Double-variant collaborative adaptive control method for virtual inertia and damping parameters of GFM inverter under weak power grid

Abstract

The invention discloses a double-variant collaborative self-adaptive control method for virtual inertia and damping parameters of a GFM inverter under a weak power grid, and the method comprises the steps: based on a photovoltaic power generation device, an inverter and an LC filter which are connected in sequence, the photovoltaic power generation device outputs a direct current, the inverter converts the direct current into an alternating current, and the high-frequency harmonic wave is filtered through the LC filter; and the grid-connected point is connected into an alternating-current power grid through an equivalent inductor. The method specifically comprises the following steps: step 1, acquiring an electrical quantity in real time based on a network-forming inverter; step 2, the DSP control module obtains a power grid short-circuit ratio and an active power change rate through voltage, current and active power data; 3, presetting a parameter matrix of virtual inertia and virtual damping corresponding to different short-circuit ratios and active power change rates; and 4, comparing whether the data change exceeds 5%, and judging whether to enter a parameter adaptation process. According to the method, secondary disturbance of parameter switching is eliminated, system frequency oscillation and electric energy quality deterioration are inhibited, and the dynamic response performance is remarkably improved.

Description

A bivariate cooperative adaptive control method for virtual inertia and damping parameters of GFM inverters under weak power grid conditions Technical Field

[0001] This invention relates to the field of power electronic control technology, and in particular to a bivariate cooperative adaptive control method for virtual inertia and damping parameters of a GFM inverter under weak grid conditions. This method is applicable to the frequency stability control of grid-type inverters in weak grid scenarios where the short-circuit ratio (SCR<=3) fluctuates dynamically. Background Technology

[0002] Currently, grid-connected inverters, by simulating the virtual inertia and virtual damping characteristics of synchronous generators, provide frequency support and oscillation suppression capabilities for weak power grids, making them key equipment for grids with high penetration of new energy sources. However, the short-circuit ratio of weak power grids dynamically changes in the range of 1.2-3.0 depending on factors such as load switching and new energy power fluctuations. At the same time, photovoltaic fluctuations and sudden load changes can cause changes in the active power fluctuation rate.

[0003] To address these issues, current solutions primarily rely on fixed virtual inertia and damping values ​​based on a single SCR parameter. This neglects the coupling effect between power fluctuations and the SCR, meaning that when the SCR drops sharply and power surges, the fixed parameters can cause frequency overshoot exceeding 0.2Hz and oscillation convergence time exceeding 1 second. While some dynamic parameter adjustment schemes incorporate SCR feedback, they employ step-like parameter switching, which can easily trigger new voltage fluctuations and lacks adaptation to power fluctuations. Furthermore, traditional control methods independently adjust inertia and damping parameters, ignoring their synergistic effect. Under dual-variable disturbances in a weak power grid, this makes it difficult to balance frequency response speed and stability, easily triggering inverter protection shutdown. Summary of the Invention

[0004] To address the aforementioned issues, this invention provides a bivariate cooperative adaptive control method for virtual inertia and damping parameters of a GFM inverter under weak power grid conditions.

[0005] The specific plan is as follows:

[0006] A bivariate cooperative adaptive control method for virtual inertia and damping parameters of a GFM inverter under weak grid conditions is proposed. This method is based on a photovoltaic power generation device, an inverter, and an LC filter connected sequentially. The photovoltaic power generation device outputs DC power, which is then converted to AC power by the inverter. High-frequency harmonics are filtered out by the LC filter, and the AC power is connected to the AC grid via the equivalent inductance at the grid connection point. The method includes the following steps:

[0007] Step 1: Based on the real-time electrical quantities at the grid connection point of the grid-connected inverter, collect the three-phase voltage, three-phase current, and output active power;

[0008] Step 2: The DSP control module calculates the short-circuit ratio (SCR) of the power grid in real time based on the collected voltage and current, and calculates the active power fluctuation rate (dP / dt) based on the collected active power.

[0009] Step 3: The DSP control module presets a two-dimensional parameter matrix of SCR and dP / dt, which includes the optimal virtual inertia J and virtual damping D under different scenarios;

[0010] Step four: Compare the current combined SCR and dP / dt data with the data from the previous sampling period. Determine if the difference in range or deviation exceeds 5%, and proceed with the parameter adaptation process. If the difference exceeds 5%, proceed with the parameter adaptation process; otherwise, maintain the current virtual inertia and virtual damping.

[0011] In step four, the DSP control module uses a bilinear interpolation algorithm to generate the target virtual inertia J and target virtual damping D values ​​from the parameter matrix based on the real-time SCR and dP / dt. Then, it uses an exponential gradient strategy to smoothly update the current J and D to the target values, with a gradient time constant of 40~60ms. The real-time J and D are then injected into the virtual synchronous machine control core of the grid-connected inverter to generate inverter drive signals, adjust the output, and feed them back to the grid connection point PCC, forming a real-time closed-loop control.

[0012] Furthermore, the acquired three-phase voltage and current signals are transformed from a three-phase stationary coordinate system to a two-phase stationary coordinate system using Parker transformation to eliminate coupling effects and filter out harmonics, obtaining the voltage and current signals in the two-phase stationary coordinate system. Then, the equivalent impedance Zep of the power grid is iteratively updated using the Kalman filter algorithm. The calculation process for the power grid short-circuit ratio (SCR) is as follows:

[0013]

[0014] in The equivalent impedance of the power grid. The effective value of the line voltage at PCC; where This refers to the rated capacity of the inverter.

[0015] Furthermore, the calculation process for the active power fluctuation rate dP / dt is as follows:

[0016] Using the sliding time window method, the active power difference between two adjacent sampling periods is taken, i.e.

[0017]

[0018] in The sampling time interval is defined as a value that takes the value of 1. ms; The active power for the current cycle. This represents the active power of the previous cycle, and the units of dP / dt are uniformly set to [missing unit]. ( (Rated active power of the GFM inverter).

[0019] Furthermore, the preset two-dimensional parameter matrix stores corresponding optimal virtual inertia and virtual damping parameter pairs in two dimensions: short-circuit ratio and active power change rate. The process of dynamically determining the target parameter value is to use an interpolation algorithm to query and calculate in the two-dimensional parameter matrix.

[0020] Furthermore, the row dimension of the matrix is ​​the SCR interval, divided into 3 levels: weak network interval ( ), medium and weak network intervals ( ), near-strong network interval ( ); The column dimension is in dP / dt level, divided into 3 levels: weak perturbation Medium disturbance Strong disturbance .

[0021] Furthermore, the application scenario of the bilinear interpolation algorithm is: when the real-time SCR or dP / dt is at the interval boundary of the parameter matrix, bilinear interpolation is triggered.

[0022] Furthermore, the smooth transition strategy is a first-order inertial element or an exponential gradual curve, and its mathematical expression is:

[0023]

[0024] in, This is a time constant, with a value ranging from 40ms to 60ms. This is the current value of the virtual inertia. The target value for virtual inertia, where the unit of virtual inertia is... .

[0025] The advantages and positive effects of this invention are:

[0026] (1) Breaking through the limitations of traditional perception based solely on SCR or a single power variable, the grid short-circuit ratio and active power fluctuation rate are collected simultaneously to construct a dual-dimensional adaptation basis for grid strength and disturbance intensity. Through dual-variable collaborative adaptation, the frequency overshoot is reduced and the oscillation convergence time is shortened, thus meeting both steady-state and dynamic requirements.

[0027] (2) The rate of change of virtual inertia and virtual impedance parameters is effectively controlled, the voltage fluctuation amplitude is reduced, there is no new oscillation, the power quality of the power grid is guaranteed, and the risk of inverter disconnection due to parameter switching is avoided.

[0028] (3) The core control module includes dual-variable sensing and coordination adaptation, which are embedded in the existing GFM control logic in the form of subroutines. No system reconstruction is required, and it can be adapted to the mainstream domestic GFM inverters.

[0029] (4) Through stable frequency support capability, the grid connection stability margin of GFM inverter in weak grid is increased by 50%, which meets the grid's requirements for accepting high-penetration new energy sources.

[0030] (5) It can not only adapt to grid-side disturbances such as load switching and SCR fluctuations, but also cope with new energy-side disturbances such as photovoltaic irradiance changes and energy storage charging and discharging switching. It can be widely used in distributed photovoltaic, energy storage power stations, microgrids and other scenarios. It does not require customized development for different scenarios and has strong versatility. Attached Figure Description

[0031] Figure 1 is a system block diagram of the present invention;

[0032] Figure 2 is a flowchart of the method of the present invention. Detailed Implementation

[0033] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0034] The English terms involved in this invention are annotated in Chinese as follows: GFM grid-type control, SCR short-circuit ratio, dP / dt power change rate, and PCC grid connection point.

[0035] Please refer to Figures 1 and 2. A bivariate cooperative adaptive control method for virtual inertia and damping parameters of a grid-connected inverter under weak power grid conditions is proposed. This method is based on a photovoltaic power generation device, an inverter, and an LC filter connected in sequence. The photovoltaic power generation device outputs DC power, the inverter converts the DC power to AC power, the LC filter removes high-frequency harmonics, and the AC power is connected to the grid via the equivalent inductance at the grid connection point. The method specifically includes the following steps:

[0036] Step 1: Based on the real-time electrical quantities at the grid connection point of the grid-connected inverter, collect the three-phase voltage, three-phase current, and output active power;

[0037] Step 2: The DSP control module calculates the short-circuit ratio (SCR) of the power grid in real time based on the collected three-phase voltage and three-phase current, and calculates the active power fluctuation rate (dP / dt) based on the collected active power.

[0038] Step 3: The DSP control module presets a two-dimensional parameter matrix of SCR and dP / dt, which includes the optimal virtual inertia J and virtual damping D under different scenarios;

[0039] Step four: Compare the current combined SCR and dP / dt data with the data from the previous sampling period. Determine if the difference in range or deviation exceeds 5%, and proceed with the parameter adaptation process. If the difference exceeds 5%, proceed with the parameter adaptation process; otherwise, maintain the current virtual inertia and virtual damping.

[0040] In step four, the DSP control module uses a bilinear interpolation algorithm to generate the target virtual inertia J and target virtual damping D values ​​from the parameter matrix based on the real-time SCR and dP / dt. Then, it uses an exponential gradient strategy to smoothly update the current virtual inertia J and virtual impedance D to the target values, with a gradient time constant of 40~60ms. The real-time J and D are then injected into the virtual synchronous machine control core of the grid-connected inverter to generate inverter drive signals, adjust the output, and feed them back to the grid connection point PCC to form real-time closed-loop control.

[0041] Preferably, the acquired three-phase voltage and current signals are transformed from a three-phase stationary coordinate system to a two-phase stationary coordinate system using Park transform to eliminate coupling effects and filter harmonics, obtaining voltage and current signals in the two-phase stationary coordinate system. Then, the equivalent impedance Zep of the power grid is iteratively updated using a Kalman filter algorithm. The calculation process for the power grid short-circuit ratio (SCR) is as follows:

[0042]

[0043] in The equivalent impedance of the power grid. The effective value of the line voltage at PCC; where This refers to the rated capacity of the inverter.

[0044] Preferably, the calculation process for the active power fluctuation rate dP / dt is as follows:

[0045] Using the sliding time window method, the active power difference between two adjacent sampling periods is taken, i.e.

[0046]

[0047] in The sampling time interval is defined as a value that takes the value of 1. ms; The active power for the current cycle. This represents the active power of the previous cycle, and the units of dP / dt are uniformly set to [missing unit]. ( (Rated active power of the GFM inverter).

[0048] Preferably, the cooperative relationship is a preset two-dimensional parameter matrix. This matrix stores the corresponding optimal virtual inertia and virtual damping parameter pairs in two dimensions: short-circuit ratio and active power change rate. The process of dynamically determining the target value of the parameters is to use an interpolation algorithm to query and calculate in the two-dimensional parameter matrix.

[0049] Preferably, the matrix is ​​characterized by having row dimensions in the SCR interval, divided into three levels: weak network interval, medium-weak network interval, and near-strong network interval; and column dimensions in the dP / dt level, divided into three levels: weak disturbance, medium disturbance, and strong disturbance.

[0050] Preferably, the application scenario of the bilinear interpolation algorithm is: when the real-time SCR or dP / dt is at the interval boundary of the parameter matrix, bilinear interpolation is triggered.

[0051] Preferably, the smooth transition strategy is a first-order inertial element or an exponential gradient curve, and its mathematical expression is:

[0052]

[0053] in, This is a time constant, with a value ranging from 40ms to 60ms.

[0054] The aforementioned process control, which features dual-variable sensing, collaborative adaptation, and smooth switching, uses the TIF28379D floating-point DSP as its control core. It is built upon the existing sensor and circuit architecture of the inverter and is written in C language to ensure real-time performance and engineering repeatability.

[0055] The structure and working principle of the present invention will be further illustrated below with reference to a preferred embodiment:

[0056] Please refer to Figures 1 and 2. A bivariate cooperative adaptive control method for virtual inertia and damping parameters of a grid-connected inverter under weak power grid conditions includes a photovoltaic power generation device, an inverter, and an LC filter connected in sequence. The photovoltaic power generation device outputs DC power, the inverter converts the DC power into AC power, the LC filter removes high-frequency harmonics, and the AC power is connected to the AC power grid through the grid connection point via the equivalent inductance. The method is characterized by further including the following steps:

[0057] Step 1: Based on the real-time electrical quantities at the grid connection point of the grid-connected inverter, collect the three-phase voltage, three-phase current, and output active power;

[0058] Step 2: The DSP control module calculates the short-circuit ratio (SCR) of the power grid in real time based on the collected three-phase voltage and three-phase current, and calculates the active power fluctuation rate (dP / dt) based on the collected active power.

[0059] Step 3: The DSP control module presets a two-dimensional parameter matrix of SCR and dP / dt, which includes the optimal virtual inertia J and virtual damping D under different scenarios;

[0060] Step four: Compare the current combined SCR and dP / dt data with the data from the previous sampling period. Determine if the difference in range or deviation exceeds 5%, and proceed with the parameter adaptation process. If the difference exceeds 5%, proceed with the parameter adaptation process; otherwise, maintain the current virtual inertia and virtual damping.

[0061] In step four, the DSP control module uses a bilinear interpolation algorithm to generate the target virtual inertia J and target virtual damping D values ​​from the parameter matrix based on the real-time SCR and dP / dt. Then, it uses an exponential gradient strategy to smoothly update the current virtual inertia J and virtual impedance D to the target values, with a gradient time constant of 40~60ms. The real-time J and D are then injected into the virtual synchronous machine control core of the grid-connected inverter to generate inverter drive signals, adjust the output, and feed them back to the grid connection point PCC to form real-time closed-loop control.

[0062] First, the GFM inverter control platform is initialized and configured, including setting the system clock frequency and interrupt cycle. The ADC interrupt cycle corresponding to signal acquisition and preprocessing is set to 100μs, the timer interrupt 1 cycle corresponding to SCR and dP / dt dual-variable calculation is set to 10ms, the timer interrupt 2 cycle corresponding to parameter adaptation and gradual switching is set to 100μs, and the signal interrupt cycle corresponding to GFM control core and PWM generation is set to 50μs. At the same time, the basic parameters of the algorithm are pre-configured, including the process noise covariance, measurement noise covariance and initial impedance estimate of Kalman filtering, and the three-dimensional parameter matrix "ZCR-dP / dt-JD" is pre-stored. The matrix elements are the optimal virtual inertia J and virtual damping D values ​​pre-tuned by small signal stability analysis. The smooth switching parameters are initialized, with the initial virtual inertia J_old=0.08kg·m², the initial virtual damping D_old=20, and the exponential gradual switching time constant τ=50ms. All kinds of interrupts are enabled to ensure the orderly execution of the process.

[0063] Furthermore, the bivariate real-time sensing step is executed in timer interrupt 1 with a period of 10ms. First, the three-phase voltage, three-phase current, and output active power of the grid connection point (PCC) are collected through the existing sampling circuit of the GFM inverter. The collected three-phase voltage and current signals are transformed from the abc three-phase stationary coordinate system to the α and β two-phase stationary coordinate system through Clark transformation to eliminate the influence of three-phase coupling. Then, the switching harmonics and grid background harmonics are filtered out by a second-order low-pass filter (cutoff frequency 200Hz) to obtain the denoised voltage and current signals. Based on the denoised signals, SCR estimation is performed. First, the grid equivalent impedance is iteratively updated through the Kalman filter algorithm. The specific process is to reconstruct the ideal grid voltage, construct the observation matrix H and the measurement vector y, update the covariance matrix through the prediction step, calculate the gain matrix and correct the impedance estimate through the update step, and calculate the grid equivalent impedance to obtain the real-time SCR.

[0064]

[0065] in The equivalent impedance of the power grid. This is the effective value of the line voltage at PCC. The effective value of the line current at PCC; where This refers to the rated capacity of the inverter.

[0066] Furthermore, the parameter coordination and adaptation process and the bivariate perception process are executed synchronously in timer interrupt 1. First, the parameter adjustment requirement is determined by comparing the current combination of SCR and dP / dt with the combination of the previous sampling period. If the difference between the intervals or the deviation exceeds 5%, the parameter adaptation process is started; otherwise, the current J and D parameters are maintained. After the adaptation process is started, the pre-stored three-dimensional parameter matrix is ​​called. When the real-time SCR or dP / dt is at the boundary of the matrix interval, the bilinear interpolation algorithm is triggered to calculate the target virtual inertia and target virtual damping. The interpolation process first determines the interval where the real-time SCR and dP / dt are located and the corresponding interpolation ratio, and then obtains continuous target parameter values ​​through two linear interpolations in the horizontal and vertical directions.

[0067] Furthermore, the smooth switching execution phase is executed in timer interrupt 2 with a period of 100μs, and an exponential gradient strategy is used to update the current J and D parameters to the target value J. new D new ,

[0068]

[0069] in, This is a time constant, with a value ranging from 40ms to 60ms. This is the current value of the virtual inertia. The target value for virtual inertia, where the unit of virtual inertia is... .

[0070] The GFM control core executes during a 50μs period signal interruption, injecting the smoothly updated real-time J and D parameters into the Virtual Synchronous Machine (VSM) control equations. It calculates the inverter angular frequency change through the rotor motion equations, updates the angular frequency, and generates dq-axis voltage reference values. These values ​​are then converted into abc-axis voltage signals via Park inverse transformation. Finally, a drive signal is generated through space vector pulse width modulation to drive the inverter power conversion module to adjust the output power. The output power is then filtered and fed back to the PCC for stable grid-connected operation.

Claims

1. A bivariate cooperative adaptive control method for virtual inertia and damping parameters of a GFM inverter under weak power grid conditions, characterized in that, Based on a photovoltaic power generation device, an inverter, and an LC filter connected in sequence, where the photovoltaic power generation device outputs DC power, the inverter converts the DC power to AC power, the LC filter removes high-frequency harmonics, and the power is connected to the AC grid through the grid connection point via the equivalent inductance; specifically, the following steps are included: Step 1, based on the real-time electrical quantities at the grid connection point of the grid-connected inverter, the three-phase voltage, three-phase current, and output active power are collected; Step 2, the DSP control module calculates the short-circuit ratio (SCR) of the grid in real time based on the collected three-phase voltage and three-phase current, and calculates the active power fluctuation rate (dP / dt) based on the collected active power; Step 3, the DSP control module presets a two-dimensional parameter matrix for SCR and dP / dt, including the optimal virtual inertia J and virtual damping D under different scenarios; Step 4, ... If the combined data of the current SCR and dP / dt is compared with the data of the previous sampling period, and the difference in interval or deviation exceeds 5%, it is determined whether to enter the parameter adaptation process. If it is higher than 5%, the parameter adaptation process is entered; otherwise, the current virtual inertia and virtual damping are maintained. In step four, the DSP control module uses a bilinear interpolation algorithm to generate the target virtual inertia J and target virtual damping D values ​​from the parameter matrix based on the real-time SCR and dP / dt. Then, the current virtual inertia J and virtual impedance D are smoothly transitioned to the target values ​​through an exponential gradient strategy, with a gradient time constant of 40~60ms. The real-time J and D are then injected into the virtual synchronous machine control core of the grid-connected inverter, thereby generating the inverter drive signal, adjusting the output and feeding it back to the grid connection point PCC to form a real-time closed-loop control.

2. The method according to claim 1, characterized in that, The calculation process for the short-circuit ratio (SCR) of the power grid is as follows: the collected three-phase voltage and current signals are transformed from a three-phase stationary coordinate system to a two-phase stationary coordinate system through Park transformation to eliminate coupling effects and filter out harmonics, obtaining the voltage and current signals in the two-phase stationary coordinate system. Then, the equivalent impedance Z of the power grid is iteratively updated using the Kalman filter algorithm. ep Then calculate SCR: in The equivalent impedance of the power grid. This is the effective value of the line voltage at PCC. This refers to the rated capacity of the inverter.

3. The method according to claim 1, characterized in that, The calculation process for the active power fluctuation rate dP / dt is as follows: using the sliding time window method, the active power difference between two adjacent sampling periods is taken, i.e. in The sampling time interval is defined as a value that takes the value of 1. ms; The active power for the current cycle. This represents the active power of the previous cycle, and the units of dP / dt are uniformly set to [missing unit]. ， This refers to the rated active power of the GFM inverter.

4. The method according to claim 1, characterized in that, The two-dimensional parameter matrix stores the corresponding optimal virtual inertia and virtual damping parameter pairs in two dimensions: short-circuit ratio and active power change rate. The process of dynamically determining the target parameter value is to use an interpolation algorithm to query and calculate in the two-dimensional parameter matrix.

5. The method according to claim 1, characterized in that, The row dimension of the two-dimensional parameter matrix is ​​the SCR interval, divided into 3 levels: weak network interval. Medium and weak network intervals ； Near strong network interval The column dimension is in dP / dt level, divided into 3 levels: weak perturbation. Medium disturbance Strong disturbance 。 6. The method according to claim 1, characterized in that, The application scenario for bilinear interpolation algorithm is: when the real-time SCR or dP / dt is at the interval boundary of the parameter matrix, bilinear interpolation is triggered.

7. The method according to claim 1, characterized in that, The smooth transition strategy is a first-order inertial element or an exponential gradient curve, and its mathematical expression is: in, This is a time constant, with a value ranging from 40ms to 60ms. This is the current value of the virtual inertia. The target value for virtual inertia, where the unit of virtual inertia is... 。

Images