A Dynamic Clustering and Aggregation Method and System for Offshore Wind Farms Based on Key Parameter Classification

Abstract

This invention discloses a method and system for dynamic clustering and aggregation of offshore wind farms based on key parameter classification in the field of power system modeling and simulation technology. The method includes: acquiring the configuration parameters, external disturbance signals, and operating status data of the offshore wind farm, and inputting them into a pre-constructed parameterized aggregation dynamic model to obtain aggregated active / reactive power; wherein, the model includes a parameter adaptive module for calculating dynamic response parameters such as virtual inertia and virtual damping based on the configuration parameters; a mechanism core module for receiving dynamic response parameters, grid feedback signals, and external disturbance signals, and calculating power angle, angular frequency, and voltage; a transfer function module for generating active and reactive power based on the power angle and voltage, outputting to the grid and providing feedback signals; and a parameter update interface for updating the internal parameters of the transfer function online. This invention achieves accurate and adaptive equivalence of wind farm dynamics under mixed configurations and topology changes.

Description

Technical Field

[0001] This invention relates to a dynamic clustering and aggregation method and system for offshore wind farms based on key parameter classification, belonging to the field of power system modeling and simulation technology. Background Technology

[0002] As the penetration rate of new energy sources continues to rise, the power system is undergoing a structural transformation from being dominated by synchronous machines to being dominated by power electronic converters. Against this backdrop, the hybrid integration of grid-connected (GFL) and grid-connected (GFM) power sources has become the mainstream approach: the former relies on grid voltage synchronization (such as photovoltaic power plants), while the latter autonomously establishes voltage / frequency support (such as energy storage and virtual synchronous machines). However, the dynamic characteristics of these two types of power sources differ fundamentally: GFLs behave as controlled current sources, while GFMs exhibit synchronous machine-like inertia, causing traditional synchronous machine aggregation models to fail.

[0003] Existing models are mostly designed for fixed configurations and lack adaptability to topological changes. There is an urgent need to establish a parameterized general model that can accurately depict the dynamic interaction between the source and the network through online reconstruction of key parameters. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic clustering and aggregation method and system for offshore wind farms based on key parameter classification, which can accurately and adaptively equivalently represent the dynamic response of offshore wind farms under mixed configurations and topology changes, and solve the key problems of insufficient accuracy and poor adaptability of traditional aggregation models.

[0005] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.

[0006] In a first aspect, the present invention provides a dynamic clustering and aggregation method for offshore wind farms based on key parameter classification, comprising:

[0007] Acquire configuration parameters, external disturbance signals, and operational status data of offshore wind farms;

[0008] The configuration parameters, external disturbance signals, and operating status data are input into a pre-constructed parametric aggregation dynamic model to obtain the aggregated active and reactive power of the offshore wind farm. The parametric aggregation dynamic model includes a core mechanism module, a transfer function module, a parameter adaptive module, and a parameter update interface.

[0009] The parameter adaptive module is used to calculate and output dynamic response parameters, including virtual inertia, virtual damping, voltage-reactive droop coefficient and time constant, based on the configuration parameters and external disturbance signals.

[0010] The transfer function module receives the power angle and voltage from the power grid, calculates the aggregated active and reactive power of the offshore wind farm, and outputs the aggregated active and reactive power to the power grid; simultaneously, it receives real-time feedback of the actual voltage from the power grid side. and actual angular frequency As an input signal, it is fed back to the transfer function module;

[0011] The core mechanism module is used to receive aggregated active and reactive power, dynamic response parameters, and input signals from the grid side, and calculate the power angle, angular frequency, and voltage of the grid based on the aggregated active and reactive power, dynamic response parameters, and input signals.

[0012] The parameter update interface is used to identify and update the internal parameters of the transfer function module online based on the real-time response data of the power grid.

[0013] In conjunction with the first aspect, the parameter adaptive module further calculates and outputs dynamic response parameters, including virtual inertia, virtual damping, voltage-reactive droop coefficient, and time constant, based on the configuration parameters and external disturbance signals.

[0014] The configuration parameters include the number of parallel units and the operating mode;

[0015] Based on the operating status data, the reference virtual inertia of the acquisition unit is obtained. Unit reference virtual damping ;

[0016] Based on the number of parallel units n, the reference virtual inertia of the unit is determined. and unit reference virtual damping Perform linear scaling to obtain the adjusted virtual inertia. Virtual damping ;

[0017] Based on the aforementioned operating mode, the Mantis Shrimp optimization algorithm is used to optimize the voltage-reactive power droop coefficient. and time constant Optimization is performed; specifically, when the operating mode is grid-connected, the optimization direction is to reduce the voltage-reactive power droop coefficient. Approaching 0 and increasing the time constant. When the operating model is a network type, the optimization direction is to increase the voltage-reactive power droop factor. Reach a positive value and decrease the time constant. .

[0018] In conjunction with the first aspect, the adjustment expressions for the virtual inertia, virtual damping, voltage-reactive droop coefficient, and time constant are as follows:

[0019] ;

[0020] in, This represents the adjusted virtual inertia; This represents the adjusted virtual damping; n represents the number of parallel units; Represents the unit's reference virtual inertia; Indicates the virtual damping of the unit reference; This represents the adjusted voltage-reactive power droop coefficient; This represents the baseline value of the droop coefficient in the network-type operation mode; This represents the maximum droop coefficient in the network-type operation mode; Indicates the network structure; Indicates the mesh type; This represents the adjusted time constant; This represents the fast response time constant in the network configuration mode; This represents the slow response time constant in mesh pattern; This represents the minimum value of the response time constant; This represents the maximum value of the response time constant.

[0021] In conjunction with the first aspect, further, based on the aggregated active and reactive power, dynamic response parameters, and input signal, the expressions for calculating the power angle, angular frequency, and voltage are as follows:

[0022] ;

[0023] in, Indicates the current working angle; Indicates the reference angular frequency; Indicates the current angular frequency; Indicates the angular frequency of the power grid; This represents the adjusted virtual inertia; Indicates mechanical power; This represents the active power injected into the power grid after the offshore wind farm is aggregated. This indicates the adjusted virtual damping; This represents the adjusted time constant; Indicates the voltage reference value; This represents the adjusted voltage-reactive power droop coefficient; This indicates the reference value for reactive power. This represents the reactive power injected into the power grid after the offshore wind farm's power generation is aggregated. Indicates the current voltage; Indicates the actual angular frequency on the power grid side; This indicates the actual voltage on the grid side.

[0024] In conjunction with the first aspect, further, based on the aforementioned power angle and voltage, the expressions for the aggregated active and reactive power of offshore wind power are obtained as follows, including:

[0025] ;

[0026] in, This represents the active power injected into the power grid after the offshore wind farm is aggregated. This represents the reactive power injected into the power grid after the offshore wind farm's power generation is aggregated. This represents the power angle-active power transfer function; This represents the voltage-active power transfer function; This represents the voltage-reactive power transfer function; Represents the power angle-reactive power transfer function; V represents the current power angle; V represents the current voltage.

[0027] In conjunction with the first aspect, the mantis shrimp optimization algorithm is further used to optimize the voltage-reactive power droop coefficient. and time constant Optimizations include:

[0028] An initial population of size N is randomly generated; wherein the initial population comprises multiple individuals, each representing a set of parameters to be optimized. , And assign a polarization type indicator with a value of 1, 2 or 3 to each individual;

[0029] The initial population is iteratively optimized until the termination condition is met; wherein, in each iteration, the following operations are performed:

[0030] For each individual, the angle between the current position and the new position is calculated as the left polarization angle, and the right polarization angle is randomly generated;

[0031] The left polarization type and right polarization type are determined based on the distances between the left polarization angle and the right polarization angle and the preset reference angle;

[0032] Calculate the angle difference between the left polarization angle and the corresponding reference angle, and the angle difference between the right polarization angle and the corresponding reference angle. By comparing the magnitude of the two angle differences, select the polarization type corresponding to the smaller angle difference to update the individual's current polarization type indicator.

[0033] Based on the updated polarization type indicator, select the appropriate search strategy to update the individual's location; the search strategy includes foraging strategy, attack strategy, and evasion / defense strategy.

[0034] Once the iterative optimization meets the termination condition, the optimal individual value is output as the optimized voltage-reactive power droop coefficient. and time constant .

[0035] In conjunction with the first aspect, the expression for determining its left-polarization type and right-polarization type further includes:

[0036] ;

[0037] ;

[0038] in, Indicates the left polarization type; Indicates the right polarization type; Indicates the left polarization angle; This represents the right polarization angle.

[0039] In conjunction with the first aspect, further, the selection of an appropriate search strategy to update the individual's location includes:

[0040] When the polarization type indicator is set to 1:

[0041] Randomly select individuals from the same iteration round;

[0042] By introducing random variables and diffusion coefficients, a foraging strategy is used to update the position of individuals in the same iteration round to obtain a first position, and the first position is used as the updated position of the current individual;

[0043] When the polarization type indicator is 2:

[0044] A global search is performed on the current individual using circular oscillation to generate the oscillation angle. ;

[0045] According to the swing angle The attack strategy is used to calculate the new position of the current individual, obtain the second position, and use the second position as the updated position of the current individual;

[0046] When the polarization type indicator is set to 3:

[0047] By introducing a random scaling factor Based on the evasion and defense strategy, the current individual is contracted or expanded to obtain a third position, and the third position is used as the updated position of the current individual.

[0048] In conjunction with the first aspect, the expressions for obtaining the first position, the second position, and the third position are as follows:

[0049] ;

[0050] ;

[0051] ;

[0052] in, Indicates the first position; This indicates the globally optimal position in the current iteration; This represents the difference between the previous iteration position and the optimal position; Indicates the diffusion coefficient; Represents a random vector; Indicates the position of the previous iteration; Indicates the second position; This indicates that the individual is thrown out along a large arc around the optimal value of the current iteration; Indicates the third position; Indicates a random scaling factor; Represents the set of n-tuples of real numbers; Indicates the swing angle. This indicates taking the maximum value. This indicates taking the absolute value.

[0053] Secondly, a dynamic clustering and aggregation system for offshore wind farms based on key parameter classification includes:

[0054] The parameter acquisition unit is used to acquire the configuration parameters, external disturbance signals and operating status data of the offshore wind farm.

[0055] The aggregation processing unit is used to input the configuration parameters, external disturbance signals, and operating status data into a pre-constructed parametric aggregation dynamic model to obtain the aggregated active and reactive power of the offshore wind farm; wherein, the parametric aggregation dynamic model includes:

[0056] The parameter adaptive module is used to calculate and output dynamic response parameters, including virtual inertia, virtual damping, voltage-reactive droop coefficient and time constant, based on the configuration parameters and external disturbance signals.

[0057] The transfer function module is used to calculate the aggregated active power and reactive power of the offshore wind farm based on the power angle and voltage, and output the aggregated active power and reactive power to the power grid; at the same time, the actual voltage and actual angular frequency fed back in real time from the power grid side are fed back to the transfer function module as input signals.

[0058] The core mechanism module is used to receive aggregated active and reactive power, dynamic response parameters, and input signals from the grid side, and to calculate the power angle, angular frequency, and voltage of the grid based on the aggregated active and reactive power, dynamic response parameters, and input signals.

[0059] The parameter update interface is used to identify and update the internal parameters of the transfer function module online based on real-time response data from the power grid.

[0060] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0061] This invention constructs a parameterized aggregation dynamic model and introduces a parameter adaptive mechanism based on the number of parallel units and operating mode. This enables high-precision equivalence of the dynamic characteristics of offshore wind farms in scenarios with mixed grid-connected and grid-structured power supply access and frequent topology changes. Simultaneously, by combining online parameter identification and the Mantis Shrimp optimization algorithm to update and optimize the operating state parameters, the model can not only dynamically track the actual operating state of the system but also autonomously optimize its dynamic response. Thus, while ensuring simulation accuracy, the adaptability, robustness, and computational efficiency of the model are significantly improved, effectively overcoming the shortcomings of traditional fixed-configuration aggregation models, such as insufficient accuracy and difficulty in applying them to large-scale power grid dynamic simulation. Attached Figure Description

[0062] Figure 1 The diagram shows a flowchart of the dynamic clustering and aggregation method for offshore wind farms provided in an embodiment of the present invention.

[0063] Figure 2 The diagram shown is a flowchart of the mantis shrimp optimization algorithm provided in an embodiment of the present invention. Detailed Implementation

[0064] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0065] Example 1

[0066] See Figure 1 This embodiment introduces a dynamic clustering and aggregation method for offshore wind farms based on key parameter classification, including:

[0067] Acquire configuration parameters, external disturbance signals, and operational status data of offshore wind farms;

[0068] Configuration parameters, external disturbance signals, and operational status data are input into a pre-constructed parametric aggregated dynamic model to obtain the aggregated active and reactive power of the offshore wind farm. The parametric aggregated dynamic model includes a core mechanism module, a transfer function module, a parameter adaptive module, and a parameter update interface.

[0069] The parameter adaptation module is used to generate a set of dynamic response parameters, including virtual inertia, based on configuration parameters and operating status data. Virtual damping Voltage-Reactive Power Droop Coefficient Time constant ;

[0070] It should be noted that, and It scales linearly with n to accurately reflect the topological changes of offshore wind farms; and Then switch according to the operating mode: set under network-type GFM >0, and take the smaller one Value, and the setting under the network type GFL ≈0, and take the larger one. value.

[0071] The transfer function module is used to receive the power angle of the power grid. and voltage By encapsulating the transfer function matrix of high-frequency dynamics and control delay within its internal structure, the aggregated active power of the offshore wind farm can be obtained. and reactive power And inject it into the power grid; at the same time, the power grid measures and feeds back the actual voltage in real time. and actual angular frequency As an input signal, it is fed back to the transfer function module to form a closed-loop dynamic response, ensuring that the model can accurately simulate the field-network interaction process;

[0072] The core mechanism module is used for the aggregated active power. and reactive power And input signals from the grid side, and based on the aggregated active power and reactive power Based on dynamic response parameters and input signals, the power angle of the power grid is calculated. angular frequency and voltage ;

[0073] The parameter update interface uses real-time response data based on the power grid, which includes the actual voltage collected by the power grid in real time. and actual angular frequency Aggregated active power and reactive power And four dynamic response parameters caused by external disturbance signals.

[0074] Based on this real-time response parameter, the internal parameters (gain) of the transfer function module are evaluated. and each time constant Online identification and updates are performed to ensure that the dynamic characteristics of the model remain highly consistent with the real system, maintaining accurate equivalence even under topological abrupt changes such as equipment switching and mode switching. In summary, this embodiment effectively solves the modeling challenge of mixed grid-connected and grid-building power sources in high-proportion renewable energy power systems through the collaborative mechanism of parameter adaptation, mechanism modeling, transfer function calculation, and online parameter identification. It achieves universal and high-precision dynamic aggregation of offshore wind farms under dynamic configuration changes.

[0075] Example 2

[0076] To address the modeling requirements of mixed grid-connected and grid-building power sources in high-proportion renewable energy power systems, this invention constructs a parameterized aggregation dynamic model. This model includes a core mechanism module, a transfer function module, a parameter adaptive module, and a parameter update interface. Through the collaborative processing of these modules, specifically including:

[0077] a. Parameter Adaptation Module

[0078] The acquired configuration parameters and external disturbance signals are input into the parameter adaptation module, which outputs the dynamically adjusted virtual inertia. Virtual damping Voltage-Reactive Power Droop Coefficient and time constant The configuration parameters include the number of parallel units n and the operating mode.

[0079] In step S1, external disturbances (such as grid voltage fluctuations, frequency deviations, equipment switching or failures, etc.) cause dynamic changes in the number of parallel units n and the operating mode. These changes are fed back to the parameter adaptive module in real time, thereby adjusting the dynamic response parameters. For example, external disturbances may cause an increase or decrease in the number of parallel units n (such as wind turbines going offline or in parallel), or external disturbances may trigger a switch in the operating mode (such as an energy storage system switching from grid-connected to grid-connected).

[0080] Furthermore, the unit's reference virtual inertia is obtained based on the operational status data. and unit reference virtual damping And by combining the number of parallel units n (i.e., performing a linear scaling operation), the adjusted virtual inertia is obtained. Virtual damping Its expression is:

[0081] (1)

[0082] (2)

[0083] in, This represents the adjusted virtual inertia; This represents the adjusted virtual damping; n represents the number of parallel units; Represents the unit's reference virtual inertia; This indicates the virtual damping of the unit reference.

[0084] Based on the operating mode, the Mantis Shrimp optimization algorithm is used to optimize the voltage-reactive power droop coefficient. and time constant Optimization is performed; specifically, when the operating mode is a grid-following GFL, its control objective is power point tracking, and the optimization direction is to reduce the voltage-reactive power droop coefficient. Approaching 0 to eliminate reactive power regulation and increasing the time constant. To mitigate voltage dynamic response; when the operating model is a network-based GFM, its control objective is to autonomously construct voltage, and the optimization direction is to increase the voltage-reactive power droop factor. To improve reactive power support capability and reduce time constant, a positive value is set. To accelerate voltage regulation.

[0085] Adjusted voltage-reactive power droop coefficient and time constant The expression is:

[0086] (3)

[0087] (4)

[0088] in, This represents the adjusted virtual inertia; This represents the adjusted virtual damping; n represents the number of parallel units; Represents the unit's reference virtual inertia; Indicates the virtual damping of the unit reference; This represents the adjusted voltage-reactive power droop coefficient; This represents the baseline value of the droop coefficient in the network-type operation mode; This represents the maximum droop coefficient in the network-type operation mode; Indicates the network structure; Indicates the mesh type; This represents the adjusted time constant; This represents the fast response time constant in the network configuration mode; This represents the slow response time constant in mesh pattern; This represents the minimum value of the response time constant; This represents the maximum value of the response time constant.

[0089] b. Core Mechanism Module: This module receives the aggregated active power. and reactive power Dynamically adjusted virtual inertia Time constant And the input signal from the power grid (the actual voltage of the power grid) and actual angular frequency The data is then input into the core mechanism module, which calculates the power angle in real time based on the following virtual power angle equation, virtual swing equation, and voltage equation. angular frequency and voltage Its expression is:

[0090] (5)

[0091] in, Indicates the angle of attack; Indicates the reference angular frequency; Indicates the current angular frequency; Indicates the angular frequency of the power grid; Represents virtual inertia; Indicates mechanical power; This represents the active power injected into the power grid after the offshore wind farm is aggregated. Indicates the voltage reference value; This indicates the reference value for reactive power. This represents the reactive power injected into the power grid after the offshore wind farm's power generation is aggregated. This indicates the current voltage.

[0092] This step utilizes a parameter adaptive mechanism to dynamically adjust parameters such as inertia, damping, and voltage, enabling it to respond to topology changes (such as the number of parallel units). It can switch between operating modes (GFL / GFM) to accurately characterize system dynamics even during configurational mutations.

[0093] c. Transfer function module

[0094] The power angle and voltage output from the core mechanism module are input into the transfer function module, which encapsulates high-frequency dynamics and control delay. Its output is the aggregated active power of the offshore wind farm. And reactive power Q, the active power The expression for reactive power Q is:

[0095] (6)

[0096] in, This represents the power angle-active power transfer function; This represents the voltage-active power transfer function; This represents the voltage-reactive power transfer function; Represents the power angle-reactive power transfer function; Indicates the current working angle.

[0097] This step achieves a high-precision mapping from system state variables to output power. By using frequency domain modeling of the transfer function module, it preserves key dynamic characteristics while avoiding the high computational burden of detailed switching models, making it suitable for large power grid simulation.

[0098] Finally, the active power output by the transfer function module will be... Reactive power Q is injected into the power grid for large-scale power grid simulation, while real-time voltage on the grid side is collected. and angular frequency It is then fed back to the transfer function module as an input signal.

[0099] This module can realistically simulate the dynamic interaction between renewable energy power plants and the power grid. Its output can affect the grid status, and changes in the grid status can be fed back into the model in real time, ensuring an accurate depiction of the dynamic interaction between the power source and the grid. d. Parameter Update Interface

[0100] This interface receives real-time response data from the power grid, including the actual voltage collected by the power grid in real time. and actual angular frequency Active power from the transfer function module and reactive power And four dynamic response parameters caused by external disturbance signals;

[0101] Based on real-time response data, the transfer function module The internal parameters of the transfer functions used to describe the dynamic mapping from power angle / voltage to active and reactive power are identified and updated online. The transfer function module contains multiple transfer functions, each representing the dynamic impact of power angle changes on active power, voltage changes on active power, voltage changes on reactive power, and power angle changes on reactive power. Its internal parameters include the steady-state gain of each transfer function. Zero-point time constant Pole time constant Pure delay time By updating these parameters in real time, the high-frequency dynamic and control delay characteristics encapsulated in the transfer function module can track the changes in the real wind farm, thereby overcoming the shortcomings of fixed parameter models in terms of poor adaptability in scenarios such as equipment aging and control strategy switching.

[0102] Among them, the transfer function module Let be a first-order lead-lag transfer function with delay, its expression is:

[0103] (7)

[0104] in, This indicates the transfer function module; Represents steady-state gain; Represents the zero-point time constant; Represents the pole time constant; Indicates the pure delay time; This represents a complex frequency variable (Laplace operator).

[0105] On the other hand, in order to further improve the dynamic performance under complex operating conditions, this invention utilizes the Mantis Shrimp optimization algorithm to optimize the voltage-reactive power droop coefficient. and time constant .

[0106] See Figure 2 The adjustment mechanism based on the operating mode is optimized in real time using the Mantis Shrimp Optimization Algorithm (MshOA). and The specific value of is determined as follows:

[0107] Step S11: Randomly generate an initial population of size N and dimension dim (number of features in the vector); wherein the initial population consists of multiple individuals, each individual representing a set of parameters to be optimized ( , );

[0108] Specifically, the initial population is generated using the following formula:

[0109] (8)

[0110] in, This represents the position of the i-th individual in the j-th dimension (i.e., the parameter); Let denote the upper bound of the j-th dimension; This represents the lower bound of the j-th dimension; ∈[0,1] follows a uniform distribution.

[0111] Subsequently, each individual is assigned a polarization type indicator with a value of 1, 2, or 3 using the following formula, which will determine the search strategy it adopts in subsequent iterations;

[0112] (9)

[0113] PTI represents the polarization type indicator: PTI=1 represents vertically linearly polarized light, PTI=2 represents horizontally polarized light, and PTI=3 represents circularly polarized light.

[0114] Step S12: Iteratively optimize the initial population until the termination condition is met; wherein, in each iteration, the following operations are performed:

[0115] Step S121: In each iteration, for each individual, calculate the current position. With new location The angle between them is taken as the left polarization angle, and a right polarization angle is randomly generated;

[0116] The expressions for the left polarization angle and the right polarization angle are:

[0117] (10)

[0118] (11)

[0119] in, Indicates the left polarization angle; Indicates the current position; Indicates the new location; represents the right polarization angle; rand represents a random value.

[0120] Step S122: Based on the left polarization angle and the right polarization angle and the preset reference angle (e.g., ... , The distance (e.g., depending on the specific circumstances) determines its left-polarization type and right-polarization type;

[0121] The expressions for the left-polarized and right-polarized types are:

[0122] (12)

[0123] (13)

[0124] in, Indicates the left polarization type; Indicates the right polarization type.

[0125] Step S123: Calculate the angle difference between the left polarization angle and the corresponding reference angle, and the angle difference between the right polarization angle and the corresponding reference angle;

[0126] The expression for this angle difference is:

[0127] (14)

[0128] (15)

[0129] in, This represents the angle difference between the left polarization angle and the corresponding reference angle; This represents the angular difference between the right polarization angle and the corresponding reference angle.

[0130] Step S124: By comparing the magnitudes of the two angle differences, select the polarization type corresponding to the smaller angle difference to update the individual's current polarization type indicator. If smaller, the left polarization type is retained. Otherwise, retain the right polarization type. The following conditions must be met:

[0131] (16)

[0132] in, Indicates the current polarization type.

[0133] Step S125: Based on the updated polarization type indicator, select the corresponding search strategy to update the individual's position; wherein, the search strategy includes foraging strategy, attack strategy and evasion / defense strategy.

[0134] (a) When the polarization type indicator is 1, the foraging strategy is selected: the irregular swimming of mantis shrimp when foraging is regarded as Brownian motion, and local search is performed by simulating Brownian motion.

[0135] Specifically, individuals in the same iteration round are randomly selected. By introducing random variables and diffusion coefficients, a foraging strategy is used to update the position of the individuals in the same iteration round to obtain the first position, and the first position is used as the updated position of the current individual.

[0136] The expression for the first position is:

[0137] (17)

[0138] in, Indicates the first position; This indicates the globally optimal position in the current iteration; This represents the difference between the previous iteration position and the optimal position; Indicates the diffusion coefficient; Represents a random vector; Indicates the position of the previous iteration.

[0139] (b) When the polarization type indicator is 2, select the attack strategy: the forelimb attack of the mantis shrimp can be regarded as a circular swing in a two-dimensional plane, and global exploration is carried out through the circular swing.

[0140] Specifically, a global search is performed on the current individual through circular oscillation to generate the oscillation angle. ;

[0141] According to the swing angle The attack strategy is used to calculate the new position of the current individual, obtain the second position, and use the second position as the updated position of the current individual.

[0142] The expression for the second position is:

[0143] (18)

[0144] in, Indicates the second position; Indicates the range of control transfer; This means that the individual is thrown along a large arc direction "around" the optimal solution, achieving remote exploration to escape local extrema.

[0145] (c) When the polarization type indicator is 3, the evasion defense strategy is selected. When the circular polarization signal is dominant, the individual chooses to "approach" or "retreat" according to the threat level, that is, to balance exploration and development by shrinking or expanding near the optimal solution.

[0146] Specifically, by introducing a random scaling factor Based on the evasion and defense strategy, the current individual shrinks and expands near the optimal point to obtain the third position, and the third position is used as the updated position of the current individual.

[0147] The expression for the third position is:

[0148] (19)

[0149] in, Indicates the third position; Indicates a random scaling factor; Let represent the set of n-tuples of real numbers, such that the third position slightly contracts or expands around the optimal point, enhancing refinement while maintaining a certain degree of diversity. This indicates taking the maximum value. This indicates taking the absolute value.

[0150] Step S126: After the iterative optimization meets the termination condition, output the optimal individual value as the optimized voltage-reactive power droop coefficient. and time constant .

[0151] Example 3

[0152] A dynamic clustering and aggregation system for offshore wind farms based on key parameter classification, comprising:

[0153] The parameter acquisition unit is used to acquire the configuration parameters, external disturbance signals and operating status data of the offshore wind farm.

[0154] The aggregation processing unit is used to input configuration parameters, external disturbance signals, and operating status data into a pre-built parametric aggregation dynamic model to obtain the aggregated active and reactive power of the offshore wind farm; wherein, the parametric aggregation dynamic model includes:

[0155] The parameter adaptive module is used to calculate and output dynamic response parameters, including virtual inertia, virtual damping, voltage-reactive droop coefficient and time constant, based on the configuration parameters and external disturbance signals.

[0156] The transfer function module is used to calculate the aggregated active and reactive power of the offshore wind farm based on the power angle and voltage, and output it to the power grid; simultaneously, the actual voltage is fed back in real time from the power grid side. and actual angular frequency As an input signal, it is fed back to the transfer function module;

[0157] The core mechanism module is used to receive aggregated active and reactive power, dynamic response parameters, and input signals from the grid side, and to calculate the power angle, angular frequency, and voltage of the grid based on the aggregated active and reactive power, dynamic response parameters, and input signals.

[0158] The parameter update interface is used to identify and update the internal parameters of the transfer function module online based on the real-time response data of the power grid.

[0159] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A dynamic clustering aggregation method for offshore wind farms based on key parameter classification, characterized in that, include: Acquire configuration parameters, external disturbance signals, and operational status data of offshore wind farms; The configuration parameters, external disturbance signals, and operating status data are input into a pre-constructed parametric aggregation dynamic model to obtain the aggregated active and reactive power of the offshore wind farm. The parametric aggregation dynamic model includes a core mechanism module, a transfer function module, a parameter adaptive module, and a parameter update interface. The parameter adaptive module is used to calculate and output dynamic response parameters, including virtual inertia, virtual damping, voltage-reactive droop coefficient, and time constant, based on the configuration parameters and external disturbance signals; wherein, the configuration parameters include the number of parallel units n and the operating mode; acquiring the reference virtual inertia according to the operating state data , unit reference virtual damping ; in combination with the number n of parallel units, the unit reference virtual inertia and the unit reference virtual damping are linearly scaled to obtain an adjusted virtual inertia , virtual damping ; Based on the operation mode, the voltage-reactive droop coefficient and time constant are optimized by using mantis shrimp optimization algorithm; wherein, when the operation mode is grid-following type, the optimization direction is to make the voltage-reactive droop coefficient tend to 0 and increase the time constant ; when the operation mode is grid-forming type, the optimization direction is to increase the voltage-reactive droop coefficient to positive value and decrease the time constant ; The transfer function module is used to receive the power angle and voltage of the power grid, and calculate the aggregated active power and reactive power of the offshore wind farm based on the power angle and voltage, and output them to the power grid; at the same time, the actual voltage and actual angular frequency fed back in real time from the power grid side are fed back to the transfer function module as input signals. The core mechanism module is used to receive aggregated active and reactive power, dynamic response parameters, and input signals from the grid side, and calculate the power angle, angular frequency, and voltage of the grid based on the aggregated active and reactive power, dynamic response parameters, and input signals. The parameter update interface is used to identify and update the internal parameters of the transfer function module online based on the real-time response data of the power grid.

2. The method for dynamic clustering and aggregation of offshore wind farms based on key parameter classification according to claim 1, characterized in that, The adjustment expressions for the virtual inertia, virtual damping, voltage-reactive droop coefficient, and time constant are as follows: ； in, This represents the adjusted virtual inertia; This represents the adjusted virtual damping; n represents the number of parallel units; Represents the unit's reference virtual inertia; Indicates the virtual damping of the unit reference; This represents the adjusted voltage-reactive power droop coefficient; This represents the baseline value of the droop coefficient in the network-type operation mode; This represents the maximum droop coefficient in the network-type operation mode; Indicates the network structure; Indicates the mesh type; This represents the adjusted time constant; This represents the fast response time constant in the network configuration mode; This represents the slow response time constant in mesh pattern; This represents the minimum value of the response time constant; This represents the maximum value of the response time constant.

3. The method for dynamic clustering and aggregation of offshore wind farms based on key parameter classification according to claim 2, characterized in that, Based on the aggregated active and reactive power, dynamic response parameters, and input signal, the expressions for calculating the power angle, angular frequency, and voltage are as follows: ； in, Indicates the current working angle; Indicates the reference angular frequency; Indicates the current angular frequency; Indicates the angular frequency of the power grid; This represents the adjusted virtual inertia; Indicates mechanical power; This represents the active power injected into the power grid after the offshore wind farm is aggregated. This indicates the adjusted virtual damping; This represents the adjusted time constant; Indicates the voltage reference value; This represents the adjusted voltage-reactive power droop coefficient; This indicates the reference value for reactive power. This represents the reactive power injected into the power grid after the offshore wind farm's power generation is aggregated. Indicates the current voltage; Indicates the actual angular frequency on the power grid side; This indicates the actual voltage on the grid side.

4. The method for dynamic clustering and aggregation of offshore wind farms based on key parameter classification according to claim 1, characterized in that, Based on the aforementioned power angle and voltage, the expressions for the aggregated active and reactive power of offshore wind power are obtained as follows, including: ； in, This represents the active power injected into the power grid after the offshore wind farm is aggregated. This represents the reactive power injected into the power grid after the offshore wind farm's power generation is aggregated. This represents the power angle-active power transfer function; This represents the voltage-active power transfer function; This represents the voltage-reactive power transfer function; Represents the power angle-reactive power transfer function; V represents the current power angle; V represents the current voltage.

5. The method for dynamic clustering and aggregation of offshore wind farms based on key parameter classification according to claim 1, characterized in that, The mantis shrimp optimization algorithm is used to optimize the voltage-reactive power droop coefficient. and time constant Optimizations include: An initial population of size N is randomly generated; wherein the initial population comprises multiple individuals, each representing a set of parameters to be optimized. , And assign a polarization type indicator with a value of 1, 2 or 3 to each individual; The initial population is iteratively optimized until the termination condition is met; wherein, in each iteration, the following operations are performed: For each individual, the angle between the current position and the new position is calculated as the left polarization angle, and the right polarization angle is randomly generated; The left polarization type and right polarization type are determined based on the distances between the left polarization angle and the right polarization angle and the preset reference angle; Calculate the angle difference between the left polarization angle and the corresponding reference angle, and the angle difference between the right polarization angle and the corresponding reference angle. By comparing the magnitude of the two angle differences, select the polarization type corresponding to the smaller angle difference to update the individual's current polarization type indicator. Based on the updated polarization type indicator, select the appropriate search strategy to update the individual's location; the search strategy includes foraging strategy, attack strategy, and evasion / defense strategy. Once the iterative optimization meets the termination condition, the optimal individual value is output as the optimized voltage-reactive power droop coefficient. and time constant .

6. The method for dynamic clustering and aggregation of offshore wind farms based on key parameter classification according to claim 5, characterized in that, The expression that determines its left polarization type and right polarization type includes: ； ； in, Indicates the left polarization type; Indicates the right polarization type; Indicates the left polarization angle; This represents the right polarization angle.

7. The method for dynamic clustering and aggregation of offshore wind farms based on key parameter classification according to claim 5, characterized in that, The step of selecting an appropriate search strategy to update the individual's location includes: When the polarization type indicator is set to 1: Randomly select individuals from the same iteration round; By introducing random variables and diffusion coefficients, a foraging strategy is used to update the position of individuals in the same iteration round to obtain a first position, and the first position is used as the updated position of the current individual; When the polarization type indicator is 2: A global search is performed on the current individual using circular oscillation to generate the oscillation angle. ; According to the swing angle The attack strategy is used to calculate the new position of the current individual, obtain the second position, and use the second position as the updated position of the current individual; When the polarization type indicator is set to 3: By introducing a random scaling factor Based on the evasion and defense strategy, the current individual is contracted or expanded to obtain a third position, and the third position is used as the updated position of the current individual.

8. The method for dynamic clustering and aggregation of offshore wind farms based on key parameter classification according to claim 7, characterized in that, The expressions for obtaining the first position, the second position, and the third position are as follows: ； ； ； in, Indicates the first position; This indicates the globally optimal position in the current iteration; This represents the difference between the previous iteration position and the optimal position; Indicates the diffusion coefficient; Represents a random vector; Indicates the position of the previous iteration; Indicates the second position; This indicates that the individual is thrown out along a large arc around the optimal value of the current iteration; Indicates the third position; Indicates the random scaling factor; Represents the set of n-tuples of real numbers; Indicates the swing angle. This indicates taking the maximum value; This indicates taking the absolute value.

9. A dynamic clustering and aggregation system for offshore wind farms based on key parameter classification, characterized in that, include: The parameter acquisition unit is used to acquire the configuration parameters, external disturbance signals and operating status data of the offshore wind farm. The aggregation processing unit is used to input the configuration parameters, external disturbance signals, and operating status data into a pre-constructed parametric aggregation dynamic model to obtain the aggregated active and reactive power of the offshore wind farm; wherein, the parametric aggregation dynamic model includes: The parameter adaptive module is used to calculate and output dynamic response parameters, including virtual inertia, virtual damping, voltage-reactive droop coefficient, and time constant, based on the configuration parameters and external disturbance signals; wherein, the configuration parameters include the number of parallel units n and the operating mode; Based on the operating status data, the reference virtual inertia of the acquisition unit is obtained. Unit reference virtual damping ; Based on the number of parallel units n, the reference virtual inertia of the unit is determined. and unit reference virtual damping Perform linear scaling to obtain the adjusted virtual inertia. Virtual damping ; Based on the aforementioned operating mode, the Mantis Shrimp optimization algorithm is used to optimize the voltage-reactive power droop coefficient. and time constant Optimization is performed; specifically, when the operating mode is grid-connected, the optimization direction is to reduce the voltage-reactive power droop coefficient. Approaching 0 and increasing the time constant. When the operating model is a network type, the optimization direction is to increase the voltage-reactive power droop factor. Reach a positive value and decrease the time constant. ; The transfer function module is used to receive the power angle and voltage of the power grid, and calculate the aggregated active power and reactive power of the offshore wind farm based on the power angle and voltage, and output them to the power grid; at the same time, the actual voltage and actual angular frequency fed back in real time from the power grid side are fed back to the transfer function module as input signals. The core mechanism module is used to receive aggregated active and reactive power, dynamic response parameters, and input signals from the grid side, and to calculate the power angle, angular frequency, and voltage of the grid based on the aggregated active and reactive power, dynamic response parameters, and input signals. The parameter update interface is used to identify and update the internal parameters of the transfer function module online based on real-time response data from the power grid.

Images