A transient frequency support control method considering the low-voltage breakdown characteristics of converters

Abstract

This invention discloses a transient frequency support control method considering the voltage underrun characteristics of a converter, relating to the field of electrical engineering. The method includes: acquiring grid voltage and frequency data through real-time monitoring sensors; detecting the voltage underrun characteristics of the converter based on the acquired grid voltage data to obtain the voltage underrun state; activating an adaptive response strategy to adjust the power command based on the voltage underrun state; initiating a virtual synchronous machine control algorithm to adjust the power output based on the adjusted power command and grid frequency data; and using a reinforcement learning algorithm to coordinate the voltage underrun adaptive response strategy and the virtual synchronous machine control algorithm to restore the grid state. This invention effectively enhances the role of the converter in grid frequency regulation and strengthens the stability and reliability of the grid frequency through the coordinated adjustment of the adaptive response strategy for the voltage underrun state and the virtual synchronous machine control algorithm.

Description

Technical Field

[0001] This invention relates to the field of electrical engineering, and in particular to a transient frequency support control method that takes into account the voltage underrun characteristics of converters. Background Technology

[0002] In recent years, with the rapid development of new energy power generation technologies, especially the integration of renewable energy sources such as wind and solar power, the scheduling and control challenges faced by power systems have been increasing. Converters, as key equipment for grid connection of renewable energy power generation, play a crucial role. In grid frequency regulation, converters not only undertake the task of regulating grid frequency fluctuations but also need to ensure the stability of their voltage and power output. However, when the grid faces sudden load fluctuations and voltage sags, the power output of converters may be affected, especially under voltage underrun conditions. Traditional converter control strategies are insufficient to effectively cope with instantaneous grid frequency fluctuations and power demands, thus affecting grid stability. Therefore, improving the voltage underrun characteristics of converters and their ability to support grid frequency has become an important research topic for enhancing power system stability.

[0003] Current grid frequency regulation methods primarily rely on traditional synchronous generator inertial response and virtual synchronous machine control algorithms. However, these methods largely depend on fixed response models and fail to fully consider dynamic grid changes and the nonlinear response characteristics of converters under voltage underrun conditions. Especially during converter voltage underruns, grid frequency and voltage changes are highly dynamic, making it difficult for traditional control algorithms to make precise adjustments in a short time, resulting in unsatisfactory frequency support. Furthermore, existing technologies typically lack flexible adaptive mechanisms, making it difficult to cope with changes in complex grid environments. To address this issue, this paper proposes a method combining real-time grid monitoring data and adaptive control strategies, introducing advanced algorithms such as virtual synchronous machines and machine learning, which promises to achieve efficient frequency support for converters under voltage underrun conditions. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a transient frequency support control method that takes into account the voltage underrun characteristics of the converter, which solves the problem of insufficient response speed and accuracy of the prior art to frequency changes during voltage underrun.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a transient frequency support control method considering the voltage underrun characteristics of a converter, comprising: acquiring grid voltage data and grid frequency data through real-time monitoring sensors; detecting the voltage underrun characteristics of the converter based on the acquired grid voltage data to obtain the voltage underrun state; activating an adaptive response strategy and adjusting the power command according to the voltage underrun state; initiating a virtual synchronous machine control algorithm based on the grid frequency data and the adjusted power command to adjust the power output; and using a reinforcement learning algorithm to coordinate the voltage underrun adaptive response strategy and the virtual synchronous machine control algorithm to restore the grid state.

[0008] As a preferred embodiment of the transient frequency support control method considering the low-voltage breakdown characteristics of the converter described in this invention, the step of collecting grid voltage data and grid frequency data through real-time monitoring sensors specifically involves:

[0009] The grid voltage data includes the instantaneous voltage values ​​of each node in the grid, the fluctuation range and amplitude of the grid voltage, and the instantaneous power of the grid.

[0010] The power grid frequency data includes the instantaneous frequency of the power grid and the deviation of the power grid frequency from the standard frequency.

[0011] As a preferred embodiment of the transient frequency support control method for considering the low-voltage ride-through characteristics of converters according to the present invention, the specific steps for detecting the low-voltage ride-through characteristics of the converter based on the collected grid voltage data and obtaining the low-voltage ride-through state are as follows:

[0012] Set the voltage under-breakdown threshold and voltage recovery threshold;

[0013] The voltage underrun threshold and voltage recovery threshold are compared with the grid voltage data to determine the voltage underrun status.

[0014] The voltage underrun status is transmitted to the dispatch control unit;

[0015] As a preferred embodiment of the transient frequency support control method considering the voltage under-breakdown characteristics of the converter according to the present invention, the specific steps of activating an adaptive response strategy and adjusting the power command based on the voltage under-breakdown state are as follows:

[0016] Based on the voltage underrun status and grid voltage and frequency data from the incoming dispatch control unit, a voltage underrun dynamic response model is established, expressed as follows:

[0017]

[0018] Among them, P dy (t) represents the power command after dynamic adjustment at time t, and P0 is the initial power setpoint. V represents the fast response term to voltage deviation. ref Here, V(t) is the set voltage reference value, V(t) is the currently acquired voltage value, and α is the voltage adjustment sensitivity coefficient. Here is the integral response term with respect to the rate of frequency change, β is the frequency adjustment gain coefficient, and f g (t′) is the power grid frequency at time t′. For the power grid frequency f g (t′) is the partial derivative with respect to time t′, where t is the current time, t′ is the dynamic time of voltage underrun, t0 is the starting point of the integration time, and d is the total differential sign;

[0019] Optimize parameters α and β according to different stages of the voltage underrun state to enhance the response performance;

[0020] According to the output P of the voltage low-down dynamic response model dy (t), combined with the actual power demand, a multivariable control algorithm is used for coordinated adjustment, and the expression is:

[0021]

[0022] Among them, P out (t) represents the output power command, γ represents the power dynamic allocation weight, and P dy (t) represents the power command after dynamic adjustment at time t, κ is the frequency deviation correction coefficient, and f ref f is the reference frequency. g (t′) is the real-time collected power grid frequency, t is the current time, t′ is the dynamic time of voltage underrun, and t0 is the starting point of the integration time;

[0023] The output power command is sent to the converter controller to control the power output.

[0024] As a preferred embodiment of the transient frequency support control method for considering the voltage under-runaway characteristics of converters according to the present invention, the step of optimizing parameters α and β according to different stages of the voltage under-runaway state to enhance the response effect includes the following specific steps:

[0025] When the voltage underrun occurs in its initial stage, adjust the parameters α and β to achieve rapid recovery;

[0026] As the voltage gradually recovers, adjust parameters α and β to balance voltage recovery and frequency support.

[0027] Once the voltage returns to the normal range, adjust parameters α and β to correct the power output.

[0028] As a preferred embodiment of the transient frequency support control method considering the low-voltage breakdown characteristics of the converter described in this invention, the specific steps of adjusting the power output by initiating a virtual synchronous machine control algorithm based on grid frequency data and the adjusted power command are as follows:

[0029] The inertia and damping characteristics of a simulated synchronous generator are used to obtain the inertia and damping compensation power, expressed as follows:

[0030]

[0031] Wherein, ΔP sync (t) represents the inertial and damping compensation power. Let J be the rate of change of the angular velocity of the virtual synchronizer, D be the equivalent inertia coefficient, and ω(t) be the instantaneous angular velocity of the virtual synchronizer. ref Used as reference angular velocity;

[0032] Based on the output power command and the inertia and damping compensation power, the final output power command is obtained, and its expression is:

[0033] P fin (t)=P out (t)+ΔP sync (t);

[0034] Calculate error data, design a feedback optimization controller, and implement integrated feedback power compensation;

[0035] As a preferred embodiment of the transient frequency support control method considering the low-voltage breakdown characteristics of the converter described in this invention, the specific steps for calculating error data, designing a feedback optimization controller, and integrating feedback power compensation are as follows:

[0036] The expressions for calculating power error and frequency error are as follows:

[0037] ΔP(t)=P fin (t)-P act (t),Δf(t)=f ref -f g (t);

[0038] Where ΔP(t) is the power error, Δf(t) is the frequency error, and P act (t) represents the actual grid power, f g (t) represents the actual power grid frequency;

[0039] Based on the error data, a feedback optimization controller is designed, with the following expression:

[0040]

[0041] Where N is the proportional gain coefficient, M is the integral gain coefficient, and ΔP fb (t) represents feedback power compensation;

[0042] Calculate the feedback compensation P based on the final output power command and the feedback power compensation. last (t), P last (t) is sent to the converter controller to drive power output.

[0043] As a preferred embodiment of the transient frequency support control method considering the voltage under-breakdown characteristics of the converter described in this invention, the method utilizes a reinforcement learning algorithm to coordinate the voltage under-breakdown adaptive response strategy and the virtual synchronous machine control algorithm to restore the grid frequency. The specific steps are as follows:

[0044] Define the power grid state space S t and action space A t ;

[0045] Design a reward function, expressed as follows:

[0046]

[0047] Where Δf(t) is the frequency error, ΔP(t) is the power error, and ΔV(t) is the voltage error. λ1 is the power error weighting coefficient, λ2 is the voltage error weighting coefficient, and λ3 is the frequency change rate weighting coefficient.

[0048] The PPO reinforcement learning algorithm was selected for reinforcement learning training.

[0049] After training, the model is deployed, and the parameters in the voltage underrun adaptive response strategy and virtual synchronous machine control algorithm are dynamically adjusted in real time.

[0050] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the transient frequency support control method considering the low-voltage underrun characteristics of the converter as described in the first aspect of the present invention.

[0051] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the transient frequency support control method considering the low-voltage underrun characteristics of the converter as described in the first aspect of the present invention.

[0052] The beneficial effects of this invention are as follows: This invention utilizes high-precision sensors to collect real-time grid voltage and frequency data, providing dynamic feedback for subsequent control. By detecting the low-voltage underrun characteristics of the converter, abnormal grid states are identified in real time, and an adaptive response strategy is initiated to adjust power commands, rapidly responding to grid changes. Combined with a virtual synchronous machine control algorithm, the characteristics of a synchronous generator are simulated, enhancing the grid's inertia and damping support, further improving frequency stability. Through reinforcement learning algorithms, the response strategy and control algorithm are collaboratively optimized, enabling dynamic adjustment of control parameters and improving the grid's adaptability under complex operating conditions. This effectively enhances the role of the converter in grid frequency regulation, strengthens the stability and reliability of the grid frequency, and possesses good technical feasibility and potential application value. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a flowchart of the transient frequency support control method that takes into account the low-voltage underrun characteristics of the converter in Example 1.

[0055] Figure 2 This is a flowchart of the voltage underrun state parameter optimization in Example 1. Detailed Implementation

[0056] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0057] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0058] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0059] Example 1, referring to Figure 1 and Figure 2This is the first embodiment of the present invention, which provides a transient frequency support control method considering the low-voltage breakdown characteristics of a converter, including:

[0060] S1: Collect grid voltage and grid frequency data through real-time monitoring sensors.

[0061] Specifically, it includes the following steps:

[0062] S1.1: Select a real-time monitoring sensor with high precision, high stability, and fast response capability.

[0063] It should be understood that high precision, high stability, and fast response capabilities can ensure the accuracy of the collected data, avoid control failures caused by errors, reduce the impact of environmental factors on sensor performance, and ensure that the control module can respond promptly to sudden power grid transient events.

[0064] Furthermore, voltage sensors need to cover the rated voltage range of the mains power grid (such as 380V or 220V) and support the measurement of sudden fluctuations. Frequency sensors can monitor 50Hz frequency fluctuations in real time with a resolution of 0.01Hz.

[0065] S1.2: Deploy real-time monitoring sensors at the input and output ends of the converter to collect data.

[0066] Specifically, additional sensors are placed at critical loads and main feeders to enhance monitoring capabilities.

[0067] S1.2.1: The collected data includes grid voltage data and grid frequency data.

[0068] Specifically, grid voltage data includes the instantaneous voltage values ​​at each node in the grid, the fluctuation range and amplitude of grid voltage, and the instantaneous power of the grid.

[0069] The power grid frequency data includes the instantaneous frequency of the power grid and the deviation of the power grid frequency from the standard frequency.

[0070] S1.3: Transmit the collected data to the central processing unit.

[0071] The collected data needs to be calibrated and synchronized in real time to ensure the accuracy of the collected data and the consistency in the time dimension, so as to facilitate subsequent analysis.

[0072] The RS485 communication protocol is selected for data transmission to ensure data integrity and real-time performance.

[0073] S1.4: Preprocess the data in the central processing unit.

[0074] Specifically, data cleaning is required to remove invalid data and outliers. Then, the data is filtered to remove high-frequency noise. Finally, data normalization and standardization are performed to facilitate subsequent algorithm processing and computation.

[0075] Ideally, the above design steps ensure accurate monitoring of the power grid status, enabling real-time acquisition and transmission of grid voltage and frequency data, thereby making the frequency-supported control method more precise and reliable. High-precision, stable, and fast-response sensors guarantee data quality while avoiding the influence of external environmental interference, enhancing the response capability to sudden power grid events. Furthermore, real-time data preprocessing improves data quality, further optimizing the execution of subsequent control algorithms and ensuring stable operation of the power grid amidst dynamic changes.

[0076] S2: Based on the collected grid voltage data, perform converter voltage underrun characteristics detection to obtain the voltage underrun status.

[0077] Specifically, it includes the following steps:

[0078] S2.1: Set the voltage underrun threshold and voltage recovery threshold.

[0079] It should be noted that, according to the power grid operation standards, the voltage underrun threshold is set at 80% of the rated voltage, and the voltage recovery threshold is set at 90% of the rated voltage.

[0080] S2.2: Compare the voltage underrun threshold and voltage recovery threshold with the grid voltage data to determine the voltage underrun state.

[0081] Specifically, the method for determining the voltage under-running state is as follows: when the grid voltage data is less than the voltage under-running threshold and remains so for more than 100ms, it is determined to be a voltage under-running state (marked as "under-running mode"). When the grid voltage data is greater than or equal to the voltage recovery threshold and remains so for more than 100ms, it is determined to be a recovery state (marked as "normal mode").

[0082] S2.3: Transmit the voltage underrun status to the dispatch control unit.

[0083] Ideally, through standardized threshold settings and precise state assessment, voltage underrun and recovery states of the power grid can be detected efficiently and accurately, and promptly transmitted to the dispatch control unit. This not only facilitates rapid identification and emergency response to power grid faults but also reduces the negative impacts of power grid fluctuations and misjudgments, improving the stability and reliability of power grid operation. Accurate determination of voltage underrun states, confirmation of recovery states, and timely transmission optimize power grid control strategies and improve overall adaptability and response speed.

[0084] S3: Based on the voltage underrun status, activate the adaptive response strategy and adjust the power command.

[0085] Specifically, it includes the following steps:

[0086] S3.1: Based on the voltage underrun status and grid voltage and frequency data from the incoming dispatch control unit, a voltage underrun dynamic response model is established, expressed as follows:

[0087]

[0088] Among them, P dy (t) represents the power command after dynamic adjustment at time t, and P0 is the initial power setpoint. V represents the fast response term to voltage deviation. ref Here, V(t) is the set voltage reference value, V(t) is the currently acquired voltage value, and α is the voltage adjustment sensitivity coefficient. Here is the integral response term with respect to the rate of frequency change, β is the frequency adjustment gain coefficient, and f g (t′) is the power grid frequency at time t′. For the power grid frequency f g (t′) is the partial derivative with respect to time t′, where t is the current time, t′ is the dynamic time of voltage underrun, t0 is the starting point of the integration time, and d is the total differential symbol.

[0089] It should be noted that, It is used to quickly respond to voltage changes; when the voltage is lower than the reference value, the model can quickly adjust the power command. By integrating historical data on frequency changes, power output is gradually adjusted to stabilize the grid frequency.

[0090] S3.2: Optimize parameters α and β according to different stages of the voltage underrun state to enhance the response effect.

[0091] Specifically, it includes the following steps:

[0092] S3.2.1: When the voltage underrun occurs in the initial stage, adjust the parameters α and β to achieve rapid recovery;

[0093] Specifically, when the grid voltage is just below the threshold (entering the low-voltage state), the parameter α needs to be increased to increase the voltage deviation response sensitivity, and the frequency response term β also needs to be increased to ensure the rapid recovery of the grid frequency.

[0094] S3.2.2: As the voltage gradually recovers, adjust parameters α and β to balance voltage recovery and frequency support;

[0095] Specifically, as the voltage gradually returns to the normal range, the sensitivity coefficient α and the frequency gain coefficient β are gradually reduced to avoid over-adjustment and excessive power fluctuations.

[0096] S3.2.3: When the voltage returns to the normal range, adjust parameters α and β to correct the power output;

[0097] Specifically, when the voltage returns to the normal range, reduce the frequency sensitivity coefficient α and the frequency gain coefficient β to avoid over-response.

[0098] S3.3: P output based on the voltage under-breakdown dynamic response model dy (t), combined with the actual power demand, a multivariable control algorithm is used for coordinated adjustment, and the expression is:

[0099]

[0100] Among them, P out (t) represents the output power command, γ represents the power dynamic allocation weight, and P dy (t) represents the power command after dynamic adjustment at time t, κ is the frequency deviation correction coefficient, and f ref f is the reference frequency. g (t′) represents the real-time power grid frequency, t represents the current time, t′ represents the dynamic time of voltage underrun, and t0 represents the starting point of the integration time.

[0101] It should be noted that during power command regulation, the frequency and reference frequency f are used as the basis for the regulation. ref The deviation is corrected by a frequency deviation correction factor κ, which is used to weight and adjust the power command. This helps to balance the impact of voltage under-drain conditions on power output. The value of γ is dynamically adjusted according to the severity and duration of the voltage under-drain, flexibly balancing the weight of voltage and frequency response.

[0102] S3.4: Send the output power command to the converter controller to control the power output.

[0103] Specifically, the calculated output power command P out (t) includes dynamic corrections for voltage and frequency response. Power commands are sent to the converter controller via the RS485 protocol control interface. The converter adjusts its output according to the received power commands to ensure that the power matches the grid frequency and stability requirements. The converter control module monitors output changes in real time to ensure that it remains within the normal range.

[0104] Ideally, through a sophisticated dynamic response model, adaptive parameter adjustment, multivariable control algorithms, and real-time power command transmission, the model can quickly and stably adjust power output when voltage underruns and frequency fluctuations occur in the power grid. This ensures grid stability while reducing the risks of over-response and power fluctuations, thus improving the overall regulation and emergency response capabilities of the power grid.

[0105] S4: Based on the grid frequency data and the adjusted power command, the virtual synchronous machine control algorithm is activated to adjust the power output.

[0106] Specifically, it includes the following steps:

[0107] S4.1: Simulate the inertia and damping characteristics of a synchronous generator to obtain the inertia and damping compensation power, expressed as:

[0108]

[0109] Wherein, ΔP sync (t) represents the inertial and damping compensation power. Let J be the rate of change of the angular velocity of the virtual synchronizer, D be the equivalent inertia coefficient, and ω(t) be the instantaneous angular velocity of the virtual synchronizer. ref This is the reference angular velocity.

[0110] Specifically, when grid frequency fluctuates, the virtual synchro machine simulates an inertial response, using inertial compensation power to offset instantaneous frequency fluctuations and prevent drastic grid frequency fluctuations. The damping component balances frequency overshoot and oscillations, quickly restoring the frequency to normal levels. This compensation function allows the virtual synchro machine to behave like a traditional generator in the short term, providing rapid power adjustment and mitigating grid fluctuations.

[0111] S4.2: Based on the output power command and the inertia and damping compensation power, obtain the final output power command, expressed as:

[0112] P fin (t)=P out (t)+ΔP sync (t);

[0113] Among them, P fin (t) represents the final output power command, which is a power command that combines inertia and damping compensation. P out (t) represents the output power command, indicating the power output after passing through the low-voltage breakdown model. ΔP sync (t) represents the inertia and damping compensation power, which can be quickly adjusted when the grid frequency changes to ensure grid stability.

[0114] S4.3: Calculate error data, design feedback optimization controller, and integrate feedback power compensation.

[0115] It should be understood that by calculating power error and frequency error, the model can optimize power compensation to ensure that the final power output matches the actual needs of the power grid.

[0116] Specifically, it includes the following steps:

[0117] S4.3.1: Calculate the power error and frequency error, using the following expressions:

[0118] ΔP(t)=P fin (t)-P act (t),Δf(t)=f ref -f g (t);

[0119] Where ΔP(t) is the power error, Δf(t) is the frequency error, and P act (t) represents the actual grid power, f g (t) represents the actual power grid frequency.

[0120] Furthermore, calculating the power error ΔP(t) helps determine whether the module provides sufficient power to meet grid demand. Calculating the frequency error Δf(t) detects whether the grid frequency deviates from the reference frequency and further derives the required frequency compensation. Based on the power and frequency errors, the control strategy of the virtual synchronizer can be further adjusted to ensure consistency between the output power and grid demand.

[0121] S4.3.2: Based on error data, design a feedback optimization controller, the expression of which is:

[0122]

[0123] Where N is the proportional gain coefficient, M is the integral gain coefficient, and ΔP fb (t) represents feedback power compensation.

[0124] Furthermore, the proportional term N·ΔP(t) directly responds to the current power error, causing the power output to adjust rapidly to reduce the error. The integral term... This feedback controller accumulates and compensates for long-term errors, eliminating deviations and improving accuracy. It also enhances stability and reduces grid frequency and power fluctuations.

[0125] S4.3.3: Calculate the feedback compensation P based on the final output power command and the feedback power compensation. last (t), P last (t) is sent to the converter controller to drive power output.

[0126] Specifically, calculate the feedback compensation P. last The expression for (t) is,

[0127] P last (t)=P fin (t)+ΔP fb (t);

[0128] By calculating the feedback compensation P last(t), the module can further optimize the power output command according to the actual state of the power grid.

[0129] Ideally, by simulating inertia and damping response, drastic changes in grid frequency fluctuations are avoided, enhancing the grid's immunity to disturbances. Real-time power error feedback control ensures that power output always meets grid demand, avoiding over- or under-regulation and improving the accuracy of grid frequency and power. The combined effect of multi-dimensional regulation mechanisms ensures the grid can quickly and smoothly recover to a stable state under dynamic environments, enhancing its adaptive capability. The introduction of a virtual synchronous machine control algorithm effectively enhances grid stability and power regulation accuracy, providing a precise and effective grid power regulation scheme.

[0130] S5: Using reinforcement learning algorithms, the voltage underrun adaptive response strategy and the virtual synchronous machine control algorithm are coordinated to restore the power grid state.

[0131] Specifically, it includes the following steps:

[0132] S5.1: Define the power grid state space S t and action space A t .

[0133] Specifically, the state space S t Various state variables representing the power grid describe its current operating state, including the following elements: frequency error Δf(t), power error ΔP(t), voltage error ΔV(t), and frequency change rate. The current power output of the power grid is P(t), and the power grid voltage is V(t).

[0134] State space S t This forms a multi-dimensional vector that integrates real-time power grid operation data, reflecting the current health status of the power grid.

[0135] Action Space A t This represents the different actions the agent can choose in the current state. Each action in the action space will determine the direction and magnitude of adjustments to various parameters in the power grid control strategy.

[0136] The main parameters include: sensitivity coefficient α for adjusting power commands, frequency response gain coefficient β, power dynamic allocation weight γ, frequency deviation correction coefficient k, inertia coefficient J of the virtual synchronizer, and damping coefficient D of the virtual synchronizer.

[0137] Action Space A t Represented as:

[0138] A t =(α,β,γ,κ,J,D);

[0139] The action space determines how to adjust the voltage underrun adaptive response strategy and the virtual synchronous machine control strategy at each moment through discrete or continuous adjustment of these parameters in order to achieve stable operation of the power grid.

[0140] S5.2: Design the reward function, the expression of which is:

[0141]

[0142] Where Δf(t) is the frequency error, ΔP(t) is the power error, and ΔV(t) is the voltage error. λ1 is the frequency change rate, λ2 is the power error weighting coefficient, λ3 is the frequency change rate weighting coefficient.

[0143] Furthermore, the frequency error penalty term -|Δf(t)|: Through the frequency error penalty, the reward function encourages the agent to minimize the grid frequency deviation. Excessive frequency error will reduce the reward, prompting the converter to adjust the frequency.

[0144] Power error term -λ1·|ΔP(t)|: This penalty term for power error encourages the converter to adjust its power output to be as close as possible to the actual demand of the power grid. The weighting coefficient λ1 controls the degree of influence of the power error.

[0145] Voltage error penalty term -λ2·|ΔV(t)|: Voltage error affects stability. The reward function, by penalizing the voltage error, prompts the converter to optimize voltage regulation and ensure that the voltage is as stable as possible within the ideal range.

[0146] Frequency change rate penalty term The rate of frequency change reflects how quickly the grid frequency changes; excessively rapid frequency changes can lead to instability. By penalizing the rate of frequency change, converters are encouraged to smooth out grid frequency changes and avoid rapid frequency fluctuations.

[0147] λ1, λ2, and λ3 are weighting coefficients that adjust the influence of different error terms. Setting these coefficients appropriately can balance the importance of various control objectives. For example, in cases of frequent voltage underruns or frequency fluctuations, the weights of voltage and frequency errors can be increased, focusing on the stability of voltage and frequency.

[0148] S5.3: Select the PPO reinforcement learning algorithm and perform reinforcement learning training.

[0149] Specifically, the reinforcement learning training process is as follows:

[0150] Initialization: Initialize the neural network model and set a random initial strategy.

[0151] Simulation training: In a simulated environment, actions are selected based on the current strategy and adjusted according to feedback from the power grid (reward function).

[0152] Policy Update: The PPO algorithm is used to optimize the current policy, update the policy to obtain higher rewards, and gradually learn the optimal method for adjusting control parameters.

[0153] Continuous iteration: Through continuous experimentation and learning, the model gradually improves its ability to adjust strategies under different power grid conditions, and ultimately optimizes the stability of the power grid under various complex conditions.

[0154] S5.4: After training is completed, the model is deployed, and the parameters in the voltage underrun adaptive response strategy and virtual synchronous machine control algorithm are dynamically adjusted in real time.

[0155] Specifically, the reinforcement learning model adjusts parameters α, β, γ, and k to improve the power output response during voltage underrun. By adjusting the inertia coefficient J and damping coefficient D, the inertia and damping characteristics of a synchronous generator are simulated, enabling rapid response to frequency fluctuations and achieving coordinated adjustment of the voltage underrun adaptive response strategy and the virtual synchronous machine control algorithm.

[0156] Ideally, by using reinforcement learning algorithms to coordinate the voltage undervoltage adaptive response strategy and the virtual synchronous machine control algorithm, real-time optimization and dynamic control of the power grid state are achieved. The control strategy is automatically adjusted based on the real-time state of the power grid, ensuring stable operation of the grid in complex dynamic environments. By reducing power, frequency, and voltage errors, and controlling the rate of frequency change, grid stability is enhanced, and the risk of grid instability is reduced. Through the reinforcement learning model, the model can continuously self-optimize and adapt, improving the stability of the power grid in long-term operation. Automated regulation reduces human intervention and improves control accuracy and response speed. It enables the grid to cope with sudden situations and complex environments, ensuring stable operation under changing conditions. This provides an efficient and intelligent power grid control strategy.

[0157] This embodiment also provides a computer device applicable to the transient frequency support control method considering the low-voltage underrun characteristics of converters, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the transient frequency support control method considering the low-voltage underrun characteristics of converters as proposed in the above embodiment.

[0158] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0159] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the transient frequency support control method considering the low-voltage breakdown characteristics of the converter as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0160] In summary, this invention utilizes high-precision sensors to collect real-time grid voltage and frequency data, providing dynamic feedback for subsequent control. By detecting the converter's low-voltage underrun characteristics, it identifies abnormal grid conditions in real time and initiates an adaptive response strategy to adjust power commands, rapidly responding to grid changes. Combined with a virtual synchronous machine control algorithm, it simulates synchronous generator characteristics, enhancing the grid's inertia and damping support, further improving frequency stability. Through reinforcement learning algorithms, it collaboratively optimizes the response strategy and control algorithm, enabling the algorithm to dynamically adjust control parameters and improve the grid's adaptability under complex operating conditions. This effectively enhances the role of the converter in grid frequency regulation, strengthens grid frequency stability and reliability, and demonstrates good technical feasibility and potential application value.

[0161] Example 2, referring to Table 1, is the second embodiment of the present invention. To further verify the technical solution of the present invention, experimental simulation data of the transient frequency support control method considering the voltage underrun characteristics of the converter are given.

[0162] This experiment designs a dynamic power adjustment strategy based on voltage under-running and frequency fluctuations. Through refined modeling and adaptive control, it can adjust the power output of the power grid in real time. First, a dynamic response model is established based on grid voltage and frequency data to quickly adjust the power command to cope with the impact of voltage under-running. Simultaneously, a multivariable control algorithm is employed to dynamically optimize power allocation based on voltage and frequency changes, ensuring grid frequency and voltage stability. By adjusting control parameters, the algorithm avoids over-response and power fluctuations, improving the grid's regulation and emergency response capabilities. Finally, the power command is precisely transmitted through the converter controller, ensuring that the grid can quickly and smoothly adjust its power output when voltage under-running and frequency fluctuations occur.

[0163] The sensor data collected is shown in Table 1 below:

[0164] Table 1 Sensor Data Acquisition Table

[0165]

[0166] In addition, set the voltage reference value V ref =380V, initial power P0=1000W.

[0167] The voltage underrun threshold is set to 80% of the voltage reference value, i.e., 304V.

[0168] The voltage recovery threshold is 90% of the voltage reference value, i.e., 342V.

[0169] The adjustment rules for the voltage sensitivity coefficient α and the frequency gain coefficient β are set as follows:

[0170] Low voltage ride-through: When V(t) < 304V, α = 0.10 and β = 0.05.

[0171] Voltage recovery: When V(t) > 342V, α = 0.05 and β = 0.02.

[0172] Voltage within the recovery range 304V < V(t) < 342V: For every 10V increase, α and β each decrease by 0.01.

[0173] The adjusted data is specifically shown in Table 2 below:

[0174] Table 2 Coefficient Adjustment Table

[0175]

[0176] According to the low voltage ride-through dynamic response model, calculate the adjusted power command P dy (t), and the expression is:

[0177]

[0178] The specific process is as follows:

[0179] Calculate the voltage deviation response term

[0180] Calculate the frequency change rate response term

[0181] Add the two terms together to obtain P dy (t).

[0182] Taking time t = 0 and t = 1 as examples, calculate P dy (t) at each time point.

[0183] Time t = 0 s:

[0184] Grid voltage V(t) = 379V; Grid frequency f g (t) = 50Hz; Frequency change rate

[0185] Voltage deviation response term:

[0186]

[0187] Frequency change rate response term:

[0188] (Since t0 = 0 and the frequency change rate is 0);

[0189] Dynamically adjusted power command:

[0190] P dy(0)=1000·0.9512+0=951.2W;

[0191] Time t = 1 second:

[0192] The grid voltage V(t) = 305V; the grid frequency f g (t) = 49.92 Hz; Rate of frequency change

[0193] Voltage deviation response term:

[0194]

[0195] Frequency change rate response term:

[0196]

[0197] Dynamically adjusted power command:

[0198] P dy (1)=1000·0.00187+(-0.00432)=1.87-0.00432≈1.8657W;

[0199] For other times, calculate according to the example above.

[0200] The calculated dynamically adjusted power command is shown in Table 3:

[0201] Table 3 Dynamic Power Adjustment Command Table

[0202] Time (seconds) Grid voltage (V) Dynamically adjusted power command (W) 0 379 951.2 1 305 1.8657 2 300 0.3315 3 338 73.29928 4 348 201.9024 5 357 316.6018

[0203] Then, a multivariable control algorithm is activated for coordinated adjustment to calculate the output power command P. out (t), the expression is:

[0204]

[0205] Set reference frequency f ref =50Hz; power dynamic allocation weight γ=0.5; frequency deviation correction coefficient k=100.

[0206] Taking times t=0 and t=1 as examples, calculate P at each time point. out (t).

[0207] Time t = 0 seconds:

[0208] P dy (0) = 951.2;

[0209] There is no change from t0 = 0 to t = 0, so the integral is 0;

[0210] P out (0)=0.5·951.2+(1-0.5)·0=0.5Pout951.2+0=475.6W;

[0211] Time t = 1 second:

[0212] P dy (1) = 1.8657;

[0213] From t0 = 0 to t = 1,

[0214]

[0215] P out (1) = 0.5·1.8657 + (1-0.5)·8 = 0.5·1.8657 + 0.5·8 = 0.93285 +

[0216] 4 = 4.93285W;

[0217] For other times, calculate according to the example above.

[0218] The calculated output power command is shown in Table 4:

[0219] Table 4 Output Power Command Table

[0220]

[0221] The calculated P out (t) is sent to the converter controller, based on the acquired frequency data and the calculated P. out (t) The virtual synchronous machine control algorithm is started to continue adjusting the power output. Finally, the reinforcement learning algorithm is used to coordinate the adjustment of the parameters in the voltage underrun adaptive response strategy and the virtual synchronous machine control algorithm to restore the grid state.

[0222] Through the above steps and data tables, the experiment can be used to control power output. By employing a sophisticated dynamic response model, adaptive parameter adjustment, multivariable control algorithms, and real-time power command transmission, it ensures that the converter can quickly and stably adjust power output when voltage underruns and frequency fluctuations occur in the power grid. While ensuring grid stability, it reduces the risks of over-response and power fluctuations, thereby improving the overall regulation and emergency response capabilities of the power grid.

[0223] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A transient frequency support control method considering the low-voltage breakdown characteristics of a converter, characterized in that, include: Real-time monitoring sensors are used to collect power grid voltage and frequency data. Based on the collected grid voltage data, the converter voltage underrun characteristics are detected to obtain the voltage underrun status. Based on the low-voltage breakdown state, an adaptive response strategy is activated to adjust the power command. The specific steps are as follows: Based on the voltage underrun status and grid voltage and frequency data from the incoming dispatch control unit, a voltage underrun dynamic response model is established, expressed as follows: Among them, P dy (t) represents the power command after dynamic adjustment at time t, and P0 is the initial power setpoint. V represents the fast response term to voltage deviation. ref Here, V(t) is the set voltage reference value, V(t) is the currently acquired voltage value, and α is the voltage adjustment sensitivity coefficient. Here is the integral response term with respect to the rate of frequency change, β is the frequency adjustment gain coefficient, and f g (t′) is the power grid frequency at time t′. For the power grid frequency f g (t′) is the partial derivative with respect to time t′, where t is the current time, t′ is the dynamic time of voltage underrun, t0 is the starting point of the integration time, and d is the total differential sign; Optimize parameters α and β according to different stages of the voltage underrun state to enhance the response performance; According to the output P of the voltage low-down dynamic response model dy (t), combined with the actual power demand, a multivariable control algorithm is used for coordinated adjustment, and the expression is: Among them, P out (t) represents the output power command, γ represents the power dynamic allocation weight, and P dy (t) represents the power command after dynamic adjustment at time t, κ is the frequency deviation correction coefficient, and f ref f is the reference frequency. g (t′) represents the real-time collected power grid frequency, and t represents the current time. ′ The dynamic time of voltage low-voltage breakdown is given by t0, which is the starting point of the integration time. The output power command is sent to the converter controller to control the power output. Based on the grid frequency data and the adjusted power command, the virtual synchronous machine control algorithm is activated to adjust the power output. The specific steps are as follows: The inertia and damping characteristics of a simulated synchronous generator are used to obtain the inertia and damping compensation power, expressed as follows: Wherein, ΔP sync (t) represents the inertial and damping compensation power. Let J be the rate of change of the angular velocity of the virtual synchronizer, D be the equivalent inertia coefficient, and ω(t) be the instantaneous angular velocity of the virtual synchronizer. ref Used as reference angular velocity; Based on the output power command and the inertia and damping compensation power, the final output power command is obtained, and its expression is: P fin (t)=P out (t)+ΔP sync (t); Calculate error data, design a feedback optimization controller, and implement integrated feedback power compensation; By using reinforcement learning algorithms, the voltage underrun adaptive response strategy and the virtual synchronous machine control algorithm are coordinated to restore the power grid state.

2. The transient frequency support control method considering the low-voltage breakdown characteristics of the converter as described in claim 1, characterized in that: The process involves collecting grid voltage and frequency data through real-time monitoring sensors. The grid voltage data includes the instantaneous voltage values ​​of each node in the grid, the fluctuation range and amplitude of the grid voltage, and the instantaneous power of the grid. The power grid frequency data includes the instantaneous frequency of the power grid and the deviation of the power grid frequency from the standard frequency.

3. The transient frequency support control method considering the low-voltage breakdown characteristics of the converter as described in claim 1, characterized in that: The process of detecting the low-voltage breakdown characteristics of the converter based on the collected grid voltage data to obtain the low-voltage breakdown state involves the following steps: Set the voltage under-breakdown threshold and voltage recovery threshold; The voltage underrun threshold and voltage recovery threshold are compared with the grid voltage data to determine the voltage underrun status. The voltage underrun status is transmitted to the dispatch control unit.

4. The transient frequency support control method considering the low-voltage breakdown characteristics of the converter as described in claim 1, characterized in that: The steps for optimizing parameters α and β based on different stages of the voltage under-runout state to enhance the response are as follows: When the voltage underrun occurs in its initial stage, adjust the parameters α and β to achieve rapid recovery; As the voltage gradually recovers, adjust parameters α and β to balance voltage recovery and frequency support. Once the voltage returns to the normal range, adjust parameters α and β to correct the power output.

5. The transient frequency support control method considering the low-voltage breakdown characteristics of the converter as described in claim 1, characterized in that: Based on the calculated error data, a feedback optimization controller is designed to comprehensively compensate for feedback power. The specific steps are as follows: The expressions for calculating power error and frequency error are as follows: ΔP(t)=P fin (t)-P act (t),Δf(t)=f ref -f g (t); Where ΔP(t) is the power error, Δf(t) is the frequency error, and P act (t) represents the actual grid power, f g (t) represents the actual power grid frequency; Based on the error data, a feedback optimization controller is designed, with the following expression: Where N is the proportional gain coefficient, M is the integral gain coefficient, and ΔP fb (t) represents feedback power compensation; Calculate the feedback compensation P based on the final output power command and the feedback power compensation. last (t), P last (t) is sent to the converter controller to drive power output.

6. The transient frequency support control method considering the low-voltage breakdown characteristics of the converter as described in claim 1, characterized in that: Using reinforcement learning algorithms, the voltage under-breakdown adaptive response strategy and the virtual synchronous machine control algorithm are coordinated to restore the grid frequency. The specific steps are as follows: Define the power grid state space S t and action space A t ; Design a reward function, expressed as follows: Where Δf(t) is the frequency error, ΔP(t) is the power error, and ΔV(t) is the voltage error. λ1 is the power error weighting coefficient, λ2 is the voltage error weighting coefficient, and λ3 is the frequency change rate weighting coefficient. The PPO reinforcement learning algorithm was selected for reinforcement learning training. After training, the model is deployed, and the parameters in the voltage underrun adaptive response strategy and virtual synchronous machine control algorithm are dynamically adjusted in real time.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the transient frequency support control method considering the low-voltage underrun characteristics of the converter as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the transient frequency support control method considering the low-voltage underrun characteristics of the converter as described in any one of claims 1 to 6.

Images