CN-121485159-B - Data driving optimal control method for grid-connected inverter

CN121485159BCN 121485159 BCN121485159 BCN 121485159BCN-121485159-B

Abstract

The invention discloses a data driving optimal control method for a grid-connected inverter. The method comprises the steps of constructing DeePC optimal controllers of grid-connected inverters on the power grid side, acquiring a data matrix of the grid-connected inverters in a preset history period and an initial track of a preset adjacent period, inputting the data matrix into the controllers, setting parameters aiming at different control targets and types of the grid-connected inverters, and outputting an optimal input data sequence after processing, so that sine wave pulse width modulation wave signals are generated and act on the grid-connected inverters to realize data driving optimal control. The method of the invention automatically perceives and adapts to the dynamic characteristics of the actual power grid through real-time data driving predictive control, and can display or enhance the control behavior of the grid-built or grid-following inverter only through modifying the cost parameter, thereby realizing multi-mode flexible, optimal and coordinated control, having excellent dynamic response and steady-state performance, being beneficial to safe and stable operation and improving the capability of adapting to complex and changeable operation conditions and power grid strength.

Inventors

HUANG LINBIN
Leng Ruohan
Cai Hanzhang
XIN HUANHAI
ZHANG GUOSHENG

Assignees

浙江大学

Dates

Publication Date: 20260508
Application Date: 20260108

Claims (8)

1. A data-driven optimal control method for a grid-connected inverter, comprising: step 1), constructing DeePC optimal controllers of grid-connected inverters at a power grid side; Step 2) acquiring input and output data sequences of the grid-connected inverter in a preset history period, constructing a data matrix, and acquiring the input and output data sequences of the grid-connected inverter in a preset adjacent period before a control moment as an initial track; step 3) inputting the data matrix and the initial track into DeePC an optimized controller, setting different weights and coupling matrixes in DeePC the optimized controller according to different control targets of the grid-connected inverter, setting different cost parameters to form different inverter types, and outputting an optimal input data sequence after processing; Step 4) obtaining frequency deviation and modulation wave signal voltage amplitude of the grid-connected inverter according to the optimal input data sequence, further generating sine wave pulse width modulation (SPWM) wave signals and acting on the grid-connected inverter to realize data driving optimal control of the grid-connected inverter; The DeePC optimizing controller optimizes the decision variable g to finally obtain an optimal input data sequence; The DeePC optimizing controller includes output cost item The method comprises the following steps: ; Wherein, the Is a vector Is of the quadratic form of (2) , =y-r, For the output data sequence of the grid-connected inverter to be optimized, For a preset reference output data sequence, Is a punishment matrix; 、 And First, second and third weights corresponding to active power control, reactive power/voltage control and voltage directional control, respectively; The frequency deviation of the grid-connected inverter is the time t, t+1, and the time t+N-1, wherein N is the length of the optimal input data sequence; And Active and reactive power of the grid-connected inverter at times t, t+1, & gt, t+N-1, respectively; is the offset; And Respectively presetting reference active power and reactive power; is an identity matrix with a dimension of N; Is of the quadratic form, s= 、 Or (b) , Is an identity matrix with the dimension of 2N; And A first coupling matrix and a second coupling matrix respectively; And The dq-axis component of the voltage at the output port of the grid-connected inverter at times t, t+1, & gt, t+n-1, respectively; And Respectively the dq-axis component of the preset reference voltage.
2. The method for data driving optimal control of grid-connected inverter according to claim 1, wherein in the step 1), deePC optimal controller is as follows: ; ; wherein g is a decision variable to be optimized; And Input and output relaxation variables, respectively; And Respectively inputting and outputting data sequences of the grid-connected inverter to be optimized; A constraint set for the input and output data sequences; Is a vector Is of the quadratic form of (2) , =U or y-R, P is a matrix, p=r or Q, And Respectively a cost matrix and a punishment matrix; Is the two norms of the vector S Square of (s=s) Or (b) , = , Is an output cost term; outputting a data sequence for a preset reference; In order to regularize the term(s), 、 And Respectively is 、 And Is a scaling factor of (2); And A first input and a first output track of the grid-connected inverter for a prediction history period respectively, And A second input track and a second output track of the grid-connected inverter for the prediction history period respectively; And Respectively the preset adjacent time periods before the control time Input and output data sequences of the grid-connected inverter.
3. The method for optimal control of a grid-connected inverter as set forth in claim 2, wherein in said step 2), an input data sequence of the grid-connected inverter for a predetermined history period T is obtained And outputting the data sequence Respectively constructing input data sequences Hankel matrix of (A) And output Hankel matrix Then respectively performing block operation to obtain input data sequence Constructed input data matrix Output data sequence Constructed output data matrix , , Finally obtaining a first input track of the grid-connected inverter in the prediction history period And a first output track A second input track And a second output track ; Is a definition symbol; Input data sequence The frequency deviation of the grid-connected inverter and the amplitude of the dq axis of the modulation wave voltage are included, and a data sequence is output Including the dq-axis component of the voltage, the dq-axis component of the current, the active power and the reactive power of the grid-tied inverter output port.
4. The method for data driving optimal control of a grid-connected inverter as set forth in claim 1, wherein in said step 3), when the control target of the grid-connected inverter is voltage-oriented control, a q-axis component of a preset reference voltage is set And a third weight Is greater than a first preset threshold; When the control target of the grid-connected inverter is the voltage amplitude, setting a second weight Greater than a second predetermined threshold, a second coupling matrix The settings were as follows: ; Wherein, the And The voltage amplitude and reactive power tracking state quantity are respectively, , The tracking voltage amplitude is indicated at 1, A value of 0 means that the tracking voltage amplitude is not targeted, When 1 is used to represent tracking reactive power, A value of 0 means that tracking reactive power is not targeted; Is Q-V sag factor; Is Cronecker product; is an identity matrix with a dimension of N; when the control target of the grid-connected inverter is active power, setting a first weight Is larger than a third preset threshold value, and a first coupling matrix is set The following are provided: 。
5. the method for data driving optimal control of a grid-connected inverter as recited in claim 1, wherein in the step 3), the cost parameter includes a first weight Second weight Third weight First coupling matrix A second coupling matrix And offset amount When the inverter type formed is a following-net type inverter GFL, the following-net type inverter GFL is set , =0, First coupling matrix And a second coupling matrix The following are provided: = ; = ; Wherein, the Is Cronecker product; is an identity matrix with a dimension of N; when the inverter type formed is a grid-type inverter GFM, the configuration is that First coupling matrix A second coupling matrix And offset amount The following are provided: = ; = ; = ; ; Wherein, the To act on Is a discrete rocking equation operator; Is virtual inertia; is an initial condition vector; optimizing the sampling time of the controller for DeePC; is a damping coefficient; is a backward differential operator; thereby realizing flexible switching between the grid-connected inverter GFL and the grid-structured inverter GFM modes.
6. The method of claim 1, wherein in step 4), the optimal input data sequence includes an optimal frequency deviation of the grid-connected inverter and an amplitude of a dq axis of the optimal modulation wave voltage, a modulation wave signal phase angle of the grid-connected inverter is obtained according to the optimal frequency deviation, the amplitude of the dq axis of the optimal modulation wave voltage and the modulation wave signal phase angle are subjected to Park inverse transformation to obtain a three-phase modulation wave signal, and the three-phase modulation wave signal is applied to a Pulse Width Modulation (PWM) generator to generate a sine wave pulse width modulation (SPWM) wave signal and is applied to the grid-connected inverter.
7. An electronic device comprising a memory and a processor coupled to each other, wherein the memory stores program data and the processor invokes the program data to perform the method of any of claims 1-6.
8. A computer readable storage medium having stored thereon program data, which when executed by a processor, implements the method of any of claims 1-6.

Description

Data driving optimal control method for grid-connected inverter Technical Field The invention relates to an inverter control method, in particular to a data driving optimal control method for a grid-connected inverter. Background In the global context of coping with climate change, there is currently a need to advance large scale grid-connection of renewable energy sources. Unlike conventional fossil fuel power plants that rely on synchronous generators, renewable energy sources are mainly connected to the grid through power electronic converters. This transition, while driving the development of clean energy sources, presents new technical challenges. With the large-scale access of renewable energy sources to an electric power system, a main stream mode is realized by realizing remote transmission of new energy bases such as wind power, photovoltaic and the like through high-voltage direct current transmission. However, the high proportion of the power electronization characteristics of the new energy units are contradictory to the weak supporting capacity of the power grid. The traditional converter control mostly adopts a multi-ring control structure based on PID, and comprises a synchronous unit, an outer ring power/voltage control, an inner ring current control and the like. These control loops are typically parameter-set based on a simplified grid model (e.g., stand-alone infinity system) and are fixed after commissioning. However, the actual grid is a highly complex, time-varying and often unknown dynamic system, with significant differences from the simplified model. Such "model mismatch" may lead to reduced control performance and even to system oscillations or instability. Studies have shown that the problem of small signal oscillations in practical operation of many wind farms is due to the inability of controllers tuned under ideal test conditions (e.g. connected to an ideal voltage source) to adapt to the dynamics of a real grid. To address this problem, researchers have attempted to employ robust control or adaptive control methods to cope with the uncertainty and unknown dynamics of the grid. However, due to the complex and variable power grid conditions, it is difficult to design a fixed controller to accommodate all possible scenarios. Disclosure of Invention In order to solve the problems in the background art, the invention provides a data driving optimal control method for a grid-connected inverter. The method can solve the problems of poor adaptability, slow response, insufficient stability, high cost and the like of the traditional control strategy in a novel power system, and dynamically adjusts the frequency, voltage and power reference value of the converter by collecting power grid operation data in real time, constructing a data prediction model and solving an optimal control sequence on line. The invention can flexibly realize smooth switching of two control modes of a follow-up network and a construction network, accurately configures dynamic response characteristics of active-frequency and reactive-voltage by designing a weight matrix and a coupling term in a cost function, effectively solves the stability problem caused by model mismatch in the traditional converter control in a complex power grid, and has stronger adaptability, robustness and comprehensive control performance. The technical scheme adopted by the invention is as follows: The data driving optimal control method for the grid-connected inverter comprises the following steps: step 1) constructing a DeePC optimized controller for introducing regularization term of the grid-connected inverter at the power grid side. Step 2) acquiring input and output data sequences of the grid-connected inverter in a preset history period, constructing a data matrix, and acquiring the input and output data sequences of the grid-connected inverter in a preset adjacent period before a control moment as an initial track. Step 3) inputting the data matrix and the initial track into DeePC an optimized controller, setting different weights and coupling matrixes in DeePC the optimized controller according to different control targets of the grid-connected inverter, setting different cost parameters to form different inverter types, and outputting an optimal input data sequence after processing. And 4) obtaining the frequency deviation and the modulating wave signal voltage amplitude of the grid-connected inverter according to the optimal input data sequence, further generating sine wave pulse width modulation SPWM (Sine Wave Pulse Width Modulation) wave signals and acting on the grid-connected inverter to realize the data driving optimal control of the grid-connected inverter. In the step 1), the DeePC optimizing controller is specifically as follows: wherein g is a decision variable to be optimized; And Input and output relaxation variables, respectively; And Respectively inputting and outputting data sequences of the grid-connected inve