New energy intelligent anti-disturbance system and control method

Abstract

The invention discloses a new energy intelligent anti-disturbance system which comprises a new energy unit, a virtual synchronous machine-broadband impedance remodeling control system, a high-frequency generator tripping anti-disturbance execution unit and a communication cooperative control platform. The virtual synchronous machine-broadband impedance remodeling control system interactively fuses a virtual synchronous machine control technology and a broadband impedance remodeling technology, and a virtual synchronous machine actively performs inertia compensation and damping adjustment collaborative broadband impedance optimization collaborative fusion. The new energy unit and the virtual synchronous machine-broadband impedance remodeling control system share data and are in smooth connection, a high-frequency generator tripping anti-disturbance execution unit is optimized through interactive fusion, data are interactively transmitted to a communication cooperative control platform, and dynamic cooperation and accurate control of the system are achieved. According to the method, the efficient anti-disturbance performance under the new energy high-frequency switching scene can be improved, stable operation of a new energy grid-connected power system is guaranteed, and key support and decision are provided for safe operation of a high-proportion new energy investment application complex power grid scene.

Description

Technical Field

[0001] The present invention relates to the technical field of new power systems and new energy grid connection, and particularly to a new energy intelligent anti-disturbance system and a control method thereof. Background Art

[0002] As an important part of the technical field of new power systems and new energy grid connection, the new energy high-frequency generator tripping disturbance control system is facing a series of key challenges and problems to be solved. With the continuous increase in the penetration rate of new energy sources such as wind power and photovoltaic power, traditional synchronous generator sets have been largely replaced, resulting in a significant decline in the equivalent inertia and primary frequency regulation ability of the power grid, posing a potential threat to system stability; specifically, the inherent low-inertia characteristics of new energy units and their insufficient adaptability under dynamic conditions make it difficult for traditional impedance reshaping technologies to cover the full frequency band (such as 20 Hz to 1 kHz), further exacerbating the risk of broadband oscillation of the power grid; in addition, new energy power stations are widely distributed and the control subjects are scattered. Existing collaborative control and communication architectures have problems such as high latency and insufficient standardization, making it difficult to meet the millisecond-level collaborative requirements, resulting in an increasingly prominent contradiction between the "low-inertia - high-fluctuation - weak anti-disturbance" characteristics and the traditional control system; at the same time, new energy power stations are limited by the limited profit space, and the promotion of technologies faces economic constraints, leading to a decline in resource utilization rate. For example, the curtailment rate of wind and light once exceeded 8%, and the economic benefits have been significantly challenged. Summary of the Invention

[0003] Object of the Invention: To overcome the deficiencies of the background art, the first object of the present invention is to disclose a new energy intelligent anti-disturbance system;

[0004] The second object is to disclose an anti-disturbance control method based on the above new energy intelligent anti-disturbance system.

[0005] Technical Solution: The new energy intelligent anti-disturbance system disclosed by the present invention includes new energy units, a virtual synchronous machine - broadband impedance reshaping control system, a high-frequency generator tripping anti-disturbance execution unit, and a communication collaborative control platform;

[0006] The virtual synchronous machine - broadband impedance reshaping control system is an interactive fusion of virtual synchronous machine control technology and broadband impedance reshaping technology. The virtual synchronous machine actively performs inertia compensation and damping regulation to synergistically optimize the broadband impedance.

[0007] The new energy units share data with the virtual synchronous machine - broadband impedance reshaping control system, are smoothly connected, and optimize the high-frequency generator tripping anti-disturbance execution unit through interactive fusion, and transmit the data interaction to the communication collaborative control platform to achieve system dynamic collaboration and precise control.

[0008] Furthermore, the new energy generator sets include wind turbine generator sets, solar photovoltaic generator sets, hydropower generator sets, biomass generator sets, gravel energy storage generator sets and tidal energy generator sets.

[0009] Furthermore, the high-frequency cutting and anti-disturbance execution unit includes a multi-source collaborative cutting execution module, an energy storage unit fast response module, a virtual impedance dynamic adjustment module and a grid-connected inverter control module, which dynamically and collaboratively optimizes them through virtual synchronous machines and wide-screen impedance reshaping to form a hierarchical control instruction chain, including a perception layer, a decision layer and an execution layer.

[0010] Furthermore, the communication collaborative control platform and the high-frequency cutting anti-disturbance execution unit coordinate and optimize the target orientation through dynamic data interaction and collaborative strategy, thereby achieving grid frequency stability control, optimized operation of new energy power stations, new energy cluster management, and fault diagnosis and early warning.

[0011] Correspondingly, a new energy intelligent anti-disturbance control method is disclosed. Based on the above-mentioned new energy intelligent anti-disturbance system, the core parameters of the system are optimized through the MOGA intelligent optimization algorithm to improve the anti-disturbance performance of the new energy grid-connected high-frequency cutting generator, including the following steps:

[0012] S1. Establish an objective function to measure the overall anti-disturbance performance index of the system;

[0013] S2, the MOGA intelligent optimization algorithm based on the NSGA-III architecture processes the high-dimensional parameter space through non-dominated sorting, and optimizes the core parameter combination of high-frequency cutting to achieve the best anti-disturbance performance of the high-frequency cutting of new energy grid-connected new energy.

[0014] Furthermore, S1 is specifically:

[0015] Based on the anti-disturbance performance in the high-frequency cutting scenario of new energy, the dynamic response, impedance characteristics and core parameters of system stability are collaboratively optimized, and the objective function F is established as an indicator to measure the overall anti-disturbance performance of the system. The smaller the value, the better the performance:

[0016]

[0017] Where: It contains four core items, corresponding to different performance dimensions;

[0018] Among them, α1·∫Δfdt is the dynamic response term, the frequency deviation integral evaluates the inertia support effect of the dynamic response capability of the virtual synchronous machine frequency, and the core parameter is the virtual inertia J VSG , damping coefficient D VSG ;

[0019] α2·ΔZ(f) is the impedance mismatch term, which reflects the ability of impedance reshaping technology to suppress broadband oscillation; it measures the actual impedance Z of the new energy unit. meas(f) Deviation sum with respect to the ideal impedance Z ref (f) Deviation sum in the high-frequency band of 10 - 1000 Hz. The key parameter is the impedance reshaping bandwidth BW Z ;

[0020] α3·P trip is the generator tripping trigger probability term, reflecting the system's anti-disturbance ability. The lower the value, the better the suppression effect of the coordinated control strategy on high-frequency generator tripping events

[0021]

[0022] In the formula: N trip Number of generator tripping triggers in simulation or actual measurement. N total is the total number of disturbances;

[0023] α4·σ P is the power volatility, characterizing the smoothness of the output power of new energy units. The standard deviation form quantifies the fluctuation amplitude. The smaller the value, the better the transient stability of the system

[0024]

[0025] In the formula: is the rated power, and T is the observation time window.

[0026] Furthermore, S2 specifically includes the following steps:

[0027] S2-1. Encode the parameters and initialize the population to generate the initial population;

[0028] For the continuous parameters J VSG , D VSG Use 16-bit binary encoding with an accuracy of 0.01. Use integer encoding for the discrete parameter control mode flag bit for VSG droop control switching; randomly initialize individuals to ensure that the constraint BW Z ≥500 Hz and introduce Latin hypercube sampling (LHS) to enhance the diversity of the initial solutions, generating the initial population;

[0029] S2-2. Apply multi-objective fitness evaluation to the objective function to calculate the optimal fitness value;

[0030] The population dynamically moves in the determined search space to find the position corresponding to the optimal fitness value as the best solution;

[0031]

[0032] S2-3. Use non-dominated sorting and reference point associated Pareto stratification and adopt the Das-Dennis method to improve and optimize the algorithm;Using non-dominated sorting and reference point associated Pareto stratification, the population is divided into levels such as Front1 optimal front and Front2 through fast non-dominated sorting, and the crowding distance is calculated to ensure the distribution of the front solutions;

[0033] Adopt the Das-Dennis method to uniformly generate 120 reference points in the three-dimensional objective space, dynamically adjust the reference point density, and increase the number of reference points in the sparse area of the solution set; Normalize the objective function values to the interval [0, 1], calculate the vertical distance from each individual to the reference point, and associate it with the nearest reference point;

[0034] S2-4. Use genetic operation to simulate binary crossover method;

[0035] Randomly select 2 individuals from those associated with each reference point, and retain the one with higher fitness;

[0036] Simulated binary crossover:

[0037] Offspring = 0.5[(1 + β)·Parent1 + (1 - β)·Parent2], β ~ U(0, 1)

[0038] In the formula: the offspring is the new individual generated from the parent individuals, Parent1 and Parent2 are the two parent individuals participating in the crossover operation, β is the crossover distribution index, and 0.5 is the normalization coefficient;

[0039] Among them, the crossover probability is p c = 0.9, and the distribution index η c = 20 controls the dispersion degree of the offspring;

[0040] Utilize polynomial mutation:

[0041] x′ = x + δ·(x U - x L ),

[0042] In the formula: x′ is the new individual after mutation, x is the original individual, δ is the mutation step size, x U is the upper bound of the individual, x L is the lower bound of the individual, r is a random number with a uniform distribution, and η m is the mutation distribution index;

[0043] Among them, the mutation probability p m = 0.1, and the distribution index η m = 20 controls the perturbation amplitude.

[0044] S2-5. Update the algorithm, introduce adaptive weights, adopt subjective weighting, use the Analytic Hierarchy Process (AHP) to construct a hierarchical model and a judgment matrix;

[0045] The objective function is decomposed into a criterion layer and an index layer. The criterion layer includes dynamic response and impedance characteristics, and the index layer includes frequency deviation and generator tripping probability;

[0046] Quantify the relative importance between indicators through a scoring method, combine objective weighting, improve the entropy weight method, and perform range normalization on time series data:

[0047]

[0048] In the formula: x′ ij is the normalized data value, x ij is the original data value, min x j is the minimum value of the j-th feature, max x j is the maximum value of the j-th feature;

[0049] Introduce a time decay factor λ, with higher weights for recent data:

[0050]

[0051] In the formula: is the objective weight of the j-th indicator, E j is the entropy value of the j-th indicator, Σ(1 - E j ) is the sum of all indicators (1 - E j ), and λ T-t is the time decay factor;

[0052] Calculate the consistency ratio CR. If CR < 0.1, it passes the test; otherwise, readjust the scale;

[0053] S2 - 6. Verify the termination condition. If the iteration count is reached or the termination condition is met, terminate the iteration; otherwise, go to S2 - 2;

[0054] S2 - 7. Obtain the optimal Pareto solution set corresponding to the optimal fitness value, that is, the core parameters corresponding to the best anti - disturbance performance, so as to achieve the best anti - disturbance performance of new energy grid connection.

[0055] Beneficial effects: Compared with the prior art, the advantages of the present invention are:

[0056] The collaborative application of virtual synchronous generator (VSG) technology and broadband impedance reshaping technology significantly improves the performance of the anti-disturbance control system for high-frequency generator tripping in new energy sources, bringing multiple benefits to grid stability and new energy utilization efficiency. VSG simulates the inertia and damping characteristics of synchronous generators to provide transient frequency support, significantly reducing the probability of high-frequency generator tripping. VSG dynamically adjusts the output of new energy units, reduces power volatility, enhances the transient stability of the system, and avoids unnecessary unit tripping. The analytic hierarchy process and improved entropy weight method are used to dynamically adjust the weights of the objective function, adapt to different disturbance types, enhance the robustness of the system, and improve the dynamic response ability of the system. By optimizing the impedance reshaping bandwidth to cover the entire frequency band from 10 Hz to 1 kHz, the impedance mismatch term is significantly reduced, the risk of sub- / supra-synchronous oscillation is suppressed, and high-frequency generator tripping events are reduced. Dynamically adjusting the impedance characteristics of the new energy unit ports improves the anti-disturbance ability of the system and reduces the tripping trigger frequency. By dynamically adjusting the power volatility, the smoothness of the output power of new energy units is improved, and the transient stability of the system is enhanced; simulated binary crossover and polynomial mutation are introduced to enhance population diversity, improve the algorithm convergence speed and global search ability; through the time decay factor and consistency ratio test, the weights of the objective function and the termination conditions are dynamically adjusted to ensure the adaptability and reliability of the algorithm, while enhancing the reliability of the entire power grid. Description of the Drawings

[0057] Figure 1 Schematic diagram of the system structure of the present invention;

[0058] Figure 2 Flowchart of the method of the present invention. Detailed Embodiments

[0059] As Figure 1 shown, the new energy intelligent anti-disturbance system includes new energy units, a virtual synchronous generator - broadband impedance reshaping control system, a high-frequency generator tripping anti-disturbance execution unit, and a communication collaborative control platform;

[0060] The new energy units include wind power generation units, solar photovoltaic power generation units, hydraulic power generation units, biomass power generation units, gravel energy storage units, and tidal energy power generation units. They become the core power sources for "active support", realizing the substitution of clean energy and low-carbon transformation, and are the key supports for building a new type of power system.

[0061] The virtual synchronous generator - broadband impedance reshaping control system is an interactive integration of virtual synchronous generator (VSG) control technology and broadband impedance reshaping technology. The virtual synchronous generator actively performs inertia compensation and damping regulation, and collaboratively optimizes broadband impedance. Through parameter dynamic coupling, it realizes "linkage regulation - collaborative design - hierarchical response", improving broadband stability and anti-disturbance ability.

[0062] The high-frequency generator cutting and anti-disturbance execution unit includes a multi-source coordinated generator cutting execution module, an energy storage unit fast response module, a virtual impedance dynamic adjustment module and a grid-connected inverter control module. It dynamically and collaboratively optimizes them through virtual synchronous machines and wide-screen impedance reshaping to form a hierarchical control instruction chain, including a perception layer, a decision layer and an execution layer. Through the three core links of the "perception-decision-execution" layer, a closed-loop control architecture is formed to realize a disturbance warning mechanism and active anti-disturbance control.

[0063] The communication collaborative control platform and the high-frequency generator switching anti-disturbance execution unit achieve goal-oriented optimization through dynamic data interaction and coordinated strategy coordination, achieving grid frequency stability control, optimized operation of new energy power plants, new energy cluster management, and fault diagnosis and early warning. This improves the grid's frequency anti-disturbance capabilities, utilizes dynamic optimization algorithms and intelligent decision-making to enhance system adaptability, achieve new energy optimization and large-scale new energy cluster management, and dynamically update strategies to form a "monitoring-diagnosis-optimization" closed loop, thereby dynamically adjusting fault diagnosis and early warning for new energy grid integration.

[0064] The new energy unit and the virtual synchronous machine-wideband impedance reshaping control system share data and are smoothly connected. By interactively fusion-optimizing the high-frequency cutting anti-disturbance execution unit, the data is interactively transmitted to the communication collaborative control platform to achieve dynamic coordination and precise control of the system.

[0065] Based on the above new energy intelligent anti-disturbance system, the core parameters of the system are optimized through the MOGA intelligent optimization algorithm to improve the anti-disturbance performance of the new energy grid-connected high-frequency cutting generator, such as Figure 2 As shown, the following steps are included:

[0066] S1. Establish an objective function to measure the overall anti-disturbance performance index of the system;

[0067] S2, the MOGA intelligent optimization algorithm based on the NSGA-III architecture, processes the high-dimensional parameter space through non-dominated sorting, optimizing the core parameter combination of high-frequency generator switching to achieve optimal anti-disturbance performance for high-frequency generator switching during renewable energy grid connection. The process is divided into three stages: initialization - evolutionary iteration - convergence output.

[0068] In S1, the anti-disturbance performance in the high-frequency cutting scenario of new energy is optimized based on the dynamic response, impedance characteristics and core parameters of system stability. The objective function F is established as an indicator to measure the overall anti-disturbance performance of the system. The smaller the value, the better the performance:

[0069]

[0070] Where: It contains four core items, corresponding to different performance dimensions;

[0071] Among them, α1·∫Δfdt is the dynamic response term, and the integral of frequency deviation evaluates the inertia support effect of the dynamic response ability of the virtual synchronous machine frequency. The core parameter is the virtual inertia J VSG and the damping coefficient D VSG ;

[0072] α2·ΔZ(f) is the impedance mismatch term, which reflects the ability of the impedance reshaping technology to suppress broadband oscillation; it measures the deviation sum of the actual impedance Z meas (f) and the ideal impedance Z ref (f) in the high-frequency band of 10 - 1000 Hz. The key parameter is the impedance reshaping bandwidth BW Z ;

[0073] α3·P trip is the generator tripping trigger probability term, which reflects the system's anti-disturbance ability. The lower the value, the better the suppression effect of the coordinated control strategy on high-frequency generator tripping events

[0074]

[0075] In the formula: N trip is the number of generator trips in simulation or measurement, and N total is the total number of disturbances;

[0076] α4·σ P is the power volatility, which characterizes the smoothness of the output power of the new energy unit. The standard deviation form quantifies the fluctuation amplitude. The smaller the value, the better the transient stability of the system

[0077]

[0078] In the formula: is the rated power, and T is the observation time window.

[0079] S2 specifically includes the following steps:

[0080] S2-1. Encode the parameters and initialize the population to generate the initial population;

[0081] For the continuous parameters J VSG and D VSG use 16-bit binary encoding with an accuracy of 0.01, and use integer encoding for the discrete parameter control mode flag bit VSG droop control switching; randomly initialize individuals to ensure that the constraint BW Z ≥500 Ηz is satisfied and introduce Latin hypercube sampling LHS to enhance the diversity of the initial solutions to generate the initial population.

[0082] S2-2. Apply multi-objective fitness evaluation to the objective function to calculate the optimal fitness value;

[0083] The population dynamically moves in a defined search space to find the position corresponding to the optimal fitness value as the best solution.

[0084] S2-3. Improve and optimize the algorithm by using non-dominated sorting to associate with the reference point for Pareto stratification and adopting the Das-Dennis method;

[0085] Using non-dominated sorting to associate with the reference point for Pareto stratification, the population is divided into levels such as Front1 optimal front and Front2 through fast non-dominated sorting, and the crowding distance is calculated to ensure the distribution of the front solutions;

[0086] Adopt the Das-Dennis method to uniformly generate 120 reference points in the three-dimensional objective space, dynamically adjust the reference point density, and increase the number of reference points in the sparse area of the solution set; normalize the objective function values to the interval [0,1], calculate the vertical distance from each individual to the reference point, and associate it with the nearest reference point.

[0087] S2-4. Use genetic operation to simulate binary crossover method;

[0088] Randomly select 2 individuals associated with each reference point and retain the one with higher fitness;

[0089] Simulated binary crossover:

[0090] Offspring = 0.5[(1 + β)·Parent1 + (1 - β)·Parent2], β ∼ U(0,1)

[0091] In the formula: the offspring is the new individual generated from the parent individuals, Parent1 and Parent2 are the two parent individuals participating in the crossover operation, β is the crossover distribution index, and 0.5 is the normalization coefficient;

[0092] Among them, the crossover probability is p c = 0.9, and the distribution index η c = 20 controls the dispersion degree of the offspring;

[0093] Use polynomial mutation:

[0094] x′ = x + δ·(x U - x L ),

[0095] In the formula: x′ is the new individual after mutation, x is the original individual, δ is the mutation step size, x U is the upper bound of the individual, x L is the lower bound of the individual, r is a random number with a uniform distribution, and η m is the mutation distribution index;

[0096] Among them, the mutation probability p m= 0.1, distribution index η m = 20 to control the disturbance amplitude.

[0097] S2-5. Update algorithm, introduce adaptive weights, adopt subjective weighting, use the Analytic Hierarchy Process (AHP) to construct a hierarchical model and a judgment matrix;

[0098] Decompose the objective function into a criterion layer and an index layer. The criterion layer includes dynamic response and impedance characteristics, and the index layer includes frequency deviation and generator tripping probability;

[0099] Quantify the relative importance between indicators through a scoring method, combine with objective weighting, improve the entropy weight method, and perform range normalization on time series data (frequency deviation sequence Δf(t)):

[0100]

[0101] In the formula: x′ ij is the normalized data value, x ij is the original data value, min x j is the minimum value of the jth feature, max x j is the maximum value of the jth feature;

[0102] Introduce a time decay factor λ, and the weight of recent data is higher:

[0103]

[0104] In the formula: is the objective weight of the jth indicator, E j is the entropy value of the jth indicator, Σ(1 - E j ) is the sum of all indicators (1 - E j ), λ T-t is the time decay factor;

[0105] Calculate the consistency ratio CR. If CR < 0.1, it passes the test; otherwise, readjust the scale.

[0106] S2-6. Verify the termination condition. If the iteration times are reached or the termination condition is satisfied, terminate the iteration; otherwise, go to S2-2.

[0107] S2-7. Obtain the optimal Pareto solution set corresponding to the optimal fitness value, that is, the core parameters corresponding to the best anti-disturbance performance, so as to make the anti-disturbance performance of new energy grid connection reach the best.

Claims

1. A new energy intelligent anti-disturbance system, characterized in that: It includes new energy units, a virtual synchronous machine-wideband impedance reshaping control system, a high-frequency generator tripping anti-disturbance execution unit, and a communication collaborative control platform; The virtual synchronous machine-wideband impedance reshaping control system is an interactive integration of virtual synchronous machine control technology and wideband impedance reshaping technology. The virtual synchronous machine actively conducts inertia compensation and damping regulation to synergistically optimize the wideband impedance; The new energy units share data with the virtual synchronous machine-wideband impedance reshaping control system and are smoothly linked. Through interactive integration, the high-frequency generator tripping anti-disturbance execution unit is optimized, and the data is interactively transmitted to the communication collaborative control platform to achieve system dynamic collaboration and precise control.

2. The new energy intelligent anti-disturbance system according to claim 1, wherein: The new energy units include wind power generation units, solar photovoltaic power generation units, hydraulic power generation units, biomass power generation units, gravel energy storage units, and tidal energy power generation units.

3. The new energy intelligent anti-disturbance system according to claim 1, characterized in that: The high-frequency generator tripping anti-disturbance execution unit includes a multi-source collaborative generator tripping execution module, an energy storage unit rapid response module, a virtual impedance dynamic regulation module, and a grid-connected inverter control module. Through virtual synchronous machines and wideband impedance reshaping, its dynamic collaboration is optimized to form a hierarchical control instruction chain, including a sensing layer, a decision-making layer, and an execution layer.

4. The new energy intelligent anti-disturbance system according to claim 1, characterized in that: The communication collaborative control platform and the high-frequency generator tripping anti-disturbance execution unit coordinate and optimize the target orientation through dynamic data interaction and collaborative strategies to achieve power grid frequency stability control, new energy power station optimal operation, new energy cluster management, and fault diagnosis and early warning.

5. A new energy intelligent anti-disturbance control method, characterized in that, Based on the new energy intelligent anti-disturbance system described in claim 1, the core parameters of the system are optimized through the MOGA intelligent optimization algorithm to improve the high-frequency generator tripping anti-disturbance performance of new energy grid connection, including the following steps: S1. Establish an objective function for measuring the overall anti-disturbance performance index of the system; S2. Based on the MOGA intelligent optimization algorithm of the NSGA-III architecture, the high-dimensional parameter space is processed through non-dominated sorting to optimize the core parameter combination of high-frequency generator tripping to make the high-frequency generator tripping anti-disturbance performance of new energy grid connection reach the best.

6. The new energy intelligent anti-disturbance control method according to claim 5, characterized in that Specifically, S1 is as follows: According to the anti-disturbance performance in the new energy high-frequency generator tripping scenario, the core parameters of dynamic response, impedance characteristics, and system stability are synergistically optimized. An objective function F is established as a measure of the overall anti-disturbance performance index of the system. The smaller the value, the better the performance: In the formula: it contains four core terms, corresponding to different performance dimensions respectively; Among them, α1·∫Δfdt is the dynamic response term, and the integral of frequency deviation evaluates the inertia support effect of the dynamic response ability of the virtual synchronous machine. The core parameters are the virtual inertia J VSG and the damping coefficient D VSG ; α2·ΔZ(f) is the impedance mismatch term, which reflects the ability of the impedance reshaping technology to suppress broadband oscillations; it measures the deviation sum of the actual impedance Z meas (f) and the ideal impedance Z ref (f) in the high-frequency band of 10 - 1000 Hz. The key parameter is the impedance reshaping bandwidth BW Z ; α3·P trip It is the probability term for generator tripping triggering, reflecting the anti-disturbance ability of the system. The lower the value, the better the suppression effect of the coordinated control strategy on high-frequency generator tripping events. where: N trip the number of times of generator tripping triggered in simulation or actual measurement, N total is the total number of disturbances; α4·σ P The power volatility characterizes the smoothness of the output power of new energy units. The standard deviation form quantifies the fluctuation amplitude. The smaller the value, the better the transient stability of the system. Where: is the rated power, and T is the observation time window.

7. The new energy intelligent anti-disturbance control method according to claim 5, characterized in that: Specifically, S2 includes the following steps: S2-1. Encode the parameters and initialize the population to generate an initial population; For the continuous parameters J VSG , D VSG Adopt 16-bit binary coding with an accuracy of 0.01, and use integer coding to control the mode flag bit of the discrete parameter VSG droop control switching; randomly initialize individuals to ensure that the constraint BW Z ≥500 Hz and introduce Latin hypercube sampling LHS to enhance the diversity of the initial solutions to generate the initial population; S2-2. Apply multi-objective fitness evaluation to the objective function to calculate the optimal fitness value; The population dynamically moves in the determined search space to find the position corresponding to the optimal fitness value as the best solution; S2-3. Use non-dominated sorting and reference point-associated Pareto stratification and adopt the Das-Dennis method to improve and optimize the algorithm; Use non-dominated sorting and reference point-associated Pareto stratification. Through fast non-dominated sorting, the population is divided into levels such as Front1 optimal front and Front2, and the crowding distance is calculated to ensure the distribution of the front solutions; Using the Das-Dennis method, 120 reference points are uniformly generated in the three-dimensional objective space, and the reference point density is dynamically adjusted to increase the number of reference points in the sparse region of the solution set; the objective function values are normalized to the interval [0, 1], and the vertical distance from each individual to the reference point is calculated and associated with the nearest reference point; S2-4. Use the genetic operation to simulate the binary crossover method; Randomly select 2 individuals associated with each reference point and retain the one with higher fitness; Simulate binary crossover: Offspring = 0.5[(1 + β)·Parent 1+(1 - β)·Parent 2], β ∼ U(0, 1) Where: the offspring is the new individual generated from the parent individuals, Parent 1 and Parent 2 are the two parent individuals participating in the crossover operation, β is the crossover distribution index, and 0.5 is the normalization coefficient; where the crossover probability is p c = 0.9, and the distribution exponent η c = 20 controls the offspring dispersion; Use polynomial mutation: where: x′ is the new individual after mutation, x is the original individual, δ is the mutation step size, x U is the upper bound of the individual, x L is the lower bound of the individual, r is a random number uniformly distributed, and ηm is the mutation distribution index; Among them, the mutation probability p m = 0.1, and the distribution index η m = 20 controls the disturbance amplitude; S2-5. Update the algorithm, introduce adaptive weights, subjectively assign weights, use the Analytic Hierarchy Process (AHP) to construct a hierarchical model and a judgment matrix; Decompose the objective function into a criterion layer and an index layer. The criterion layer includes dynamic response and impedance characteristics, and the index layer includes frequency deviation and generator tripping probability; Quantify the relative importance between indicators by the scoring method, combine with objective weighting, improve the entropy weight method, and perform range normalization on the time series data: where: x i ′ j is the normalized data value, x ij is the original data value, minx j is the minimum value of the j-th feature, maxx j is the maximum value of the j-th feature; Introduce a time decay factor λ, with higher weights for recent data: In the formula: is the objective weight of the j-th index, E j is the entropy value of the j-th index, Σ(1 - E j ) is the sum of all indices (1 - E j ), and λ T-t is the time decay factor; Calculate the consistency ratio CR. If CR < 0.1, it passes the test; otherwise, readjust the scale; S2-6. Verify the termination condition. If the iteration number is reached or the termination condition is satisfied, terminate the iteration; otherwise, go to S2-2; S2-7. Obtain the optimal Pareto solution set corresponding to the optimal fitness value, that is, the core parameters corresponding to the best anti-disturbance performance, so as to achieve the best anti-disturbance performance of new energy grid connection.

Images