Dynamic reactive power control method and device for new energy power plants

Abstract

This invention relates to a dynamic reactive power control method and device for new energy power plants. The method includes the following steps: calculating a dynamic safety domain of virtual impedance based on a real-time estimated value of grid impedance; generating a virtual impedance control quantity under the constraints of the dynamic safety domain based on a multi-timescale reinforcement learning algorithm; obtaining a reactive power output error value and correcting the virtual impedance control quantity using the reactive power output error value; and controlling the reactive power output of the power plant's inverter based on the corrected virtual impedance control quantity. The technical solution provided by this invention predicts grid impedance using a multi-timescale reinforcement learning algorithm, thereby generating a virtual impedance control quantity. The upper-level prediction model can predict impedance change trends in advance, providing a forward-looking benchmark for lower-level dynamic correction. The lower-level dynamic correction model can achieve instantaneous disturbance suppression and rapid fine-tuning of the virtual impedance, thus obtaining a more accurate virtual impedance control quantity.

Description

Technical Field

[0001] This invention relates to the field of power system control, and in particular to a method and apparatus for dynamic reactive power control in new energy power plants. Background Technology

[0002] As the penetration rate of new energy sources in the power system continues to increase, reactive power control in new energy power plants is becoming increasingly critical for maintaining the safe and stable operation of the power system and providing large-scale grid security. Traditional reactive power control methods for new energy power plants typically design control strategies based on fixed grid model parameters and control models with single characteristics.

[0003] However, the actual operation of the power grid is complex and variable. Grid impedance changes in real time due to factors such as load fluctuations, grid topology adjustments, and the randomness of renewable energy generation output. Control strategies based on fixed parameters are difficult to adapt to these dynamic changes, easily leading to inaccurate reactive power control and an inability to effectively maintain system voltage stability. Furthermore, the operation of renewable energy plants involves both responding to rapid power fluctuations on short timescales and ensuring stable system operation on long timescales. Single-characteristic control models cannot comprehensively coordinate control objectives across different timescales, making it difficult to achieve optimal reactive power control in actual operation.

[0004] Therefore, how to achieve optimal control of reactive power in new energy power plants and improve control accuracy has become an urgent problem to be solved. Summary of the Invention

[0005] Based on the above-mentioned situation of the prior art, the purpose of this invention is to provide a dynamic reactive power control method and device for new energy power plants. By using a multi-timescale reinforcement learning algorithm to predict the grid impedance, it achieves precise control of reactive power of new energy power plants, providing a strong guarantee for the safe operation of large-scale power grids.

[0006] To achieve the above objectives, according to one aspect of the present invention, a method for dynamic reactive power control in new energy power plants is provided, comprising the steps of:

[0007] The dynamic security domain of the virtual impedance is calculated based on the real-time estimated value of the grid impedance.

[0008] Based on a multi-timescale reinforcement learning algorithm, a virtual impedance control quantity is generated under the constraint of the dynamic security domain; the multi-timescale reinforcement learning algorithm includes an upper-layer prediction model and a lower-layer dynamic correction model.

[0009] Obtain the reactive power output error value, and use the reactive power output error value to correct the virtual impedance control quantity;

[0010] Reactive power output control of the inverters in the power plant is performed based on the modified virtual impedance control quantity.

[0011] Furthermore, the dynamic security domain includes an impedance constraint; the impedance constraint is expressed as:

[0012] X min ≤X virtual ≤X max

[0013] Among them, X virtual X represents the virtual impedance. min X represents the minimum value of the virtual impedance. max This represents the maximum value of the virtual impedance.

[0014] Furthermore, the minimum value of the virtual impedance is expressed as:

[0015]

[0016] The maximum value of the virtual impedance is expressed as:

[0017]

[0018] Where, ΔV max Indicates the upper limit of voltage deviation, ΔV min Indicates the lower limit of voltage deviation, ΔQ max V represents the maximum permissible change in reactive power. pcc X represents the voltage at the point of common coupling, and X represents the equivalent reactance of the power grid.

[0019] Furthermore, the upper-level model includes an LSTM-PINN hybrid prediction model, which is used to predict future grid impedance. The hybrid prediction model is expressed as follows:

[0020]

[0021] Among them, Z grid (tn:t) represents the historical impedance sequence, n represents the time step, and P pv ω represents the real-time photovoltaic output value. LSTM and ω PINN These represent the weight coefficients of the LSTM model and the PINN model, respectively. This represents the predicted value of the grid impedance in the k-th second.

[0022] Furthermore, the lower-level virtual impedance dynamic correction model includes the SAC model, and the state space s and action space a of the SAC model are represented as follows:

[0023]

[0024] a=ΔX virtual ∈[-0.1,0.1]pu

[0025] Among them, Q current Indicates the current reactive power output; ΔX virtual This represents the virtual impedance correction amount; This represents the predicted value of the grid impedance in the next k steps as predicted by the upper-level model;

[0026] The reward function of the SAC model is expressed as:

[0027]

[0028] Where α represents the voltage deviation penalty coefficient, β represents the reactive power deviation penalty coefficient, η represents the rate of change penalty coefficient, λ represents the safety over-limit penalty coefficient, and ΔV pcc ΔQ represents the voltage deviation value, ΔQ represents the reactive power deviation value, Ω represents the dynamic safety domain, and Violation(Ω) represents the degree of violation of the dynamic safety domain constraints.

[0029] Furthermore, the upper-level prediction model and the lower-level dynamic correction model are nested on a time scale. The upper-level prediction model updates the virtual impedance reference value every first cycle, and the lower-level dynamic correction model updates the virtual impedance correction amount every second cycle. The outputs of the upper-level prediction model and the lower-level dynamic correction model are fused to obtain the virtual impedance control amount.

[0030] The first period is shorter than the second period.

[0031] Furthermore, the virtual impedance control quantity is corrected using the reactive power output error value, including:

[0032] Calculate the virtual impedance adjustment value based on the reactive power output error value;

[0033] The virtual impedance control value is corrected based on the virtual impedance adjustment value.

[0034] Furthermore, the virtual impedance adjustment value ΔX(k) is expressed as:

[0035]

[0036] Among them, K p K represents the proportionality coefficient. i denoted by , e(k) represents the integral coefficient, and e(k) represents the reactive power output error value.

[0037] According to another aspect of the present invention, a dynamic reactive power control device for new energy power plants is provided, comprising:

[0038] The dynamic security domain calculation module is used to calculate the dynamic security domain of virtual impedance based on the real-time estimated value of grid impedance.

[0039] The virtual impedance control quantity calculation module is used to generate virtual impedance control quantities under the constraints of the dynamic security domain based on a multi-timescale reinforcement learning algorithm; the multi-timescale reinforcement learning algorithm includes an upper-layer prediction model and a lower-layer dynamic correction model.

[0040] The virtual impedance control quantity correction module is used to obtain the reactive power output error value and use the reactive power output error value to correct the virtual impedance control quantity.

[0041] The reactive power output control module is used to control the reactive power output of the inverters in the power plant based on the corrected virtual impedance control quantity.

[0042] In summary, this invention provides a dynamic reactive power control method and device for new energy power plants. The method includes the following steps: calculating the dynamic security domain of virtual impedance based on real-time estimates of grid impedance; generating a virtual impedance control quantity under the constraints of the dynamic security domain using a multi-timescale reinforcement learning algorithm; the multi-timescale reinforcement learning algorithm includes an upper-level prediction model and a lower-level dynamic correction model; obtaining reactive power output error values ​​and using these errors to correct the virtual impedance control quantity; and controlling the reactive power output of the power plant's inverter based on the corrected virtual impedance control quantity. The technical solution provided by this invention predicts grid impedance using a multi-timescale reinforcement learning algorithm, thereby generating a virtual impedance control quantity. The upper-level prediction model can predict impedance change trends in advance, providing a forward-looking benchmark for lower-level dynamic correction. The lower-level dynamic correction model can suppress instantaneous disturbances and achieve rapid fine-tuning of the virtual impedance, thus obtaining a more accurate virtual impedance control quantity. Based on this, the reactive power output error value is used to correct the virtual impedance control quantity, forming a hierarchical control architecture of "coarse adjustment + fine adjustment", which realizes precise control of reactive power of new energy power plants and provides a strong guarantee for the safe operation of large-scale power grids. Attached Figure Description

[0043] Figure 1 This is a flowchart of the dynamic reactive power control method for new energy power plants provided in an embodiment of the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0045] It should be noted that, unless otherwise defined, the technical or scientific terms used in one or more embodiments of the present invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in one or more embodiments of the present invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the element or object listed following the word and its equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0046] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings. An embodiment of the present invention provides a dynamic reactive power control method for new energy power plants. In this embodiment, the new energy power plant is a photovoltaic power plant. Figure 1 The flowchart of the dynamic reactive power control method for new energy power plants according to an embodiment of the present invention is shown below. Figure 1 As shown, the control method includes the following steps:

[0047] S202. Calculate the dynamic security domain of the virtual impedance based on the real-time estimated value of the grid impedance.

[0048] S2021. Collect voltage, current, frequency, and phase angle data from the power grid's point of common coupling (PCC) to calculate a real-time estimate of the current power grid impedance. Perform a synchronous coordinate transformation (dq transformation) on the collected PCC voltage and current data to obtain the dq-axis coordinates of voltage and current. Construct a linear regression model:

[0049]

[0050] Among them, V d V q I d and I q These represent the dq-axis components of voltage and current, respectively. Represents the observed value. Represents the regression matrix. Let R and X represent the parameter vector, where R and X represent the real and imaginary parts of the grid impedance, respectively.

[0051] Each time a sample is updated, the regression matrix and the observed values ​​at the current time step are substituted into the recursive formula:

[0052] θ k+1 =θ k +K k+1 (Y k+1 -Φ k+1T θ k )

[0053] The Kalman gain update formula is:

[0054]

[0055] Where P represents the covariance matrix and λ is the forgetting factor, typically 0.99.

[0056] The initial values ​​of the parameter vector are θ0 = [R0, X0]. T The values ​​of R0 and X0 can be set according to the nominal impedance of the power grid, for example, R0 = 0.01 pu and X0 = 0.05 pu; the initial value of the covariance matrix can be set as a diagonal matrix.

[0057] P0 = σ 2 I

[0058] Where I represents the identity matrix, and σ takes a large value, for example, 10. 6 This indicates that the initial estimate has high uncertainty. Based on a predetermined sampling period (e.g., 0.1 ms), the parameters are updated each time sampled data is received, thus obtaining the current estimated value of the grid equivalent impedance Z. grid (k):

[0059] Z grid (k)=R(k)+jX(k)

[0060] S2022. Based on the estimated equivalent impedance of the power grid, calculate the dynamic security domain of the virtual impedance. The dynamic security domain of the virtual impedance is an impedance constraint generated based on the voltage constraint, used to constrain the values ​​of the virtual impedance control variables. This impedance constraint is expressed as:

[0061] X min ≤X virtual ≤X max

[0062] Among them, X virtual X represents the virtual impedance. min and X max These represent the minimum and maximum values ​​of the virtual impedance obtained based on voltage constraints. A photovoltaic power station connects to the grid through virtual impedance, and its equivalent circuit satisfies the equivalent circuit equation:

[0063] V pcc =V grid +(R+jX)·I pcc +jX virtual ·I pcc

[0064] Among them, V grid V represents the grid-side voltage.pcc I represents the voltage at the point of common coupling. pcc This represents the point of common coupling current. Ignoring the active component, the photovoltaic reactive power approximately satisfies:

[0065] Q out ≈V pcc ·I Q

[0066] Among them, I Q This represents the vertical component of the current (i.e., the reactive component). Substituting the current into the equivalent circuit equation, the voltage deviation can be expressed as:

[0067]

[0068] Among them, P out This represents the active power output of photovoltaic systems (usually decoupled from reactive power control). Therefore, when the active power effect is ignored (P... out Under conditions of slow change, voltage deviation is mainly related to reactive power:

[0069] ΔV pcc ≈K V ·Q out

[0070] Among them, K V The sensitivity coefficient is expressed by the following formula:

[0071]

[0072] Therefore, X is calculated based on the voltage constraint. min and X max :

[0073]

[0074] Where, ΔV max Indicates the upper limit of voltage deviation, ΔV min Indicates the lower limit of voltage deviation, ΔQ max This indicates the maximum permissible change in reactive power.

[0075] S204. Based on a multi-timescale reinforcement learning algorithm, a virtual impedance control quantity is generated under the constraints of the dynamic security domain. The multi-timescale reinforcement learning algorithm in this embodiment includes an upper-layer prediction model and a lower-layer dynamic correction model. The upper-layer prediction model predicts the grid impedance change in the next k seconds (e.g., k=3) based on an LSTM-PINN hybrid model, outputs the predicted grid impedance value, and obtains the virtual impedance reference value X accordingly. virtual_base The hybrid model can be represented as:

[0076]

[0077] The PINN loss function includes physical constraint terms:

[0078]

[0079] in, This represents the predicted grid impedance at the k-th second. In the LSTM model, the input is the historical impedance sequence Z. grid (tn:t), where n represents the time step, yields the predicted grid impedance value for the k-th second. In the PINN model, the real-time photovoltaic output value P is input. pv Real-time voltage value V at the point of common coupling with the power grid pcc This yields the predicted increment of the grid impedance in the k-th second, ω. LSTM and ω PINN These represent the weight coefficients of the LSTM model and the PINN model, respectively. In this embodiment, ω LSTM =0.7, ω PINN =0.3. In the physical constraint terms, the first term is used to constrain the mean square error between the predicted voltage and the measured voltage, and the second term is used to constrain the residual between the voltage derivative and the theoretical value of the physical model. For example, the voltage differential equation based on Kirchhoff's laws. By jointly training the above hybrid model and minimizing the prediction error and physical residual using the physical constraint terms, the predicted value of the power grid impedance can be obtained, thus reflecting the trend change of the power grid impedance.

[0080] Based on the predicted grid impedance, the virtual impedance reference value X is obtained according to the following formula. virtual_base :

[0081]

[0082] Among them, Z target The target total impedance, Z, can be obtained based on the voltage-reactive power droop characteristic. This is achieved by setting the target total impedance Z. target This ensures that the reactive power-voltage characteristics of the photovoltaic power plant meet the droop control requirements.

[0083] Q = K·(V) ref -V pcc )

[0084]

[0085] Where K represents the droop coefficient. The maximum reactive power is determined as Q based on grid dispatch requirements or equipment capacity. max The allowable voltage deviation range is set to ΔV. max Therefore, the target total impedance can be expressed as:

[0086]

[0087] The lower-level dynamic correction model uses the SAC (Flexible Actor-Critic) model to generate the virtual impedance correction amount ΔX. virtual The state space s and action space a of the SAC (Flexible Actor-Critic) model are defined as follows:

[0088]

[0089] a=ΔX virtual ∈[-0.1,0.1]pu

[0090] Where R and X represent the equivalent resistance and equivalent reactance of the power grid, respectively, and Q is the real and imaginary part of the equivalent impedance of the power grid, which can be the estimated value of the equivalent impedance of the power grid obtained in step S2021. current Indicates the current no reactive power output, ΔX virtual This represents the virtual impedance correction amount, i.e., the output of the lower-level dynamic correction model; This represents the predicted grid impedance for the next k steps, as predicted by the upper-level model. The reward function r is:

[0091]

[0092] Where α represents the voltage deviation penalty coefficient, β represents the reactive power deviation penalty coefficient, η represents the rate of change of action penalty coefficient, λ represents the safety limit violation penalty coefficient, Ω represents the dynamic safety domain, and Violation(Ω) represents the degree of violation of the dynamic safety domain constraints, which can be quantified as, for example, a percentage of the degree of violation. In this embodiment, the typical values ​​of each penalty coefficient are α = 1.0, β = 0.5, η = 0.2, and λ = 10.0. The trained SAC (Flexible Actor-Critic) model generates action a based on the real-time state s, thus obtaining the virtual impedance correction amount.

[0093] The upper-level prediction model and the lower-level dynamic correction model are nested on a time scale. The upper-level prediction model updates the virtual impedance reference value every first cycle, and the lower-level dynamic correction model updates the virtual impedance correction amount every second cycle, with the first cycle being shorter than the second cycle. For example, the upper-level prediction model receives the latest prediction result every second and updates the virtual impedance reference value X. virtual_base The lower-level dynamic correction model generates a virtual impedance correction amount ΔX every 10 seconds based on the real-time status. virtual The outputs of the upper-level prediction model and the lower-level dynamic correction model are fused, that is, the virtual impedance reference value and the virtual impedance correction amount are merged to obtain the virtual impedance control amount:

[0094] X virtual =X virtual_base ·(1+tanh(ΔX virtual )).

[0095] The multi-timescale reinforcement learning algorithm provided in this invention allows the upper-layer prediction model to predict impedance change trends in advance, providing a forward-looking benchmark for lower-layer dynamic correction. For example, when an increasing impedance trend is predicted, the virtual impedance benchmark value can be increased in advance to reserve reactive power capacity; when a decreasing trend is detected, the virtual impedance benchmark value can be decreased to avoid overcompensation. The lower-layer dynamic correction model can achieve instantaneous disturbance suppression (such as cloud cover, switching operations, etc.) and can achieve rapid fine-tuning of virtual impedance through virtual impedance control quantities.

[0096] S206. Obtain the reactive power output error value, and use the reactive power output error value to correct the virtual impedance control quantity. The reactive power output error value e(k) represents the reactive power prediction value Q. pred (k) and the measured reactive power value Q meas The difference of (k):

[0097] e(k) = Q pred (k)-Q meas (k)

[0098] In this embodiment of the invention, the reactive power prediction value is obtained through moving average prediction. To improve prediction accuracy, other more complex algorithms can also be used. Based on the moving average prediction, using the reactive power data from the most recent N sampling points (e.g., N=5), the reactive power prediction value for the next moment is obtained using the aforementioned reactive power data:

[0099]

[0100] Among them, Q pred (k+1) represents the predicted reactive power value at the next time step, Q(ki) represents the reactive power value at the most recent sampling point, and the weighting coefficient w i It can be set to give higher weight to the nearest time.

[0101] Based on the reactive power output error value, an incremental PI feedback method is used to correct the virtual impedance. The virtual impedance adjustment value ΔX(k) is expressed as:

[0102]

[0103] Among them, K p This represents the proportionality coefficient, typically ranging from 0.05 to 0.1. The proportional term is used for rapid response to the current error, immediately adjusting the virtual impedance; K i This represents the integral coefficient, typically ranging from 0.01 to 0.05. The integral term is used to eliminate historical accumulated errors and avoid steady-state deviations. In this embodiment of the invention, the proportional term directly uses the current error e(k), while the integral term uses the historical accumulated error. To eliminate long-term steady-state deviation. If e(k) > 0, it indicates that the measured reactive power is insufficient, and X needs to be reduced. virtual To increase reactive power output; if e(k) < 0, it indicates that the measured reactive power is excessive and X needs to be increased. virtual To reduce reactive power output.

[0104] S208. Reactive power output control of the power plant inverters is performed based on calibrated virtual impedance control quantities. In multi-inverter scenarios, the virtual impedance values ​​of each inverter can be allocated according to the contribution weight of each inverter. The final virtual impedance control quantity is input into the inverter control loop to generate a reactive power reference value, and dynamic tracking output is achieved through a PI controller.

[0105] An embodiment of the present invention also provides a dynamic reactive power control device for new energy power plants, comprising:

[0106] The dynamic security domain calculation module is used to calculate the dynamic security domain of virtual impedance based on the real-time estimated value of grid impedance.

[0107] The virtual impedance control quantity calculation module is used to generate virtual impedance control quantities under the constraints of the dynamic security domain based on a multi-timescale reinforcement learning algorithm; the multi-timescale reinforcement learning algorithm includes an upper-layer prediction model and a lower-layer dynamic correction model.

[0108] The virtual impedance control quantity correction module is used to obtain the reactive power output error value and use the reactive power output error value to correct the virtual impedance control quantity.

[0109] The reactive power output control module is used to control the reactive power output of the inverters in the power plant based on the corrected virtual impedance control quantity.

[0110] The specific steps for each module in the dynamic reactive power control device for new energy power plants in this embodiment of the present invention to realize its function are the same as the steps provided in the dynamic reactive power control method for new energy power plants in the above embodiment of the present invention, and their repeated description will be omitted here.

[0111] In summary, this invention relates to a dynamic reactive power control method and device for new energy power plants. The method includes the following steps: calculating the dynamic security domain of virtual impedance based on real-time estimates of grid impedance; generating a virtual impedance control quantity under the constraints of the dynamic security domain using a multi-timescale reinforcement learning algorithm; the multi-timescale reinforcement learning algorithm includes an upper-level prediction model and a lower-level dynamic correction model; obtaining reactive power output error values ​​and using these errors to correct the virtual impedance control quantity; and controlling the reactive power output of the power plant's inverter based on the corrected virtual impedance control quantity. The technical solution provided by this invention predicts grid impedance using a multi-timescale reinforcement learning algorithm, thereby generating a virtual impedance control quantity. The upper-level prediction model can predict impedance change trends in advance, providing a forward-looking benchmark for lower-level dynamic correction. The lower-level dynamic correction model can suppress instantaneous disturbances and achieve rapid fine-tuning of the virtual impedance, thus obtaining a more accurate virtual impedance control quantity. Based on this, the virtual impedance control quantity is corrected by using the reactive power output error value, forming a hierarchical control architecture of "coarse adjustment + fine adjustment", which realizes precise control of reactive power of new energy power plants.

[0112] It should be understood that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the invention (including the claims) is limited to these examples. Within the framework of this invention, technical features of the above embodiments or different embodiments can also be combined, steps can be implemented in any order, and many other variations exist regarding different aspects of one or more embodiments of the invention as described above, which are not provided in the details for the sake of brevity. The specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A dynamic reactive power control method for new energy power plants, characterized in that, Including the following steps: The dynamic security domain of the virtual impedance is calculated based on the real-time estimated value of the grid impedance. Based on a multi-timescale reinforcement learning algorithm, a virtual impedance control quantity is generated under the constraints of the dynamic security domain. The multi-timescale reinforcement learning algorithm includes an upper-layer prediction model and a lower-layer dynamic correction model. The upper-layer prediction model includes an LSTM-PINN hybrid prediction model, which is used to predict future grid impedance. The hybrid prediction model is expressed as follows: in, This represents the historical impedance sequence, where n represents the time step. This indicates the real-time photovoltaic power output value. and These represent the weight coefficients of the LSTM model and the PINN model, respectively. This represents the predicted grid impedance value at the k-th second; the lower-level dynamic correction model includes the SAC model, and the state space s and action space a of the SAC model are represented as follows: in, This indicates that there is currently no work being done; This represents the virtual impedance correction amount; This represents the predicted value of the grid impedance for the next k steps, as predicted by the upper-level prediction model. The reward function of the SAC model is expressed as: in, This represents the voltage deviation penalty coefficient. This represents the reactive power deviation penalty coefficient. This represents the penalty coefficient for the rate of change of action. This represents the penalty coefficient for exceeding safety limits. Indicates the voltage deviation value. Indicates the reactive power deviation value; Represents a dynamic security domain, Violation( The symbol () indicates the degree of violation of the constraints of the dynamic security domain. Obtain the reactive power output error value, and use the reactive power output error value to correct the virtual impedance control quantity; Reactive power output control of the inverters in the power plant is performed based on the modified virtual impedance control quantity.

2. The method according to claim 1, characterized in that, The dynamic security domain includes impedance constraints; the impedance constraints are expressed as follows: in, Represents virtual impedance. This represents the minimum value of the virtual impedance. This represents the maximum value of the virtual impedance.

3. The method according to claim 2, characterized in that, The minimum value of the virtual impedance is expressed as: The maximum value of the virtual impedance is expressed as: in, Indicates the upper limit of voltage deviation. This indicates the lower limit of voltage deviation. This represents the maximum permissible change in reactive power. X represents the voltage at the point of common coupling, and X represents the equivalent reactance of the power grid.

4. The method according to claim 3, characterized in that, The upper-level prediction model and the lower-level dynamic correction model are nested on a time scale. The upper-level prediction model updates the virtual impedance reference value every first cycle, and the lower-level dynamic correction model updates the virtual impedance correction amount every second cycle. The outputs of the upper-level prediction model and the lower-level dynamic correction model are fused to obtain the virtual impedance control amount. The first period is shorter than the second period.

5. The method according to any one of claims 1-4, characterized in that, The virtual impedance control quantity is corrected using the reactive power output error value, including: Calculate the virtual impedance adjustment value based on the reactive power output error value; The virtual impedance control value is corrected based on the virtual impedance adjustment value.

6. The method according to claim 5, characterized in that, Virtual impedance adjustment value Represented as: in, Represents the proportionality coefficient. Represents the integral coefficient. This indicates the reactive power output error value.

7. A dynamic reactive power control device for new energy power plants, characterized in that, include: The dynamic security domain calculation module is used to calculate the dynamic security domain of virtual impedance based on the real-time estimated value of grid impedance. A virtual impedance control quantity calculation module is used to generate virtual impedance control quantities under the constraints of the dynamic security domain based on a multi-timescale reinforcement learning algorithm. The multi-timescale reinforcement learning algorithm includes an upper-layer prediction model and a lower-layer dynamic correction model. The upper-layer prediction model includes an LSTM-PINN hybrid prediction model, which is used to predict future grid impedance. The hybrid prediction model is expressed as follows: in, This represents the historical impedance sequence, where n represents the time step. This indicates the real-time photovoltaic power output value. and These represent the weight coefficients of the LSTM model and the PINN model, respectively. This represents the predicted grid impedance value at the k-th second; the lower-level dynamic correction model includes the SAC model, and the state space s and action space a of the SAC model are represented as follows: in, This indicates that there is currently no work being done; This represents the virtual impedance correction amount; This represents the predicted value of the grid impedance for the next k steps, as predicted by the upper-level prediction model. The reward function of the SAC model is expressed as: in, This represents the voltage deviation penalty coefficient. This represents the reactive power deviation penalty coefficient. This represents the penalty coefficient for the rate of change of action. This represents the penalty coefficient for exceeding safety limits. Indicates the voltage deviation value. Indicates the reactive power deviation value; Represents a dynamic security domain, Violation( The symbol () indicates the degree of violation of the constraints of the dynamic security domain. The virtual impedance control quantity correction module is used to obtain the reactive power output error value and use the reactive power output error value to correct the virtual impedance control quantity. The reactive power output control module is used to control the reactive power output of the inverters in the power plant based on the corrected virtual impedance control quantity.

Images