CN-122000950-A - Reactive power output adjusting method for high-voltage static synchronous compensator

CN122000950ACN 122000950 ACN122000950 ACN 122000950ACN-122000950-A

Abstract

The application discloses a reactive output adjusting method of a high-voltage static synchronous compensator, which comprises a three-layer closed-loop framework of a sensing layer, a control layer and an execution layer, and further comprises a layered multi-target model prediction MPC control unit, wherein the control unit comprises a data preprocessing module, an online parameter identification module, an upper global reactive power distribution module, a lower bridge arm level MPC optimization module, a multi-target weight self-adaptation module and an AI error compensation module. The structure can reduce the calculation complexity of the MPC algorithm, reduce single-period control delay to within 5ms and meet the real-time control requirement of high-voltage SVG under the premise of ensuring reactive power regulation precision.

Inventors

LUO CIWEI
LIN SHIBIN
YANG WENJING
Guan Yinqia

Assignees

杭州耐立电气有限公司

Dates

Publication Date: 20260508
Application Date: 20260313

Claims (10)

1. The reactive output adjusting method of the high-voltage static synchronous compensator is characterized by comprising a three-layer closed-loop framework of a sensing layer, a control layer and an execution layer, and further comprising a hierarchical multi-target model prediction control MPC control unit, wherein the control unit comprises a data preprocessing module, an online parameter identification module, an upper global reactive power distribution module, a lower bridge arm level MPC optimization module, a multi-target weight self-adaptation module and an AI error compensation module, and concretely comprises the following steps: s1, a sensing layer collects power grid side operation data and SVG equipment state data, and original data are transmitted to a data preprocessing module for processing to obtain standardized effective data; s2, an online parameter identification module receives standardized effective data, identifies parameters of a power grid and SVG key models in real time, and transmits the identified parameters to a corresponding downstream module; S3, synchronously receiving the total reactive power demand of the power grid, the state data of each bridge arm of the SVG and the identification parameters in the standardized effective data by the upper global reactive power distribution module, distributing the total reactive power demand into reference reactive power values of each bridge arm, and transmitting the reference reactive power values to the corresponding downstream modules; s4, the multi-target weight self-adaptive module receives working condition parameters and real-time values of all control sub-targets in the standardized effective data, calculates and determines dynamic weights, and transmits the dynamic weights to the downstream optimization module; s5, a lower bridge arm MPC optimization module receives a bridge arm reference reactive value, key model parameters and dynamic weights, and calculates and obtains a predicted reactive value; S6, the AI error compensation module synchronously receives the bridge arm reference reactive value, the key model parameter and the predicted reactive value, outputs error compensation quantity based on the BP neural network model after off-line training, corrects the predicted reactive value through the compensation quantity to obtain a final control signal and transmits the final control signal to an execution layer; and S7, the execution layer receives the final control signal, drives the SVG main circuit to output target reactive power, and simultaneously feeds back the SVG actual output reactive power and the equipment running state to the sensing layer to form complete closed-loop control.
2. The reactive output adjusting method of the high-voltage static synchronous compensator according to claim 1, wherein the specific implementation process of the parameter identification in the step S2 comprises the following steps that an online parameter identification module receives standardized effective data output by a data preprocessing module, wherein the standardized effective data comprises grid voltage, SVG output current and bridge arm capacitor voltage, an improved extended Kalman filter EKF algorithm is adopted to identify key model parameters in real time, the key model parameters comprise grid impedance, SVG equivalent inductance and equivalent capacitance, and the identified parameters are respectively transmitted to an upper-layer global reactive power distribution module, a lower-layer bridge arm level optimization module and an AI error compensation module.
3. The reactive output adjusting method of the high-voltage static synchronous compensator according to claim 2, wherein the improved extended kalman filter EKF algorithm of the online parameter identification module in S2 has a core improvement point of an adaptive kalman gain adjusting strategy, and the specific implementation process includes: S21, defining a convergence judging index, wherein the convergence judging index comprises a residual error between an observed value and a predicted value obtained based on a prediction model; s22, setting a residual error threshold value which is the product of the observed value and a preset proportionality coefficient; S23, adjusting Kalman gain according to the magnitude relation between the residual error and the threshold value, and increasing the Kalman gain to accelerate the algorithm convergence speed when the absolute value of the residual error is larger than the threshold value; S24, updating key model parameters of the power grid and the SVG according to a preset identification period, wherein the key model parameters comprise power grid impedance, SVG equivalent inductance and equivalent capacitance.
4. The reactive power output adjusting method of the high-voltage static synchronous compensator according to claim 1, wherein the specific implementation process of reactive power distribution in the step S3 comprises the following steps that an upper global reactive power distribution module receives state data output by a data preprocessing module, wherein the state data comprises a power grid total reactive power demand, capacitor voltage and switching device temperature of each bridge arm, a dynamic distribution algorithm based on capacity weighting and state constraint is combined with the key model parameters output by the step S2, the capacitor voltage balance weight and the temperature weight of each bridge arm are calculated first, comprehensive weight is obtained, the total reactive power demand is distributed into reference reactive power values of each bridge arm according to the ratio of the comprehensive weight, and the reference reactive power values are transmitted to a lower bridge arm MPC optimizing module and an AI error compensating module, so that reasonable distribution of reactive power demands and equipment state protection are achieved.
5. The reactive power output adjusting method of the high-voltage static synchronous compensator according to claim 4, wherein the capacity weighted combination state constraint dynamic allocation algorithm of the upper global reactive power allocation module in S3 comprises the following specific implementation steps: S31, calculating the balance weight of the capacitance voltage of each bridge arm, wherein the weight is inversely proportional to the balance degree of the capacitance voltage of each bridge arm, and the balance degree of the capacitance voltage of a single bridge arm and the average value of the capacitance voltages of all bridge arms is calculated; S32, calculating the temperature weight of each bridge arm, determining the weight by adopting a piecewise linear function according to the difference value between the real-time temperature of the bridge arm switching device and the safety operation threshold value, wherein the weight is reset to zero when the temperature exceeds the safety threshold value as the temperature is closer to the safety threshold value; S33, carrying out product operation on the capacitance-voltage balance weight and the temperature weight of the same bridge arm to obtain the comprehensive weight of each bridge arm; And S34, distributing the total reactive power demand of the power grid to the corresponding bridge arms based on the proportion of the comprehensive weight of each bridge arm to the total weight of all the bridge arms, so as to obtain the reference reactive power value of each bridge arm, wherein the total sum of the reference reactive power values of all the bridge arms is equal to the total reactive power demand of the power grid.
6. The reactive power output adjusting method of the high-voltage static synchronous compensator according to claim 1, wherein the multi-target weight self-adaptive module in S4 comprises an AHP fusion entropy weight strategy, and the specific implementation process comprises: s41, establishing a hierarchical structure comprising a target layer, a criterion layer and a scheme layer, wherein the target layer is multi-target optimization, the criterion layer is each control sub-target, the scheme layer is each switch state, a judgment matrix is constructed by adopting a preset scale method, and subjective basic weights of each control sub-target are obtained by calculating feature vectors of the judgment matrix and carrying out consistency inspection; S42, carrying out standardized processing on real-time values of all control sub-targets collected under different working conditions to eliminate dimension influence, and calculating information entropy of each control sub-target based on an information entropy principle, wherein the smaller the information entropy is, the larger the fluctuation of the target data is, the more information is contained, and the larger the corresponding objective weight is; s43, adopting a linear weighting fusion mode to combine the subjective basic weight and the objective weight, and distributing the duty ratio of the subjective basic weight and the objective weight through a preset proportionality coefficient to obtain the dynamic weight of each control sub-target; s44, repeating the steps according to a preset period to update the dynamic weight, so as to ensure that the weight can adapt to the working condition change.
7. The reactive output adjusting method of the high-voltage static synchronous compensator according to claim 1, wherein the lower bridge arm stage MPC optimizing module in S5 includes a low-complexity strategy of switch state grouping, simplified model, and quick solution, and the specific implementation process includes: S51, dividing the switch states of a plurality of submodules of a single bridge arm of the SVG multilevel topology into three types of magnetism increasing groups, magnetism reducing groups and holding groups, wherein the magnetism increasing groups comprise key switch states capable of increasing the output voltage of the bridge arm to increase reactive power output, the magnetism reducing groups comprise key switch states capable of reducing the output voltage of the bridge arm to reduce reactive power output, the holding groups comprise key switch states capable of stabilizing the output voltage of the bridge arm to maintain the reactive power output unchanged, and only the key switch states in the three types are selected for subsequent prediction; S52, based on kirchhoff voltage law, neglecting secondary parameters including parasitic resistance and stray capacitance of the submodule, constructing a bridge arm level low-order simplified model, and reducing the complexity of model operation; And S53, solving an optimal switching state by adopting a sparse linear programming algorithm, constructing a comprehensive objective function based on the dynamic weight output by the multi-objective weight self-adaptive module, including reactive compensation precision and switching loss targets, using the optimal switching state of the previous control period as an initial value to narrow a search range, and quickly solving to obtain a predicted reactive value.
8. The reactive output adjustment method of a high-voltage static synchronous compensator according to claim 1, wherein the training process of the BP neural network model of the AI error compensation module in S6 comprises: s601, building an SVG simulation platform, simulating three typical working conditions of steady state, transient state and complex state, and collecting input parameters and output parameters under each working condition, wherein the input parameters comprise bridge arm reference reactive values, SVG equivalent inductances, equivalent capacitances and grid impedances, the output parameters comprise MPC predicted reactive values and SVG actual output reactive values, and error data are obtained through differences of the MPC predicted reactive values and the SVG actual output reactive values; s602, denoising and normalizing the acquired input parameters and error data, and dividing a training set and a testing set; And S603, constructing a BP neural network, wherein the BP neural network comprises an input layer, an implicit layer and an output layer, determining the node number and the activation function of each layer, training the neural network by adopting a preset loss function and an optimizer, verifying the performance of the model by a test set, if the performance does not meet the preset requirement, retraining after the fitting is restrained by adopting a preset technology, and solidifying the model to a control chip after the training is completed.
9. The reactive output adjustment method of a high-voltage static synchronous compensator according to claim 1, wherein the BP neural network model of the AI error compensation module in S6 comprises the following on-line operation process: S611, carrying out normalization processing on key model parameters output by an online parameter identification module and bridge arm reference reactive values output by an upper global reactive power distribution module, and then inputting the normalized values into a trained BP neural network model to obtain an error compensation quantity; S612, correcting the predicted reactive value output by the lower bridge arm MPC optimization module by using the error compensation quantity to obtain a final control signal, so that SVG reactive compensation errors meet the preset precision requirement.
10. The reactive power output adjusting method of the high-voltage static synchronous compensator according to claim 1, wherein the executing layer in the step S7 comprises a switching device driving circuit, an SVG main circuit and a protection circuit, wherein the driving circuit receives a final control signal and generates driving pulses to drive the SVG main circuit to output target reactive power, the protection circuit monitors overcurrent, overvoltage and overtemperature states of the SVG main circuit in real time, outputs protection signals when abnormal states are monitored, and cuts off the driving circuit to protect equipment.

Description

Reactive power output adjusting method for high-voltage static synchronous compensator Technical Field The invention relates to the technical field of high voltage electricity, in particular to a reactive power output adjusting method of a high voltage static synchronous compensator. Background The high-voltage static synchronous compensator SVG is core equipment for reactive compensation and voltage stabilization control of a power grid, and the control performance of the SVG directly relates to the power quality and the operation stability of the power grid. The model predictive control MPC has the advantages of strong multi-constraint processing capability, excellent nonlinear control effect and the like, and becomes a main flow research direction of SVG reactive power regulation. However, in practical engineering application, the prior art has the following core bottlenecks, and it is difficult to meet the high performance requirement of the high-voltage scene: 1. The real-time performance is insufficient, and the high-voltage SVG adopts modular multilevel converters MMC, neutral point clamped NPC and other multilevel topologies. The traditional MPC needs to traverse all switching states, taking 20 submodules of a single bridge arm of the MMC as an example, the number of the switching states is 2 20, the calculation complexity is exponentially increased, the single-period control delay is usually more than 8ms, and the response requirement of the high-voltage SVG millisecond level less than or equal to 5ms cannot be met; 2. The problem of poor multi-target cooperativity is that the existing scheme takes reactive compensation precision as a single control target, and key indexes such as switching loss, power grid voltage harmonic distortion rate THD, multi-level topological capacitor voltage balance and the like are ignored. The partial multi-target MPC scheme adopts a fixed weight weighting summation mode, and the weights cannot be dynamically adjusted according to working conditions, so that the running efficiency of equipment is low, and the performance of a power grid side is unstable; 3. The problem of insufficient robustness is that the traditional MPC relies on an accurate mathematical model, but the high-voltage power grid has complex working conditions such as impedance fluctuation, load impact, voltage sag and the like, model parameters are easy to deviate from design values, so that control accuracy is reduced and even a system is unstable, and meanwhile, the traditional error correction means are mostly fixed compensation, and self-adaptive correction is realized without combining a data driving technology. In the prior art, a solution to the real-time problem mostly adopts a fixed model to reduce the order or simplify a switch state set, but the control precision loss is easy to cause, a solution to the multi-objective problem lacks a dynamic weight adjustment mechanism, and a solution to the robustness problem does not form a cooperative mechanism of parameter identification and error compensation. Therefore, there is a need to develop a reactive output adjusting method for a high-voltage static synchronous compensator to solve the problems in the prior art. Disclosure of Invention The invention aims to provide a reactive output adjusting method of a high-voltage static synchronous compensator, which can reduce the calculation complexity of an MPC algorithm on the premise of ensuring reactive adjustment precision, reduce single-period control delay to within 5ms, meet the real-time control requirement of high-voltage SVG, and has simple structure and convenient use so as to solve the problems in the background technology. In order to achieve the above purpose, the present invention provides the following technical solutions: The reactive output adjusting method of the high-voltage static synchronous compensator comprises a three-layer closed loop architecture of a sensing layer, a control layer and an execution layer, and further comprises a hierarchical multi-target model prediction control MPC control unit, wherein the control unit comprises a data preprocessing module, an online parameter identification module, an upper global reactive power distribution module, a lower bridge arm MPC optimization module, a multi-target weight self-adaptation module and an AI error compensation module, and concretely comprises the following steps: s1, a sensing layer collects power grid side operation data and SVG equipment state data, and original data are transmitted to a data preprocessing module for processing to obtain standardized effective data; s2, an online parameter identification module receives standardized effective data, identifies parameters of a power grid and SVG key models in real time, and transmits the identified parameters to a corresponding downstream module; S3, synchronously receiving the total reactive power demand of the power grid, the state data of each bridge arm of the SVG a