A dynamic hierarchical partitioning frequency control method

Abstract

The dynamic hierarchical partition frequency modulation control method of the application uses real-time operation data of the power grid to evaluate the system inertia, and simultaneously carries out N-1 fault analysis of the power grid, calculates the initial frequency change rate and the maximum frequency deviation value of the 500kV and above nodes of the power grid under N-1 fault, dynamically partitions according to the initial frequency change rate and the maximum frequency deviation value combined with the topology structure of the power grid, and divides into a low-inertia weak frequency modulation area, a low-inertia strong frequency modulation area, a high-inertia weak frequency modulation area and a high-inertia strong frequency modulation area. According to the partition result combined with the frequency adaptability, the transmission section transmission limit and the voltage level in the region, the frequency modulation control structure of each region is divided into an upper layer global optimization layer and a lower layer local execution layer, different frequency modulation parameters, optimization targets and execution objects are configured for different regions, the upper layer adopts MPC optimization secondary frequency modulation instruction, the lower layer adopts DMPC or PID controller to execute local unit power adjustment, and the frequency stability and the anti-disturbance ability of the whole system are improved, and the frequency collapse risk is inhibited.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power grid control, in particular to a dynamic hierarchical zoning frequency modulation control method. BACKGROUND

[0002] With large-scale access of new energy to the power grid, the power system presents the characteristics of "low inertia and high fluctuation", and the traditional power system frequency modulation control mode faces severe challenges. The existing technology mainly has the following problems: the current power system does not carry out effective zoning frequency modulation, and cannot reflect the inertia space distribution difference in real time. When a large disturbance occurs in the low inertia area, due to the homogenization of the frequency modulation control parameter setting, the low inertia area cannot maximize the use of its frequency modulation potential, cannot suppress its frequency fluctuation, and further causes power oscillation to spread to the high inertia area, eventually leading to system collapse. At the same time, the secondary frequency modulation control level coordination efficiency is low, the provincial power grid secondary frequency modulation adopts tie-line power frequency deviation control, which cannot effectively suppress the system power oscillation caused by the current source and load time and space distribution imbalance and system power electronic. The parameter configuration and the regional characteristics are mismatched, which aggravates the frequency oscillation risk. Different time scale frequency modulation functions (inertia response, primary frequency modulation, secondary frequency modulation) cannot effectively coordinate with each other, and there is a reverse regulation situation. New type of frequency modulation resources (network type new energy, flexible DC, energy storage / hybrid energy storage) cannot coordinate frequency modulation, which aggravates the system risk. In summary, the power system based on fixed topology zoning and homogenization control parameters aggravates the frequency instability risk of new type of power system. SUMMARY

[0003] In view of this, the present application provides a dynamic hierarchical zoning frequency modulation control method, which dynamically divides different frequency modulation areas, and constructs a hierarchical control architecture of an upper layer global optimization layer and a lower layer local execution layer, thereby reducing the frequency instability risk of the power system.

[0004] The technical scheme of the present application is as follows:

[0005] A dynamic hierarchical zoning frequency modulation control method, comprising the following steps:

[0006] Step S1, collecting real-time operation data of the power grid for inertia evaluation, and obtaining equivalent inertia;

[0007] Step S2, carrying out power grid N-1 fault analysis, and calculating the initial frequency change rate and the maximum frequency deviation value of the 500kV and above nodes of the power grid under N-1 fault;

[0008] Step S3, dynamically zoning based on the equivalent inertia, the initial frequency change rate and the maximum frequency deviation value in combination with the power grid topology structure, and obtaining a plurality of regions;

[0009] Step S4, dividing the frequency regulation structure of the region into an upper global optimization layer and a lower local execution layer according to the frequency adaptability, the transmission limit of the transmission section and the voltage level;

[0010] Step S5, configuring different parameters, optimization targets and execution objects for different regions;

[0011] Step S6, the upper global optimization layer generates the total frequency regulation power demand instruction of the region and the safety constraint set of the key section / node in the region to the lower local execution layer by using the centralized model predictive control, and the lower local execution layer coordinates the local execution unit by using the distributed model predictive control or the coordinated controller with constraint optimization.

[0012] Preferably, the specific steps of step S1 include:

[0013] Step S11, collecting the real-time operation data of the power grid provided by the wide-area measurement system or the advanced measurement system;

[0014] Step S12, taking the 500kV and above nodes or stations as the basic unit, calculating the equivalent inertia based on the generator rotor motion equation or the system frequency response model by using the real-time operation data of the power grid;

[0015] Step S13, improving the calculation accuracy of the equivalent inertia by using the state estimation or the data-driven method.

[0016] Preferably, the specific steps of step S2 include: simulating and calculating the dynamic response of all 500kV and above nodes of the power grid after N-1 opening fault by using the real-time topology and operation state, based on the power system simulation software or the online fast simulation module, and calculating the initial frequency change rate and the maximum frequency deviation of the whole network under each fault scenario.

[0017] Preferably, the specific steps of step S3 include:

[0018] Step S31, setting the inertia threshold and the frequency vulnerability threshold;

[0019] Step S32, when the equivalent inertia is less than the inertia threshold and the initial frequency change rate or the maximum frequency deviation value is greater than the frequency vulnerability threshold, dividing into the low-inertia weak frequency regulation area;

[0020] Step S33, when the equivalent inertia is less than the inertia threshold and the initial frequency change rate and the maximum frequency deviation value are less than or equal to the frequency vulnerability threshold, dividing into the low-inertia strong frequency regulation area;

[0021] Step S34, when the equivalent inertia is greater than or equal to the inertia threshold and the initial frequency change rate or the maximum frequency deviation value is greater than the frequency vulnerability threshold, dividing into the high-inertia weak frequency regulation area;

[0022] Step S35, when the equivalent inertia is greater than or equal to the inertia threshold value, and the initial frequency change rate and the maximum frequency deviation value are less than or equal to the frequency vulnerability threshold value, the high-inertia emphasis frequency area is divided.

[0023] Preferably, the specific step of step S4 comprises: dividing the frequency modulation control structure of the area into an upper global optimization layer and a lower local execution layer based on the voltage level after the frequency adaptability of each area is utilized, the transmission section transmission limit constraint is complied with, and efficient decoupling is achieved.

[0024] Preferably, the specific step of step S5 of configuring different parameters for different areas comprises:

[0025] Step S51, configuring a virtual inertia constant and a damping coefficient higher than those of other areas in the low-inertia weak modulation frequency area;

[0026] Step S52, reducing the dead zone and increasing the modulation difference coefficient slope in the low-inertia weak modulation frequency area and the low-inertia emphasis frequency area, and adopting relatively conservative parameters in the high-inertia weak modulation frequency area and the high-inertia emphasis frequency area;

[0027] Step S53, in the upper global optimization layer, setting different target function weights and constraint condition tightness.

[0028] Preferably, the specific step of step S5 of configuring different optimization targets for different areas comprises:

[0029] Step S54, in the low-inertia weak modulation frequency area and the low-inertia emphasis frequency area, the primary goal is to quickly suppress the frequency deviation of the area center / key node, and the local fast modulation resource is maximized to prevent frequency collapse diffusion, and the secondary goal is to meet the area power balance and tie line planning;

[0030] Step S55, in the high-inertia weak modulation frequency area and the high-inertia emphasis frequency area, the power instruction of the low-inertia area is optimized under the premise of meeting the area power balance and safety, and the target includes minimizing the system frequency deviation, controlling the tie line power deviation, and economy.

[0031] Preferably, the specific step of step S5 of configuring different execution objects for different areas comprises:

[0032] Step S56, in the low-inertia weak modulation frequency area, the network-type new energy and energy storage are preferentially called to provide inertia response and fast support of the primary frequency modulation level;

[0033] Step S57, in the low-inertia emphasis frequency area and the high-inertia weak modulation frequency area, the energy storage, network-type new energy, and traditional units are coordinated to call, and the connection between the primary frequency modulation and the secondary frequency modulation is focused on.

[0034] Step S58, in the high-inertia frequency emphasis zone, the secondary frequency modulation is borne by the conventional unit group, and the new resource is used as a supplement or backup.

[0035] Preferably, the upper global optimization layer establishes a prediction model including a regional frequency dynamic model, a unit / resource regulation characteristic model, and a power grid safety constraint model, the input of the prediction model is a regional real-time frequency deviation, a tie-line power deviation, a regional available frequency modulation capacity, a key section transmission margin, a voltage out-of-limit information, and a state feedback from the lower local execution layer, and the output is an optimized regional total frequency modulation power demand instruction and a safety constraint set of a key section / node in the region.

[0036] Preferably, the local execution unit includes a conventional thermal / water power unit, a grid-constructed new energy power station, an energy storage power station, and a flexible HVDC converter station.

[0037] Compared with the prior art, the application has the following beneficial effects:

[0038] ①By real-time sensing of the power grid inertia distribution and the fault frequency vulnerability, accurate dynamic zoning is performed, and frequency modulation parameters are configured differently in different regions, so that the low-inertia region can maximize the excavation and release of its frequency modulation potential.

[0039] ②A hierarchical architecture of the upper global optimization layer and the lower local execution layer is constructed, and a closed loop is formed between the upper layer and the lower layer through information interaction, wherein the upper layer can set different optimization objectives according to the dynamic characteristics and safety requirements of different regions, and perform secondary frequency modulation instruction optimization under multiple constraints, and the lower layer realizes rapid and accurate allocation and execution of the instruction among heterogeneous resources. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only preferred embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0041] Fig. 1 A flowchart of a dynamic hierarchical zoning frequency modulation control method of the application;

[0042] Fig. 2 A flowchart of step S1 of a dynamic hierarchical zoning frequency modulation control method of the application;

[0043] Fig. 3 A flowchart of step S3 of a dynamic hierarchical zoning frequency modulation control method of the application;

[0044] Fig. 4 A flowchart of step S5 of a dynamic hierarchical zoning frequency modulation control method of the application. DETAILED DESCRIPTION

[0045] In order to better understand the technical content of the present application, a specific embodiment is provided below, and the present application is further described in conjunction with the drawings.

[0046] Referring to Figs. 1 to 4 The dynamic hierarchical partitioning frequency control method provided by the present application comprises the following steps:

[0047] Step S1, collecting real-time operation data of the power grid for inertia evaluation to obtain equivalent inertia; the specific steps comprise:

[0048] Step S11, collecting real-time operation data of the power grid provided by the wide-area measurement system or the advanced measurement system, wherein the real-time operation data of the power grid comprises data such as generator speed / power, new energy output, load, key node voltage / frequency, network topology, etc.

[0049] Step S12, taking 500kV and above nodes or stations as basic units, and based on the generator rotor motion equation or the system frequency response model, the equivalent inertia is calculated by using the real-time operation data of the power grid (for example, the generator output rate of change, the system frequency rate of change);

[0050] Step S13, using state estimation or data-driven methods (Kalman filtering, machine learning) to improve the calculation accuracy of the equivalent inertia.

[0051] Step S2, performing N-1 fault analysis of the power grid, and calculating the initial frequency change rate and the maximum frequency deviation value of the 500kV and above nodes of the power grid under N-1 fault; the specific steps comprise: using the real-time topology and operation state, based on the power system simulation software (BPA, PSASP, PSS / E) or the online fast simulation module, simulating and calculating the dynamic response of all 500kV and above nodes of the power grid after N-1 opening fault, and calculating the initial frequency change rate and the maximum frequency deviation of the whole network under each fault scenario, focusing on the first swing dynamic after the fault occurs; after obtaining the equivalent inertia, the initial frequency change rate and the maximum frequency deviation of the whole network, the dynamic partitioning can be performed.

[0052] Step S3, performing dynamic partitioning based on the equivalent inertia, the initial frequency change rate and the maximum frequency deviation value in combination with the topology structure of the power grid to obtain a plurality of regions; the specific steps comprise:

[0053] Step S31, setting an inertia threshold and a frequency vulnerability threshold;

[0054] Step S32, when the equivalent inertia is less than the inertia threshold, and the initial frequency change rate or the maximum frequency deviation value is greater than the frequency vulnerability threshold, dividing into a low-inertia weak frequency modulation region (LLR), wherein the low-inertia weak frequency modulation region has low inertia and serious frequency deterioration after disturbance.

[0055] Step S33, when the equivalent inertia is less than the inertia threshold value, and the initial frequency change rate and the maximum frequency deviation value are less than or equal to the frequency vulnerability threshold value, the area is divided into a low-inertia emphasis frequency area (LHR), the low-inertia emphasis frequency area has low inertia, and only has strong local frequency modulation capability or a favorable topology;

[0056] Step S34, when the equivalent inertia is greater than or equal to the inertia threshold value, and the initial frequency change rate or the maximum frequency deviation value is greater than the frequency vulnerability threshold value, the area is divided into a high-inertia weak frequency modulation area (HLR), the low-inertia emphasis frequency area has high inertia, but the topology or frequency modulation resource distribution causes local vulnerability;

[0057] Step S35, when the equivalent inertia is greater than or equal to the inertia threshold value, and the initial frequency change rate and the maximum frequency deviation value are less than or equal to the frequency vulnerability threshold value, the area is divided into a high-inertia emphasis frequency area (HHR), the high-inertia emphasis frequency area has high inertia and strong anti-interference capability.

[0058] The partition logic is to compare the equivalent inertia, the initial frequency change rate and the maximum frequency deviation with the preset inertia threshold value and the frequency vulnerability threshold value, wherein the frequency vulnerability threshold value includes the threshold values corresponding to the initial frequency change rate and the maximum frequency deviation, and four areas can be divided according to different comparison results, and the partition result is not fixed, and is periodically (such as several minutes) or event triggered (such as after a large disturbance) updated according to the real-time evaluation results of steps S1 and S2, the power grid operation mode, the new energy output, the load change, the network topology change (after planned maintenance or fault), the spot transaction cycle and the like.

[0059] Step S4, the frequency modulation control structure of the area is divided into an upper layer global optimization layer and a lower layer local execution layer in combination with the frequency adaptability, the transmission limit of the transmission section and the voltage level; the specific steps are as follows: the frequency modulation control structure of the area is divided into the upper layer global optimization layer and the lower layer local execution layer after the frequency adaptability of each area is utilized, the transmission limit of the transmission section is constrained, and efficient decoupling is realized based on the voltage level, wherein the frequency adaptability drives the parameter differential configuration of the control layers of different areas, the transmission limit of the transmission section constrains the action range of the local execution layer, and the voltage level difference determines the decoupling mode of the upper and lower layers.

[0060] The function of the upper layer global optimization layer is to be responsible for the centralized optimization calculation of the secondary frequency modulation instruction of the area, the control period is second-level or minute-level, and the core technology is to adopt centralized model predictive control (MPC), and the function of the lower layer local execution layer is to receive the upper layer instruction, quickly and accurately distribute the instruction to each frequency modulation execution unit (conventional power supply, new energy, energy storage and flexible straight) in the area, and perform local closed-loop control, and the control period is millisecond-level to second-level.

[0061] Step S5, configuring different parameters, optimization objectives and execution objects for different regions;

[0062] The specific steps of configuring different parameters for different regions include:

[0063] Step S51, setting virtual inertia constant and virtual coefficient matching the dynamic characteristics of the region for grid-forming new energy, virtual synchronous machine, etc., for example, configuring higher virtual inertia constant and damping coefficient in the low inertia weak frequency modulation area than in other areas to simulate stronger "inertia wall" effect and suppress initial frequency change rate;

[0064] Step S52, differentially setting primary frequency modulation dead zone, modulation difference coefficient (R) and power limit of various types of units / resources, for example, reducing the dead zone and increasing the modulation difference coefficient slope in the low inertia weak frequency modulation area and the low inertia strong frequency modulation area to make the local resources more sensitive to small frequency changes and have larger action range to fully release the frequency modulation potential, or using relatively conservative parameters in the high inertia weak frequency modulation area and the high inertia strong frequency modulation area, or flexibly adjusting according to inter-regional support demand;

[0065] Step S53, in the upper global optimization layer, setting different target function weights and constraint condition tightness to realize parameter differentiation.

[0066] The specific steps of configuring different optimization objectives for different regions include:

[0067] Step S54, in the low inertia weak frequency modulation area and the low inertia strong frequency modulation area, the primary objective is to quickly suppress the frequency deviation of the regional center / key node, maximize the use of local fast frequency modulation resources (such as energy storage and grid-forming new energy) to prevent frequency collapse and diffusion, and the secondary objective is to meet regional power balance and tie-line planning;

[0068] Step S55, in the high inertia weak frequency modulation area and the high inertia strong frequency modulation area, more emphasis is placed on inter-regional power support coordination, and under the premise of meeting regional power balance and safety, the power instruction for supporting the low inertia area is optimized, and the target includes minimizing the system frequency deviation, tie-line power deviation control (TBC or AGC standard mode) and economy (considering adjustment cost).

[0069] The specific steps of configuring different execution objects for different regions include:

[0070] Step S56, in the low inertia weak frequency modulation area, grid-forming new energy and energy storage (especially power type) are preferentially called to provide inertia response and fast support of primary frequency modulation level, and the local execution layer needs to have super-fast response capability;

[0071] Step S57, in the low inertia strong frequency modulation area and the high inertia weak frequency modulation area, energy storage (energy type / mixed), grid-forming new energy and traditional units are coordinated to call, and emphasis is placed on the connection of primary frequency modulation and secondary frequency modulation.

[0072] Step S58, in the high inertia frequency area, the secondary frequency modulation is borne by the conventional unit, and the new resource is supplemented or reserved.

[0073] By real-time sensing of the inertia distribution and fault frequency vulnerability of the power grid, accurate dynamic partitioning is performed, and frequency modulation parameters (such as reducing the dead zone, increasing the modulation difference coefficient slope, and enhancing the virtual inertia) are differentially configured in different regions (especially in the low inertia weak frequency modulation area LLR), so that the low inertia area can maximize the excavation and release of its frequency modulation potential (especially local fast resources such as energy storage and grid-forming new energy).

[0074] A hierarchical architecture of an upper global optimization layer and a lower local execution layer is constructed, the upper centralized MPC sets different optimization objectives according to the dynamic characteristics and safety requirements of different regions (such as the LLR area prioritizing frequency and the HHR area focusing on economic support), and performs secondary frequency modulation instruction optimization under multiple constraints (section, voltage, etc.); the lower distributed MPC / PID realizes fast and accurate allocation and execution of instructions among heterogeneous resources.

[0075] The hierarchical architecture of the upper global optimization layer and the lower local execution layer naturally decouples control tasks of different time scales, the lower local execution layer focuses on the response of local fast resources (such as energy storage and grid-forming converters) in the extremely fast time scale (inertia response and primary frequency modulation), and the upper global optimization layer is responsible for centralized optimization and coordination in the second-minute level (secondary frequency modulation). Through differential parameter configuration (such as local resources in the LLR area focusing on fast response) and optimization objective setting, the action direction of inertia response, primary frequency modulation, and secondary frequency modulation is consistent, the intensity is complementary, and the time sequence is connected.

[0076] Step S6, the upper global optimization layer generates regional total frequency modulation power demand instructions and a safety constraint set of key sections / nodes in the region to the lower local execution layer using centralized model predictive control, and the lower local execution layer uses a distributed model predictive control or a coordinated controller with constraint optimization to coordinate local execution units, which include traditional thermal / water power units, grid-forming new energy power stations, energy storage power stations, and flexible HVDC converter stations.

[0077] Preferably, the upper global optimization layer establishes a prediction model including a regional frequency dynamic model, a unit / resource regulation characteristic model, and a power grid safety constraint model (section flow inequality constraint, voltage upper and lower limit constraint), the input of the prediction model is the regional real-time frequency deviation, the tie-line power deviation, the regional available frequency modulation capacity (including the state of traditional units, new energy, and energy storage), the key section transmission margin, the voltage out-of-limit information, and the state feedback from the lower local execution layer, and the output is the optimized regional total frequency modulation power demand instruction and the safety constraint set of key sections / nodes in the region for the lower layer to refer, and the coordination mechanism between the upper layer and the lower layer is:

[0078] Upper layer global optimization layer: directly consider the dynamic characteristics and cost of various resources in the optimization model, realize the economic-security coordinated optimization across resource types.

[0079] Lower layer local execution layer: optimal combination and distribution according to resource characteristics (such as energy storage responding fastest to lead the way, thermal power providing sustained support).

[0080] The lower layer local execution layer adopts a distributed model predictive control (DMPC) or a coordinated controller with constraint optimization for scenarios requiring coordination of multiple local execution units (such as multiple units in a station or hybrid energy storage), optimally distributes according to the dynamic characteristics (response speed, adjustment cost, SOC state) of each unit under the premise of meeting the total frequency modulation power demand instruction of the upper layer domain and the safety constraint set of key sections / nodes in the region, and adopts a high-performance PID controller or direct power control for single local execution units or scenarios requiring extremely fast response (such as single energy storage or grid-forming converter) to achieve millisecond-level accurate tracking.

[0081] The traditional units, grid-forming new energy, energy storage (hybrid energy storage), and flexible DC are seamlessly integrated into a unified control framework. The upper layer global optimization layer considers the dynamic characteristics and cost of various resources during optimization; the lower layer local execution layer optimally combines and distributes according to the resource characteristics (speed, cost, state) (such as energy storage responding fastest to lead the way, and traditional units providing sustained support).

[0082] The dynamic hierarchical zoning frequency modulation control method of the application can accurately match the regional dynamic characteristics, differentially configure frequency modulation parameters, fully release the frequency modulation potential of low inertia regions, and effectively suppress frequency fluctuation diffusion through real-time evaluation of power grid inertia distribution and fault frequency characteristics, dynamic division of differentiated frequency modulation regions, and construction of a hierarchical control architecture of "upper layer global optimization layer + lower layer local execution layer". The hierarchical collaborative mechanism eliminates multi-time scale frequency modulation conflicts, significantly improves the secondary frequency modulation coordination efficiency, seamlessly integrates grid-forming new energy, energy storage, and other heterogeneous resources for collaborative participation, synchronously embeds power grid safety constraints, systematically enhances the frequency stability and disturbance resistance of high-proportion new energy power grids, and realizes frequency rapid recovery and safe and stable operation.

[0083] The above only describes the preferred embodiments of the application and is not intended to limit the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the application shall be included in the protection scope of the application.

Claims

1. A dynamic hierarchical zoned frequency control method, characterized by, The method comprises the following steps: Step S1, collecting real-time operation data of the power grid to perform inertia evaluation, and obtaining equivalent inertia; Step S2, performing N-1 fault analysis of the power grid, and calculating initial frequency change rate and maximum frequency deviation of 500kV and above nodes of the power grid under N-1 fault; Step S3, performing dynamic partitioning based on equivalent inertia, initial frequency change rate and maximum frequency deviation in combination with the topology structure of the power grid, and obtaining a plurality of regions; Step S4, dividing the frequency modulation control structure of the region into an upper global optimization layer and a lower local execution layer in combination with frequency adaptability, transmission section transmission limit and voltage level; Step S5, configuring different parameters, optimization objectives and execution objects for different regions; Step S6, the upper global optimization layer adopts centralized model predictive control technology to generate total frequency modulation power demand instructions and safety constraint sets of key sections / nodes in the region, and sends them to the lower local execution layer, and the lower local execution layer adopts distributed model predictive control or a coordination controller with constraint optimization to coordinate local execution units; The specific steps of step S4 are as follows: after the frequency adaptability of each region is utilized, the transmission section transmission limit constraint is followed, and efficient decoupling is realized based on the voltage level, the frequency modulation control structure of the region is divided into an upper global optimization layer and a lower local execution layer; The local execution unit comprises a thermal / water turbine unit, a network-structured new energy power station, an energy storage power station and a flexible HVDC converter station.

2. The dynamic hierarchical zoned FM control method of claim 1, wherein, The specific steps of step S1 comprise: Step S11, collecting real-time operation data of the power grid provided by a wide area measurement system or an advanced measurement system; Step S12, taking 500kV and above nodes or stations as basic units, and calculating equivalent inertia based on a generator rotor motion equation or a system frequency response model and using real-time operation data of the power grid; Step S13, using state estimation or a data-driven method to improve the calculation accuracy of equivalent inertia.

3. The method of claim 1, wherein, The specific steps of step S2 comprise: using real-time topology and operation state, simulating and calculating the dynamic response of all 500kV and above nodes of the power grid after N-1 opening fault based on a power system simulation software or an online fast simulation module, and calculating the initial frequency change rate and the maximum frequency deviation of the whole network under each fault scenario.

4. The method of claim 1, wherein, The specific steps of step S3 comprise: Step S31, setting an inertia threshold and a frequency vulnerability threshold; Step S32, when the equivalent inertia is less than the inertia threshold and the initial frequency change rate or the maximum frequency deviation is greater than the frequency vulnerability threshold, dividing into a low-inertia weak frequency modulation region; Step S33, when the equivalent inertia is less than the inertia threshold and the initial frequency change rate and the maximum frequency deviation are less than or equal to the frequency vulnerability threshold, dividing into a low-inertia strong frequency modulation region; Step S34, when the equivalent inertia is greater than or equal to the inertia threshold and the initial frequency change rate or the maximum frequency deviation is greater than the frequency vulnerability threshold, dividing into a high-inertia weak frequency modulation region; Step S35, when the equivalent inertia is greater than or equal to the inertia threshold and the initial frequency change rate and the maximum frequency deviation are less than or equal to the frequency vulnerability threshold, dividing into a high-inertia strong frequency modulation region.

5. The method of claim 4, wherein, The specific steps of configuring different parameters for different areas in step S5 include: Step S51, configuring higher virtual inertia constant and damping coefficient in the low-inertia weak frequency modulation area than in other areas; Step S52, reducing the dead zone and increasing the modulation difference coefficient slope in the low-inertia weak frequency modulation area and the low-inertia strong frequency modulation area, and adopting relatively conservative parameters in the high-inertia weak frequency modulation area and the high-inertia strong frequency modulation area; Step S53, in the upper global optimization layer, setting different target function weights and constraint condition tightness.

6. The method of claim 4, wherein, The specific steps of configuring different optimization targets for different areas in step S5 include: Step S54, in the low-inertia weak frequency modulation area and the low-inertia strong frequency modulation area, the primary goal is to quickly suppress the frequency deviation of the area center / key node, and to maximize the use of local fast frequency modulation resources to prevent frequency collapse diffusion, and the secondary goal is to meet the regional power balance and tie line plan; Step S55, in the high-inertia weak frequency modulation area and the high-inertia strong frequency modulation area, under the premise of meeting the regional power balance and safety, optimizing the power command of the low-inertia area, the target includes minimizing the system frequency deviation, tie line power deviation control, and economy.

7. The method of claim 4, wherein, The specific steps of configuring different execution objects for different areas in step S5 include: Step S56, in the low-inertia weak frequency modulation area, preferentially calling new energy and energy storage to provide inertia response and primary frequency modulation level fast support; Step S57, in the low-inertia strong frequency modulation area and the high-inertia weak frequency modulation area, coordinating the call of energy storage, grid-connected new energy, and traditional units, focusing on the connection of primary frequency modulation and secondary frequency modulation; Step S58, in the high-inertia strong frequency modulation area, traditional units bear secondary frequency modulation, and new resources are used as supplements or backups.

8. The method of claim 1, wherein, The upper global optimization layer establishes a prediction model including a regional frequency dynamic model, a unit / resource regulation characteristic model, and a power grid safety constraint model, the input of the prediction model is the regional real-time frequency deviation, the tie line power deviation, the regional available frequency modulation capacity, the key section transmission margin, the voltage out-of-limit information, and the state feedback from the lower local execution layer, and the output is the optimized regional total frequency modulation power demand instruction and the safety constraint set of the key section / node in the region.

Images