Low frequency load shedding method of joint voltage control in new energy high proportion power system

Abstract

This invention belongs to the field of power system safety and stability control technology, and relates to a low-frequency load shedding method with joint voltage control in power systems with a high proportion of renewable energy. The method includes: 1. System modeling and coupling characteristic analysis, quantifying the mapping relationship between active power shedding and voltage rise at renewable energy grid connection points during low-frequency load shedding; 2. Frequency-voltage coordinated load shedding strategy and optimization decision-making, achieving coordinated active power frequency control and reactive power voltage control through optimized redistribution of load shedding points and reactive power pre-control of SVG (Static Var Generator) at renewable energy power plants; 3. Multi-round dynamic correction and fault self-healing verification, adaptively adjusting the action delay and load shedding amount in subsequent rounds, and verifying the effectiveness of the strategy. This invention can accurately quantify the voltage fluctuation risk corresponding to different load shedding actions, avoiding transient overvoltage phenomena in the power grid caused by active power load shedding and reactive power voltage imbalance from the control source, and improving voltage safety and stability under fault conditions in power grids with high renewable energy penetration.

Description

Technical Field

[0001] This invention belongs to the field of power system safety and stability control technology, specifically involving a low-frequency load shedding method for combined voltage control in power systems with a high proportion of renewable energy. By coordinating and optimizing the active power cut-off decision and the dynamic response of reactive power voltage in the low-frequency load shedding process, it prevents the overvoltage disconnection of renewable energy units and further frequency deterioration caused by reactive power excess due to load shedding. Background Technology

[0002] Low-frequency load shedding is a core fallback measure for emergency frequency control in power systems. As the last line of defense for ensuring grid frequency stability, it plays a crucial role in scenarios where grid faults such as generator tripping or DC blocking cause large-scale active power deficits. Traditional low-frequency load shedding strategies take system frequency stability as the sole control objective. The core setting logic is to perform load shedding operations according to a preset fixed cycle and fixed load amount based on the magnitude of the grid frequency drop. This strategy is designed based on the grid architecture dominated by traditional synchronous machines, focusing only on the matching relationship between active power deficit and frequency response, without fully considering the grid's physical characteristics that active load shedding will simultaneously cause changes in the system's reactive power operating state during the load shedding operation.

[0003] With the advancement of new power system construction and the large-scale grid connection of new energy units such as wind power and photovoltaics, the proportion of new energy in the power system continues to increase. This has fundamentally changed the power grid's power supply structure, inertia level, and reactive power support characteristics, highlighting the inadequacy of traditional low-frequency load shedding strategies. Currently, wind farms and photovoltaic power plants are generally equipped with a large number of static var generators (SVMs) and parallel capacitor banks to meet grid connection voltage requirements, providing routine support for grid connection voltage stability. When a grid fault triggers traditional low-frequency load shedding, cutting off some active load, the overall active power consumption of the system decreases significantly. However, the reactive power compensation equipment at new energy power plants remains in operation and does not shut down synchronously with the active load.

[0004] This operational characteristic can lead to an imbalance between active and reactive power supply and demand in the power grid, resulting in a significant accumulation of redundant reactive power. This, in turn, can cause a sharp rise in transient voltage at the grid connection points of renewable energy units, easily exceeding the safe operating voltage threshold for renewable energy units. This can trigger overvoltage protection actions, causing large-scale disconnection of renewable energy units from the grid. The mass disconnection of renewable energy units will, in turn, exacerbate the active power deficit in the system, disrupt the power balance of the grid after low-frequency load shedding, trigger a secondary frequency drop in the system, and even trigger subsequent rounds of low-frequency load shedding. Ultimately, this creates a vicious chain reaction of "load shedding—transient overvoltage—renewable energy disconnection—frequency drop again—repeated load shedding," significantly expanding the scope of grid faults and seriously threatening the transient safety and stability of power systems with a high proportion of renewable energy.

[0005] Current research on frequency stability control of new energy power systems, both domestically and internationally, mostly focuses on optimizing system frequency response models, improving frequency regulation control strategies for new energy units, and enhancing the frequency regulation capability of new energy grid connection. It only focuses on solving the single problem of grid frequency drop, without taking into account the dynamic changes in reactive power of new energy power plants and the dynamic response characteristics of grid voltage during the entire process of low-frequency load shedding. It has failed to form an effective coordinated control scheme to address the cascading problems of "overvoltage induced by load shedding and unit disconnection worsening the fault".

[0006] The existing technology system lacks a low-frequency load shedding optimization method that can simultaneously consider reactive voltage constraints and actively block the chain reaction of frequency and voltage instability during low-frequency load shedding operations, making it difficult to adapt to the safety and stability control requirements of power systems with a high proportion of new energy. Summary of the Invention

[0007] This invention provides the following technical solution: a low-frequency load shedding method for combined voltage control in a power system with a high proportion of new energy sources, comprising the following steps: Step 1: System modeling and coupling characteristic analysis, including establishing a system frequency response model, a transient voltage characteristic model of new energy power plants, a load shedding-voltage rise sensitivity model, and a new energy unit overvoltage disconnection threshold model, and quantifying the mapping relationship between active power cut-off and voltage rise at the new energy grid connection point during low-frequency load shedding.

[0008] Step 2, Frequency-Voltage Coordinated Load Shedding Strategy and Optimization Decision: Based on the model in Step 1, frequency recovery requirements and voltage safety constraints are evaluated simultaneously before each round of low-frequency load shedding. Coordination of active frequency control and reactive voltage control is achieved through load shedding point optimization redistribution and reactive power pre-control of static var generators (SVG) in new energy power plants.

[0009] Step 3: Multi-round dynamic correction and fault self-healing verification. The action delay and load shedding amount of subsequent rounds are adaptively adjusted according to the real-time frequency change rate and voltage change rate, and the effectiveness of the strategy is verified through typical fault scenarios.

[0010] The specific technical solution is as follows: Step 1. System Modeling and Coupling Characteristic Analysis. This step establishes a frequency response model for a power system with a high proportion of renewable energy and a transient voltage characteristic model for renewable energy power plants. It quantifies the mapping relationship between active power cut-off during low-frequency load shedding and the voltage rise at the grid connection point, providing a theoretical basis for subsequent collaborative decision-making. The specific implementation is as follows: Step 1-1: Data Acquisition and Status Awareness. System frequency for accessing the main network SCADA / PMU system. Rate of change of frequency Voltage amplitude of each bus and phase angle Simultaneously acquire the voltage at the grid connection point of the new energy power station (j=1,2,...,M, where M is the total number of renewable energy power stations, and the superscript 0 indicates the steady-state value before the fault, in pu), real-time reactive power output of the power station's reactive power compensation device (Static Var Generator / Capacitor Bank). and available reactive power reserves and the active power of each load shedding node. With reactive power (MVA, i=1,2,...,L, where L is the total number of load-switching nodes). Multi-source heterogeneous data is time-scaled and normalized using a standardized interface (the baseline value is the system's rated capacity). With rated voltage This ensures the synchronization and consistency of subsequent modeling.

[0011] Step 1-2: System Frequency Response Aggregate Modeling. Establish an equivalent frequency response model for a power system with a high proportion of renewable energy, equating the system to an aggregated frequency response transfer function including renewable energy units and traditional synchronous machines:

[0012] in, System frequency deviation , For the system active power deficit , The equivalent inertial time constant of the system , Load damping coefficient , Equivalent frequency regulation coefficient for new energy generating units , The frequency response transfer function of the new energy unit. (Characterizing the dynamic response characteristics of the active power output of new energy units to frequency deviation,) (This is the Laplace operator). This model characterizes the dynamic evolution of the system frequency under power deficit excitation.

[0013] Steps 1-3: Modeling the transient voltage characteristics of renewable energy power plants. Establish a coupling model between the grid connection point voltage and the reactive power of the renewable energy power plant. For the first... A new energy power station, grid connection point voltage satisfy:

[0014] in, For the first Equivalent voltage (pu) on the power grid side of each station. For the first Total reactive power output (PU, including the sum of reactive power output of new energy units and reactive power output of SVG) of each power station. For the first The equivalent reactive load (pu, including the equivalent reactive power loss of the step-up transformer and collector lines within the station). For the first Short-circuit capacity (MVA) at each station's grid connection point. This is a standardized baseline value (1.0 pu for simplified dimensionless calculations). When low-frequency load shedding occurs, the reactive power demand on the system side decreases, but the reactive power compensation equipment at the renewable energy power plant remains operational, leading to... Relative increase The corresponding increase.

[0015] Steps 1-4: Load shedding-voltage rise sensitivity modeling. Define the first... Active power cut-off amount per load node (pu, with) (based on) and the first Voltage rise at grid connection point of each new energy power station Influence factors between :

[0016] The negative sign maps "reduced load" to the positive physical quantity of "increased voltage," that is... A positive value (pu / MW·reference capacity, dimensionless after normalization) has the physical meaning of "the first The first unit active load shedding caused by the load node The magnitude of voltage rise at the grid connection point of each new energy power station. Based on the inverse Jacobian matrix of the Newton-Raphson power flow equations, the sensitivity of voltage at each node to load power can be analytically calculated, thus obtaining the complete influence factor matrix. This matrix was repeatedly used in subsequent rounds of voltage pre-evaluation.

[0017] Steps 1-5: Modeling the overvoltage disconnection threshold for new energy generating units. Determine the high-voltage protection action threshold for new energy generating units. (PU, usually 1.1~1.3PU, varies depending on the model and protection settings), defining the first Voltage safety margin for a given load shedding scheme at a given power station:

[0018] in, For the first The steady-state voltage (pu) before the fault at each station. For the first The active power cutoff amount (pu) of each load node in the current decision. When The station voltage has a safety margin; when The system indicated that the station was at risk of overvoltage disconnection and that voltage coordination correction needed to be initiated.

[0019] Steps 1-6: Model Validation and Parameter Identification. The model is validated using time-domain simulation or measured data. The equivalent inertia in the frequency response model is identified using the least squares method. With damping coefficient and voltage sensitivity coefficient matrix Each element in the model is considered. The model accuracy requirements are frequency response error ≤ 0.05Hz and voltage prediction error ≤ 0.02pu, to ensure the reliability of subsequent optimization decisions.

[0020] Step 2. Frequency-Voltage Coordinated Load Shedding Strategy and Optimization Decision. This step is based on the coupling model and influence factor matrix established in Step 1. The design incorporates a coordinated load shedding strategy that considers voltage safety constraints at renewable energy power plants. During each load shedding round, the determination of active power cut-off amounts is synchronized with an assessment of the risk of voltage exceeding limits at the grid connection point, achieving joint optimization of frequency recovery and voltage safety. The specific implementation is as follows: Step 2-1: Mainnet frequency drop detection and startup criteria. Real-time monitoring of system frequency. With frequency change rate When satisfied and When the low-frequency load shearing start signal is triggered, the fault time is recorded. and initial power deficit estimate .in, This is the low-frequency load shedding start-up frequency (Hz, usually taken as 49.0~49.5Hz). This is the threshold for the rate of change of frequency (Hz / s, typically taken as -0.5 to -1.0 Hz / s).

[0021] Step 2-2: Preliminary tuning of the multi-round load shedding scheme. Estimate the total load shedding amount based on the system frequency response model. (pu):

[0022] in, For the rated frequency ( ), To estimate the lowest frequency , and This is a proportional coefficient tuned based on the system inertia level. Following the traditional low-frequency load shedding strategy... The motions are assigned to N rounds (N is usually 4 to 6 rounds), and the frequency of each round is fixed. (Hz, k=1,2,..., N ,satisfy ) and initial load shedding The setting is determined according to the principle of step-decreasing adjustment.

[0023] Steps 2-3: Pre-assessment of voltage over-limit in each round. In the... Before each operation (at which point the current grid connection voltage of each substation is known), (The superscript cur indicates the real-time measurement value, in units of pu), based on the influence factor matrix established in step 1. Predict the grid connection point voltage after load shedding in this round:

[0024] in, For the first The set of load nodes to be removed in each round ( (Pre-determined by the traditional stepped tuning scheme). No. The number of nodes to be removed (pu) in this round. To predict the voltage value (pu) after load shedding, define the safety margin parameter. (PU, typically 0.02~0.05 PU, is used to allow for redundancy in measurement error and model uncertainty), if any site satisfy If the load shedding point redistribution in step 2-4 is determined to be at risk of overvoltage disconnection from the grid, then the coordinated regulation of SVG pre-control in step 2-5 must be performed.

[0025] Steps 2-4: Optimized Redistribution of Load Shedding Points under Voltage Constraints. For rounds with a risk of voltage exceeding limits, optimized redistribution of load shedding points is performed to control the voltage rise within a safe range while ensuring the total load shedding amount. The following optimization problem is constructed:

[0026] st

[0027]

[0028]

[0029] in, For the first The removal weight of each load node (dimensionless, set according to the importance of the load, with important loads given a larger weight to reduce the probability of being removed first). For the first The maximum active power that can be cut off at a load node at the current operating point (pu, which is the actual active power load of the node at the current operating point). (per unit value) Let be the optimal load shedding amount (pu) for each node to be solved. This optimization problem is a linear programming problem, which is solved using the simplex method to obtain the optimal load shedding allocation scheme that satisfies the voltage constraint. .

[0030] Step 2-5: Generation of reactive power pre-control commands for SVG at renewable energy power plants. Before each load shedding cycle, simultaneously assess whether it is necessary to adjust the reactive power output of the SVG at the renewable energy power plant in advance to increase the voltage safety margin. If the voltage prediction value optimized in Step 2-4 is still close to the safety boundary (i.e., If the load shedding action is not performed, an SVG reactive power pre-absorption command is issued to the target power station to pre-reduce the grid connection point voltage to a safe range. The required pre-absorbed reactive power... (pu, a positive value indicates reactive power absorption) Derived from the equivalent circuit of the power station:

[0031] in, For the first The sensitivity of the voltage at the grid connection point of each power station to the reactive power output of the SVG is determined by the equivalent impedance of the power station. ( Approximate expression:

[0032] For the first The sum of the short-circuit impedance of the step-up transformer and the equivalent reactance of the collector line at each station ( The parameters are obtained offline from the station equipment and verified in the parameter identification steps 2-1 to 2-6. The constraints are as follows: ,in For the first The minimum reactive power output (pu, i.e., maximum absorption capacity, usually a negative value, determined by the rated capacity of the SVG) allowed for each power station SVG. This represents the current reactive power output (PU) of the SVG. This pre-control command is issued synchronously with the load shedding command, preceding the load shedding action. =Executes within 50~100Ms, ensuring that the grid connection point voltage is in a state of sufficient safety margin at the moment of load shedding.

[0033] Steps 2-6: Verification of Collaborative Decision-Making Effectiveness. After each round of actions (let's say the first round is...), The timing of each round of action is (Unit: Ms), real-time monitoring system frequency recovery trajectory and voltage at grid connection points of each new energy power station. Calculation of frequency recovery indicators:

[0034] in, For the first Frequency recovery time (s) after a round of actions. To ensure the safe restoration of the target frequency value (Hz, typically 49.5Hz), the voltage at all substation grid connection points must be verified to meet the requirements. If any metric is not met, the next round is triggered. The collaborative decision-making process (round) uses the measured response information of the current round (actual minimum frequency, actual voltage peak) as the input for dynamic correction in step 3-3.

[0035] Step 3. Multi-round dynamic correction and fault self-healing verification. This step utilizes a frequency-voltage dual-state driven multi-round dynamic correction mechanism to achieve adaptive adjustment of action settings and load shedding amounts in each round, and verifies the effectiveness of the strategy through typical fault scenarios. The specific implementation is as follows: Step 3-1: Real-time monitoring of frequency and voltage change rates. After each round of operation, continuously monitor the system frequency change rate. Voltage variation rate at each substation grid connection point Define the frequency acceleration index:

[0036] in, Frequency acceleration .when When the value remains negative and its absolute value increases, it indicates that the downward trend in frequency has not yet been contained, and subsequent rounds of actions need to be accelerated to quickly restore the active power balance.

[0037] Step 3-2: Adaptive Correction of Action Delay in Each Round. Based on real-time frequency and voltage status, dynamically adjust the action delay of subsequent rounds. :

[0038] in, For the first Initial tuning delay for each round ( (Pre-defined by the traditional step-tuning scheme). Correction factor for the rate of change of frequency ( (positive number) The voltage change rate correction factor ( (positive number). The physical logic is: when the frequency drops too quickly ( (negative values ​​and large absolute values) A negative value shortens the delay to accelerate load shedding; when the voltage rise rate is too fast ( (positive value) If the value is positive, the delay should be appropriately extended to allow time for the SVG pre-control response.

[0039] Step 3-3: Dynamic correction of load shedding amount for subsequent rounds. The load shedding amount for subsequent rounds is dynamically corrected according to the actual frequency response effect of the executed rounds:

[0040] where, is the correction coefficient (0<y<1, dimensionless), is the remaining power deficit re-estimated based on the real-time frequency response (pu, deduced back from the measured frequency nadir and the frequency response models of Step 3-1 to Step 3-2), is the sum of the actual shedding amounts of the previous rounds (pu, is the number of the executed rounds), is the corrected load shedding amount for the th round (pu). The corrected needs to be substituted back into the optimization model in Step 2-4 to verify the voltage constraint, and the original is replaced with to re-solve the optimal allocation scheme.

[0041] Step 3-4: Timing coordination and communication verification. A two-level coordination architecture of master station and slave stations is designed: the master station (dispatching center) is responsible for global frequency monitoring and round decision-making, and each new energy station slave station is responsible for local voltage monitoring and SVG execution. Define the action timing error between the master station and the th slave station:

[0042] where, is the timestamp (Ms) at which the master station issues the decision instruction for the th round, is the timestamp (Ms) at which the th station slave station receives and feeds back the execution confirmation. It is required that the timing errors of all stations satisfy ( is usually set to 100Ms), and the redundant communication channels (optical fiber main channel + 5G backup channel) are used to ensure the reliability of instruction transmission, so that the communication success rate can be improved.

[0043] Step 3-5: Fault recovery effect evaluation. After all N rounds of actions are completed, the total system frequency recovery time (s, that is, the time difference from the fault moment to the time when the frequency first recovers to and remains stably) and the maximum voltage of each station (pu, that is, the peak value of the grid connection point voltage of each station during the entire fault transient process) are calculated. The comprehensive recovery performance index is defined as:

[0044] in, and Weighting coefficients ( (dimensionless) The target recovery time (in seconds, usually 5-10 seconds, determined according to system size and scheduling procedures). The smaller the value, the better the frequency-voltage coordinated recovery effect.

[0045] Steps 3-6: Verification of Typical Fault Scenario Library. A scenario library is constructed containing typical power deficit faults such as unit tripping, DC blocking, and large-scale grid disconnection of renewable energy, covering different operating modes with renewable energy penetration rates ranging from 10% to 50%. The proposed coordinated load reduction strategy is tested in each scenario to verify the frequency recovery time. Peak voltage at grid connection points of each substation (i.e., no overvoltage protection is triggered throughout the entire process), critical load retention rate Key indicators, etc. Among them, critical load is defined as a weighting factor. In the top 30 percentile (i.e. Larger critical loads), critical load retention rate = (Total active power of critical loads not disconnected) / (Total active power of critical loads before the fault) × 100%. The setting parameters are determined through offline simulation. , , , Optimize strategies (e.g., to improve their adaptability to different scenarios).

[0046] The beneficial effects of this invention are: 1. This invention establishes a quantitative sensitivity model of load shedding and transient voltage rise at renewable energy grid connection points, transforming the risk of renewable energy units disconnecting from the grid due to overvoltage into a voltage safety margin constraint. This upgrades low-frequency load shedding decisions from a single frequency dimension to a multi-dimensional frequency-voltage coupling approach. Traditional load shedding strategies cannot predict transient overvoltage issues caused by load shedding, easily leading to mass disconnections of renewable energy units. This invention can accurately quantify the voltage fluctuation risks corresponding to different load shedding actions, avoiding grid transient overvoltage phenomena caused by active power load shedding and reactive power voltage imbalance from the control source. It completely blocks the vicious chain reaction of "load shedding - voltage over-limit - renewable energy disconnection," significantly improving voltage safety and stability under grid fault conditions with high renewable energy penetration.

[0047] 2. This invention designs a frequency-voltage coordinated load shedding strategy, simultaneously verifying the frequency recovery effect and voltage safety boundary during each round of low-frequency load shedding decision-making. For predicted voltage exceedance scenarios, by optimizing the active-reactive load distribution ratio and linking it with the rapid reactive power absorption regulation of SVG at renewable energy power plants, it achieves coordinated operation of active frequency control and reactive voltage control, solving the technical challenge of decoupling active and reactive power control in traditional load shedding methods. This ensures effective compensation for system active power deficits and suppresses significant frequency drops, while accurately constraining the grid connection point voltage within a safe operating range, avoiding the over-frequency voltage loss defects caused by single-dimensional control, and achieving dual stability of grid frequency and voltage under fault conditions.

[0048] 3. This invention proposes a dynamic cycle correction mechanism based on frequency and voltage dual-state driving, which can adaptively adjust the action delay and load shedding amount of the load shedding cycle according to the real-time frequency change rate and voltage change rate. Compared with the traditional coarse load shedding logic with fixed cycles, fixed delays, and fixed load shedding amounts, this invention has a strong adaptability to operating conditions, and can adapt to the low inertia, strong fluctuations, and tight frequency-voltage coupling characteristics of new energy power grids. It avoids the waste of grid resources and unnecessary power outages caused by excessive load shedding, while preventing secondary frequency drops and multi-level cascading load shedding problems caused by under-load shedding, significantly improving the accuracy, flexibility, and adaptability of low-frequency load shedding strategies.

[0049] 4. This invention effectively solves the problem of frequency and voltage instability under high power deficit faults in power systems with a high proportion of new energy sources through integrated and coordinated control of frequency and voltage joint decision-making. It significantly reduces the risk of new energy units disconnecting from the grid and the grid frequency continuing to deteriorate after a fault, effectively reduces the range of grid fault disturbances, and improves the safety and stability control capability and fault self-healing capability of new power systems under emergency faults. It provides a solid technical guarantee for the safe and reliable operation of power grids with a high proportion of new energy sources and has extremely strong engineering application value and promotion prospects. Attached Figure Description

[0050] Figure 1 This is an overview diagram of the architecture of the low-frequency load shedding method with combined voltage control in a power system with a high proportion of new energy sources according to the present invention. Figure 2 This is a flowchart illustrating the modeling process of the system frequency response and voltage coupling characteristics of new energy power stations in this invention. Figure 3 This is a flowchart of the frequency-voltage coordinated load shedding strategy optimization decision-making process of the present invention; Figure 4 This is a flowchart of the multi-round dynamic correction and fault self-healing process of the present invention; Figure 5 This is a schematic diagram illustrating the verification results of a typical failure scenario of the present invention. Detailed Implementation

[0051] The relevant technologies of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0052] like Figures 1-5 As shown, the low-frequency load shedding method for combined voltage control in a power system with a high proportion of new energy sources in this embodiment includes the following steps: Step 1. System Modeling and Coupling Characteristic Analysis: Step 1-1: Data Acquisition and Status Awareness. The system frequency is collected in real time by connecting to the main network SCADA / PMU measurement system via the scheduling data network. (Unit: Hz), Rate of change of frequency (Unit: Hz / s), voltage amplitude of each 220kV and above busbar (Per-unit value, reference voltage is the rated voltage of the bus) and phase angle (Unit: rad). Voltage at the grid connection point of each new energy power station is synchronously acquired through the remote control terminal of the new energy power station. ( =1,2,..., M , M The total number of renewable energy power stations, with the superscript 0 indicating the steady-state value before the fault, in units of pu; and the real-time reactive power output of the dynamic reactive power compensation devices (using static var generators, SVG, as an example) within the power station. (Unit: pu, based on the rated capacity of the power station) and available reactive power reserve (Unit: pu), and the active power of each load shedding node (load nodes equipped with low-frequency load shedding devices in distribution networks of 110kV and below voltage levels). (Unit: MW) and reactive power (Unit: Mvar, i=1,2,..., L , L (Total number of load shelvable nodes). All the above multi-source heterogeneous data are uniformly time-aligned (based on GPS / BeiDou time synchronization) and normalized. The active power per-unit reference is taken as the system's rated capacity. The voltage per-unit standard is based on the rated voltage of each voltage level. This ensures the synchronization and consistency of subsequent modeling.

[0053] Step 1-2: System Frequency Response Aggregate Modeling. The high-proportion renewable energy power system is equated to an aggregated frequency response model including both renewable energy units and traditional synchronous generator units. Based on the system's moment of inertia distribution and the virtual inertia response characteristics of the renewable energy units, a model is established to address the power deficit. to frequency deviation Transfer function:

[0054] in, System frequency deviation (unit: Hz). The system active power deficit (per unit value, in terms of...) (Based on) Laplace operator (unit: ); The system equivalent inertial time constant (unit: s) is obtained by weighted aggregation of the inertial constants of synchronous units across the entire network and the virtual inertial constants of new energy units. The weighting weight is the proportion of the rated capacity of each unit to the total capacity of the system. Load damping coefficient (unit: This reflects the self-regulating characteristic of the load's active power as a function of frequency, typically ranging from 1.0 to 2.0. ; Equivalent frequency regulation coefficient for new energy generating units (unit: This characterizes the comprehensive capability of new energy power plants to participate in system frequency regulation; The frequency response transfer function of the renewable energy unit reflects the dynamic response characteristics of the unit's active power output to frequency deviation, and can usually be approximated as a first-order inertial element. ,in The frequency regulation response time constant of the new energy unit is given (unit: s, typical value is 0.5~2.0s). This transfer function model provides a quantitative calculation basis for the subsequent prediction of total load shedding.

[0055] Steps 1-3: Modeling the transient voltage characteristics of renewable energy power plants. Establish a coupling model between the grid connection point voltage and the reactive power of the renewable energy power plant. For the first... For a new energy power station, the resistance of the internal collector lines of the station is ignored (typically...). (The resistive component has a relatively small impact on the voltage) Grid connection point voltage satisfy:

[0056] in, For the first The equivalent voltage on the grid side of each station (per unit value, taken as the equivalent value of the voltage of the upstream bus at the grid connection point after voltage division by equivalent impedance, obtained by state estimation); For the first The total reactive power output (per unit value) of each power station is the sum of the reactive power output of each new energy unit in the station and the reactive power output of the SVG. For the first The equivalent reactive load (per unit value) of each station, including the reactive losses of the step-up transformer, the reactive losses of the collector lines and the reactive consumption of the auxiliary equipment in the station, can be calculated offline by the station's reactive balance equation or identified online through state estimation. For the first The short-circuit capacity (unit: MVA) of each power station's grid connection point is provided by the power grid operation department; This is a standardized baseline value (1.0 pu, used to maintain dimensional consistency in the formula). When low-frequency load shedding occurs, the equivalent reactive load on the system side is... The load decreased (because the removed load usually includes some reactive load), but the reactive power compensation equipment (SVG / capacitor bank) in new energy power plants remained in operation due to its faster response speed. The voltage at the grid connection point was not reduced synchronously. This model forms the physical basis for the load shedding-voltage mapping.

[0057] Steps 1-4: Load Shedding-Voltage Rise Sensitivity Modeling. Based on the transient voltage model established in Steps 1-3, the impact of load shedding on the voltage at the grid connection point of renewable energy power plants is further quantified. Define the... Active power cut-off amount per load node (per unit value) and the number Voltage rise at grid connection point of each new energy power station Influence factors between (per unit value) :

[0058] in, , The partial derivative of voltage with respect to the active power of the load is solved by linearizing the Newton-Raphson power flow equations for the entire system at the pre-fault operating point. Specifically, the Jacobian matrix of the power flow equations is... The blocks are [ ; Then, the voltage sensitivity to the active power of the load can be expressed as:

[0059] The first in the sensitivity matrix Line number The column element is Its value is negative (indicating that an increase in load leads to a decrease in voltage, and a decrease in load leads to an increase in voltage), hence the introduction of a negative sign makes... Positive values. Influence factor matrix. The calculations are performed online and cached before a fault occurs, and then directly invoked in subsequent rounds of voltage pre-assessment to avoid redundant power flow calculations.

[0060] Steps 1-5: Modeling the overvoltage disconnection threshold for new energy units. Based on the grid connection agreements of each new energy power station and the low-voltage / high-voltage ride-through technical requirements of the units, determine the first... High voltage protection activation threshold of new energy generating units at each power station The per-unit value is typically 1.1 to 1.3 pu, depending on the specific model and protection settings; generally, wind power uses 1.15 pu and photovoltaic uses 1.10 pu. For simplicity, this invention will be uniformly referred to as pu from now on. During project implementation, each station will take values ​​according to their actual settings. Definition 1 Each site is given a load shedding scheme Voltage safety margin under (column vector, containing the active power cut-off amounts of all L load nodes):

[0061] in, For the first Steady-state voltage (per unit value) before a fault at each station. The load shedding scheme caused the first The total voltage increase (per unit value) at the grid connection point of each substation. When At that time, the substation voltage has a safety margin, and the new energy units can operate safely; when When this occurs, it indicates that the grid connection point voltage will reach or exceed the high voltage protection action threshold, posing a risk of grid disconnection of new energy units, and the voltage collaborative correction mechanism needs to be activated.

[0062] Steps 1-6: Model Validation and Parameter Identification. Detailed electromagnetic or electromechanical transient models of a power system with a high proportion of renewable energy are built using an offline time-domain simulation platform (such as PSD-BPA or PSS / E). Different power deficit faults (such as single-unit tripping, bipolar blocking, etc.) are simulated under typical operating conditions. The simulated frequency dynamic curves are compared with the output of the frequency response aggregation model from Steps 1-2. The equivalent inertia is identified using the least squares method. With damping coefficient :

[0063] in, For simulating the system frequency trajectory (unit: Hz). For aggregation models in the same Frequency output under excitation (unit: Hz). The simulation duration is typically 10 seconds. Simultaneously, the voltage dynamics after the load shedding operation in the simulation are compared with the voltage predicted by the sensitivity model in steps 1-4. If the voltage prediction error exceeds 0.02 pu, the influence factor matrix is ​​recalibrated using measured data. The corresponding elements in the model. Model accuracy requirements: maximum frequency response deviation ≤ 0.05Hz, maximum voltage prediction deviation ≤ 0.02pu. Only after meeting the accuracy requirements can the model be put into online application.

[0064] Step 2. Frequency-Voltage Coordinated Load Shedding Strategy and Optimization Decision. This step is based on the system frequency response model and influence factor matrix established in Step 1. and voltage safety margin of each station In the decision-making process for each low-frequency load shedding cycle, the determination of active power cut-off is synchronously coupled with voltage safety constraints, achieving joint optimization of frequency recovery and voltage safety. The specific implementation is as follows: Step 2-1: Main Network Frequency Drop Detection and Startup Criteria. A low-frequency load shedding startup monitoring module is deployed at the scheduling master station to monitor the system frequency at 20ms intervals. (Unit: Hz) and rate of change of frequency (unit: ) Perform rolling judgment. When the system frequency Reduce to low-frequency load reduction start-up frequency (Unit: Hz, typically set between 49.0~49.5Hz, the specific setting depends on the system's normal frequency fluctuation range and low-frequency load shedding procedures) and frequency change rate Less than the threshold (unit: When a frequency fluctuation (typically -0.5 to -1.0 Hz / s, used to distinguish between normal frequency fluctuations and severe power deficit faults) occurs, a severe power deficit fault is determined to have occurred in the system, and a low-frequency load shedding start signal is immediately triggered. The time of the fault is recorded. (Unit: ms, based on GPS timing system), and estimate the initial power deficit based on the initial frequency change rate. (per unit value), where This is the instantaneous value of the rate of change of frequency at the start-up moment.

[0065] Step 2-2: Preliminary tuning of the multi-round load shedding scheme. Based on the system frequency response aggregation model in Step 1, estimate the total load shedding required to restore the system frequency to a safe level. (per unit value):

[0066] in, The rated frequency is 50Hz. To estimate the lowest frequency (in Hz), the frequency response model is used in the initial... The minimum point of the time-domain response curve under excitation is determined; The maximum absolute value of the rate of change of frequency after the fault (unit: ); (unit: )and (unit: ) is a proportionality coefficient, determined offline based on the system inertia level and tuning procedure. For high-inertia systems ( ) Take a value of 0.15~0.20. For low inertia systems ( ) Take a value of 0.25~0.35 .Will The load is allocated to N unloading cycles (N is usually 4 to 6 cycles, depending on the system size and scheduling procedures), and the frequency setpoint for each cycle is... (unit: k=1,2,..., N ,satisfy The frequency interval between adjacent rounds is typically 0.1~0.2Hz, and the initial load shedding is... (Per unit value) is set according to the principle of step-decreasing: the first round of operation has the highest frequency but the smallest load shedding (approximately accounting for...). (10%~15%), the frequency of subsequent rounds decreases while the load shedding increases in each round (the last round accounts for approximately 10%~15%). (25%~30%), the sum of the initial load shedding amounts for each round equals .

[0067] Steps 2-3: Pre-assessment of voltage over-limit in each round. Before the action in the kth round (at which point the current voltage of each renewable energy power station's grid connection point is known). (The superscript "cur" indicates the real-time measurement value, in units of pu), based on the influence factor matrix established in step 1. The predicted load shedding amount according to the traditional scheme (i.e., the original setting scheme that only aims at frequency recovery and does not consider voltage constraints) is... The voltage at the grid connection point after the connection is:

[0068] in, The set of load nodes to be removed in the initial tuning scheme of the kth round is the subset of nodes pre-specified according to the step-decreasing principle in step 2-2. The sets of load shedding nodes in adjacent rounds do not overlap, i.e. ); For the first The proportion of nodes removed in the initial scheme (dimensionless, satisfying...) Typically, the load is distributed proportionally to the load size of each node, i.e. ); That is, the first The initial planned cutoff amount for each node in this round. Define the safety margin parameter. (per unit value, usually taken as) To allow for redundancy in measurement error and model uncertainty, the lower limit is used when the system measurement accuracy is high, and the upper limit is used when the measurement conditions are poor. For any site... satisfy If the traditional scheme is executed in this round, it will be determined that the voltage at the grid connection point of the new energy power station will approach or exceed the high voltage protection threshold, posing a risk of grid disconnection. Therefore, it is necessary to immediately implement the load shedding point optimization and redistribution in step 2-4 and the SVG reactive power pre-control coordinated regulation in step 2-5.

[0069] Step 2-4: Optimized redistribution of load shedding points under voltage constraints. For round k identified in Step 2-3 as having a risk of voltage exceeding limits, the original coarse-grained load shedding scheme (allocated proportionally according to load capacity) is upgraded to an optimized redistribution considering voltage constraints. The following linear programming model is constructed to optimize and adjust the specific shedding amounts at each node to control voltage rise, while ensuring the total load shedding meets frequency recovery requirements:

[0070] st

[0071]

[0072]

[0073] in, This is the set of candidate load nodes for this round (consistent with the initial scheme, but the allocation ratio has been re-optimized). The first one to be solved Optimal load shedding for each node (per unit, decision variable, unit: pu); For the first The cutoff weight for each load node (dimensionless, pre-set based on load importance, such as hospitals, transportation hubs, etc.). Take the larger value, typical industrial load Taking the middle value can interrupt the load. Take the smaller value to prioritize important loads while meeting voltage constraints. For the first The current actual active power of each load node (per unit value, i.e., the upper limit of the load that can be cut off at the current operating point, which cannot exceed the current actual load). This optimization model is a typical linear programming problem, with a number of variables, etc. (Typically dozens to hundreds of constraints), the number of constraints equals M+2 (the number of new energy power stations plus equality and inequality constraints), the scale is relatively small, and the standard simplex method can complete the solution in milliseconds (≤50Ms) to obtain the optimal load shedding scheme that satisfies the voltage constraints. .

[0074] Step 2-5: Generation of reactive power pre-control commands for SVG at renewable energy power plants. Before each round of load shedding, simultaneously assess whether it is necessary to adjust the reactive power output of the SVG at the renewable energy power plant in advance to further increase the voltage safety margin. If the voltage prediction value optimized in Step 2-4 is still close to the safety boundary (i.e., there are power plants...) satisfy If the load shedding command indicates insufficient safety margin but not yet exceeded limit, then an SVG reactive power pre-absorption command is sent to the corresponding renewable energy power station substation simultaneously with the load shedding command. This pre-lowers the grid connection point voltage, ensuring that the instantaneous voltage peak during load shedding falls within the safe range. The required pre-absorbed reactive power... (Per-unit value, a positive value indicates that the SVG absorbs reactive power from the system, unit: pu) Calculated according to the following formula:

[0075] in, The predicted voltage after load shedding after optimization in steps 2-4 (unit: pu); For the first The sensitivity of the voltage at the grid connection point of each power station to the reactive power output of the SVG is determined by the equivalent impedance of the power station. (unit: Approximately expressed as:

[0076] For the first The sum of the short-circuit impedance of the step-up transformer at each station (referred to the high-voltage side) and the equivalent reactance of the collection lines (unit: This approximation is obtained from the parameters on the nameplates of the station equipment or through offline equivalent calculations. This approximation is based on high-voltage transmission lines. Its physical properties, within the engineering accuracy requirements (error ≤ 10%), meet the application requirements. The constraints are... ,in This represents the current reactive power output of the SVG (per unit value; a positive value indicates reactive power output). For the first The minimum per-unit reactive power output allowed by the SVG at each power station (a per-unit value, i.e., the maximum inductive reactive power absorption capacity, typically 0.8 to 1.0 times the rated capacity of the SVG, determined by the equipment's factory parameters) ensures that the pre-absorption command is within the power station equipment's regulation capacity range. This pre-control command is issued synchronously with the load shedding command (transmitted via the same communication message), and is issued in advance of the load shedding circuit breaker's operation. =50~100ms, the substation performs the SVG setting switch, ensuring that the grid connection point voltage is in a state of sufficient safety margin at the moment of load shedding. This timing arrangement takes advantage of the SVG's response speed (the SVG's full response time from receiving the command to outputting reactive power change is ≤30ms), and can complete pre-control within the circuit breaker's operating window.

[0077] Steps 2-6: Verification of Collaborative Decision-Making Effectiveness. After each round of load shedding action is completed (let the action time of the kth round be...), (Unit: ms, determined by the return signal from the circuit breaker auxiliary contacts), the dispatch master station continuously monitors the dynamic trajectory of the system frequency response and the voltage at the grid connection points of each new energy power station. The frequency recovery index is calculated as follows:

[0078] in, The system frequency is restored to a safe value after the k-th round of action. (Unit: Hz, typically 49.5Hz) Required recovery time (unit: seconds). Simultaneously, the maximum voltage at the grid connection point of each renewable energy power station during the load shedding transient process is verified. Does it meet the requirements? If the frequency recovery time exceeds expectations or the voltage constraint is violated, the coordination effect of the current round is deemed unsatisfactory, and the measured response information of the current round (including the actual lowest frequency point) needs to be provided. Actual voltage peak The feedback input is fed into the dynamic correction module in step 3-3, which adaptively adjusts the load shedding amount in subsequent rounds to ensure that the final control effect meets the safety requirements.

[0079] Step 3. Multi-round dynamic correction and fault self-healing verification. This step uses a frequency-voltage dual-state driven multi-round dynamic correction mechanism to adaptively adjust the action delay and load shedding of subsequent rounds based on the measured response effect of the previous rounds. Comprehensive verification using a typical fault scenario library ensures the strategy's adaptability to different operating modes and fault types. The specific implementation is as follows: Step 3-1: Real-time monitoring of frequency and voltage change rates. After each round of action, the dispatch master station continuously monitors the system frequency change rate at 10-ms intervals. (unit: Voltage variation rate at grid connection points of various new energy power plants (unit: Calculate the frequency acceleration index (i.e., the acceleration due to frequency change):

[0080] in, Frequency acceleration (unit: ), consisting of two adjacent sampling periods The difference is obtained by dividing the sampling period value. This indicator is used to determine whether the frequency drop trend has been effectively curbed: when When the value remains negative and its absolute value shows an increasing trend, it indicates that the frequency descent rate is still accelerating, the system inertia support is insufficient, or new power disturbances have occurred, requiring a faster pace of subsequent cycles; when When the frequency changes from negative to positive, it indicates that the downward trend has been effectively curbed, and the pace of subsequent rounds can be maintained or slowed down to avoid over-cutting.

[0081] Step 3-2: Adaptive Correction of Action Delay in Each Round. Based on the real-time frequency and voltage change rates monitored in Step 3-1, the action delay of subsequent rounds is dynamically adjusted. (Unit: seconds) This makes the delay no longer fixed, but adaptively changes with the dynamic response state of the system:

[0082] in, The initial setting delay (in seconds) for the k-th round is given in advance by the traditional low-frequency load shedding setting procedure according to the step time limit principle (usually the first round has a delay of 0.2 seconds, and subsequent rounds increase by 0.2~0.3 seconds). Correction factor for rate of change of frequency (unit: Take positive values, typically 0.05~0.10. When the frequency drop rate is too fast (df / dt is negative and the absolute value is large), A negative value makes the actual delay less than the initial tuned delay, thus accelerating subsequent rounds of action; Voltage change rate correction factor (unit: Take positive values, typically 0.05~0.10. When the voltage rise rate is too fast ( (positive and relatively large) A positive value ensures that the actual delay is greater than the initial set delay, allowing more adjustment time for the SVG pre-control and absorption response, thus preventing voltage over-limit. This correction mechanism achieves differentiated adaptive control by "rapidly reducing load shedding when the frequency drops" versus "temporarily reducing load shedding and allowing the SVG to adjust first when the voltage rises quickly."

[0083] Step 3-3: Dynamic Adjustment of Load Shedding in Subsequent Rounds. Based on the actual frequency response of the rounds already executed, dynamically adjust the load shedding for subsequent rounds to avoid over- or under-shearing according to a fixed setting. Utilize the actual load shedding amounts from each of the rounds already executed (l=1,2,...,k-1). The lowest measured frequency , the actual power shortage is back-calculated through the frequency response aggregation model of steps 3-1 to 3-2 (per unit value):

[0084] This equation is the inverse solution of the frequency response model, where the measured frequency trajectory is substituted into the time-domain solution of the transfer function to back-deduce the power shortage. Then the corrected load shedding amount of the k-th round is:

[0085] Herein, is a correction coefficient (0<y<1, dimensionless, typical value 0.6~0.8, used to avoid oscillation caused by excessive single correction); is the sum of the actual shedding amounts in the previous rounds (per unit value), and the difference ( ) is the remaining power shortage to be shed. The corrected is taken as the new target load shedding amount for the k-th round, replacing the original , which is re-substituted into the optimization model of step 2-4, the right-hand side term of the equality constraint of the original optimization model is replaced with , and the optimal allocation scheme is solved again to ensure that the corrected load shedding scheme remains a feasible solution under the voltage safety constraints of each new energy station.

[0086] Step 3-4: Time sequence coordination and communication verification. A two-level coordination architecture of master station and slave stations is adopted to realize reliable instruction distribution and execution confirmation: the scheduling master station (located at provincial or regional dispatch) is responsible for global frequency monitoring, round decision-making and optimization solution, and each new energy station slave station (located at the station end) is responsible for local voltage monitoring and SVG instruction execution. Define the action time sequence error between the master station and the th slave station as:

[0087] Herein, is the moment when the master station completes encapsulation and issues the k-th round decision instruction (unit: ms, based on the master station system clock); is the moment when the th station slave station receives the instruction and feeds back the execution confirmation (unit: ms). The system requires that the time sequence error of all stations satisfies ( is usually set to 100ms, determined according to the measured statistical value of communication network delay, generally the delay of optical fiber network ≤30Ms, the delay of 5G network ≤50Ms, (A 100Ms redundancy is provided). To achieve this requirement, fiber optic communication is used as the main transmission channel, and 5G wireless communication is used as a backup hot standby channel (automatically switching when the main channel is interrupted, with a switching time ≤20Ms). Both channels continuously send heartbeat packets to monitor the link status, ensuring that commands can still be transmitted normally when a single channel fails, thus improving the overall communication success rate.

[0088] Steps 3-5: Fault Recovery Effectiveness Evaluation. This evaluation is conducted after all N rounds of load shedding operations have been completed (with the completion time of the last round of circuit breaker operation as the endpoint, denoted as ). This involves a comprehensive quantitative evaluation of the recovery effectiveness from this fault. The total system frequency recovery time is calculated. (Unit: seconds): From the moment of failure The frequency recovered to its first level =49.5Hz and the time difference between the points where the frequency remains at or above that value, i.e. Simultaneously, the peak voltage at the grid connection point of each renewable energy power station was recorded during the entire fault transient process. (Unit: pu), the maximum value of the peak voltage at each substation is taken as the basis for the overall voltage security assessment. Define the comprehensive recovery performance index. (Dimensionless):

[0089] in, and Weighting coefficients ( Dimensionless, set according to the relative importance of scheduling operation to frequency security and voltage security, and entered when system frequency stability issues are prominent. Take a value of 0.7~0.8, and enter when the risk of new energy sources disconnecting from the grid is prominent. Take 0.6~0.7). The target recovery time (in seconds) is determined based on the system scale. For regional power grids (equivalent inertia) For ≥5s), take 5~8s; for provincial power grids (equivalent inertia 3~5s), take 8~10s. A smaller value indicates a better frequency-voltage coordinated recovery effect. If the value is ≤1.0, the control effect is considered to meet the requirements in this scenario. If the value is greater than 1.0, the tuning parameters need to be optimized and adjusted.

[0090] Steps 3-6: Verification of Typical Fault Scenario Library. To fully verify the robustness and adaptability of the collaborative load reduction strategy proposed in this invention, a typical fault scenario library covering different fault types, different operating modes, and different new energy penetration levels is constructed, including but not limited to: (1) large-capacity thermal power unit tripping fault (single unit tripping or double unit tripping in succession, power deficit of 500~1000MW); (2) bipolar blocking fault of high-voltage DC transmission system (power deficit of 1000~3000MW); (3) large-scale power drop caused by meteorological changes in new energy power plants (power deficit of 300~800MW); (4) combined faults (such as DC blocking + new energy grid disconnection cascading faults). Each scenario further covers different operating modes with new energy penetration rates of 10%~50% (penetration rate step size of 10%, a total of 5 levels), with a total of no less than 20 typical test conditions. The collaborative load reduction strategy proposed in this invention is run under each condition, and the following core indicators are recorded and verified: (a) frequency recovery time ≤10s; (b) Peak voltage at grid connection points of each substation throughout the entire process (i.e., no high-voltage protection is triggered at any of the stations); (c) Critical load retention rate The critical load is defined as the weighting factor. High-weighted load nodes located in the top 30 percentile (i.e. Larger, critical loads, such as hospitals, government offices, data centers, and transportation hubs. = (Total active power of critical loads not removed) / (Total active power of critical loads before the fault) × 100%. For extreme operating conditions that do not meet the indicators, offline reinforcement learning methods are used to adjust the tuning parameter set (including the start-up frequency). proportionality coefficient and Correction coefficient and Correction coefficient Safety margin The optimization is performed, with the objective function being to minimize the overall performance index. The optimal parameter combination is obtained through multiple rounds of iterative training and then updated offline to the parameter library of the online control system, continuously improving the adaptability of the strategy to different scenarios.

[0091] In summary, this invention proposes a joint voltage control low-frequency load shedding method for power systems with a high proportion of renewable energy. By constructing a coupling characteristic model of frequency response and renewable energy station voltage, a load shedding optimization allocation framework considering voltage safety constraints is established. This incorporates the reactive power pre-control link of the renewable energy station's SVG (Static Var Generator) and constructs a multi-round dynamic correction mechanism driven by both frequency and voltage states. Under frequency fault disturbance scenarios, this method can balance rapid system frequency recovery with renewable energy station voltage safety, prioritizing the supply of important loads. It solves the problem that traditional low-frequency load shedding methods do not consider the risk of renewable energy station voltage exceeding limits and are prone to triggering large-scale grid disconnection cascading faults. The method's optimization model is small in scale and has a fast solution speed, adapting to the existing communication architecture between the dispatch master station and station substations. Dual-channel redundancy design ensures reliable command transmission. After offline verification and parameter optimization in multiple typical fault scenarios, it can effectively improve the safe and stable control capability of power systems with a high proportion of renewable energy in response to large-power deficit faults, demonstrating promising engineering application prospects.

[0092] It should be emphasized that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A low-frequency load shedding method using combined voltage control in a power system with a high proportion of new energy sources, characterized in that, Includes the following steps: Step 1: System modeling and coupling characteristic analysis, including establishing a system frequency response model, a transient voltage characteristic model of new energy power plants, a load shedding-voltage rise sensitivity model, and a new energy unit overvoltage disconnection threshold model, and quantifying the mapping relationship between active power cut-off and voltage rise at the new energy grid connection point during low-frequency load shedding. Step 2, Frequency-Voltage Coordinated Load Shedding Strategy and Optimization Decision: Based on the model in Step 1, frequency recovery requirements and voltage safety constraints are evaluated simultaneously before each round of low-frequency load shedding. Coordination of active frequency control and reactive voltage control is achieved through load shedding point optimization redistribution and reactive power pre-control of static var generators (SVG) in new energy power plants. Step 3: Multi-round dynamic correction and fault self-healing verification. The action delay and load shedding amount of subsequent rounds are adaptively adjusted according to the real-time frequency change rate and voltage change rate, and the effectiveness of the strategy is verified through typical fault scenarios.

2. The low-frequency load shedding method for combined voltage control in a power system with a high proportion of new energy sources according to claim 1, characterized in that, Step 1 specifically includes the following sub-steps: Step 1-1: Real-time acquisition of system frequency, frequency change rate, voltage at grid connection points of each new energy power station, output of reactive power compensation device, and power of load shedding nodes through the data acquisition system; Step 1-2: Establish the equivalent frequency response transfer function model of a power system with a high proportion of new energy sources; Steps 1-3: Establish a coupling model between the grid connection point voltage of the new energy power plant and the reactive power of the power plant; Steps 1-4: Calculate the influence factor matrix of load shedding and voltage rise based on the Jacobian matrix of the power flow equation; Steps 1-5: Define the voltage safety margin according to the high-voltage protection settings of the new energy unit; Steps 1-6: Verify the model accuracy through time-domain simulation and parameter identification.

3. The low-frequency load shedding method with combined voltage control in a power system with a high proportion of new energy sources according to claim 2, characterized in that, The equivalent frequency response transfer function model in steps 1-2 is as follows: in, For system frequency deviation, This is due to the system's active power deficit. The system's equivalent inertial time constant. For the Laplace operator, This is the load damping coefficient. This refers to the equivalent frequency regulation coefficient for new energy generating units. This is the frequency response transfer function for new energy power generation units.

4. The low-frequency load shedding method with combined voltage control in a power system with a high proportion of new energy sources according to claim 2, characterized in that, The voltage model for the grid connection point of the new energy power station in steps 1-3 is as follows: in, Let the voltage at the grid connection point of the j-th power station be . For the first Equivalent voltage on the power grid side of each station For the first Each station was ultimately unproductive. For the first The equivalent reactive load of each station For the first Short-circuit capacity of each station's grid connection point This is the standardization baseline value.

5. The low-frequency load shedding method for combined voltage control in a power system with a high proportion of new energy sources according to claim 1, characterized in that, Step 2 specifically includes the following sub-steps: Step 2-1: Monitor the system frequency and rate of frequency change. When the following conditions are met... and Low-frequency load shedding is triggered at certain times; Step 2-2: Estimate the total load shedding based on the frequency response model and allocate it to each round according to the step principle; Steps 2-3: Before each round of operation, predict the voltage after load shedding based on the influence factor matrix to determine whether there is an overvoltage risk; Steps 2-4: For rounds with voltage over-limit risk, construct an optimized load shedding and redistribution model; Steps 2-5: Generate SVG reactive power pre-absorption instructions to adjust reactive power output before load shedding. Steps 2-6: Verify the frequency recovery and voltage safety effect after each round of operation.

6. The low-frequency load shedding method with combined voltage control in a power system with a high proportion of new energy sources according to claim 5, characterized in that, The optimized redistribution model in steps 2-4 is as follows: st in, For the first The removal weight of each load node For the first The maximum active power that can be cut off at each load node under the current operating point. The optimal load shedding amount for each node is to be determined. This is the original load. Since the grid connection voltage is known, As the impact factor, The overvoltage protection threshold, For safety margin.

7. The low-frequency load shedding method for combined voltage control in a power system with a high proportion of new energy sources according to claim 5, characterized in that, The calculation formula for the SVG reactive power pre-absorption command in steps 2-5 is as follows: in, Let J be the reactive power that the j-th power station needs to absorb. To predict the voltage at the grid connection point, For the first The sensitivity of the voltage at the grid connection point of a power station to the reactive power output of the SVG.

8. The low-frequency load shedding method for combined voltage control in a power system with a high proportion of new energy sources according to claim 1, characterized in that, Step 3 specifically includes the following sub-steps: Step 3-1: Monitor the real-time rate of change of frequency and rate of change of voltage; Step 3-2: Adaptively adjust the delay of subsequent rounds of actions based on frequency and voltage change rate; Step 3-3: Dynamically adjust the load shedding amount for subsequent rounds based on the measured response of the already activated rounds; Steps 3-4: Ensure that the timing error of instruction transmission is less than the set threshold through a two-level collaborative architecture of master station and substation; Steps 3-5: Calculate the total frequency recovery time and peak voltage, and evaluate the overall recovery performance indicators; Steps 3-6: Verify the robustness of the strategy through a library of typical failure scenarios.

9. The low-frequency load shedding method for a power system with a high proportion of new energy sources using combined voltage control as described in claim 8, characterized in that, The action delay correction formula in step 3-2 is: in, For the first Initial tuning delay for each round This is the correction factor for the rate of change of frequency. This is the voltage change rate correction factor. The rate of change of frequency, This is the voltage at the grid connection point of the first new energy power station.

10. The low-frequency load shedding method with combined voltage control in a power system with a high proportion of new energy sources according to claim 8, characterized in that, The formula for correcting the load shearing amount in step 3-3 is as follows: in, For the revised version The load shedding amount per cycle, This is the original load. For correction factor, This is the remaining power deficit re-estimated based on the real-time frequency response. This is the sum of the actual amount removed in the previous rounds.

Images