A method and system for optimizing the configuration of a hybrid energy storage system for a wind farm in a mountainous area

Abstract

This invention relates to the field of renewable energy grid connection and energy storage optimization configuration technology, and provides a method and system for optimizing the configuration of a hybrid energy storage system in a mountainous wind farm. The method includes: collecting multi-source data from the wind farm; calculating the wind turbine terrain correction coefficient and power deviation signal based on terrain DEM data; decomposing the power deviation signal into essential mode function components; determining the optimal boundary point based on spectral entropy and decomposing it into high- and low-frequency power components; generating power allocation instructions for energy storage devices using a temperature-lifetime coupling model in conjunction with ambient temperature data; generating a capacity configuration scheme using the MADDPG algorithm based on the power allocation instructions and grid parameters; and outputting the optimal scheme after feasibility verification and comprehensive evaluation. This invention effectively solves problems such as configuration deviation, unreasonable power decomposition, and poor adaptability to weak grids caused by complex terrain in mountainous areas, significantly improving the scientific, economic, and reliable nature of the configuration.

Description

Technical Field

[0001] This invention belongs to the field of renewable energy grid connection and energy storage optimization configuration technology, and relates to an optimization configuration method and system for a hybrid energy storage system in a mountainous wind farm. Background Technology

[0002] With the continuous growth of wind power installed capacity, configuring energy storage systems has become an effective means to mitigate power fluctuations and improve power quality. Existing energy storage capacity configuration methods mainly include those based on statistical laws, signal decomposition, optimization algorithms, and multi-agent collaboration. However, existing technologies still have significant shortcomings in addressing the complex terrain conditions of mountainous areas: on the one hand, existing methods often treat wind farms as uniform power sources, failing to fully consider the differentiated impacts of terrain factors such as altitude, slope, and shading on wind turbine output, leading to significant deviations in energy storage capacity configuration; on the other hand, power decomposition parameters are usually fixed and cannot adapt to terrain complexity and environmental temperature changes, resulting in unreasonable power allocation. Furthermore, the impact of drastic temperature changes in mountainous areas on the performance of energy storage devices has not been accurately characterized, and the weak characteristics of mountain power grids, such as small short-circuit capacity and weak voltage support, are not included in the constraints, making it difficult to balance the economy and reliability of the configuration scheme in actual operation. Therefore, there is an urgent need for an energy storage optimization configuration method that can adapt to terrain characteristics, accurately consider environmental factors, and adapt to weak power grid environments. Summary of the Invention

[0003] The purpose of this invention is to solve the technical problems of insufficient consideration of terrain factors, poor adaptability of power decomposition, inaccurate assessment of energy storage life and insufficient adaptability to weak power grids in the prior art, and to provide an optimized configuration method and system for a hybrid energy storage system in a mountainous wind farm.

[0004] To achieve the above objectives, the present invention employs the following technical solution: The first aspect of this invention provides an optimized configuration method for a hybrid energy storage system in a mountainous wind farm, comprising the following steps: Collect multi-source data from mountain wind farms, including the output power of each wind turbine, topographic DEM data, ambient temperature data, and power grid parameters; Based on the terrain DEM data, the terrain correction coefficient for each wind turbine is calculated, and the power deviation signal is calculated based on the output power of each wind turbine. The power deviation signal is decomposed into several essential mode function components; Calculate the spectral entropy of each essential mode function component based on the essential mode function components, and determine the optimal boundary point based on the spectral entropy; based on the optimal boundary point, decompose each essential mode function component into low-frequency power components and high-frequency power components; The low-frequency power component, high-frequency power component, and ambient temperature data are input into the temperature-lifetime coupling model to generate power allocation instructions for each energy storage device. Based on the power allocation command and the grid parameters, the MADDPG algorithm is used to make optimization decisions and generate a capacity configuration scheme; the feasibility of the capacity configuration scheme is checked; this step is iterated to generate a set of feasible configuration schemes that meet the feasibility check. For each of the feasible configuration schemes in the set, a comprehensive cost and mitigation effect index evaluation is performed, and the optimal capacity configuration scheme is output.

[0005] As an example, the formula for calculating the terrain correction factor is:

[0006] in, For the first Terrain correction factor for typhoon generators; For the first Altitude correction factor for typhoon generators; For the first Slope correction factor for typhoon turbines; For the first Shading correction factor for typhoon generators.

[0007] As an example, the formula for calculating the power deviation signal is:

[0008] in, express The power deviation signal at any given time; express The total equivalent power at any given moment; This is the reference power for grid connection.

[0009] As an example, VMD decomposition is used to decompose the power deviation signal into several essential mode function components; the VMD decomposition is described as follows:

[0010] The constraints are: ;in, The number of modes determined adaptively; For the first One essential modal function component; For the first The center frequency of the mode; is the _ ... The terrain-adaptive weighting coefficients for each modality are determined by the terrain complexity; is the Dirac delta function; is the imaginary unit; represents the convolution operation; This is the partial derivative with respect to time; express Norm.

[0011] As an example, the formula for calculating the spectral entropy is:

[0012] in, For the first The spectral entropy of each essential modal function component Normalized power spectral density: ; For the Fourier transform of the essential modal function components, This represents the number of frequency points.

[0013] As an example, the formula for determining the optimal boundary point is:

[0014] in, It is the first The spectral entropy of each essential modal function component.

[0015] As an example, the allocation rule expression for decomposing each essential mode function component into low-frequency power components and high-frequency power components is as follows:

[0016] in, Low-frequency power components; It represents the high-frequency power component.

[0017] As an example, the temperature-lifetime coupling model includes: Lithium-ion battery capacity decay model:

[0018] Temperature model of supercapacitor internal resistance:

[0019] Low-temperature efficiency model for hydrogen storage systems:

[0020] in: This refers to the rated capacity of the lithium battery. For optimal operating temperature, The temperature decay coefficient is These are the parameters for cyclic aging. The reference internal resistance of the supercapacitor. For temperature coefficient, The rated efficiency of the electrolytic cell, The critical start-up temperature, This is the low-temperature efficiency attenuation coefficient.

[0021] A second aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for optimizing the configuration of the hybrid energy storage system in a mountainous wind farm.

[0022] A third aspect of the present invention provides an optimized configuration system for a hybrid energy storage system in a mountainous wind farm, comprising: The data acquisition module is used to collect multi-source data from mountain wind farms. The multi-source data includes the output power of each wind turbine, terrain DEM data, ambient temperature data, and power grid parameters. The terrain perception module is used to calculate the terrain correction coefficient of each wind turbine based on the terrain DEM data, and to calculate the power deviation signal based on the output power of each wind turbine. The signal decomposition module is used to decompose the power deviation signal into several essential mode function components; The spectrum analysis module is used to calculate the spectral entropy of each essential mode function component based on the essential mode function components, determine the optimal boundary point based on the spectral entropy, and decompose each essential mode function component into low-frequency power components and high-frequency power components based on the optimal boundary point. The energy storage allocation module is used to input the low-frequency power component, high-frequency power component and ambient temperature data into the temperature-lifetime coupling model to generate power allocation instructions for each energy storage device. The multi-agent optimization module is used to make optimization decisions based on the power allocation command and the grid parameters using the MADDPG algorithm, generate capacity configuration schemes, perform feasibility checks on the capacity configuration schemes, and iteratively generate a set of feasible configuration schemes that meet the feasibility checks. The target evaluation module is used to evaluate the comprehensive cost and mitigation effect indicators of each scheme in the set of feasible configuration schemes, and output the optimal capacity configuration scheme.

[0023] Compared with the prior art, the present invention has the following beneficial effects: This invention discloses an optimized configuration method for a hybrid energy storage system in a mountainous wind farm. By collecting multi-source data, the method sequentially performs terrain correction, power deviation calculation, adaptive signal decomposition, spectral entropy frequency division, temperature-lifetime coupling modeling, multi-agent optimization, feasibility verification, and comprehensive evaluation, forming a complete energy storage capacity configuration process. This process can adapt to complex mountainous terrain and environmental temperature variations, significantly improving power decomposition accuracy and the rationality of energy storage allocation. Ultimately, it outputs a configuration scheme that synergistically optimizes economic efficiency and mitigation effects, solving the problem that traditional methods cannot simultaneously consider terrain influences, equipment lifespan, and grid adaptability.

[0024] Furthermore, by introducing the product of three terrain correction factors—altitude, slope, and shading—it is possible to accurately quantify the differentiated impact of complex mountainous terrain on the output of each wind turbine, avoiding configuration deviations caused by treating the wind farm as a uniform power source. This improves the accuracy of the equivalent power calculation for the entire field and lays a reliable foundation for subsequent power deviation signal calculation.

[0025] Furthermore, by calculating the difference between the equivalent power across the entire field and the grid-connected reference power, a power deviation signal reflecting the comprehensive impact of terrain is obtained. This signal is directly used as the input to the decomposition module, enabling subsequent frequency decomposition to accurately reflect the power fluctuation characteristics caused by mountainous terrain, thus avoiding misjudgments of power fluctuations caused by neglecting terrain factors in traditional methods.

[0026] Furthermore, variational mode decomposition (VMD) is employed to adaptively decompose the power deviation signal. By introducing terrain-adaptive weighting coefficients and the number of adaptive modes K, the decomposition process can be dynamically adjusted according to the complexity of the terrain. Compared to fixed-parameter decomposition methods, this technique can more accurately separate different frequency components, improving the physical interpretability of the decomposition results and the accuracy of subsequent frequency division.

[0027] Furthermore, by calculating the spectral entropy of each IMF component, the uncertainty of energy distribution in the frequency domain can be quantitatively characterized. A smaller spectral entropy indicates a more concentrated frequency component, while a larger one indicates a more dispersed one. This characteristic provides an objective basis for automatically determining the optimal boundary point between low and high frequencies, avoiding the subjectivity and inadequacy of manually setting the boundary frequency. The optimal boundary point is determined by finding the maximum difference in spectral entropy between adjacent IMF components. This method does not require a preset threshold and can adaptively identify inflection points where frequency components undergo significant changes, thus ensuring that the separation of low-frequency and high-frequency power components conforms to both the signal's inherent characteristics and the response characteristics required by different types of energy storage devices.

[0028] Furthermore, by establishing three temperature-lifetime coupled models—lithium battery capacity decay, supercapacitor internal resistance change, and hydrogen storage system low-temperature efficiency—the impact of drastic temperature changes in mountainous areas on the performance of various energy storage devices can be accurately characterized. This model makes energy storage lifetime prediction more accurate and the full life-cycle economic assessment more realistic and reliable, avoiding the problem of overly optimistic or conservative configuration schemes caused by ignoring temperature effects. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A block diagram of the optimized configuration method for a hybrid energy storage system in a mountainous wind farm provided by the present invention; Figure 2 The flowchart illustrates the optimized configuration method for a hybrid energy storage system in a mountainous wind farm, as provided by this invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0032] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0033] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0034] The present invention will now be described in further detail with reference to the accompanying drawings: See Figure 1 and Figure 2 This embodiment provides an optimized configuration method for a hybrid energy storage system in a mountainous wind farm. Addressing the challenges of complex terrain, variable climate, and weak power grids in mountainous areas, this method constructs a closed-loop process from data perception to optimization decision-making to achieve scientific planning of the hybrid energy storage system's capacity configuration. Specifically, it includes the following steps: Step S1: Collect multi-source data of the mountain wind farm. The multi-source data includes the output power of each wind turbine, terrain DEM data, ambient temperature data and power grid parameters.

[0035] Specifically, data acquisition is the foundation of the entire optimization configuration. In mountainous wind farms, due to the undulating terrain and microclimate, a single data source is insufficient to comprehensively reflect the operating status of the wind farm. This step involves real-time acquisition of output power using sensors deployed at the wind turbines, obtaining high-precision digital elevation model (DEM) data of the terrain through a geographic information system, and simultaneously acquiring ambient temperature and grid parameters such as voltage and frequency at the grid connection point. After cleaning and time-series alignment, this data forms a preprocessed dataset, providing accurate input for subsequent terrain sensing and power prediction. It should be understood that the acquisition frequency of multi-source data can be adjusted according to actual needs; for example, power data can be acquired at the second level, while terrain data can be updated periodically as static or quasi-static data.

[0036] Step S2: Calculate the terrain correction coefficient for each wind turbine based on the terrain DEM data, and calculate the power deviation signal based on the output power of each wind turbine.

[0037] This step aims to eliminate the differential impact of terrain factors on wind turbine output. In mountainous wind farms, altitude, slope, and shading effects can lead to significant differences in the wind energy capture capacity of turbines at different locations. Simply summing the power of each turbine would result in a deviation in the overall power prediction. Therefore, this step introduces a terrain correction coefficient to weight and correct the output power of each turbine, thereby calculating the equivalent power of the entire farm. The power deviation signal is obtained by comparing the equivalent power of the entire farm with the grid-connected reference power. It reflects the difference between the actual output of the wind farm and the grid demand, and is the target signal that needs to be smoothed by the energy storage system. This step quantifies the terrain features and integrates them into the power signal processing, providing a more physically realistic input source for subsequent signal decomposition.

[0038] Step S3: Decompose the power deviation signal into several essential mode function components.

[0039] Power deviation signals typically contain fluctuations of multiple frequencies, including low-frequency trend fluctuations caused by terrain and high-frequency random fluctuations caused by turbulence. This step employs signal decomposition technology to break down the complex power deviation signal into several relatively independent essential mode function components. This decomposition method can isolate the fluctuation characteristics at different time scales, allowing for targeted processing of fluctuations with different frequency characteristics using energy storage media with different properties. For example, low-frequency components are suitable for energy storage, while high-frequency components are suitable for power storage. This decomposition avoids the lifespan degradation problem caused by frequent charging and discharging of a single energy storage medium.

[0040] Step S4: Calculate the spectral entropy of each essential mode function component based on the essential mode function components, and determine the optimal boundary point based on the spectral entropy; based on the optimal boundary point, decompose each essential mode function component into low-frequency power components and high-frequency power components.

[0041] This step addresses the problem of unreasonable power allocation caused by fixed boundary frequencies in traditional methods. Spectral entropy quantifies the complexity and uncertainty of a signal in the frequency domain. By calculating the spectral entropy of each component, the frequency distribution characteristics of each component can be objectively evaluated. The optimal boundary point is determined based on the principle of maximizing the difference in spectral entropy between adjacent components. This principle adaptively identifies the boundary between low-frequency trends and high-frequency fluctuations. Based on this boundary point, the essential mode function components are reorganized into low-frequency power components and high-frequency power components, achieving refined power allocation. This adaptive allocation mechanism enables the energy storage system to dynamically adjust its strategy according to actual signal characteristics, improving the adaptability of the configuration scheme.

[0042] Step S5: Input the low-frequency power component, high-frequency power component, and ambient temperature data into the temperature-lifetime coupling model to generate power allocation instructions for each energy storage device.

[0043] The drastic temperature variations in mountainous areas significantly impact the performance and lifespan of energy storage devices. For example, low temperatures reduce the usable capacity of lithium batteries, while high temperatures accelerate aging. This step establishes a temperature-lifespan coupled model, incorporating ambient temperature as a key variable into power allocation decisions. Based on current temperature conditions, the model adjusts parameters such as the charging and discharging efficiency and capacity decay rate of the energy storage devices, thereby generating reasonable power allocation commands. This not only ensures effective power smoothing but also extends the lifespan of the energy storage system through operational strategies, reducing total lifespan costs.

[0044] Step S6: Based on the power allocation command and the grid parameters, the MADDPG algorithm is used to make optimization decisions and generate a capacity configuration scheme.

[0045] The optimal configuration of hybrid energy storage systems is a complex problem with multiple objectives and constraints. This step employs the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm for solution. This algorithm sets up multiple intelligent agents, representing different energy storage media such as lithium batteries, supercapacitors, and hydrogen storage systems. These agents interact and learn in a collaborative environment. By designing a reward function that incorporates cost, technical performance, lifetime, and grid adaptability, the algorithm guides each agent to explore the optimal capacity configuration strategy. Compared to traditional centralized optimization algorithms, the MADDPG algorithm offers higher solution efficiency and stronger robustness, making it particularly suitable for handling high-dimensional, nonlinear optimization problems such as those in mountainous wind farms.

[0046] Step S7: Perform a feasibility test on the capacity configuration scheme; iterate this step to generate a set of feasible configuration schemes that meet the feasibility test.

[0047] Mountainous power grids typically exhibit weaknesses such as small short-circuit capacity and weak voltage support. Directly applying optimization algorithms to generate solutions may lead to voltage exceeding limits or frequency instability during actual operation. Therefore, this step introduces a feasibility verification process, imposing constraints such as voltage stability, short-circuit capacity, virtual inertia, and low-voltage ride-through on the generated capacity configuration schemes. If a scheme fails to meet these constraints, an iterative mechanism is used to adjust the optimization parameters and re-determine the solution until a set of feasible configuration schemes that satisfy all safety constraints is generated. This step ensures the feasibility of the configuration scheme in practical engineering.

[0048] Step S8: Evaluate the comprehensive cost and mitigation effect of each scheme in the set of feasible configuration schemes, and output the optimal capacity configuration scheme.

[0049] The feasible configuration scheme set contains multiple schemes that meet security constraints, and their overall performance needs further evaluation. This step constructs an evaluation index system that includes overall cost (investment, operation and maintenance, losses, etc.) and mitigation effects (power volatility, tracking error, etc.). By quantitatively evaluating each scheme, the scheme with the best overall index is selected as the final output. This ensures the optimal balance between economics and technicality in the configuration scheme.

[0050] Step S9: Deploy the optimal capacity configuration scheme to the energy storage control system and realize online monitoring and adaptive updates.

[0051] The operating environment and equipment status of wind farms are dynamically changing. This step deploys the optimization results to the actual control system and establishes an online monitoring mechanism. When a performance deviation exceeds a threshold, an adaptive update process is triggered to re-execute the optimization configuration, thereby achieving optimal operation of the energy storage system throughout its entire lifecycle.

[0052] Through the synergistic effect of steps S1 to S9 above, this embodiment constructs a complete optimized configuration system for a hybrid energy storage system in a mountainous wind farm. From data acquisition, terrain perception, signal decomposition, power allocation to optimization decision-making and feasibility verification, each link is closely connected, fully considering the actual constraints such as the complexity of mountainous terrain, environmental temperature sensitivity, and grid weakness, thus achieving scientific, refined, and adaptive energy storage configuration.

[0053] An embodiment of the present invention provides an optimized configuration method for a hybrid energy storage system in a mountainous wind farm, which is implemented according to the following steps: Step 1: Collect multi-source data from the mountain wind farm, including the output power of each wind turbine, terrain DEM data, ambient temperature data, and grid parameters. After data cleaning and time-series alignment, output a preprocessed dataset. Step 2: Input the preprocessed dataset obtained in Step 1 into the terrain perception module, calculate the terrain correction coefficient for each wind turbine and perform weighted fusion, and output the equivalent power and power deviation signal of the entire field; The formula for calculating the terrain correction factor is: (1) in, For the first Terrain correction factor for typhoon generators; For the first Altitude correction factor for typhoon generators; For the first Slope correction factor for typhoon turbines; For the first Shading correction factor for typhoon generators.

[0054] The altitude correction factor is: (2) The slope correction factor is: (3) The occlusion correction factor is: (4) in, For the first Typhoon altitude For slope, The area to be covered.

[0055] The formula for calculating the equivalent power across the entire field is: (5) The power deviation signal is: (6) in, For grid connection reference power, As input for step 3.

[0056] Step 3: Input the power deviation signal obtained in Step 2 into the terrain adaptive variational mode decomposition module, adaptively determine the number of modes and perform decomposition, and output several essential mode function components; The adaptive formula for determining the number of modes in terrain-adaptive variational mode decomposition is as follows: (7) in, To adapt the number of modes, As the reference mode number, For adjustment coefficients, The standard deviation of the topographic factor reflects the complexity of the terrain. The reference standard deviation; The formula for correcting the temperature at the boundary frequency is: (8) in, As the reference boundary frequency, For temperature sensitivity coefficient, For reference temperature, The ambient temperature.

[0057] The improved VMD variational problem is: (9) The constraints are: (10) Decomposed output Individual eigenmode functions As input for step 4.

[0058] Step 4: Input the essential mode function components obtained in Step 3 into the spectrum analysis module, calculate the spectral entropy of each component and determine the optimal boundary point, and output the low-frequency power component and the high-frequency power component. The formula for calculating spectral entropy is: (11) in, For the first Spectral entropy of each IMF component Normalized power spectral density: (12) Fourier transform of the IMF components Number of frequency points; The formula for determining the optimal boundary point is: (13) The power allocation rules are as follows: (14) in, and As input for step 5; Low-frequency power components; It represents the high-frequency power component.

[0059] Step 5: Input the low-frequency power component and high-frequency power component obtained in Step 4 into the energy storage distribution module, and combine them with the temperature-lifetime coupling model to output the power distribution command for each energy storage device. Temperature-lifetime coupled energy storage models include: Lithium-ion battery capacity decay model: (15) Temperature model of supercapacitor internal resistance: (16) Low-temperature efficiency model for hydrogen storage systems: (17) in: This refers to the rated capacity of the lithium battery. For optimal operating temperature, The temperature decay coefficient is These are the parameters for cyclic aging. The reference internal resistance of the supercapacitor. For temperature coefficient, The rated efficiency of the electrolytic cell, The critical start-up temperature is -10℃. This is the low-temperature efficiency attenuation coefficient.

[0060] The energy storage state equation is: (18) The output power allocation instructions for each energy storage device are used as input for step 6.

[0061] Step 6: Input the power allocation command obtained in Step 5 and the grid parameters collected in Step 1 into the multi-agent optimization module. Each agent makes collaborative decisions and iterative updates to output the capacity configuration scheme. The multi-agent optimization module is configured with 4 smart agents, and the multi-objective reward function is as follows: (19) in: (20) (twenty one) The policy update uses the MADDPG algorithm: (twenty two) The formula for accelerating transfer learning is: (twenty three) Scene similarity: (twenty four) Output capacity configuration scheme As input for step 7.

[0062] Step 7: Input the capacity configuration scheme obtained in Step 6 into the constraint verification module, apply the weak adaptability constraint of the mountain power grid to perform feasibility verification, and output a set of feasible configuration schemes; Mountainous power grid weakness adaptability constraints include: Voltage stability constraints: (25) Short-circuit capacity constraints: (26) Virtual inertia constraint: (27) Low voltage ride-through constraint: (28) in, For the short-circuit capacity of the grid connection point, To allow for power fluctuations, Rated voltage, This is the rated frequency.

[0063] The output set of feasible configuration schemes serves as input for step 8.

[0064] Step 8: Input the set of feasible configuration schemes obtained in Step 7 into the target evaluation module, calculate the overall cost and the smoothing effect index, and output the optimal capacity configuration scheme; The objective evaluation function is: (29) The investment cost takes into account the additional factor for transportation and installation in mountainous areas: (30) Maintenance costs take into account the difficulty of maintenance in mountainous areas: (31) in, Additional coefficient for transportation and installation in mountainous areas. The difficulty of operation and maintenance in mountainous areas, For the cost of wind curtailment, Revenue from power generation and ancillary services.

[0065] The evaluation indicators for the stabilization effect are: (32) The optimal capacity configuration scheme is output as input for step 9.

[0066] Step 9: Deploy the optimal capacity configuration scheme obtained in Step 8 to the energy storage control system and realize online monitoring and adaptive updates.

[0067] The online update mechanism is as follows: Performance evaluation metrics: (33) when Time-triggered model update: (34) in, This is the performance deviation threshold. For learning rate, This is the policy loss function.

[0068] One embodiment of the present invention provides a computer-readable storage medium. The computer-readable storage medium stores a computer program that, when executed by a processor, implements the optimized configuration method for a hybrid energy storage system in a mountainous wind farm as described in the above embodiment.

[0069] Specifically, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this embodiment, a computer-readable storage medium may include, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory, magnetic storage (such as magnetic tape, hard disk), optical storage (such as compact optical disc read-only memory CD-ROM, digital versatile optical disc DVD), or any combination of the above media. It should be understood that a computer-readable storage medium is not limited to the specific forms listed above; any medium capable of storing a computer program and being read by a processor should be included within the scope of protection of this invention.

[0070] When the computer program is executed by the processor, the processor will sequentially call the code instructions stored in the storage medium according to a preset logical order, and perform the following steps: First, the processor reads the collected multi-source data of the mountain wind farm, calls the terrain perception algorithm module in the storage medium, and calculates the terrain correction coefficient and power deviation signal of each wind turbine; next, it calls the signal decomposition module, executes the VMD decomposition algorithm, and decomposes the power deviation signal into several essential mode function components; then, it calls the spectrum analysis module, calculates the spectrum entropy and determines the optimal boundary point, and reconstructs the high and low frequency power components; further, it calls the energy storage allocation module, combines the temperature-lifetime coupling model to generate power allocation instructions; finally, it calls the multi-agent optimization module, uses the MADDPG algorithm for iterative optimization, and outputs the optimal capacity configuration scheme after feasibility verification and target evaluation. In this way, this embodiment solidifies the complex mathematical model and optimization algorithm into executable program code, enabling wind farm operation and maintenance personnel to easily deploy the method, significantly improving the automation level and decision-making efficiency of the configuration of the hybrid energy storage system in the mountain wind farm, and ensuring the scientificity and reliability of the configuration scheme.

[0071] This embodiment provides an optimized configuration system for a hybrid energy storage system in a mountainous wind farm. This system implements the aforementioned optimized configuration method for a hybrid energy storage system in a mountainous wind farm. Through modular design, the system encapsulates complex data processing, signal decomposition, and optimization decision-making logic into independent functional units, improving the system's maintainability and scalability.

[0072] Specifically, the system includes a data acquisition module, a terrain sensing module, a signal decomposition module, a spectrum analysis module, an energy storage and distribution module, a multi-agent optimization module, and a target evaluation module.

[0073] The data acquisition module is used to collect multi-source data from mountain wind farms. The multi-source data includes the output power of each wind turbine, terrain DEM data, ambient temperature data, and power grid parameters.

[0074] Specifically, the data acquisition module is the perception layer of the entire system. In practical applications, this module may include power sensors deployed at the wind turbine, temperature sensors arranged within the wind farm, and interface units for communication with the power grid dispatch system. This module acquires data in real time via wired or wireless networks (such as fiber optic ring networks or 4G / 5G private networks), performs preliminary filtering and analog-to-digital conversion, transforming the raw physical quantities into digital signals for subsequent processing. It should be understood that the data acquisition module is not limited to sensor hardware; it may also include devices such as data acquisition cards and remote terminal units (RTUs).

[0075] The terrain perception module is used to calculate the terrain correction coefficient of each wind turbine based on the terrain DEM data, and to calculate the power deviation signal based on the output power of each wind turbine.

[0076] This module receives terrain DEM data and wind turbine power data output from the data acquisition module. At the hardware level, the terrain perception module can be composed of a microprocessor or a digital signal processor (DSP), internally storing pre-loaded terrain correction algorithm programs. By executing the program, this module calculates the terrain correction coefficient for each wind turbine, then weights these to obtain the equivalent power for the entire field, and subtracts the grid-connected reference power to obtain the power deviation signal. This process achieves the digital quantification of the impact of complex mountainous terrain, providing a corrected input source for subsequent signal processing.

[0077] The signal decomposition module is used to decompose the power deviation signal into several essential mode function components.

[0078] The signal decomposition module is connected to the terrain sensing module and receives the power deviation signal as input. This module integrates variational mode decomposition (VMD) algorithm logic. During processing, this module adaptively adjusts the decomposition parameters according to the terrain complexity, decomposing the non-stationary power deviation signal into multiple essential mode function components with different frequency scales. This decomposition decouples the complex power fluctuation characteristics, facilitating subsequent differentiated processing based on different frequency characteristics.

[0079] The spectrum analysis module is used to calculate the spectral entropy of each essential mode function component based on the essential mode function components, determine the optimal boundary point based on the spectral entropy, and decompose each essential mode function component into low-frequency power components and high-frequency power components based on the optimal boundary point.

[0080] This module receives the essential mode function components output by the signal decomposition module. Its core function is to perform frequency domain feature analysis on the decomposed components. By calculating the spectral entropy, this module can quantify the complexity of each component and automatically search for the position with the largest difference in spectral entropy between adjacent components as the optimal dividing point. Based on this dividing point, the spectrum analysis module reconstructs the essential mode function components into low-frequency power components and high-frequency power components, thereby realizing adaptive frequency division of the power signal.

[0081] The energy storage allocation module is used to input the low-frequency power component, high-frequency power component and ambient temperature data into the temperature-lifetime coupling model to generate power allocation instructions for each energy storage device.

[0082] The energy storage distribution module is connected to both the spectrum analysis module and the data acquisition module to obtain power component and ambient temperature data. This module has pre-built temperature-lifetime coupling models for the lithium battery, supercapacitor, and hydrogen storage system. During operation, the module adjusts the performance parameters of each energy storage medium based on the current ambient temperature and, combined with the frequency characteristics of the power component, generates power distribution commands that match the characteristics of each energy storage medium. This ensures that the energy storage system operates at its optimal state under different temperature conditions, extending the equipment's lifespan.

[0083] The multi-agent optimization module is used to make optimization decisions based on the power allocation command and the grid parameters using the MADDPG algorithm, generate capacity configuration schemes, perform feasibility checks on the capacity configuration schemes, and iteratively generate a set of feasible configuration schemes that meet the feasibility checks.

[0084] This module is the core decision-making unit of the system. Internally, it constructs a multi-agent collaborative environment, setting up multiple intelligent agents corresponding to different energy storage media. This module receives instructions from the energy storage allocation module and grid parameters, and uses the MADDPG algorithm for reinforcement learning training to continuously explore the optimal capacity configuration strategy. Simultaneously, this module incorporates feasibility verification logic, imposing constraints such as voltage stability and short-circuit capacity on the generated schemes, and iteratively filters out a set of feasible configuration schemes that meet the safety constraints. This design distributes the processing of complex optimization problems, significantly improving solution efficiency.

[0085] The target evaluation module is used to evaluate the comprehensive cost and mitigation effect indicators of each scheme in the set of feasible configuration schemes, and output the optimal capacity configuration scheme.

[0086] The target evaluation module is connected to the multi-agent optimization module. For each option in the feasible configuration set, this module calculates its total cost over its entire lifecycle (including investment and maintenance costs) and power mitigation performance indicators (such as volatility and tracking error). Through comprehensive evaluation, this module ranks the options from multiple dimensions and ultimately outputs the capacity configuration option with the best overall performance, achieving the optimal balance between economy and technology.

[0087] Through the collaborative work of the aforementioned modules, this embodiment constructs a complete optimized configuration system for a hybrid energy storage system in a mountainous wind farm. The modules communicate via a data bus or network interface, achieving fully automated processing from data sensing, feature extraction, power allocation to optimization decision-making. This system not only accurately adapts to the complex terrain and variable climate of mountainous areas but also effectively ensures safe and stable operation in environments with weak power grids, significantly improving the scientific rigor and engineering practicality of wind farm energy storage configuration.

[0088] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing configuration of a hybrid energy storage system for a mountainous wind farm, characterized in that, Includes the following steps: Collect multi-source data from mountain wind farms, including the output power of each wind turbine, topographic DEM data, ambient temperature data, and power grid parameters; Based on the terrain DEM data, the terrain correction coefficient for each wind turbine is calculated, and the power deviation signal is calculated based on the output power of each wind turbine. The power deviation signal is decomposed into several essential mode function components; Calculate the spectral entropy of each essential mode function component based on the essential mode function components, and determine the optimal boundary point based on the spectral entropy; based on the optimal boundary point, decompose each essential mode function component into low-frequency power components and high-frequency power components; The low-frequency power component, high-frequency power component, and ambient temperature data are input into the temperature-lifetime coupling model to generate power allocation instructions for each energy storage device. Based on the power allocation command and the grid parameters, the MADDPG algorithm is used to make optimization decisions and generate a capacity configuration scheme. The feasibility of the proposed capacity configuration scheme shall be verified. Iterate through this step to generate a set of feasible configuration schemes that satisfy the feasibility test; For each of the feasible configuration schemes in the set, a comprehensive cost and mitigation effect index evaluation is performed, and the optimal capacity configuration scheme is output.

2. The optimized configuration method for a hybrid energy storage system in a mountainous wind farm according to claim 1, characterized in that, The formula for calculating the terrain correction factor is as follows: in, For the first Terrain correction factor for typhoon generators; For the first Altitude correction factor for typhoon generators; For the first Slope correction factor for typhoon turbines; For the first Shading correction factor for typhoon generators.

3. The optimized configuration method for a hybrid energy storage system in a mountainous wind farm according to claim 1, characterized in that, The formula for calculating the power deviation signal is: in, express The power deviation signal at any given time; express The total equivalent power at any given moment; This is the reference power for grid connection.

4. The optimized configuration method for a hybrid energy storage system in a mountainous wind farm according to claim 1, characterized in that, The power deviation signal is decomposed into several essential mode function components using VMD decomposition; the VMD decomposition is described as follows: The constraints are: ;in, The number of modes determined adaptively; For the first One essential modal function component; For the first The center frequency of the mode; is the _ ... The terrain-adaptive weighting coefficients for each modality are determined by the terrain complexity; is the Dirac delta function; is the imaginary unit; represents the convolution operation; This is the partial derivative with respect to time; express Norm.

5. The optimized configuration method for a hybrid energy storage system in a mountainous wind farm according to claim 1, characterized in that, The formula for calculating the spectral entropy is: in, For the first The spectral entropy of each essential modal function component Normalized power spectral density: ; For the Fourier transform of the essential modal function components, This represents the number of frequency points.

6. The optimized configuration method for a hybrid energy storage system in a mountainous wind farm according to claim 5, characterized in that, The formula for determining the optimal dividing point is: in, It is the first The spectral entropy of each essential modal function component.

7. The optimized configuration method for a hybrid energy storage system in a mountainous wind farm according to claim 5, characterized in that, The allocation rule expression for decomposing each essential mode function component into low-frequency power components and high-frequency power components is as follows: in, Low-frequency power components; It represents the high-frequency power component.

8. The method for optimizing the configuration of a hybrid energy storage system for a mountain wind farm according to claim 1, characterized in that, The temperature-lifetime coupling model includes: Lithium-ion battery capacity decay model: Temperature model of supercapacitor internal resistance: Low-temperature efficiency model for hydrogen storage systems: in: The rated capacity of the lithium battery, For optimal operating temperature, The temperature decay coefficient is These are the parameters for cyclic aging. The reference internal resistance of the supercapacitor. For temperature coefficient, The rated efficiency of the electrolytic cell, The critical start-up temperature, This is the low-temperature efficiency attenuation coefficient.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the method for optimizing the configuration of a hybrid energy storage system for a mountain wind farm as described in any one of claims 1-8.

10. An optimized configuration system for a hybrid energy storage system in a mountainous wind farm, characterized in that, include: The data acquisition module is used to collect multi-source data from mountain wind farms. The multi-source data includes the output power of each wind turbine, terrain DEM data, ambient temperature data, and power grid parameters. The terrain perception module is used to calculate the terrain correction coefficient of each wind turbine based on the terrain DEM data, and to calculate the power deviation signal based on the output power of each wind turbine. The signal decomposition module is used to decompose the power deviation signal into several essential mode function components; The spectrum analysis module is used to calculate the spectral entropy of each essential mode function component based on the essential mode function components, determine the optimal boundary point based on the spectral entropy, and decompose each essential mode function component into low-frequency power components and high-frequency power components based on the optimal boundary point. The energy storage allocation module is used to input the low-frequency power component, high-frequency power component and ambient temperature data into the temperature-lifetime coupling model to generate power allocation instructions for each energy storage device. The multi-agent optimization module is used to make optimization decisions based on the power allocation command and the grid parameters using the MADDPG algorithm, generate capacity configuration schemes, perform feasibility checks on the capacity configuration schemes, and iteratively generate a set of feasible configuration schemes that meet the feasibility checks. The target evaluation module is used to evaluate the comprehensive cost and mitigation effect indicators of each scheme in the set of feasible configuration schemes, and output the optimal capacity configuration scheme.

Images