Load harmonic suppression method and system for remote microgrid

Abstract

The invention relates to the technical field of harmonic suppression, and discloses a load harmonic suppression method and system for a remote micro-grid, and the method comprises the steps: analyzing the stator flux linkage data of a motor in the remote micro-grid to determine a filtering parameter and SVG reactive data, analyzing the SVG reactive data to obtain a DC side capacitor, and carrying out the electromagnetic torque simulation. Generating stator flux linkage response data and determining a response speed; determining permittivity parameters according to the direct-current side capacitance, and calculating a damping coefficient in combination with the response speed and a preset model; carrying out harmonic suppression processing on the voltage transient data based on the damping coefficient, and if the suppression effect does not reach the standard, decomposing the voltage transient data to optimize SVG impedance matching parameters; extracting an angle deviation in the voltage transient data to determine the switching frequency of the converter valve, further constructing an electromagnetic torque ripple response model to optimize SVG reactive data, and generating a load harmonic suppression strategy in combination with optimized filtering parameters and the switching frequency; and the voltage stability and the electric energy quality of the independent micro-grid in the remote area are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of harmonic suppression, in particular to a load harmonic suppression method and system for remote micro-grid. BACKGROUND

[0002] The load harmonic problem of the micro-grid in remote areas is particularly prominent due to the simple grid structure, dispersed equipment and large load fluctuation. Harmonics can cause motor overheating, protection misoperation, communication interference and other problems, which seriously affect the safety of the grid and power quality. The existing harmonic suppression methods mostly rely on centralized filtering devices or static reactive power compensation equipment, which are difficult to adapt to the dynamic changing working environment of the micro-grid.

[0003] Therefore, how to improve the response speed and accuracy of the existing harmonic suppression method for the remote micro-grid has become a technical problem to be solved by the person skilled in the art. SUMMARY

[0004] The present application provides a load harmonic suppression method and system for remote micro-grid, which solves the problem of how to improve the response speed and accuracy of the existing harmonic suppression method for the remote micro-grid.

[0005] To solve the above technical problems, the present application provides a load harmonic suppression method for remote micro-grid, comprising:

[0006] Obtaining stator flux linkage data of a motor in a remote micro-grid, and performing AC filter characteristic matching according to the stator flux linkage data to obtain filter parameters to determine SVG reactive power data;

[0007] Analyzing the SVG reactive power data to obtain DC side capacitor data to simulate the electromagnetic torque of the motor, generate stator flux linkage dynamic response data to determine the response speed;

[0008] Determining the capacitance rate parameter according to the DC side capacitor data, combining with the response speed and a preset control loop model to calculate a damping coefficient;

[0009] Based on the damping coefficient, obtaining voltage transient data of the remote micro-grid to perform harmonic suppression processing based on the filter parameters to calculate an inhibition fitness value;

[0010] When the inhibition fitness value does not meet the standard, decomposing the voltage transient data to adjust the impedance matching parameters of the SVG to obtain optimized impedance parameters and corresponding optimized voltage transient data;

[0011] extract angle deviation data in the optimized voltage transient data to determine a switching frequency of the converter valve, construct an electromagnetic torque fluctuation response model in combination with the optimized impedance parameter to optimize the SVG reactive power data to obtain target reactive power data;

[0012] optimize the filter parameter according to the target reactive power data to obtain target filter parameter and combine the target filter parameter with the switching frequency to generate a load harmonic suppression strategy.

[0013] The second aspect of the present application provides a load harmonic suppression system for a remote micro-grid, comprising:

[0014] a characteristic matching module configured to obtain stator flux data of a motor in the remote micro-grid and perform AC filter characteristic matching according to the stator flux data to obtain filter parameter to determine SVG reactive power data;

[0015] an electromagnetic simulation module configured to analyze the SVG reactive power data to obtain DC side capacitor data to perform electromagnetic torque simulation on the motor and generate stator flux dynamic response data to determine response speed;

[0016] a coefficient quantification module configured to determine a capacitor rate parameter according to the DC side capacitor data to combine the capacitor rate parameter with the response speed and a preset control loop model to calculate a damping coefficient;

[0017] a fitness calculation module configured to obtain voltage transient data of the remote micro-grid based on the damping coefficient to perform harmonic suppression processing based on the filter parameter to calculate a suppression fitness value;

[0018] an impedance matching module configured to decompose the voltage transient data to adjust impedance matching parameter of the SVG to obtain optimized impedance parameter and corresponding optimized voltage transient data when the suppression fitness value is not up to standard;

[0019] a data optimization module configured to extract angle deviation data in the optimized voltage transient data to determine a switching frequency of the converter valve, construct an electromagnetic torque fluctuation response model in combination with the optimized impedance parameter to optimize the SVG reactive power data to obtain target reactive power data;

[0020] a strategy generation module configured to optimize the filter parameter according to the target reactive power data to obtain target filter parameter and combine the target filter parameter with the switching frequency to generate a load harmonic suppression strategy.

[0021] Compared with the prior art, the embodiment of the present application has the following advantages:

[0022] A series of key parameters such as filter parameters and SVG reactive power data are derived from the stator flux linkage dynamics of the motor in the remote micro-grid, and the accurate acquisition of these parameters helps to improve the overall performance of the motor system in the remote micro-grid; the damping coefficient and dynamic response data are introduced through the simulation and calculation of the derived key parameters, so that the micro-grid can foresee and quickly offset the disturbance caused by load mutation or intermittent new energy, prevent system oscillation, and improve the toughness of the micro-grid; the harmonic suppression processing is performed according to the calculated damping coefficient, and the impedance matching parameters and the switching frequency of the converter valve are dynamically adjusted when the suppression effect is not up to standard, which can effectively reduce the harmonic interference in the system, enhance the stability of the micro-grid system, and reduce the risk of equipment failure caused by harmonics; based on the adjustment result, an electromagnetic torque fluctuation response model is constructed to optimize the SVG reactive power data and filter parameters, so as to generate an effective current harmonic component suppression scheme, which can more accurately perform reactive power compensation, effectively suppress load harmonics, improve power quality, and ensure the normal operation of various equipment in the remote micro-grid; the entire scheme forms a complete data-driven closed loop from data acquisition, simulation analysis, parameter calculation to effect evaluation and re-optimization, which is a highly integrated, adaptive and data-based comprehensive solution, and can effectively solve the problems of voltage instability, high harmonic content and poor anti-interference ability of the remote independent micro-grid, improve power quality, and ensure the normal operation of the remote micro-grid. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, the drawings described in the following are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0024] Figure 1 is a flowchart of a load harmonic suppression method for a remote micro-grid provided by an embodiment of the present application;

[0025] Figure 2 is a structural diagram of a load harmonic suppression system for a remote micro-grid provided by an embodiment of the present application;

[0026] Reference signs:

[0027] Among them, 10, characteristic matching module; 20, electromagnetic simulation module; 30, coefficient quantization module; 40, fitness calculation module; 50, impedance matching module; 60, data optimization module; 70, strategy generation module. DETAILED DESCRIPTION

[0028] With reference to the drawings and embodiments, the technical solutions in the embodiments of the present application are described clearly and completely. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. The purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0029] In the description of the present application, the terms "first", "second", "third" and the like are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second", "third" and the like can explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "multiple" is two or more. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0030] In the description of the present application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present application have the same meaning as generally understood by those skilled in the art. The terms used in the description of the present application are only used to describe specific embodiments, and are not intended to limit the present application. For those of ordinary skill in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0031] In an embodiment, as shown in Figure 1 The first aspect of the present application provides a load harmonic suppression method for a remote microgrid, comprising:

[0032] S1, obtaining stator flux data of a motor in a remote microgrid, and performing AC filter characteristic matching according to the stator flux data to obtain filter parameters to determine SVG reactive power data;

[0033] In an embodiment, step S1 comprises:

[0034] Obtaining stator flux data of a motor in a remote microgrid, and setting a saturation threshold by magnetic core saturation modeling to filter the stator flux data and then perform polynomial fitting processing to obtain a nonlinear leakage reactance curve of the motor;

[0035] Extracting feature data in the nonlinear leakage reactance curve to perform AC filter characteristic matching to obtain filter model data and filter output values as the filter parameters, and determining the amount of reactive power demand according to the filter output values;

[0036] Determine an adjustment step length based on the reactive power demand, and determine a step length adjustment range using linear interpolation centered on the adjustment step length, to combine with the reactive power demand and the adjustment step length to obtain the SVG reactive power data.

[0037] Specifically, the flux linkage data of the motor stator winding in the remote micro-grid is collected by a Hall sensor, the sampling frequency is greater than or equal to 5 kHz, the time length is greater than or equal to 10 seconds, three working conditions of no-load, 50% load and rated load are covered, and the stator flux linkage data is obtained; wherein, the motor is at least two or less, each motor is assigned a label, and the corresponding data is labeled; in addition, the stator flux linkage data can also be obtained by simulating the electromagnetic field of the transformer in the remote micro-grid through a finite element analysis software, the rated voltage of the transformer is set to 10 kV, the rated current is 100 A, and the frequency is 50 Hz, wherein the transformer is also at least two. At the same time, in order to more accurately describe the saturation phenomenon of the magnetic core, based on the collected stator flux linkage data, the Jiles-Atherton magnetization model is used to describe the magnetization process of the magnetic core through the hysteresis loop for modeling, and the saturation flux density is set to 8T and the coercive force is set to 800A / m. Through the model, the magnetic flux density of the magnetic core under different magnetic field strengths is calculated, and it is found that when the magnetic field strength reaches 600A / m, the magnetic flux density starts to increase nonlinearly, indicating that the magnetic core enters the saturation state. The saturation threshold (preferably 5T) is set according to the historical data, and the collected stator flux linkage data is analyzed based on the saturation threshold, the data reaching the flux linkage saturation threshold is screened out, and the selected data is subjected to polynomial fitting processing (the stator current amplitude and the corresponding leakage reactance are subjected to quadratic polynomial fitting), so that the unique nonlinear leakage reactance curve of the corresponding equipment (motor or transformer) can be obtained. This curve reveals the characteristic that the leakage reactance decreases with the increase of the saturation depth.

[0038] The dynamic change data of the nonlinear leakage reactance curve is collected, and the sliding window method (window size 5s) is used to analyze the curve trend, and then 3 types of core feature parameters are extracted: 1, static characteristics: leakage reactance peak value under rated load (such as 17Ω for the first device and 19Ω for the second device); dynamic characteristics: leakage reactance slope change rate when the load increases from 50% to 100%; critical characteristics: load rate corresponding to the leakage reactance slope mutation point; and the resonance frequency corresponding to the leakage reactance slope mutation point is selected as the filtering target, and a second-order low-pass filter model is selected, and the transfer function is: H(s) = 1 / (s 2 + 2πsf c +(2πf c ) 2 ), wherein f cFor the cut-off frequency, the resonant frequency (preferably 100 Hz) is often taken, and then the filter model data (cut-off frequency, transfer function coefficient) is determined; the leakage reactance characteristic parameters are substituted into the filter model for AC filter characteristic matching, so that the filter output value, that is, the leakage reactance fluctuation amplitude reduction data after filtering, can be obtained. By connecting a static var generator (SVG) between two same devices (transformers or motors), the active power can be calculated by real-time monitoring of system voltage and current, and then the reactive power demand can be calculated by combining the filter output value with the transformer terminal voltage deviation, that is, the SVG compensation amount Q: Q=X / U 2 -Ptan(arccosφ); wherein U is the terminal voltage; X is the leakage reactance after filtering (i.e. the filter output value); P is the active power; φ is the power factor, and the power factor is preferably 0.85 when the terminal voltage deviation is between ±5%, and the power factor is determined according to historical data or expert experience when the voltage deviation exceeds ±5%.

[0039] According to the rule that the step size is not more than 10% of the compensation amount and is not less than the minimum adjustment unit (preferably 10 kVar) of the SVG, the adjustment step size can be determined; then the linear interpolation method is used to optimize the step size adjustment range, that is, the adjustment range is expanded by ±20% around the adjustment step size, the intermediate step size value is completed by interpolation, and the upper and lower boundary values of the optimized range are extracted, which are combined with the adjustment step size, the reactive power demand and the step size adjustment range, so that the SVG reactive power data can be obtained.

[0040] The present application accurately depicts the nonlinear change curve of the leakage reactance of the motor when the magnetic core is saturated due to large current through magnetic core saturation modeling and polynomial fitting, thereby laying an accurate data foundation for all subsequent operations and avoiding control errors caused by model distortion; AC filter characteristic matching is performed according to the real-time extracted leakage reactance characteristic data, so that the design of the filter can adapt to the current actual operating state of the motor, thereby achieving optimal harmonic suppression effect under any operating condition and improving the robustness of the system; after the reactive power demand is determined, the adjustment step size and the step size adjustment range are determined based on the reactive power demand, so that the SVG reactive power compensation output is smooth and gradual, thereby effectively avoiding the fact that the SVG itself becomes a new disturbance source due to the dramatic change of the SVG output, and greatly enhancing the stability of the micro-grid operation; based on the nonlinear change law of the leakage reactance, the AC filter characteristic and the reactive power compensation demand are matched, thereby solving the dynamic adaptation problem of the SVG compensation step size and the filter characteristic, and deeply coupling the physical characteristics of the motor and the control strategy of the SVG, thereby generating smooth reactive power control data that can efficiently suppress harmonics and greatly guarantee the stability of the system through high-precision modeling and adaptive algorithms.

[0041] S2, analyzing the SVG reactive power data to obtain DC side capacitor data for electromagnetic torque simulation of the motor, and generating stator flux linkage dynamic response data to determine the response speed;

[0042] In an embodiment, step S2 comprises:

[0043] The SVG reactive power data is analyzed by using fast Fourier transform to obtain a harmonic frequency to determine the DC side capacitor data;

[0044] The initial torque value is determined by combining the motor torque formula with the DC side capacitor data as a constraint, and the electromagnetic torque fluctuation finite element simulation is performed according to the initial torque value to obtain a torque dynamic change curve;

[0045] The torque dynamic change curve is smoothed by using a moving average method to obtain the stator flux dynamic response data, and the stator flux dynamic response data is fitted to the first order to obtain the response speed.

[0046] Specifically, based on the SVG reactive power data, a plurality of groups of reactive power time series data are collected by step adjustment, and fast Fourier transform processing is performed thereon to obtain frequency distribution characteristics, that is, the main harmonic frequency is concentrated in 50Hz-150Hz, and the 100Hz harmonic amplitude accounts for 60%. The switching frequency can be determined based on the harmonic frequency and the switching frequency matching principle. Since the switching frequency matching principle is 5 times higher than the main harmonic frequency (to avoid harmonic aliasing), and is within the rated frequency range (1kHz-2kHz), the minimum switching frequency is 500Hz, combined with the rated range, it is preliminarily set to 1.5kHz, and simulation verification is performed on the frequency, that is, under 1.5kHz, the switching harmonic output by the converter valve does not overlap with the system harmonic, and the loss is controllable (that is, IGBT switching loss≤5kW), so that it can be used as the final set switching frequency. Then, the DC side capacitor data is determined according to the harmonic frequency. Since the core function of the DC side capacitor is to suppress voltage ripple, the ripple size directly affects the stability of the converter valve output current, and further affects the stator flux fluctuation, the present application pre-sets the DC side voltage ripple coefficient δ≤5% (such as when the DC side voltage is 400V, the ripple is ≤20V), and the ripple frequency is consistent with the main harmonic frequency (100Hz), so according to the ripple suppression requirement: C=Q step / 2πf h U dc δ, wherein C is the DC side capacitor data, Q step is the reactive power step, f h is the main harmonic frequency, and U dc is the DC side rated voltage.

[0047] The DC side capacitor value will affect the AC current amplitude output by the converter valve, and further change the initial value of the motor electromagnetic torque. The present application takes the DC side capacitor data as a constraint, and calculates the capacitor reactance X C (X C= 1 / 2pif h C), and combining the motor torque formula to calculate the initial torque value: T e = 1.5p(ψ d i q - ψ q i d ); p is the number of pole pairs, ψ is the flux, and i is the current. In addition, the initial torque value can also be corrected according to the inverse proportion of the circuit and the capacitive reactance, combined with the reference torque under the rated load, but it needs to be noted that the correction result needs to be checked for overload, and the correction process can be ended after meeting the standard. Subsequently, based on the initial torque value, a finite element simulation tool (such as ANSYS Maxwell) is used to simulate the dynamic change of torque and capture the influence of converter valve parameters on torque fluctuation: first, input parameters such as motor parameters: stator inductance 0.5H, number of pole pairs 2, rated power 500kVA; converter valve parameters such as switching frequency 1.5kHz, PWM modulation mode (SVPWM); load parameters such as non-linear load proportion 30% (rectifier), linear load proportion 70% (asynchronous motor); then set the simulation boundary, define the grid accuracy of the stator winding 1mm and the grid accuracy of the magnetic core 0.5mm for grid division, set the simulation step 0.001s (to ensure that high-frequency harmonics are captured), set the simulation time 10s (covering 2 base wave periods) to perform simulation, and obtain the torque-time dynamic curve, that is, the torque dynamic change curve, the torque of which fluctuates within the initial torque value ±3Nm, and the fluctuation frequency is consistent with the 100Hz harmonic.

[0048] The present application adopts the sliding average method (window size 0.02s) to smooth the torque dynamic change curve to derive the flux response trend and obtain the stator flux dynamic response data, the flux amplitude of which changes sinusoidally with the torque fluctuation, and from which the core indicators are extracted: flux fluctuation amplitude, flux response time (delay from torque change to flux change) and fluctuation frequency (consistent with the main harmonic), if these characteristics exceed the standard (i.e. exceed the corresponding preset threshold), the switching frequency of the converter valve is reduced and the DC side capacitor data is increased, and the adjustment amount of the two is preferably one tenth of the original data, or it can be determined according to actual or experience, until the re-extracted characteristics meet the standard, and the stator flux dynamic response data is first-order fitted (numerical differentiation), that is, the flux change rate, that is, the response speed, can be calculated.

[0049] The application determines the parameters of the converter valve and the DC side through the frequency characteristics of the reactive power, simulates the electromagnetic torque dynamics based on the finite element simulation, reversely deduces the flux linkage response from the torque fluctuation, ensures that the flux linkage characteristics meet the standards through parameter iteration optimization, and solves the problem of quantifying the influence of the converter valve parameters on the flux linkage response by associating the converter valve switching frequency, the DC side capacitance rate and the stator flux linkage dynamics through the reactive step length; the influence of the converter valve parameters on the motor flux linkage is verified through simulation, which ensures that the parameters meet the reactive power regulation requirements and can suppress the flux linkage fluctuation, ensures the adaptation of the converter valve switching frequency, the DC side capacitance rate and the reactive step length, and avoids the flux linkage fluctuation exceeding the standard caused by blind parameter setting.

[0050] S3, determining a capacitance rate parameter according to the DC side capacitance data, combining the response speed and a preset control loop model to calculate a damping coefficient;

[0051] In an embodiment, step S3 comprises:

[0052] determining the capacitance rate parameter according to the DC side capacitance data, and inputting the capacitance rate parameter and the response speed into a preset PI control loop model to calculate a delay time through a step response method;

[0053] inputting the delay time and the capacitance rate parameter into a pre-constructed damping coefficient quantification model to generate damping coefficient space distribution data;

[0054] verifying the damping coefficient space distribution data through the Routh criterion, and outputting the damping coefficient space distribution data that passes the verification as the damping coefficient, otherwise adjusting the capacitance rate parameter to recalculate the damping coefficient.

[0055] Specifically, since the capacitance rate parameters (reactance and ripple coefficient) of the DC side directly affect the system impedance and then change the damping characteristics, the key parameters need to be extracted through the voltage signal. The application collects the time domain signal of the DC side voltage (sampling frequency 10 kHz, time length 5 s), adopts FFT decomposition to extract the reactance value and the voltage ripple coefficient (the ratio of the peak-to-peak ripple voltage to the DC rated voltage), and verifies the capacitance rate parameters obtained through decomposition through the DC side capacitance data. If the error between them is less than a preset threshold, it means that the data is valid, otherwise the capacitance rate parameters are measured again. Then the capacitance influence weight is determined according to the capacitance rate parameters. Based on engineering experience, the influence weight k of the capacitance rate on the damping is C inversely proportional to the ripple coefficient, that is, k C =k1×(1-δ), where k1 is a first proportional coefficient, preferably 0.4.

[0056] The control loop delay (the time difference between the instruction issuance and the execution) is the main factor causing the damping to decrease, and the application determines the delay time based on a preset PI control loop model, which is a typical unit negative feedback PI control system model and includes a PI controller, a controlled object, and an output, etc. The transfer function Gc(s) of the PI controller is the sum of a proportional term and an integral term: Gc(s) = Kp + Ki / s, wherein Kp is a proportional gain, Ki is an integral gain, and s is a Laplace operator. In the time domain, the control law is: u(t) = Kp x e(t) + Ki x ∫e(t)dt, u(t) is the output of the controller, and e(t) is an error signal. The controlled object is a first-order inertia plus pure delay link: Gp(s) = (K / (τ x s + 1)) x e (-L×s) , K is the gain of the process, τ is the time constant of the process, and L is the delay time. The closed-loop transfer function Φ(s) = C(s) / R(s) = Gc(s) x Gp(s) / (1 + Gc(s) x Gp(s)). A neutral proportional gain and an integral gain that will not cause severe oscillation are set based on engineering experience or preliminary experiments using the critical proportional degree method. A unit step signal is applied to the constructed closed-loop system as an input R(s), and the flux response speed and the permittivity parameters are substituted. The output response C(t) of the system is solved by numerical calculation (for example, in Simulink, Python, etc.), and the step response curve obtained by simulation is drawn. The point with the maximum slope on the response curve, i.e. the inflection point, is found. A tangent line is drawn at the inflection point. The intersection point of the tangent line and the time axis (horizontal axis, C(t) = 0) corresponds to the delay time. The intersection point of the tangent line and the horizontal line of the steady-state value of the system corresponds to the time minus L, which is the time constant.

[0057] In addition, a reactive power step signal (for example, from 400 kVar to 440 kVar, with a step of 40 kVar) can also be injected into the electromagnetic torque fluctuation finite element simulation model, and the time difference between the control instruction output time and the flux response start time is recorded. The average value obtained by repeating three times is taken as the delay time. Finally, the delay time is used to determine the delay influence weight k τ : k τ = k2 / (1 + τ), wherein τ is the delay time, and k2 is a second proportional coefficient, which is preferably 0.6.

[0058] The calculated delay time and the permittivity parameters are input into the pre-constructed damping coefficient quantification model to obtain the damping coefficient D. The damping coefficient quantification model is: D = k τ / τ + k C / X C; 6 load intervals are divided according to 0% (empty load) -> 20% -> 50% -> 80% -> 100% (rated) -> 120% (overload), and the damping coefficient in each load interval is repeatedly calculated to generate the damping coefficient spatial distribution data.

[0059] A system characteristic equation s 3 + 2s 2 +(1+D)s+0.5=0, the damping coefficient is substituted into the system characteristic equation, if the real part of all roots of the characteristic equation is negative, it is determined to be stable, and it is explained that it passes the verification, and the damping coefficient spatial distribution data that passes the verification is output as the finally calculated damping coefficient, otherwise the DC side capacitance value (the adjustment amount is preferably 50%, which can also be determined according to actual data or expert experience) is increased to reduce the capacitance reactance, and then the damping coefficient is recalculated until it is determined to be stable.

[0060] The damping coefficient spatial distribution data is calculated by the preset PI control loop model and the damping coefficient quantization model, which changes the traditional mode of relying on the personal experience of engineers; the Routh criterion is introduced for verification, which greatly reduces the risk of system oscillation or instability caused by improper parameters during on-site debugging; based on the correlation between flux linkage response, capacitance rate and control delay, the damping coefficient is calculated and the stability is verified, thereby solving the problem of system oscillation caused by control delay and how to quantify the stability of the remote microgrid under weak inertia.

[0061] S4, based on the damping coefficient, obtain voltage transient data of the remote microgrid to perform harmonic suppression processing based on the filter parameter to calculate an inhibition fitness value;

[0062] In an embodiment, step S4 comprises:

[0063] Based on the stator flux linkage dynamic response data, simulation transient voltage data is determined, and the simulation transient voltage data is corrected based on the damping coefficient and the delay time to obtain corrected transient voltage data;

[0064] When the damping coefficient is lower than a stability threshold, real-time voltage transient data of the remote microgrid is obtained to be combined with the corrected transient voltage data to obtain a transient voltage data set;

[0065] The transient voltage data set is subjected to 3-layer decomposition processing by using a db4 wavelet to obtain load harmonic features, and the load harmonic features are subjected to harmonic suppression processing by using a filter having the filter parameter to obtain filtered features;

[0066] The inhibition fitness value is determined according to the load harmonic features and the filtered features.

[0067] Specifically, because the electromagnetic torque of a motor is directly bound to the stator flux linkage and current, torque fluctuations inevitably cause fluctuations in flux linkage and current, leading to transient voltage changes. This invention directly calculates simulated transient voltage data based on the flux linkage change rate and current corresponding to the stator flux linkage dynamic response data, using the motor stator voltage equation; where the motor stator voltage u... s Satisfy: u s =R s i s +dψ / dt, where R s i is the stator winding resistance. s dψ / dt is the stator current; dψ / dt is the flux linkage rate of change, obtained by numerical differentiation of the stator flux linkage dynamic response data; however, the voltage data obtained from the simulation do not consider the effects of control loop delay and system damping. This invention uses a delay time to perform an initial correction on the simulated transient voltage data: first, the phase shift caused by the delay time is calculated. Based on the calculated phase shift, the simulated voltage is corrected by phase rotation using the frequency domain phase rotation method to obtain the time-domain corrected voltage data. Then, the time-domain corrected voltage data is further corrected based on the damping coefficient. This correction process is as follows: A(t) = e -Dωt A(t) is the attenuation factor, ω is the angular frequency of the voltage fluctuation, and t is time. By multiplying the voltage amplitude of the time-domain corrected voltage data at each moment by the attenuation factor at the corresponding time, the voltage transient data after secondary correction can be obtained. Then, the steady-state value is corrected by comparing the voltage data after secondary correction with the rated voltage and taking the larger value as the final corrected voltage data, which is the corrected voltage transient data that includes the system damping and delay characteristics.

[0068] When the damping coefficient is below the stability threshold (preferably 6.0) or there is a sudden load change (e.g., the load suddenly increases from 50% to 100%), high-speed transient data acquisition is automatically initiated. This involves deploying a high-precision transient voltage sensor at the common connection point of two identical devices, with a bandwidth ≥1MHz, a sampling frequency ≥20kHz, a sampling duration of 10s, and covering two fundamental cycles to simultaneously acquire voltage and current signals and ensure complete capture of the transient process. Real-time voltage transient data from the remote microgrid is then acquired and filtered for noise reduction (using a 50Hz notch filter to remove fundamental interference) and synchronously calibrated (aligned with the timestamp of the current signal). Finally, measured voltage transient data (in the format of a time-voltage value time series array) is output. Combining this with the corrected transient voltage data yields the transient voltage dataset. It should be noted that regardless of simulation derivation or actual measurement, the final voltage transient data must meet two requirements: high time resolution and inclusion of harmonic information. The final form is a time series array containing transient characteristics such as voltage peak value, decay time, and fluctuation frequency, and implicitly includes 5th and 7th harmonic components.

[0069] The db4 wavelet is used for 3-layer decomposition processing of the transient voltage data set, and load harmonic characteristics, i.e., amplitude characteristics: 5th harmonic amplitude and 7th harmonic amplitude, and distribution characteristics: harmonic amplitude proportion are obtained; then a filter with filter parameters, i.e., a second-order low-pass filter model is used for harmonic suppression processing of the load harmonic characteristics, and filtered characteristics: 5th harmonic amplitude and 7th harmonic amplitude are obtained.

[0070] The amplitude difference of the two kinds of harmonics before and after filtering is calculated, if both are greater than a preset difference threshold (0.5A), the filter characteristic does not need to be adjusted, otherwise the cutoff frequency is reduced from 100Hz to 95Hz (decreased by 5Hz successively, or the frequency adjustment amount can be determined according to expert experience), and re-filtering is performed until the difference meets the standard; finally, the fitness F (taking into account the suppression rate and response speed) is calculated by using a weighted formula: F = 0.7 x (A 前 -A 后 ) / A 前 + 0.3 x (1-t 响应 / t 额定 ), wherein A 前 and A 后 are the amplitudes before and after filtering, t 响应 is the magnetic chain response time, and t 额定 is the rated response time (preferably 0.05s).

[0071] The present application introduces a damping coefficient and a delay time to reflect the control effect, and constructs a high-fidelity, multi-scenario transient voltage data set based on the modified transient voltage and real-time voltage transient data under the actual power grid working condition, thereby improving the comprehensiveness and reliability of analysis; the wavelet decomposition can accurately separate the fundamental wave, each characteristic harmonic and random interference in the signal, thereby obtaining clear and reliable load harmonic characteristics; the suppression fitness value which comprehensively measures the attenuation degree of the filter to the main harmonics in the power grid under a specific filter parameter is used to realize objective and quantitative evaluation of the filtering effect; by quantifying the suppression effect of the filter characteristic on the harmonics, the fitness index which fuses the suppression rate and dynamic performance is introduced to avoid the disadvantages of only looking at the amplitude and not looking at the stability.

[0072] S5、in the case that the suppression fitness value does not meet the standard, decomposing the voltage transient data to adjust the impedance matching parameter of the SVG, and obtaining an optimized impedance parameter and corresponding optimized voltage transient data;

[0073] In an embodiment, step S5 comprises:

[0074] in the case that the suppression fitness value does not meet the standard, decomposing the transient voltage data set to obtain the main harmonic frequency, and determining the harmonic amplitude proportion according to the main harmonic frequency, so as to determine the impedance adjustment direction based on the resonance principle;

[0075] Based on the impedance adjustment direction, the gradient descent method is used to iteratively adjust the impedance matching parameters of the SVG to obtain optimized impedance parameters and corresponding optimized voltage transient data;

[0076] According to the optimized voltage transient data, the suppression fitness value is recalculated, and when the new suppression fitness value does not meet the standard, the impedance parameter change trend is determined based on the optimized impedance parameters;

[0077] Based on the impedance parameter change trend, the linear regression is used to optimize the parameter adjustment step to recalculate the new suppression fitness value until it meets the standard, and the finally obtained optimized impedance parameters and corresponding optimized voltage transient data are output.

[0078] Specifically, in the remote micro-grid, the impedance matching of the SVG with the transformer and the load in the system directly determines the harmonic transmission efficiency - if the impedance is mismatched (such as the SVG equivalent impedance and the system impedance at the harmonic frequency are not resonant), even if the filter characteristic is designed reasonably, the harmonics will still be transmitted through the system impedance path, resulting in attenuation of the suppression effect. Therefore, the present application realizes resonance suppression through impedance parameter adjustment, and tracks the load harmonic spectrum change through iterative algorithm to avoid the rigidity of fixed parameters.

[0079] When the calculated suppression fitness value does not meet the standard (preferably 90%), the transient voltage data set is decomposed by fast Fourier transform to obtain the main harmonic frequency (i.e. the amplitude corresponding to the 5th and 7th harmonics), and the harmonic amplitude ratio in the main harmonic frequency is calculated, i.e. the amplitude ratio of the two harmonics; then the impedance adjustment direction (increase / decrease L or C) is determined based on the resonance principle: for example, the 5th harmonic frequency is 250Hz, and the SVG equivalent impedance needs to be resonant with the harmonic frequency (i.e. X L = X C ), the current X L is calculated according to the current SVG parameters, and the impedance adjustment direction is determined according to the comparison result of X L and the resonant size of the harmonic frequency to match the resonance condition; the capacitance is adjusted first (because the capacitance adjustment has less effect on the system stability, and the capacitance value can be easily adjusted continuously by parallel connection), then the reactance is adjusted (the reactance is usually a fixed value, which needs to be adjusted by tapping or converter), and finally the adjustment object and its corresponding adjustment amount are determined according to the engineering practice. And after the calculated suppression fitness value meets the standard, the transient voltage data set is optimized as the optimized voltage transient data.

[0080] Based on the determined adjustment object and its corresponding adjustment amount, with the goal of minimizing the amplitude of the main harmonic, the learning rate, the upper limit of the number of iterations, the initial impedance parameter, the termination condition and other iteration parameters are set. The gradient descent method is used to iteratively adjust the impedance matching parameters of the SVG to obtain the optimized impedance parameters. The voltage transient dataset is then reconstructed based on the capacitance in the optimized impedance parameters as the optimized voltage transient data.

[0081] Next, the optimized voltage transient data is decomposed again, and the suppression fitness value is recalculated based on the decomposition results. If the new suppression fitness value still does not meet the standard, the trend of impedance parameter change is extracted, and the impedance-amplitude relationship is obtained by linear regression. Based on this relationship, the matching degree between the direction of impedance parameter change and the direction of harmonic amplitude change is statistically analyzed, and then it is determined whether the current step size adjustment direction is reasonable. That is, if the relationship indicates that in the past iterations, each decrease of L resulted in a decrease in harmonic amplitude, and the consistency is 100%, it means that the step size direction is correct and there is no need to change the direction. Only the amplitude needs to be optimized (e.g., the learning rate is increased by 0.01 from the original 0.01 in each iteration, with a fixed step size of 0.01 in the initial stage and a value range of 0.01-0.05). Otherwise, it means that the step size direction deviates from the resonance condition, and the resonance impedance needs to be recalculated through historical trend analysis to correct the direction (e.g., the original adjustment of L is changed to increase C). Then, the new suppression fitness value is recalculated based on the optimization results until it meets the standard, and the final optimized impedance parameters and their corresponding optimized voltage transient data are output.

[0082] This invention determines the direction of impedance adjustment based on the resonance principle, enabling the system to "understand" which harmonic is causing the problem and decide whether the impedance of the SVG should be increased or decreased to effectively suppress that harmonic. This fundamentally avoids the blindness of parameter adjustment and greatly improves optimization efficiency. It employs an iterative optimization algorithm to ensure the convergence and optimality of parameter adjustment. The introduction of trend analysis and predictive optimization allows the system to "learn" patterns from past adjustments. Through linear regression analysis, the system can predictively adjust reactive power compensation strategies, thereby more profoundly impacting the harmonic environment of the power grid. Based on the harmonic frequency distribution, iterative adjustment of the SVG impedance parameters solves the mismatch between filter characteristics and harmonic frequencies.

[0083] S6. Extract the angle deviation data from the optimized voltage transient data to determine the switching frequency of the converter valve, and construct an electromagnetic torque fluctuation response model in combination with the optimized impedance parameters to optimize the SVG reactive data and obtain the target reactive data.

[0084] In one embodiment, extracting angle deviation data from the optimized voltage transient data to determine the switching frequency of the converter valve, and constructing an electromagnetic torque fluctuation response model in conjunction with the optimized impedance parameters, includes:

[0085] performing Fourier transform on the optimized voltage transient data to extract a rotor angular velocity feature and a power angle feature;

[0086] determining an angular velocity deviation based on the rotor angular velocity feature and an angle offset based on the power angle feature;

[0087] correcting the power angle feature by using a linear difference method according to the angular velocity deviation and the angle offset to obtain a power angle correction feature;

[0088] inputting the power angle correction feature into a pre-constructed deviation frequency linear regression model to perform processing and outputting a switching frequency of the converter valve;

[0089] inputting the switching frequency, the optimized impedance parameter and a stator parameter of the motor into a synchronous motor mathematical model to construct the electromagnetic torque fluctuation response model.

[0090] Specifically, the application performs Fourier transform on the optimized voltage transient data to extract a rotor angular velocity feature: an angular velocity deviation frequency corresponding to voltage frequency fluctuation, and a power angle feature: a phase difference between voltage and current; then, the difference between the rotor angular velocity feature and a preset angular velocity feature threshold (preferably 3000) is taken as an angular velocity deviation, and the difference between the power angle feature and a preset angle feature threshold (preferably 30°) is taken as an angle offset.

[0091] The two deviation data obtained are compared with the deviation threshold values corresponding thereto respectively, when the angular velocity deviation exceeds the preset deviation threshold (40), the offset is corrected by using a linear interpolation method, that is, the historical data of the angular velocity deviation and the angle offset are obtained and are interpolated to obtain a corrected angle offset corresponding to the angular velocity deviation, and the corrected angle offset is taken as a power angle correction feature; if the angle offset exceeds, a normal angular velocity deviation corresponding to the angle offset is determined, and normal angle offset data of the normal angular velocity deviation in the historical data are obtained to be taken as a corrected angle offset, that is, a power angle correction feature; if neither of them exceeds, the angle offset is taken as a power angle correction feature for the following processing; if both of them exceed, it indicates that the accuracy of the optimized voltage transient data is not enough, and the voltage transient data set is reconstructed for calculation.

[0092] A deviation frequency linear regression model is constructed based on the historical angular velocity deviation Δω and the corresponding switching frequency f: f = 0.01 x Δω + 1.1, and the angular velocity deviation corresponding to the power angle correction feature is input into the model, so that the switching frequency of the converter valve can be output, and the dynamic adjustment range of the switching frequency can be determined in combination with the frequency fluctuation allowable range (± 0.1 kHz). Finally, based on the mathematical model of the synchronous motor, the stator parameters (inductance, resistance), impedance parameters and switching frequency parameters (switching frequency and its dynamic adjustment range) are used to construct an electromagnetic torque fluctuation response model: T e (t) = k x ψ s (t) x i s (t) x sin (Δδ (t)), wherein, T e (t) is torque fluctuation data, k is a constant, i s (t) is the stator current, and Δδ (t) is the power angle offset.

[0093] The application can accurately extract the core features reflecting the system synchronous stability by Fourier transform processing on the optimized voltage transient data, accurately capture the most critical phase and frequency dynamics in the power grid transient process, innovatively combine the angular velocity deviation and the angle offset, and correct the power angle feature by using the linear interpolation method to obtain more accurate power angle correction features, effectively compensate for the errors or lags that may exist in single angle measurement, and make the angle signal used to determine the switching frequency higher in quality and more reliable. By constructing an electromagnetic torque fluctuation response model, the electromagnetic torque fluctuation of the motor under specific grid conditions (optimized voltage transient) and specific control strategies (optimized impedance and adaptive switching frequency) can be simulated and predicted with high precision, so that the optimal configuration can be found, the torque fluctuation can be inhibited from the source, and mechanical damage to the motor shaft and load can be prevented. The rotor and power angle deviation are extracted based on the voltage transient feature to solve the switching frequency of the converter valve and the system synchronization problem.

[0094] In an embodiment, the SVG reactive power data is optimized to obtain target reactive power data, including:

[0095] Load data of the remote microgrid is obtained, and several harmonic components are extracted from the load data to be input into the electromagnetic torque fluctuation response model for simulation, and torque fluctuation data is output;

[0096] When it is determined that the torque fluctuation data exceeds the preset threshold, the gradient descent method is used to iteratively adjust the optimized impedance parameters until the obtained torque fluctuation data does not exceed the preset threshold, and the final torque fluctuation data is output;

[0097] Determine reactive power change data based on the final torque fluctuation data, and optimize the SVG reactive data according to the reactive power change data to obtain preliminary reactive optimization data;

[0098] When it is determined that the preliminary reactive optimization data matches the torque fluctuation, the support vector regression is used to perform boundary prediction on the reactive power change data, and output reactive optimization boundary data;

[0099] According to the reactive optimization boundary data, the switching frequency is adjusted, the electromagnetic torque fluctuation response model is updated based on the adjustment result, and the preliminary reactive optimization data is optimized according to the updated electromagnetic torque fluctuation response model to obtain the target reactive data.

[0100] Specifically, the load data of the remote micro-grid is obtained, a plurality of harmonic components are extracted from the load data through fast Fourier transform, and the extracted harmonic components are input into the electromagnetic torque fluctuation response model for simulation to output torque fluctuation data; When it is determined that the fluctuation amplitude of the torque fluctuation data exceeds the preset threshold (±3Nm) (the torque fluctuation exceeds the standard is that the SVG impedance parameter and the harmonic frequency are mismatched), the gradient descent method is used to iteratively adjust the obtained optimized impedance parameter, and the electromagnetic torque fluctuation response model is optimized based on the iteratively adjusted impedance parameter, so that the simulation is performed through the optimized electromagnetic torque fluctuation response model until the obtained torque fluctuation data does not exceed the preset threshold, and the final torque fluctuation data is output; If the fluctuation amplitude does not exceed the preset threshold, the torque fluctuation data is taken as the final torque fluctuation data.

[0101] The step size of the reactive power directly affects the compensation accuracy of the SVG: too large step size easily leads to compensation overshoot (aggravating torque fluctuation), and too small step size easily leads to response lag (inhibiting harmonics not in time), the present application fits the dynamic change of the reactive power by the optimized torque fluctuation data, determines the preliminary step size accuracy, that is, the reactive power change data P e is inversely deduced by the optimized torque data and the motor power formula P e = T e ×ω, and the fluctuation range and adjustment smoothness of the reactive power change curve data are analyzed, the preliminary accuracy is calculated according to the step size accuracy = ±(fluctuation range x 10%), so as to realize the optimization of the SVG reactive data and obtain the preliminary reactive optimization data.

[0102] The algorithm determines whether the model, under the change in compensation amount corresponding to the step size accuracy, can stabilize the new torque fluctuation within the preset fluctuation threshold. If so, the initial accuracy is qualified, and the initial reactive power optimization data matches the torque fluctuation. If not, the accuracy does not meet the requirements, and the SVG reactive power data is re-optimized until the two match. Subsequently, reactive power change trends and step size accuracy data under historical operating conditions are collected (e.g., 5 sets of samples: fluctuation ±1kVar → accuracy ±0.1kVar; fluctuation ±2kVar → accuracy ±0.2kVar). The Support Vector Regression (SVR) algorithm is used to predict the boundary, that is, the radial basis function (RBF) is used as the kernel function, the penalty coefficient is 10, and the gamma value is 0.1. Then, the current reactive power fluctuation trend is used as input, and the upper and lower boundaries of the step size accuracy are predicted by the trained SVR, which are used as the reactive power optimization boundary data.

[0103] The switching frequency range is adjusted based on the reactive power optimization boundary data output by the SVR. Specifically, the ratio of the absolute value of the boundary data to the maximum allowable voltage fluctuation deviation (preferably 5%) is used as the risk factor R. A range adjustment coefficient kopt is then determined based on this risk factor: kopt = k0 × (1 - γ × R), where γ is the risk-step correlation coefficient, an engineering calibration value typically between 0.11 and 0.15, and k0 is the initial adjustment coefficient, preferably 0.1. Finally, the range adjustment coefficient is multiplied by the original switching frequency range to obtain the adjusted switching frequency range. The switching frequency is then adjusted synchronously based on this range adjustment coefficient. Finally, the parameters of the electromagnetic torque fluctuation response model are updated using the adjusted switching frequency parameters. The updated model is then re-simulated, and the preliminary reactive power optimization data is optimized based on the simulation results. The re-optimized step size accuracy and step size boundary, combined with the reactive power demand, are used as the target reactive power data.

[0104] This invention establishes a direct correlation between load harmonics and torque fluctuations, enabling the system to predict potential mechanical risks based on load changes before torque fluctuations actually occur and damage the motor shaft system, thus achieving a shift from passive response to proactive prevention. It uses torque stability as a hard constraint to achieve self-convergent optimization of impedance parameters; introduces intelligent boundary prediction to improve the robustness and stability of the reactive power optimization strategy; and updates the response model to address the matching problem between reactive power step size accuracy and torque fluctuations.

[0105] S7. Optimize the filter parameters based on the target reactive power data to obtain the target filter parameters and combine them with the switching frequency to generate a load harmonic suppression strategy;

[0106] In one embodiment, optimizing the filter parameters based on the target reactive power data to obtain the target filter parameters includes:

[0107] Collect current power angle dynamic data of the remote micro-grid, and perform sensitivity analysis on the filtering parameters when the current power angle dynamic data exceeds a power angle offset threshold, to determine filtering parameters to be optimized;

[0108] Construct a constraint condition according to the target reactive power data, and construct a target function according to the damping coefficient, the suppression fitness value and the power angle offset threshold, to optimize the filtering parameters to be optimized by a multi-objective optimization algorithm, to obtain optimized filtering parameters;

[0109] Based on the optimized filtering parameters, perform full-condition simulation verification, and output the optimized filtering parameters as the target filtering parameters when the verification is passed.

[0110] Specifically, the current power angle dynamic data of the remote micro-grid is obtained by a PMU synchronous phasor measurement device, and when the current power angle dynamic data exceeds a power angle offset threshold (determined according to the micro-grid synchronization stability requirement, preferably ±5°), the optimization priority is to ensure system synchronization (i.e. to suppress angle offset), the influence sensitivity of a second-order low-pass filter model with filtering parameters on the center frequency and the cutoff frequency is analyzed by simulation control variable method, the parameters that need to be corrected are selected, and the frequency parameter with the highest sensitivity is taken as the filtering parameter to be optimized.

[0111] Since the filtering parameters need to be within the rated range of the device, the constraint condition is constructed based on the target reactive power data, the harmonic suppression fitness is ≥90%, the power angle offset is ≤±5°, and the damping coefficient D is ≥6.0 as the target function, the particle swarm optimization (PSO) algorithm is used to realize multi-objective parameter optimization, i.e. the filtering parameters to be optimized are coded as particle dimensions, and the weighted scores of the three targets are taken as the fitness function (0.5, 0.25, 0.25 respectively), the initial particle swarm is randomly generated 50 groups of parameter combinations, and the particle position is updated by individual optimal + global optimal to perform iterative optimization of the particles, until all targets are met, and the optimized filtering parameters are output.

[0112] The optimized filtering parameters are substituted into the micro-grid joint simulation model, and the effect is verified under four working conditions of no-load, light load, rated load and load mutation, if a certain working condition does not meet the standard (for example, the total harmonic distortion rate THD = 6% > 5% when the load mutates), the multi-objective parameter optimization step is adjusted to adjust the parameter weight (increase the weight of the harmonic suppression fitness), and the optimization is re-performed until the verification is passed, and the optimized filtering parameters obtained finally are output as the target filtering parameters. Finally, the target filtering parameters are combined with the switching frequency to generate a load harmonic suppression strategy to control the remote micro-grid to execute, and realize harmonic suppression. It should be noted that the preferred values of the above threshold values, coefficients and the like are only one embodiment, which can be adjusted according to actual application and expert experience.

[0113] The present application quickly locates the filter parameters to be optimized through sensitivity analysis, avoids "blanket" search on all parameters, greatly improves the pertinence and calculation efficiency of the optimization process, and realizes the optimal allocation of resources; the target reactive power data is used as a constraint, and the damping coefficient (stability), the suppression fitness value (filter effect) and the power angle offset threshold (stability margin) are used to jointly construct the objective function, so as to ensure the unity of the filter parameters and the global target of the system; the filter characteristics are optimized based on the step accuracy and the angle offset, so as to solve the matching problem of the filter and the control delay, and the target step accuracy is used as a benchmark, and the power angle offset is used as a warning signal, and the threefold balance of harmonic suppression-system synchronization-equipment safety is realized through the quantitative correction of the filter parameters.

[0114] The present application is based on how to improve the response speed and accuracy of the existing harmonic suppression method for remote microgrids, and designs a load harmonic suppression method for remote microgrids. A series of key parameters such as filter parameters and SVG reactive power data are derived from the stator flux dynamic of the motor in the remote microgrid. The accurate acquisition of these parameters helps to improve the overall performance of the motor system in the remote microgrid. The simulation and calculation of the derived key parameters introduce the damping coefficient and the dynamic response data, so that the microgrid can predict and quickly offset the disturbance caused by load mutation or intermittent new energy, prevent system oscillation, and improve the toughness of the microgrid. According to the calculated damping coefficient, the harmonic suppression processing is performed, and when the suppression effect is not up to standard, the impedance matching parameters and the switching frequency of the converter valve are dynamically adjusted, which can effectively reduce the harmonic interference in the system, enhance the stability of the microgrid system, and reduce the risk of equipment failure caused by harmonics. Based on the adjustment result, an electromagnetic torque fluctuation response model is constructed to optimize the SVG reactive power data and the filter parameters, so as to generate an effective current harmonic component suppression scheme, which can more accurately perform reactive power compensation, effectively suppress load harmonics, improve power quality, and ensure the normal operation of various equipment in the remote microgrid. The entire scheme forms a complete data-driven closed loop from data acquisition, simulation analysis, parameter calculation to effect evaluation and re-optimization, which is a highly integrated, adaptive and data-based comprehensive solution. It can effectively solve the problems of voltage instability, high harmonic content and poor anti-interference ability faced by remote independent microgrids, improve power quality, and ensure the normal operation of remote microgrids.

[0115] It should be noted that although each step in the above flowchart is displayed in sequence according to the direction of the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, the execution of these steps has no strict order limitation, and these steps can be executed in other orders.

[0116] In another embodiment, as Figure 2As shown, the second aspect of the present application provides a load harmonic suppression system for a remote micro-grid, comprising:

[0117] a characteristic matching module 10 configured to obtain stator flux data of a motor in the remote micro-grid, and perform AC filter characteristic matching according to the stator flux data to obtain filter parameters to determine SVG reactive power data;

[0118] an electromagnetic simulation module 20 configured to analyze the SVG reactive power data to obtain DC side capacitor data to perform electromagnetic torque simulation on the motor, and generate stator flux dynamic response data to determine a response speed;

[0119] a coefficient quantification module 30 configured to determine a capacitor rate parameter according to the DC side capacitor data, to combine the response speed and a preset control loop model to calculate a damping coefficient;

[0120] a fitness calculation module 40 configured to obtain voltage transient data of the remote micro-grid based on the damping coefficient to perform harmonic suppression processing based on the filter parameters to calculate a suppression fitness value;

[0121] an impedance matching module 50 configured to, when the suppression fitness value is not up to standard, decompose the voltage transient data to adjust impedance matching parameters of the SVG to obtain optimized impedance parameters and corresponding optimized voltage transient data;

[0122] a data optimization module 60 configured to extract angle deviation data in the optimized voltage transient data to determine a switching frequency of a converter valve, and combine the optimized impedance parameters to construct an electromagnetic torque fluctuation response model to optimize the SVG reactive power data to obtain target reactive power data;

[0123] a strategy generation module 70 configured to optimize the filter parameters according to the target reactive power data to obtain target filter parameters and combine the target filter parameters with the switching frequency to generate a load harmonic suppression strategy.

[0124] It should be noted that the above-described various modules of the load harmonic suppression system for the remote micro-grid can be realized in whole or in part by software, hardware, and combinations thereof. The above-described various modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so as to be called and executed by a processor to perform the operations corresponding to the above-described various modules. For specific limitations of the load harmonic suppression system for the remote micro-grid, refer to the limitations of the load harmonic suppression method for the remote micro-grid described above, both of which have the same functions and effects, and will not be described here.

[0125] In conclusion, the application relates to the technical field of harmonic suppression, and discloses a load harmonic suppression method and system for a remote micro-grid, which analyzes stator flux linkage data of a motor in the remote micro-grid to determine filter parameters and SVG reactive power data, analyzes the SVG reactive power data to obtain a DC side capacitor to perform electromagnetic torque simulation, generates stator flux linkage response data and determines a response speed; determines a capacitance rate parameter according to the DC side capacitor, combines the response speed and a preset model to calculate a damping coefficient; performs harmonic suppression processing on voltage transient data based on the damping coefficient, and if the suppression effect is not up to standard, decomposes the voltage transient data to optimize SVG impedance matching parameters; extracts an angle deviation in the voltage transient data to determine a switching frequency of a converter valve, and then constructs an electromagnetic torque fluctuation response model to optimize the SVG reactive power data, combines optimized filter parameters and the switching frequency to generate a load harmonic suppression strategy; and improves voltage stability and power quality of the independent micro-grid in the remote area.

[0126] Each embodiment in the specification is described in a progressive manner, and the same or similar parts of each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the related parts can be referred to the part of the method embodiment. It should be noted that, the technical features of the above embodiments can be combined arbitrarily, in order to make the description simple, not all possible combinations of the technical features of the above embodiments are described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the specification.

[0127] The above-described embodiments only express several preferred embodiments of the application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be noted that, for ordinary skilled in the art, without departing from the technical principles of the application, some improvements and replacements can be made, and these improvements and replacements should be considered as the protection scope of the application. Therefore, the protection scope of the patent of the application should be subject to the protection scope of the claims.

Claims

1. A method for suppressing load harmonics in remote microgrids, characterized in that, The method comprises the following steps: acquiring stator flux linkage data of a motor in a remote micro-grid, and performing AC filter characteristic matching according to the stator flux linkage data to obtain filter parameters to determine SVG reactive power data; analyzing the SVG reactive power data to obtain DC side capacitor data to simulate electromagnetic torque of the motor, generate stator flux linkage dynamic response data to determine response speed; determining a capacitor rate parameter according to the DC side capacitor data, combining the response speed and a preset control loop model to calculate a damping coefficient; based on the damping coefficient, acquiring voltage transient data of the remote micro-grid to perform harmonic suppression processing based on the filter parameters to calculate an inhibition fitness value; when the inhibition fitness value is not up to standard, decomposing the voltage transient data to adjust impedance matching parameters of the SVG to obtain optimized impedance parameters and corresponding optimized voltage transient data; extracting angle deviation data in the optimized voltage transient data to determine a switching frequency of a converter valve, combining the optimized impedance parameters to construct an electromagnetic torque fluctuation response model to optimize the SVG reactive power data to obtain target reactive power data; optimizing the filter parameters according to the target reactive power data to obtain target filter parameters and combining the switching frequency to generate a load harmonic suppression strategy.

2. The method for load harmonic suppression for remote microgrid according to claim 1, wherein, The method comprises the following steps: acquiring stator flux linkage data of a motor in a remote micro-grid, and performing AC filter characteristic matching according to the stator flux linkage data to obtain filter parameters to determine SVG reactive power data, comprising: acquiring stator flux linkage data of a motor in a remote micro-grid, and setting a saturation threshold through magnetic core saturation modeling to filter the stator flux linkage data and then perform polynomial fitting processing to obtain a nonlinear leakage reactance curve of the motor; extracting feature data in the nonlinear leakage reactance curve to perform AC filter characteristic matching to obtain filter model data and filter output values as the filter parameters, and determining reactive power demand according to the filter output values; 3. The method for load harmonic suppression for remote microgrid according to claim 1, wherein, determining an adjustment step based on the reactive power demand, and determining a step adjustment range using a linear interpolation method with the adjustment step as the center, combining the reactive power demand and the adjustment step to obtain the SVG reactive power data. The method comprises the following steps: analyzing the SVG reactive power data to obtain DC side capacitor data to simulate electromagnetic torque of the motor, generate stator flux linkage dynamic response data to determine response speed, comprising: analyzing the SVG reactive power data using fast Fourier transform to determine harmonic frequencies to determine the DC side capacitor data; 4. The method for load harmonic suppression for remote microgrid according to claim 1, wherein, determining an initial torque value based on the DC side capacitor data and a motor torque formula, and performing electromagnetic torque fluctuation finite element simulation according to the initial torque value to obtain a torque dynamic change curve; smoothing the torque dynamic change curve using a moving average method to obtain the stator flux linkage dynamic response data, and first-order fitting the stator flux linkage dynamic response data to obtain the response speed. The method comprises the following steps: determining a capacitor rate parameter according to the DC side capacitor data, combining the response speed and a preset control loop model to calculate a damping coefficient, comprising: The capacitance rate parameter is determined according to the direct-current side capacitance data, and the capacitance rate parameter and the response speed are input into a preset PI control loop model to calculate a delay time by a step response method; The delay time and the capacitance rate parameter are input into a pre-constructed damping coefficient quantification model to generate damping coefficient spatial distribution data; The damping coefficient spatial distribution data is verified by a Routh criterion, and the damping coefficient spatial distribution data that passes the verification is output as the damping coefficient, otherwise the capacitance rate parameter is adjusted to recalculate the damping coefficient.

5. The method for load harmonic mitigation for remote microgrid as claimed in claim 4, wherein, The voltage transient data of the remote micro-grid is obtained based on the damping coefficient to perform harmonic suppression processing based on the filter parameter to calculate a suppression fitness value, including: Simulation transient voltage data is determined based on the stator flux linkage dynamic response data, and the simulation transient voltage data is corrected based on the damping coefficient and the delay time to obtain corrected transient voltage data; When the damping coefficient is lower than a stable threshold, real-time voltage transient data of the remote micro-grid is obtained to be combined with the corrected transient voltage data to obtain a transient voltage data set; The transient voltage data set is subjected to 3-layer decomposition processing by a db4 wavelet to obtain load harmonic features, and the load harmonic features are subjected to harmonic suppression processing by a filter with the filter parameter to obtain filtered features; The suppression fitness value is determined according to the load harmonic features and the filtered features.

6. The method for load harmonic suppression for remote microgrid according to claim 5, wherein, When the suppression fitness value is not up to standard, the voltage transient data is decomposed to adjust impedance matching parameters of an SVG to obtain optimized impedance parameters and corresponding optimized voltage transient data, including: When the suppression fitness value is not up to standard, the transient voltage data set is decomposed to obtain a main harmonic frequency, and a harmonic amplitude proportion is determined according to the main harmonic frequency to determine an impedance adjustment direction based on resonance principle; Based on the impedance adjustment direction, the impedance matching parameters of the SVG are iteratively adjusted by a gradient descent method to obtain optimized impedance parameters and corresponding optimized voltage transient data; A suppression fitness value is recalculated according to the optimized voltage transient data, and when a new suppression fitness value is not up to standard, an impedance parameter change trend is determined based on the optimized impedance parameters; Based on the impedance parameter change trend, a parameter adjustment step is optimized by linear regression to recalculate a new suppression fitness value until it is up to standard, and the optimized impedance parameters and corresponding optimized voltage transient data are finally output.

7. The method for load harmonic mitigation for remote microgrid as claimed in claim 6 wherein, The angle deviation data in the optimized voltage transient data is extracted to determine a switching frequency of a converter valve, and an electromagnetic torque fluctuation response model is constructed in combination with the optimized impedance parameters, including: The optimized voltage transient data is subjected to Fourier transform processing to extract rotor angular velocity features and power angle features; An angular velocity deviation is determined based on the rotor angular velocity features, and an angle offset is determined based on the power angle features; According to the angular velocity deviation and the angle offset, the power angle characteristic is corrected by using a linear difference method to obtain a power angle correction characteristic; The power angle correction characteristic is input into a pre-constructed deviation frequency linear regression model for processing, and the switching frequency of the converter valve is output; The switching frequency, the optimized impedance parameter and the stator parameter of the motor are input into a synchronous motor mathematical model to construct the electromagnetic torque fluctuation response model.

8. The method for load harmonic suppression for remote microgrid according to claim 7, characterized in that, The SVG reactive power data is optimized to obtain target reactive power data, including: Obtain the load data of the remote micro-grid, and extract several harmonic components from the load data to input into the electromagnetic torque fluctuation response model for simulation, and output torque fluctuation data; When it is determined that the torque fluctuation data exceeds a preset threshold, the gradient descent method is used to iteratively adjust the optimized impedance parameter until the obtained torque fluctuation data does not exceed the preset threshold, and the final torque fluctuation data is output; Based on the final torque fluctuation data, determine the reactive power change data, and optimize the SVG reactive power data according to the reactive power change data to obtain preliminary reactive power optimization data; When it is determined that the preliminary reactive power optimization data matches the torque fluctuation, the support vector regression is used to predict the boundary of the reactive power change data, and the reactive power optimization boundary data is output; According to the reactive power optimization boundary data, the switching frequency is adjusted, the electromagnetic torque fluctuation response model is updated based on the adjustment result, and the preliminary reactive power optimization data is optimized according to the updated electromagnetic torque fluctuation response model to obtain the target reactive power data.

9. The method for load harmonic mitigation for remote microgrid as claimed in claim 1 wherein, The target filter parameter is obtained by optimizing the filter parameter according to the target reactive power data, including: Collect the current power angle dynamic data of the remote micro-grid, and when the current power angle dynamic data exceeds the power angle offset threshold, perform sensitivity analysis on the filter parameter to determine the filter parameter to be optimized; According to the target reactive power data, construct a constraint condition, and according to the damping coefficient, the suppression fitness value and the power angle offset threshold, construct a target function, so as to optimize the filter parameter to be optimized by a multi-objective optimization algorithm to obtain an optimized filter parameter; Based on the optimized filter parameter, full working condition simulation verification is performed, and when the verification is passed, the optimized filter parameter is output as the target filter parameter.

10. A load harmonic suppression system for remote microgrids, characterized in that, Including: The characteristic matching module is used to obtain the stator flux data of the motor in the remote micro-grid, and the AC filter characteristic matching is performed according to the stator flux data to obtain the filter parameter to determine the SVG reactive power data; The electromagnetic simulation module is used to analyze the SVG reactive power data to obtain the DC side capacitor data to simulate the electromagnetic torque of the motor, generate the stator flux dynamic response data to determine the response speed; The coefficient quantization module is used to determine the capacitance rate parameter according to the DC side capacitor data, and combine the response speed and the preset control loop model to calculate the damping coefficient; An adaptability calculation module is configured to obtain voltage transient data of the remote micro-grid based on the damping coefficient, perform harmonic suppression processing based on the filter parameter, and calculate an inhibition adaptability value; An impedance matching module is configured to decompose the voltage transient data to adjust impedance matching parameters of the SVG to obtain optimized impedance parameters and corresponding optimized voltage transient data when the inhibition adaptability value is not up to standard; A data optimization module is configured to extract angle deviation data in the optimized voltage transient data to determine a switching frequency of a converter valve, construct an electromagnetic torque fluctuation response model in combination with the optimized impedance parameters, and optimize the SVG reactive power data to obtain target reactive power data; A strategy generation module is configured to optimize the filter parameter according to the target reactive power data, obtain target filter parameters, and generate a load harmonic suppression strategy in combination with the switching frequency.

Images