is determined based on the reactive power reference value.

[0064] Based on the sixth formula, the outer-loop dynamic formula for instantaneous reactive power is further obtained. The outer-loop dynamic formula for instantaneous reactive power is:

[0065] in, Represents the first-order tracking stage. Represents the instantaneous reactive power at the rectifier input. This represents the time when the rectifier experiences a disturbance. The reference value representing the q-axis current component. This represents the preset coefficient.

[0066] For example, it is possible to configure the controller. Perform pre-configuration, such as... .

[0067] Step 1013: The controller determines the target power balance equation based on the rectifier's basic parameters and historical output voltage, instantaneous active power, and instantaneous reactive power.

[0068] The basic parameters of a rectifier can include the number of sub-modules in the bridge arms. Assume the average capacitance of each sub-module is... Then, the total energy stored in the capacitors of all sub-modules of the rectifier can be determined using the seventh formula. The seventh formula is:

[0069] in, This represents the total energy stored in the capacitors of all sub-modules of the rectifier. This represents the average capacitance of each submodule. This represents the number of sub-modules in the bridge arm of the rectifier. This represents the DC bus voltage output by the rectifier.

[0070] The initial power balance equation when transmission system losses are ignored is:

[0071] in, This represents the total energy stored in the capacitors of all sub-modules of the rectifier. This represents the time when the rectifier experiences a disturbance. Represents the instantaneous active power at the rectifier input. This represents the instantaneous reactive power at the rectifier output.

[0072] By rearranging the seventh formula and the initial power balance equation, we finally obtain the target power balance equation, which is:

[0073] in, This represents the average capacitance of each submodule. This represents the number of sub-modules in the bridge arm of the rectifier. This represents the DC bus voltage output by the rectifier, i.e., the first output voltage. Represents a time variable. Represents the instantaneous active power at the rectifier input. This represents the instantaneous active power at the rectifier output.

[0074] Step 1014: The controller expresses the voltage state variable using the first output voltage and determines the state vector based on the voltage state variable and instantaneous reactive power.

[0075] Specifically, the voltage state variable can be the square of the first output voltage, and the state vector determined based on the voltage state variable and the instantaneous reactive power is as follows:

[0076] in, Represents the first output voltage. Represents voltage state variables. This represents the instantaneous reactive power at the input of the rectifier.

[0077] Based on the outer-loop dynamic formula and initial power balance equation of instantaneous reactive power, it can be seen that the outer-loop dynamic of instantaneous reactive power is structurally consistent with the dynamic of the first output voltage. However, when regulating the first output voltage in a power transmission system, it is necessary to comprehensively consider many uncertainties such as nonlinear influences and the degradation of capacitance over time. Traditional PI controllers cannot integrate these uncertainties into the output control command when controlling the first output voltage. Therefore, a first-order nonlinear affine state-space form is established in the controller to improve its adaptability to the aforementioned uncertainties. The first-order nonlinear affine state-space form is as follows:

[0078] in, The variable represents the rate of change of the state vector, and u represents the control vector. and This represents an unknown nonlinear function, used to replace the influence of transmission system parameters, transmission system operating conditions, and unmodeled dynamics. and Cannot be obtained directly; This represents the disturbance vector, i.e., the change that occurs in the external environment. It cannot be determined directly.

[0079] Specifically, the forms of the state vector, control vector, and disturbance vector are as follows:

[0080] in, Represents the first output voltage. Represents voltage state variables. Represents the instantaneous reactive power at the rectifier input. The reference value representing the d-axis current component. The reference value representing the q-axis current component. This represents an unknown disturbance in the DC voltage path. Unknown disturbances representing reactive power channels.

[0081] From the fourth formula as well as achievable ,in, represent .

[0082] In the above embodiments, the controller converts the electrical quantities in the three-phase stationary coordinate system of the AC side at the sending end into DC quantities in the two-phase synchronous rotating coordinate system, i.e., the dq coordinate system. Then, combining Kirchhoff's voltage law, the first-order tracking element, and the energy storage dynamics of the submodule capacitor, it establishes a state vector expression covering voltage state variables and instantaneous reactive power. Finally, it constructs a first-order nonlinear affine state space form to describe the unknown nonlinear dynamics and disturbances of the transmission system. This effectively eliminates the shortcomings of traditional PI controllers in integrating uncertainties such as capacitor aging, nonlinear loads, and parameter perturbations. It provides accurate system dynamic characteristics and state inputs for subsequent neural network control models, thereby improving the controller's modeling accuracy and adaptive control capability for complex operating conditions and laying a solid model foundation for suppressing voltage fluctuations.

[0083] In some embodiments, step 103 includes: Step 1031: The main neural network includes multiple first neurons. The controller determines multiple first basis functions based on the basic parameters of each first neuron, and determines the first basis function vector based on the multiple first basis functions.

[0084] The basic parameters of each first neuron include a center vector and a width parameter. The controller can determine multiple first basis functions based on the center vector and width parameter of each first neuron. The first basis function can be defined as:

[0085] in, This represents the j-th first basis function determined by the j-th first neuron in the main neural network. Represents the first input vector. This represents the center vector of the j-th first neuron. The width parameter represents the j-th first neuron.

[0086] The first basis function vector can be obtained by transposing multiple first basis functions. This first basis function vector can also be a Gaussian function vector. The first basis function vector is:

[0087] in, Represents the first basis function vector. denoted by , represents the first basis function determined based on the first neuron in the main neural network, and l represents the number of first neurons in the main neural network.

[0088] Step 1032: The controller determines the first update weight based on the first basis function vector, gain data, first input vector, first historical weight, and pre-configured first weight update law.

[0089] Furthermore, in some embodiments, the first weight update law is determined based on Lyapunov stability theory, and the first weight update law is:

[0090] in, Represents the first updated weight. This represents the pre-configured first learning rate. Represents the first basis function vector. Represents the first input vector. Represents the error transformation gain. This represents the pre-configured first damping coefficient. Represents the first historical weight.

[0091] The controller substitutes the obtained first basis function vector, gain data, first input vector, and first historical weight into the first weight update law pre-designed and configured in the controller to obtain the first updated weight.

[0092] In the above embodiments, the controller uses multiple first neurons in the main neural network to determine the first basis function vector, and combines it with the first weight update law designed based on Lyapunov stability theory to achieve adaptive approximation of unknown nonlinear functions and online weight update. This avoids the problem of insufficient dynamic fitting of the transmission system by the traditional fixed parameter model. It can also dynamically adjust the first update weight according to the real-time input state vector, tracking error and gain data, thereby effectively improving the estimation accuracy of the neural network control model for the nonlinear characteristics of the rectifier and the unmodeled dynamics, enhancing the robustness of the control system under parameter perturbation and disturbance conditions, and providing reliable adaptive compensation capability for suppressing voltage fluctuations.

[0093] In some embodiments, step 104 includes: Step 1041: The perturbation observation neural network includes multiple second neurons. The controller determines multiple second basis functions based on the basic parameters of each second neuron, and determines the second basis function vector based on the multiple second basis functions.

[0094] The basic parameters of each second neuron also include a center vector and a width parameter. The controller can determine multiple second basis functions based on the center vector and width parameter of each second neuron. The second basis function can be defined as:

[0095] in, This represents the m-th second basis function determined by the m-th second neuron in the perturbation observation neural network. Represents the second input vector. This represents the center vector of the m-th second neuron. The width parameter represents the m-th second neuron.

[0096] The second basis function vector can be obtained by transposing multiple second basis functions. This second basis function vector can also be a Gaussian function vector. The second basis function vector is:

[0097] in, Represents the second basis function vector. represents the first second basis function determined based on the first second neuron in the perturbation observation neural network, and n represents the number of second neurons in the perturbation observation neural network.

[0098] Step 1042: The controller determines the second update weight based on the second basis function vector, the second input vector, the second historical weight, and the pre-configured second weight update law.

[0099] Furthermore, in some embodiments, the second weight update law is determined based on Lyapunov stability theory, and the second weight update law is:

[0100] in, Represents the second update weight. This represents the pre-configured second learning rate. Represents the second basis function vector. Represents the second input vector. This represents the pre-configured first damping coefficient. This represents the second historical weight.

[0101] The controller substitutes the determined second basis function vector, the obtained second input vector, the obtained second historical weights, and the pre-configured second learning rate and first damping coefficient into the pre-designed and configured second weight update law in the controller to obtain the second updated weights.

[0102] In the above embodiments, the controller generates a second basis function vector based on multiple second neurons in the disturbance observation neural network. Then, combined with a second weight update law designed based on Lyapunov stability theory, it achieves adaptive approximation and real-time estimation of external disturbances to the transmission system. This enables dynamic adjustment of the second update weights based on the second input vector without prior knowledge of the specific model or statistical characteristics of the external disturbance, effectively improving the controller's online estimation accuracy for fast time-varying disturbances and nonlinear uncertainties. Furthermore, the weight update law designed based on Lyapunov theory ensures the closed-loop stability and convergence of the disturbance observation process, providing accurate disturbance compensation terms for rectifier control. This further enhances the voltage fluctuation suppression capability of the control method and strengthens the robustness of the transmission system in complex environments.

[0103] For example, a dual-ended MMC-HVDC transmission system is used as an example for operating condition verification calculations. It is assumed that the rated power of this dual-ended MMC-HVDC transmission system is 800MW, and the DC bus voltage is... AC power grid frequency 50Hz, number of MMC submodules per bridge arm Submodule capacitor Bridge arm inductance .

[0104] When the voltage fluctuation suppression control method provided in this application embodiment is used in the dual-ended MMC-HVDC transmission system, and the transmission system is under active power step interference conditions, during simulation time... At that time, the active power command of the inverter station on the AC side of the receiving end will increase by a step. Figure 5This diagram shows a comparison of the control effects of traditional PI control and the control method provided in this application embodiment on the DC bus voltage output of the rectifier, under the condition that the transmission system parameters remain unchanged. Figure 5 It can be seen that when using traditional PI control, the DC bus voltage exhibits a maximum overshoot of approximately 11.2kV and a convergence time as long as 265ms. However, after adopting the voltage fluctuation suppression control method provided in this application embodiment, the maximum overshoot is reduced to 4.9kV, which is 56.25% lower than the overshoot under traditional PI control; the convergence time is shortened to 62ms, which is 76.6% shorter than the convergence time under traditional PI control. Figure 6 This diagram compares the control effects of conventional PI control and the control method provided in this embodiment on the instantaneous reactive power of the input rectifier, assuming constant transmission system parameters. When the transmission system is under reactive power step operation, and the reactive power jumps from -200MVar to 280MVar, according to... Figure 6 It is known that the overshoot generated by traditional PI control is about 9.1 MVar and the convergence time is 50 ms; while when the control method for suppressing voltage fluctuations provided in the embodiments of this application is adopted, the rectifier does not generate obvious overshoot and the convergence time is only 31 ms, realizing fast decoupling control of reactive power.

[0105] The transmission system parameters were adjusted to further verify the effectiveness of the voltage fluctuation suppression control method provided in the embodiments of this application. Specifically, the capacitance values ​​of all MMC submodules were artificially reduced to 90% of their rated values, while the bridge arm inductance was reduced to 70% of its rated value. Figure 6 This paper presents a comparison chart showing the control effects of traditional PI control and the control method provided in the embodiments of this application on the DC bus voltage output of the rectifier after adjusting the transmission system parameters. Figure 7 This diagram shows a comparison of the control effects of traditional PI control and the control method provided in this embodiment on the instantaneous reactive power of the input rectifier after adjustments to the transmission system parameters. Figure 6 and Figure 7 It is evident that the voltage overshoot of traditional PI control deteriorates to over 12kV; while the response curve obtained by the control method for suppressing voltage fluctuations provided in this application almost completely overlaps with that when the transmission system parameters remain unchanged, with the maximum overshoot and convergence time remaining essentially constant. The results demonstrate that this invention, through the combination of a dual RBF network and a BLF, can suppress DC voltage fluctuations under both power step disturbances on the receiving end AC side and unknown perturbations of system parameters, exhibiting superior adaptability and robustness.

[0106] Furthermore, as a specific implementation of the above-mentioned control method for suppressing voltage fluctuations, embodiments of this application provide a control device for suppressing voltage fluctuations. For example... Figure 8As shown, the control device 800 for suppressing voltage fluctuations includes: a data processing module 801, a gain determination module 802, a variable fitting module 803, a disturbance fitting module 804, and a vector output module 805.

[0107] The data processing module 801 is used to acquire the first output data of the sending-end AC side and the first input data of the input rectifier. The first output data includes the first output voltage and the first output current. Based on the first output data and the first input data, the target power balance equation between the sending-end AC side and the rectifier is determined, and the state vector is determined based on the target power balance equation and the first output data.

[0108] The gain determination module 802 is used to determine the tracking error based on the state vector and pre-configured target parameters, and to determine the error transformation gain based on the tracking error and a pre-configured first function.

[0109] The variable fitting module 803 is used to determine a first input vector based on the state vector, tracking error, and the rate of change of the state vector determined according to the state vector; determine a first basis function vector based on the pre-configured main neural network; and determine a first update weight based on the first basis function vector, the first input vector, gain data, and the acquired first historical weight.

[0110] The perturbation fitting module 804 is used to determine the second input vector based on the tracking error and the first output current, determine the second basis function vector based on the pre-configured perturbation observation neural network, determine the second update weight based on the second input vector, the second basis function vector and the obtained second historical weight, and determine the perturbation vector based on the pre-determined first auxiliary state variable, the observer gain and the second update weight.

[0111] The vector output module 805 outputs the final control vector based on the first input vector, tracking error, gain data, first update weight, and disturbance vector.

[0112] In some embodiments, the gain determination module 802 is further configured to implement: The gain determination module 802 determines the time-varying error boundary based on the pre-configured initial error limit, the pre-configured expected steady-state error tolerance, the pre-configured convergence rate factor, and the time when the rectifier is disturbed.

[0113] The gain determination module 802 limits the range of tracking error based on the time-varying error boundary, and determines the error transformation gain based on the tracking error and the time-varying error boundary.

[0114] In some embodiments, the first output data includes a first output voltage and a first output current, the first input data includes a second output voltage, and the data processing module 801 is further configured to implement: The data processing module 801 acquires the first output data from the AC side of the sending end and the first input data from the input rectifier. Based on the first output data and the pre-configured first coordinate system, the data processing module 801 determines the current component and the first voltage component of the first output data on the corresponding coordinate axis, and the second voltage component of the first input data on the corresponding coordinate axis.

[0115] The data processing module 801 determines the first-order tracking element based on the current component, the first voltage component, and the second voltage component, and determines the instantaneous active power and instantaneous reactive power of the rectifier based on the current component and the first voltage component.

[0116] The data processing module 801 determines the target power balance equation based on the rectifier's basic parameters and historical output voltage, instantaneous active power, and instantaneous reactive power.

[0117] The data processing module 801 expresses the voltage state variable using the first output voltage and determines the state vector based on the voltage state variable and instantaneous reactive power.

[0118] In some embodiments, the variable fitting module 803 is also used to implement: The main neural network includes multiple first neurons. The variable fitting module 803 determines multiple first basis functions based on the basic parameters of each first neuron, and determines the first basis function vector based on the multiple first basis functions.

[0119] The variable fitting module 803 determines the first update weight based on the first basis function vector, gain data, first input vector, first historical weight, and pre-configured first weight update law.

[0120] In some embodiments, the first weight update law in the variable fitting module 803 is determined based on Lyapunov stability theory, and the first weight update law is:

[0121] in, Represents the first updated weight. This represents the pre-configured first learning rate. Represents the first basis function vector. Represents the first input vector. Represents the error transformation gain. This represents the pre-configured first damping coefficient. Represents the first historical weight.

[0122] In some embodiments, the perturbation fitting module 804 is further configured to implement: The perturbation observation neural network includes multiple second neurons. The perturbation fitting module 804 determines multiple second basis functions based on the basic parameters of each second neuron, and determines the second basis function vector based on the multiple second basis functions.

[0123] The perturbation fitting module 804 determines the second update weight based on the second basis function vector, the second input vector, the second historical weight, and the pre-configured second weight update law.

[0124] In some embodiments, the second weight update law in the perturbation fitting module 804 is determined based on Lyapunov stability theory, and the second weight update law is:

[0125] in, Represents the second update weight. This represents the pre-configured second learning rate. Represents the second basis function vector. Represents the second input vector. This represents the pre-configured first damping coefficient. This represents the second historical weight.

[0126] The control device 800 for suppressing voltage fluctuations in this application embodiment can be an electronic device or a component within an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), television (TV), ATM, or self-service machine, etc. This application embodiment does not specifically limit the specific type of device.

[0127] The control device for suppressing voltage fluctuations provided in this application embodiment can achieve... Figures 1 to 7 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.

[0128] This application also provides an electronic device, such as... Figure 9 As shown, the electronic device 900 includes a processor 901 and a memory 902. The memory 902 stores a program or instructions that can run on the processor 901. When the program or instructions are executed by the processor 901, they implement the various steps of the control method embodiment for suppressing voltage fluctuations described above and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0129] The memory 902 can be used to store software programs and various data. The memory 902 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.). Furthermore, the memory 902 may include volatile memory or non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM). The memory 902 in this embodiment includes, but is not limited to, these and any other suitable types of memory.

[0130] Processor 901 may include one or more processing units; optionally, processor 901 integrates an application processor and a modem processor, wherein the application processor mainly handles operations involving the operating system, user interface, and applications, and the modem processor mainly handles wireless communication signals, such as a baseband processor. It is understood that the aforementioned modem processor may also not be integrated into processor 901.

[0131] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the control method embodiment for suppressing voltage fluctuations described above, and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0132] This application also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described control method embodiments for suppressing voltage fluctuations, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0133] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0134] This application also provides a computer program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the control method embodiment for suppressing voltage fluctuations described above, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0135] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0136] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A control method for suppressing voltage fluctuations, characterized in that, A rectifier used in a power transmission system, wherein the power transmission system further includes a sending-end AC side, the sending-end AC side being electrically connected to the rectifier, and the control method for suppressing voltage fluctuations includes: Acquire the first output data of the AC side at the sending end and the first input data of the rectifier, determine the target power balance equation between the AC side at the sending end and the rectifier based on the first output data and the first input data, and determine the state vector based on the target power balance equation and the first output data; The tracking error is determined based on the state vector and the pre-configured target parameters, and the error transformation gain is determined based on the tracking error and the pre-configured first function. Based on the state vector, tracking error, and the rate of change of the state vector determined according to the state vector, a first input vector is determined; a first basis function vector is determined based on the pre-configured main neural network; and a first update weight is determined based on the first basis function vector, the first input vector, the gain data, and the obtained first historical weight. The second input vector is determined based on the tracking error and the first output data. The second basis function vector is determined based on the pre-configured perturbation observation neural network. The second update weight is determined based on the second input vector, the second basis function vector, and the obtained second historical weight. The perturbation vector is determined based on the pre-determined first auxiliary state variable, the observer gain, and the second update weight. The final control vector is output based on the first input vector, the tracking error, the gain data, the first update weight, and the disturbance vector.

2. The control method for suppressing voltage fluctuations according to claim 1, characterized in that, The step of determining the error transformation gain based on the tracking error and a pre-configured first function includes: The time-varying error boundary is determined based on the pre-configured initial error limit, the pre-configured expected steady-state error tolerance, the pre-configured convergence rate factor, and the time when the rectifier is disturbed. The tracking error range is limited based on the time-varying error boundary, and the error transformation gain is determined based on the tracking error and the time-varying error boundary.

3. The control method for suppressing voltage fluctuations according to claim 1, characterized in that, The first output data includes a first output voltage and a first output current, and the first input data includes a second output voltage. The steps of acquiring the first output data from the AC side of the sending end and the first input data to the rectifier, determining the target power balance equation between the AC side of the sending end and the rectifier based on the first output data and the first input data, and determining the state vector based on the target power balance equation and the first output data include: Acquire the first output data of the AC side of the sending end and the first input data of the rectifier. Based on the first output data and the pre-configured first coordinate system, determine the current component and the first voltage component of the first output data on the corresponding coordinate axis, and the second voltage component of the first input data on the corresponding coordinate axis. A first-order tracking element is determined based on the current component, the first voltage component, and the second voltage component, and the instantaneous active power and instantaneous reactive power of the rectifier are determined based on the current component and the first voltage component. The target power balance equation is determined based on the rectifier's basic parameters and historical output voltage, the instantaneous active power, and the instantaneous reactive power. The voltage state variable is expressed using the first output voltage, and the state vector is determined based on the voltage state variable and the instantaneous reactive power.

4. The control method for suppressing voltage fluctuations according to claim 1, characterized in that, The process of determining the first basis function vector based on the pre-configured main neural network, and determining the first update weight based on the first basis function vector, the first input vector, the gain data, and the acquired first historical weights, includes: The main neural network includes multiple first neurons, and multiple first basis functions are determined based on the basic parameters of each first neuron, and a first basis function vector is determined based on the multiple first basis functions; The first update weight is determined based on the first basis function vector, the gain data, the first input vector, the first historical weight, and the pre-configured first weight update law.

5. The control method for suppressing voltage fluctuations according to claim 4, characterized in that, The first weight update law is determined based on Lyapunov stability theory, and the first weight update law is: in, Represents the first updated weight. This represents the pre-configured first learning rate. This represents the first basis function vector. This represents the first input vector. This represents the error transformation gain. This represents the pre-configured first damping coefficient. This represents the first historical weight.

6. The control method for suppressing voltage fluctuations according to claim 1, characterized in that, The process of determining the second input vector based on the tracking error and the first output current, determining the second basis function vector based on the pre-configured perturbation observation neural network, and determining the second update weight based on the second input vector, the second basis function vector, and the obtained second historical weights includes: The perturbation observation neural network includes multiple second neurons, each of which determines multiple second basis functions based on the basic parameters of its respective second neuron, and determines a second basis function vector based on the multiple second basis functions. The second update weight is determined based on the second basis function vector, the second input vector, the second historical weight, and the pre-configured second weight update law.

7. The control method for suppressing voltage fluctuations according to claim 6, characterized in that, The second weight update law is determined based on Lyapunov stability theory, and the second weight update law is as follows: in, This represents the second updated weight. This represents the pre-configured second learning rate. Represents the second basis function vector. Represents the second input vector. This represents the pre-configured first damping coefficient. This represents the second historical weight.

8. A control device for suppressing voltage fluctuations, characterized in that, A rectifier used in a power transmission system, the power transmission system further including a sending-end AC side, the sending-end AC side being electrically connected to the rectifier, and the control device for suppressing voltage fluctuations comprising: The data processing module is used to acquire the first output data of the AC side of the sending end and the first input data of the rectifier, determine the target power balance equation between the AC side of the sending end and the rectifier based on the first output data and the first input data, and determine the state vector based on the target power balance equation and the first output data. A gain determination module is used to determine the tracking error based on the state vector and pre-configured target parameters, and to determine the error transformation gain based on the tracking error and a pre-configured first function. The variable fitting module is used to determine a first input vector based on the state vector, the tracking error, and the rate of change of the state vector determined according to the state vector; determine a first basis function vector based on a pre-configured main neural network; and determine a first update weight based on the first basis function vector, the first input vector, the gain data, and the acquired first historical weight. The perturbation fitting module determines a second input vector based on the tracking error and the first output data, determines a second basis function vector based on a pre-configured perturbation observation neural network, determines a second update weight based on the second input vector, the second basis function vector, and the obtained second historical weight, and determines a perturbation vector based on a pre-determined first auxiliary state variable, observer gain, and the second update weight. The vector output module is used to output a final control vector based on the first input vector, the tracking error, the gain data, the first update weight, and the disturbance vector.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a program or instructions that run on the processor, the program or instructions being executed by the processor to implement the steps of the control method for suppressing voltage fluctuations as described in any one of claims 1 to 7.

10. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or instructions are executed by the processor, they implement the steps of the control method for suppressing voltage fluctuations as described in any one of claims 1 to 7.