Offshore energy island site selection and sizing method

Abstract

This invention relates to the field of comprehensive utilization and optimization of marine renewable energy, and in particular to a method for site selection and capacity determination of offshore energy islands. This method, through systematic analysis at the frequency domain level, constructs an index system capable of comprehensively diagnosing the complementary characteristics of wind, photovoltaic, and wave energy resources, quantifying system output stability, and assessing resource contribution, and further forms a unified comprehensive optimization index. This comprehensive optimization index can be directly applied to the equipment configuration, energy storage capacity planning, and operation strategy formulation of energy islands, thereby providing a scientific and reliable decision-making basis for the planning, design, and operation management of energy islands.

Description

Technical Field

[0001] This invention relates to the field of comprehensive utilization and optimization of marine renewable energy, and in particular to a method for site selection and capacity determination of marine energy islands. Background Technology

[0002] With the rapid development and large-scale application of renewable energy sources such as wind, solar, and wave energy, the integrated offshore energy development model—energy island—has attracted much attention due to its ability to intensively utilize multiple resources. However, wind, solar, and wave energy resources all exhibit significant randomness and intermittency in their time series, resulting in large fluctuations and insufficient stability in the overall output power of the energy island, thus posing a severe challenge to its reliable grid connection and continuous power supply.

[0003] Currently, assessment methods for resource complementarity largely rely on simple statistical indicators (such as mean and variance) or pairwise cross-correlation analysis. While these methods can reflect some time-domain correlation characteristics to a certain extent, they struggle to fully reveal the coupling mechanisms and dynamic interactions of multi-energy systems at the frequency domain level, and are even less effective in assessing the complementary potential at different time scales (such as intraday, weather, and seasonal). Due to the lack of in-depth analysis of system spectral characteristics, phase relationships, and energy contribution distribution, existing assessment results are insufficient to provide accurate and reliable basis for energy island equipment capacity configuration, energy storage system selection, and operation scheduling strategies.

[0004] Chinese patent application CN116341952A discloses a method for assessing the joint development of marine wind and wave resources. This patent achieves a macro-level assessment of the potential for joint development of wind and wave energy by calculating the inherent stability, correlation coefficient, and threshold-based complementarity coefficient of wind and wave energy. However, its shortcomings are: (1) the assessment objects are limited to wind and wave energy resources and do not cover photovoltaic resources; (2) the coefficient of variation and correlation coefficient used are time-domain indicators, which cannot reveal the resource coupling mechanism and dynamic complementary potential at different time scales in the frequency domain; (3) the main function is positioned at macro-level potential classification, and no optimization model is established that is connected with specific equipment capacity configuration, energy storage planning, and economic operation indicators.

[0005] Chinese patent application CN118396433A discloses a site selection method for joint development of marine energy. After constructing an index system covering resource reserves, stability, complementarity and natural conditions for macro-screening, the patent further uses cross-spectral analysis to compare representative test sites to determine the optimal site. However, its shortcomings are: (1) Although cross-spectral analysis is introduced, the core evaluation system still relies on traditional time-domain statistical indicators and fails to construct a systematic evaluation framework with frequency domain analysis as the core; (2) Only the wind energy and wave energy of a few pre-selected sites are analyzed for complementarity, without extending to wind, solar and wave variables, and without dividing frequency bands for multi-scale diagnosis; (3) Only the optimal site is output, without further combining frequency domain analysis with energy island grading.

[0006] Chinese patent application CN120765124A discloses a method for selecting sites for arrayed development of wave energy. This patent achieves the optimal selection of sites for arrayed development of wave energy through steps such as GIS spatial analysis, screening of evaluation index system, wave direction statistical analysis, and simulation of energy-concentrating structure gain. However, its shortcomings are: (1) it focuses on the arrayed development site selection of a single wave energy resource and does not involve the complementary analysis of multiple resources such as wind energy and photovoltaics; (2) the evaluation system and site selection are mainly optimized for wave energy resource endowment and engineering constraints, and a comprehensive evaluation system involving multi-energy systems is not constructed; (3) it does not conduct in-depth analysis of resource sequences from the frequency domain perspective and cannot provide differentiated basis for energy storage configuration based on the fluctuation characteristics of different time scales.

[0007] In conclusion, traditional assessment systems are insufficient when dealing with the complex systems engineering of energy islands, and there is an urgent need to propose a method for the site selection and capacity determination of offshore energy islands. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for site selection and capacity determination of offshore energy islands. This method involves conducting systematic analysis at the frequency domain level to construct an index system capable of comprehensively diagnosing the complementary characteristics of wind, photovoltaic, and wave energy resources, quantifying system output stability, and assessing resource contribution. Furthermore, it forms a unified comprehensive optimization index. This comprehensive optimization index can be directly applied to the equipment configuration, energy storage capacity planning, and operation strategy formulation of energy islands, thereby providing a scientific and reliable decision-making basis for the planning, design, and operation management of energy islands.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A method for site selection and capacity determination of offshore energy islands includes the following steps:

[0011] (1) Obtain ERA5 multidimensional reanalysis data of the target sea area to construct the power time series, and complete the preprocessing work of mean removal, normalization and stationarity test;

[0012] (2) Construct a three-variable power spectrum matrix and use the matrix inversion method to calculate the partial coherence function in order to eliminate the coupling interference of the third variable and analyze the basic data for evaluating the true complementarity between resources;

[0013] (3) A dual masking function of phase and gain is introduced to correct the evaluation of true complementarity, and the entire spectrum is divided into three characteristic frequency bands: intraday, weather and seasonal.

[0014] (4) The subjective and objective weights are calculated by integrating the analytic hierarchy process and the entropy weight method. The optimal weight vector under Nash equilibrium is solved by using the game theory combined weighting model, thereby constructing a three-dimensional evaluation framework covering complementarity, stability and contribution, and forming a comprehensive complementarity index.

[0015] (5) Based on the spatial distribution characteristics of the comprehensive complementarity index, high-value areas are delineated to identify priority candidate sites for the construction of the energy island;

[0016] (6) Establish a two-level optimization model with the goal of minimizing the levelized cost of electricity and maximizing the comprehensive complementarity index. Introduce physical constraints and adaptive frequency domain response strategies based on the energy integral of each frequency band spectrum. Use a multi-objective evolutionary algorithm to solve the optimal capacity configuration of wind-solar-wave-storage.

[0017] In step (1), ERA5 or similar reanalysis data of the target sea area are obtained, and the wind data mainly includes meridional wind speed. u and zonal wind speed v Key data on light include solar irradiance. G Wave data mainly includes significant wave height. H s Spectral peak period T p .

[0018] In step (2), based on the preprocessed power time series, the autospectral density and cross-spectral density are calculated to construct the spectral matrix of the three variables of wind, solar and wave. S ( f ).

[0019] In step (3), for any two resources X and Y, their relative position spectra are... F XY ( f ) is its cross-spectral density S XY ( f The argument of a frequency is described by its angle of reference. f Above, resources X and Y Phase lag relationship between them; gain spectrum G XY ( f ) indicates frequencyf Above, resources Y Relative to resources X The degree of linear dependence; constructing phase masks using Gaussian kernels respectively. M Φ ( f ) and gain mask M G ( f The above values ​​are decomposed and aggregated into frequency bands. The entire spectrum is divided into three frequency bands according to different time scales of interest in engineering (intraday, weather, season). Then, the spectral indices (such as partial coherence, phase mask, gain mask, etc.) in each frequency band are averaged to obtain representative indices for each frequency band.

[0020] Step (5) Calculate the value of each grid point within the simulation area. CSD The index was plotted using MATLAB tools. CSD Exponential spatial distribution cloud map. Based on the cloud map, delineate... CSD Areas with higher indices represent potential development zones for the energy island. Representative test sites were selected within these potential development zones to analyze their zoning diagnostic results (i.e.,...). C (b) , S (b) , D (b) The specific details of the diagnosis are as follows:

[0021] If the intraday frequency band complementarity at a certain point C (1) High stability indicates good short-term complementarity, which can reduce the need for power-type energy storage configurations. If the frequency band stability at a certain point... S (2) High values ​​indicate large weather-scale fluctuations, necessitating sufficient energy storage capacity. If the seasonal frequency band contribution at a certain point... D (3) Equilibrium (entropy) H (3) (High) indicates balanced resource utilization and allows for more diverse equipment configurations.

[0022] Step (6): Within the priority candidate sites designated in step (5), based on the frequency band diagnostic results obtained in step (3), construct a two-layer optimization model considering frequency domain characteristic coupling to determine the optimal configuration of wind turbines, photovoltaic, wave energy devices, and energy storage systems. Detailed calculation steps are as follows:

[0023] 1) Define decision variables;

[0024] 2) Construct a multi-objective optimization function.

[0025] 3) Before executing the two-level optimization algorithm, the following four types of key parameters need to be preset: equipment technical parameters, economic model parameters, spectral analysis feature constraint thresholds, and optimization algorithm parameters.

[0026] 4) When constructing a two-level optimization model, it is first necessary to establish a set of basic constraints to ensure the energy conservation of the system and the safe operation of the equipment. These constraints include: real-time power balance constraints, energy storage system state and operation constraints, power supply reliability constraints, decision variable boundary constraints, and setting physical constraints based on frequency band characteristics.

[0027] 5) Develop a frequency domain response operation strategy:

[0028] 6) The above model is solved using the multi-objective optimization algorithm (NSGA-II) to obtain the Pareto optimal solution set, and the knee point is selected as the final recommended equipment configuration scheme.

[0029] The beneficial effects of this invention are:

[0030] 1) Based on the overall planning of wind energy, photovoltaic energy and wave energy, this invention constructs a three-dimensional evaluation system of complementarity, stability and contribution at the frequency domain level, which effectively breaks through the limitations of traditional time domain statistics and single cross-correlation analysis, and realizes a comprehensive and accurate assessment of the time domain and frequency domain coupling characteristics of multi-energy resources;

[0031] 2) This invention eliminates the common driving effect of the third variable by using partial coherence analysis, thereby eliminating spurious correlation interference and making the evaluation results of the complementarity between wind energy, photovoltaic and wave energy resources more consistent with the real coupling relationship, thus significantly improving the credibility of the evaluation conclusions.

[0032] 3) This invention incorporates phase mask and gain mask into the evaluation scoring system, effectively avoiding the problem of "artificially high" complementary evaluation results, while realizing quantitative constraints on power amplitude matching, thus enhancing the guiding value of the evaluation results for engineering practice;

[0033] 4) Based on the frequency band decomposition method, this invention realizes the complementarity and stability diagnosis of multiple time scales such as intraday, weather and season, and can provide differentiated basis for energy storage capacity selection and operation scheduling strategies at different scales.

[0034] 5) The Comprehensive Complementary Index (CSD) constructed in this invention can simultaneously realize the integrated decision-making of macro-level site selection and micro-level equipment capacity determination for energy islands. Combined with the three-dimensional evaluation model of game theory combination weighting, it ensures the scientific balance of site selection and configuration schemes in terms of economy and technology.

[0035] 6) The dual-layer optimization model established in this invention incorporates frequency domain feature constraints and power supply reliability thresholds, and combines the NSGA-II algorithm with frequency domain response strategies. This achieves a balance between economic efficiency (optimization of levelized cost of electricity (LCOE)) and technical efficiency (maximization of CSD exponent), while also ensuring the power supply stability of the system throughout its entire life cycle.

[0036] 7) This invention adopts a weighting strategy of "global weighting and local mapping", which not only captures the resource distribution characteristics of the entire sea area, but also avoids the computational redundancy caused by grid-by-grid weighting, thus taking into account both evaluation accuracy and engineering calculation efficiency. Attached Figure Description

[0037] Figure 1 This is a diagram showing the energy distribution characteristics at different time scales in an embodiment of the present invention;

[0038] Figure 2 This is a spectral analysis diagram from an embodiment of the present invention;

[0039] Figure 3 This is a complementary polar coordinate diagnostic diagram under the weather frequency band in an embodiment of the present invention;

[0040] Figure 4 This is a power continuity curve diagram in an embodiment of the present invention;

[0041] Figure 5 This is a comparison chart of the spectrum smoothing effect in embodiments of the present invention;

[0042] Figure 6 This is a spectral attenuation efficiency diagram in an embodiment of the present invention;

[0043] Figure 7 This is a probability distribution diagram of the system modulation effect in an embodiment of the present invention;

[0044] Figure 8 This is a diagram showing the spatial trajectory of the battery in an embodiment of the present invention;

[0045] Figure 9 This is a typical weekly chart showing high volatility in an embodiment of the present invention. Detailed Implementation

[0046] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0047] The structures, proportions, and sizes illustrated in the accompanying drawings are merely for illustrative purposes and to aid those skilled in the art in understanding and reading the invention. They are not intended to limit the scope of the invention and therefore have no substantial technical significance. Any modifications to the structure, changes in proportions, or adjustments to size, provided they do not affect the effectiveness or purpose of the invention, should still fall within the scope of the technical content disclosed herein. Furthermore, the terms "upper," "lower," "left," "right," "middle," and "one" used in this specification are merely for clarity and not intended to limit the scope of the invention. Changes or adjustments to their relative relationships, without substantially altering the technical content, should also be considered within the scope of the invention's implementation.

[0048] A method for site selection and capacity determination of offshore energy islands includes the following steps:

[0049] 1. Obtain ERA5 multidimensional reanalysis data of the target sea area to construct a power time series, and complete the preprocessing work of mean removal, normalization and stationarity test;

[0050] Obtain ERA5 data or reanalysis of similar data for the target sea area. Wind data mainly includes meridional wind speed. u and zonal wind speed v Key data on light include solar irradiance. G Wave data mainly includes significant wave height. H s Spectral peak period T p .

[0051] The power conversion of the above data is calculated using the following formula:

[0052] Wind power density (unit: W / m³) 2 ): , where ρ a Air density (unit: kg / m³) 3 ), U Wind speed (unit: m / s) .

[0053] Photovoltaic power density (unit: W / m²) 2 ): P solar = the G , of which or Photovoltaic conversion efficiency (dimensionless) can be determined based on the selected photovoltaic technology type, and typically ranges from 0.15 to 0.22. G Total solar irradiance (unit: W / m²) 2 ).

[0054] Photovoltaic day and night are segmented and processed according to solar irradiance. G ( t Determine day and night, and correct the photovoltaic power sequence. P solar ( t )

[0055]

[0056] in G threshold To determine the irradiance threshold at night (typically close to 10-20 W / m²), 2 ).

[0057] Wave power density (unit) KW / m): ,in r Seawater density (units) g / cm 3 g is the acceleration due to gravity (unit: g). m / s 2 ), H s Significant wave height (unit) m ), T e Energy cycle (unit) s ), T e ≈0.9T p .

[0058] Power density time series for each resource (wind, solar, wave) P ( t Convert it to a zero-mean sequence, and the calculation method is as follows:

[0059] ,

[0060] in, The mean of the sequence. N The sequence length is given.

[0061] The Min-Max normalization method is used to normalize the mean-removed sequence. P´ ( t Mapping to the range [0,1] yields a normalized sequence. P norm ( t ):

[0062] ,

[0063] Where, max( P´ ) and min(P´ ) are sequences P´ ( t The maximum and minimum values ​​of ).

[0064] The stationarity test is performed using the ADF test method from the unit root test. The calculation method is as follows:

[0065] ,

[0066] in, The sequence is a first-order difference sequence; α is a constant term; βt is a time trend term; γ is a key parameter, the null hypothesis is γ=0 (the sequence has a unit root and is non-stationary); p is the lag order, usually chosen according to the AIC or BIC criterion; ε t This represents the white noise error term. If the test statistic is less than the critical value of the given significance level, the null hypothesis is rejected, and the series is considered stationary. If the test result indicates that the series is non-stationary, differencing is required until the stationarity test is passed.

[0067] After the above preprocessing, the standardized power time series that can be used for spectral analysis is finally obtained:

[0068] .

[0069] 2. Construct a three-variable power spectrum matrix and use the matrix inversion method to calculate the partial coherence function in order to eliminate the coupling interference of the third variable and analyze the basic data for evaluating the true complementarity between resources;

[0070] Based on the preprocessed power time series, the autospectral density and cross-spectral density are calculated to construct the spectral matrix of wind, solar, and wave variables. S ( f The detailed calculation steps are as follows:

[0071] First, it is necessary to calculate the autocovariance function of each resource power sequence and the cross-covariance function between each pair of sequences.

[0072] Wind power autocovariance function:

[0073] .

[0074] in, k For time lag (k=0,1,2…,M,M<N). The mean of the wind power series. N The sequence length is given.

[0075] Optical power autocovariance function:

[0076] .

[0077] in, The mean of the optical power sequence is denoted as . N The sequence length is given.

[0078] Wave power autocovariance function:

[0079] .

[0080] in, The mean of the wave power sequence is denoted as . N The sequence length is given.

[0081] Wind-light cross-covariance function:

[0082] .

[0083] Wind-wave cross-covariance function:

[0084] .

[0085] Light-wave cross-covariance function:

[0086] .

[0087] To reduce spectral leakage, the covariance function is windowed using the Hanning window calculation method:

[0088] .

[0089] The window function ω(k) is:

[0090] .

[0091] By performing Fourier transforms on the windowed autocovariance and crosscovariance functions, the power spectral density estimate is obtained:

[0092] The wind power autospectral density is calculated as follows:

[0093] ,in f For frequency, i The imaginary unit;

[0094] The optical power autospectral density is calculated as follows:

[0095] ;

[0096] Wave power autospectral density is calculated as follows .

[0097] Wind-light cross-spectral density calculation is as follows ;

[0098] Wind-wave cross-spectral density calculation is as follows ;

[0099] The cross-spectral density of light and wave is calculated as follows: .

[0100] The calculated autospectral density and cross-spectral density are used to construct the spectral matrix of the three variables of wind, light, and waves in the following form. S ( f ):

[0101] ,

[0102] in, S ij ( f ) is a variable i and j Cross-spectral density, diagonal elements S ii ( f ) represents the self-spectral density. Wherein, based on the symmetry of the cross-spectrum, we have , , .

[0103] When the spectral matrix is ​​close to singular, a pseudo-inverse or a small regularization term is used to ensure computational stability.

[0104] Partial coherence is calculated by inverting the spectral matrix. c XY-Z ( f To eliminate the influence of the third variable Z, the detailed calculation is as follows:

[0105] For each frequency point f , spectral matrix S ( f Find the inverse to obtain the inverse matrix. S -1 ( f ):

[0106]

[0107] Here, the superscript indicates an element in the inverse matrix, not the exponent; det( S ( f )) is the determinant of the spectral matrix, adj( S ( f )) is the adjoint matrix of the spectral matrix.

[0108] For any two resources X and Y In controlling third resources Z The influence, its partial coherence function c XY-Z ( f It can be calculated using the elements of the inverse matrix:

[0109] The value range is [0,1]. c XY-Z ( f The closer the value is to 1, the better the effect of controlling for Z. X and Y The stronger the correlation at that frequency.

[0110] Wind-light coherence calculation is as follows ;

[0111] Wind-wave coherence calculation is as follows ;

[0112] The optical-wave polarization coherence calculation is as follows .

[0113] This allows us to obtain the partial coherence spectra of three resource pairs: wind-solar, wind-wave, and solar-wave.

[0114] 3. A dual masking function of phase and gain is introduced to correct the complementarity evaluation, and the entire spectrum is divided into three characteristic frequency bands: intraday, weather, and seasonal.

[0115] For any two resources X and Y, their mutual bit spectrum F XY ( f ) is its cross-spectral density S XY ( f The argument of a frequency is described by its angle of reference. f Above, resources X and Y The phase lag relationship between them. Mutual position spectrum. F XY ( f The calculation formula is as follows:

[0116]

[0117] Wind-light phase spectrum values ;

[0118] Wind-wave phase spectrum value ;

[0119] Light-wave phase spectrum value .

[0120] The value range is [-π, π]. F XY ( f If ) < 0, it means that at frequency f Above, resources X Phase ahead of resources Y ;if F XY ( fIf ) > 0, it means that at frequency f Above, resources X Phase lags behind resources Y ;if F XY ( f If )≈π or π, it means that the two are nearly out of phase at that frequency; if F XY ( f If )≈0, it means that the two are close to being in phase at that frequency.

[0121] Phase mask function M Φ ( f This is used to reward peak shifts (phase differences close to π) and penalize phase differences close to 0. The phase masking function converts phase information into weighted coefficients that can be used for quantitative evaluation, with values ​​ranging from [0,1]. The phase masking function M... Φ ( f Calculation method:

[0122] ,

[0123] Among them, wrap π For phase wrapping operations, the phase difference is constrained to [-π, π]; σ Φ The parameters used to control the width of the reward interval are typically set to [π / 4, π / 3].

[0124] For any resource X and Y Its gain spectrum G XY ( f ) indicates frequency f Above, resources Y Relative to resources X The degree of linear dependence. Gain spectrum. G XY ( f Calculation formula:

[0125] ,

[0126] in Cross-spectral density S XY (f) The magnitude (amplitude) S XX ( f ) is a resource X The autospectral density.

[0127] Wind-light gain spectral values ;

[0128] Wind-wave gain spectrum ;

[0129] Light-wave gain spectrum .

[0130] G XY ( f The value range is [0, +∞]. G≈0， Resource Y is almost unresponsive to fluctuations in resource X at this frequency; G < 1, the fluctuation amplitude of resource Y at this frequency is smaller than that of resource X; G≈ 1. The fluctuation amplitude of resource Y at this frequency is comparable to that of resource X, indicating a good amplitude match; G > 1, the fluctuation amplitude of resource Y at this frequency is greater than that of resource X.

[0131] Gain mask function M G ( f This transforms the matching degree of power amplitude between resources into a weighting coefficient used to correct complementarity scores, thereby avoiding "artificially high complementarity" caused by amplitude mismatch. Gain mask function M G ( f Calculation method:

[0132] ,

[0133] in m G σ is the target gain ratio, usually taken as 0. G This is a tolerance parameter that controls the rate at which the weight decreases; it is typically set to 0.5 to 1.0.

[0134] The above values ​​are then subjected to frequency band decomposition and aggregation. The entire spectrum is divided into three frequency bands according to different time scales of engineering interest (intraday, weather, season). Then, the spectral indices (such as partial coherence, phase mask, gain mask, etc.) within each frequency band are averaged to obtain representative indices for each band. The specific calculation method is as follows:

[0135] Frequency band decomposition, based on the operational requirements at different time scales in engineering practice, divides the continuous spectrum into three characteristic frequency bands: F1, F2, and F3. First, based on the sampling frequency... f s and sequence length N Determine the frequency resolution Establish frequency axis ,in k =0,1,2,…,[N / 2]; then convert the frequency to a period and establish a period axis. Then the periodic boundary is converted into a frequency boundary:

[0136] Intraday frequency band F1: Period T < 1 day (frequency) f＞ 1 / 86400 Hz ),

[0137] Weather band F2: Period 1≤T≤10 days (1 / 864000≤ f≤ 1 / 86400 Hz ),

[0138] Seasonal frequency band F3: period T > 10 days ( f＜ 1 / 86400 Hz );

[0139] Finally, each frequency point f k If allocated to each frequency band, f＞ 1 / 86400 Hz ,but f k ∈F1, if 1 / 864000≤ f ≤ 1 / 86400 Hz ,but f k ∈F2, if f＜ 1 / 86400 Hz ,but f k ∈F3.

[0140] The partial coherence calculated in steps 2 and 3 c XY-Z ( f Phase mask M Φ ( f Gain Mask M G （f） Perform equal-quantity band aggregation. Calculate the arithmetic mean within each frequency band:

[0141] ,

[0142] Q∈{ c XY-Z M Φ， M G}, where b∈{1,2,3} represents the frequency band. F b Let b be the set of all frequency points belonging to frequency band b. For frequency band F bThe number of frequency points within the band. From this, the aggregated partial coherence, phase mask, and gain mask values ​​for each frequency band can be obtained:

[0143] Band convergence calculation of polarimetric spectra, wind-light polarimetric coherence ;

[0144] Wind-wave partial coherence ;

[0145] Light-wave partial coherence .

[0146] Phase masking by band aggregation calculation, wind-light phase mask ;

[0147] Wind-Wave Phase Mask ;

[0148] Light-wave phase mask .

[0149] Band-wise aggregation calculation of gain mask, wind-light gain mask ;

[0150] Wind-Wave Gain Mask ;

[0151] Light-wave gain mask .

[0152] 4. The analytic hierarchy process (AHP) and entropy weight method are integrated to calculate subjective and objective weights. The optimal weight vector under Nash equilibrium is solved using a game theory-based combined weighting model. This constructs a three-dimensional evaluation framework covering complementarity, stability, and contribution, and forms a comprehensive complementarity index.

[0153] The specific calculation steps for constructing the three-dimensional indicator system are as follows:

[0154] For each frequency band, calculate the average value of the three types of indicators:

[0155] ,

[0156] Complementarity C: First, calculate the complementarity sub-fraction within the frequency band. The total complementarity C is C for each frequency band. (b） Weighted sum: .in, To determine the weighting of complementarity across frequency bands, the spectral energy proportion method is used to calculate the proportion of the power spectral density integral value in each frequency band to the total energy of the entire frequency band. .

[0157] Stability S: First, calculate the combined power spectrum. , where ω i Set the initial weights for each resource (all can be set to 1).

[0158] Then calculate the smoothness of the combined power spectrum. ,in Var ma for P mix ( f The maximum possible variance is used for normalization. Then, phase consistency is calculated. .

[0159] Then calculate gain consistency: .

[0160] The overall stability is the weighted sum of the stability components: ,

[0161] in The weights of each component are assigned using a normalization method, and preset based on the emphasis on different characteristics according to the engineering application scenario. .

[0162] Contribution D: Total contribution .

[0163] in, To assign weights to each frequency band, the spectral energy proportion method is used to calculate the proportion of the power spectral density integral value in each frequency band to the total energy of the entire frequency band. . H （b） This represents the entropy balance of resources within frequency band b. λ represents the average grey relational degree of each resource within frequency band b. H and λ Γ For structural weights, a normalization method is adopted, and λ is preset according to the degree of attention given to different characteristics based on the engineering application scenario. H =0.6 (primary objective), λ Γ =0.4 (auxiliary indicator).

[0164] Entropy balance H （b） It is an index that quantifies the evenness of resource contribution distribution, with a value range of [0,1]. The closer it is to 1, the more even the contribution ratios of the three resources are, indicating good system balance, no over-reliance on any one resource, and strong anti-interference ability. First, calculate the resources. i ( i The integral spectral energy of ∈{wind,solar,wave} within frequency band b Then calculate its frequency band energy percentage. Then calculate the frequency band entropy equalization. , where M is the number of resource types, which is 3 in this case.

[0165] Grey relational degree Ci Characterizing the power sequence of each resource P i ( t ) and mixed power sequences P mix ( t The degree of similarity in development trends is measured in the range of (0,1]. A value closer to 1 indicates a higher correlation between the resource power change trend and the total system power. The detailed calculation steps are as follows: Assume the reference sequence is a mixed power sequence. P mix ( t The comparison sequence is the power sequence of each resource. P i ( t ),in i ∈{wind,solar,wave}, t =1,2,…, n Initialization: Eliminating the influence of dimensions. , , t =1,2,…, n ; Calculate the absolute difference sequence: Determine the extreme values , ; Calculate the correlation coefficient Where the resolution coefficient ρ = 0.5; calculate the grey relational degree. .

[0166] To analyze the contribution characteristics of each resource to the total system power at different time scales, it is necessary to calculate the grey relational degree within each characteristic frequency band b∈{1,2,3}. C i (b) First, the original power sequence is passed through a digital bandpass filter processor based on cutoff characteristics in the corresponding frequency band to obtain the sub-sequence within the frequency band. P (b) mix ( t ), P i (b) ( t Subsequently, the aforementioned subsequences are used as new reference and comparison sequences, respectively. Following the same calculation method, the average gray correlation degree of the frequency band is finally calculated. .

[0167] To overcome the limitations of a single weighting method, a strategy combining subjective weighting (Analytic Hierarchy Process, AHP) and objective weighting (Entropy Weighting, EWM) is adopted. Considering that sea area site selection involves a large number of grid points, to avoid computational redundancy in point-by-point calculations and to capture the resource distribution characteristics of the entire sea area, this step adopts a "global weighting, local mapping" method. The specific steps are as follows:

[0168] a) Constructing a global evaluation matrix: Extracting all [representations] within the simulation region. m Complementarity of individual grid points C ,stability S and contribution D The calculation results are used to construct an m*3 order evaluation matrix. R .

[0169] b) Calculate the subjective weight vector W sub Based on the preferences of energy island engineering construction for various performance dimensions (e.g., in offshore scenarios, system stability is preferred over resource contribution), a hierarchical analysis discrimination matrix is ​​constructed, and the subjective weight vector is obtained after consistency verification. .

[0170] c) Calculate the objective weight vector W obj This study utilizes the entropy weight method to mine the distribution characteristics of data across the entire sea area. Firstly, the evaluation matrix... R Perform normalization processing and calculate the information entropy of the i-th indicator. e j :

[0171] .

[0172] in, p ij For the first i The grid point at the th j The feature weights of each indicator are then used to calculate the objective weight vector. ,in .

[0173] d) Calculate the optimal weights based on game theory. W To minimize the deviation between the overall weight and the subjective and objective weights, and to achieve Nash equilibrium, a combinatorial weighting model based on game theory is constructed.

[0174]

[0175] Where α1 and α2 are the linear combination coefficients of subjective and objective weights, respectively. The optimal combination coefficients α*1 and α*2 are solved using the Lagrange multiplier method and then normalized. .

[0176] The final global weight vector is obtained as follows:

[0177] .

[0178] e) Calculate the comprehensive complementarity index. CSD Using the global weight vector, for each grid point i By performing linear weighting, the comprehensive complementarity index at that point is obtained:

[0179] .

[0180] 5. Based on the spatial distribution characteristics of the comprehensive complementarity index, high-value areas are delineated to identify priority candidate sites for the construction of the energy island;

[0181] Calculate the value of each grid point within the simulation area. CSD The index was plotted using MATLAB tools. CSD Exponential spatial distribution cloud map. Based on the cloud map, delineate... CSD Areas with higher indices represent potential development zones for the energy island. Representative test sites were selected within these potential development zones to analyze their zoning diagnostic results (i.e.,...). C (b) ,S (b) ,D (b The specific details of the diagnosis are as follows:

[0182] If the intraday frequency band complementarity at a certain point C (1) High stability indicates good short-term complementarity, which can reduce the need for power-type energy storage configurations; if the frequency band stability at a certain point is... S (2) High values ​​indicate large weather-scale fluctuations, necessitating sufficient energy storage capacity; if the seasonal frequency band contribution at a certain point... D (3) (entropy H (3 High balance indicates balanced resource utilization, allowing for more diverse equipment configurations.

[0183] 6. Within the priority candidate sites identified in step five, based on the frequency band diagnostic results obtained in step three, a two-level optimization model is established with the objectives of minimizing the levelized cost of electricity (LCOE) and maximizing the comprehensive complementarity index. Physical constraints based on the energy integral of each frequency band spectrum and an adaptive frequency domain response strategy are introduced, and a multi-objective evolutionary algorithm is used to solve for the optimal capacity configuration of wind-solar-wave-storage.

[0184] 1) Define decision variables. Let the decision vector be... X =[P wind ,P solar ,P wave E bat,P bat These represent the installed capacity of wind power, photovoltaic power, and wave energy, respectively, as well as the energy capacity (battery cells, battery racks, electrolyte or physical storage medium, which determines the system's ability to continuously supply power in response to weather-scale energy shortages) and power capacity (energy storage converter PCS, transformer, and supporting electrical switching equipment, which determines the system's ability to smooth out intraday high-frequency fluctuations).

[0185] 2) Construct a multi-objective optimization function. Objective 1 is to achieve economic optimization (Min LCOE):

[0186]

[0187] in f 1( X This represents the Levelized Cost of Energy (LCOE), typically expressed in yuan / kWh. It is a core indicator for measuring the economic feasibility of energy island projects, representing the average net present value cost required to generate one kilowatt-hour of electricity. I total This represents the total initial investment cost of the energy island system, expressed in yuan. The calculation covers the purchase and installation costs of each power generation and energy storage device, specifically expressed as follows:

[0188] ,

[0189] in C wind , C solar ...and this is the unit capacity investment cost (CAPEX) for the corresponding equipment.

[0190] O & M t This represents the operation and maintenance cost in year t, expressed in yuan. It is typically calculated as a percentage (e.g., 1%-3%) of the initial investment and is used to cover equipment maintenance, material consumption, and labor costs. E load , t represents the actual power supply (or grid-connected power) that the system provides to meet load demand in year t, in kWh. This value is determined by the system operation simulation strategy and reflects the effective output of the energy island. r represents the project's discount rate, which is dimensionless and reflects the time value of money. It is usually set based on the industry benchmark rate of return, typically between 6% and 8%. N lifeThis represents the entire project lifetime of the energy island system, expressed in years. It is usually based on the design life of the main power generation equipment (such as wind turbines), typically taken as 20 years.

[0191] Objective 2 is technically optimal (Max CSD):

[0192] max f 2( X )= CSD ( Pmix ( X ))

[0193] in, P mix ( X ) for configuration X The system's hybrid power output sequence is given by an objective function that dynamically maximizes the system's overall complementary performance in the frequency domain.

[0194] .in P i This refers to the installed capacity; p i norm The unit-normalized output curve; CSD(•) represents the global evaluation model operator established in step four. This function maps the mixed power sequence to a single evaluation score, integrating phase complementarity, amplitude stability and energy balance information in the frequency domain.

[0195] 3) Before executing the two-level optimization algorithm, the following four types of key parameters need to be preset:

[0196] (a) Equipment technical parameters: Set the cut-in / cut-out / rated wind speed of the candidate wind turbine ( v in ,v out ,v rated ) and standard power curve; set the photoelectric conversion efficiency of photovoltaic modules. or pv and inverter efficiency or inv Setting the wave energy device at different effective wave heights H s With spectral peak period T p The power matrix is ​​set below; the charge and discharge efficiency of the electrochemical energy storage system is set. or bat Maximum depth of discharge DOD max and cycle life.

[0197] (b) Economic model parameters: Set the unit kilowatt investment cost (CAPEX), annual operation and maintenance cost ratio (OPEX), project discount rate (recommended 6%-8%), and system life cycle parameters for various types of power generation equipment and energy storage systems. N life (20 years is recommended).

[0198] (c) Spectral Analysis Feature Constraint Thresholds: Set logical thresholds based on the diagnostic results from step five. This includes: seasonal frequency band correlation threshold. C min (When a certain resource ( C ( F 3 )< C min At the same time, limit its planned capacity limit); weather frequency band fluctuation energy margin coefficient or safe (Used to determine the minimum energy storage capacity required to smooth out weather-scale fluctuations).

[0199] (d) Optimize algorithm parameters: Set the population size, maximum number of iterations, crossover probability and mutation probability, and convergence criterion of Pareto front for multi-objective evolutionary algorithms (such as NSGA-II).

[0200] 4) When constructing a two-level optimization model, it is first necessary to establish a set of basic constraints to ensure the energy conservation of the system and the safe operation of the equipment, specifically including:

[0201] (a) Real-time power balance constraint: At any time t, the sum of the total power generation and energy storage discharge of the system must be equal to the sum of the load demand, energy storage charging power and power curtailment / load shedding power to ensure real-time supply and demand balance.

[0202]

[0203] in P i,gen ( t The value represents the actual output power of the wind, solar, and wave power generation equipment at time t. Note: P i,gen ( t )< P i It is limited by the rated installed capacity. P dis ( t ), P ch ( t ) represent the discharge power and charging power of the energy storage system at time t, respectively. P load( t ) represents the load demand at time t. P curt ( t ) represents the power curtailment at time t (when power generation is in excess and energy storage is full). P loss ( t ) represents the unloaded power at time t (when the power supply is insufficient and the energy storage is empty).

[0204] (b) Energy storage system status and operational constraints: Simulate the energy flow inside the energy storage battery, constraining its state of charge (SOC) to not exceed limits, and ensuring that the charging and discharging power does not exceed the inverter's rated value:

[0205] Energy state recursive equation (SOC evolution):

[0206]

[0207] State of charge (SOC) upper and lower limits constraints:

[0208]

[0209] Usually taken SOC min =10%, SOC max =90% to extend battery life.

[0210] Charge and discharge power constraints (and decision variables) P bat hook up):

[0211] ,

[0212] Charge-discharge mutual exclusion constraint:

[0213] in U ch , U dis ∈{0,1} are state variables to ensure that the battery is not in a charging and discharging state at the same time.

[0214] (c) Power supply reliability constraints: Limit the power supply failure rate (LPSP) throughout the entire life cycle to ensure that the power supply quality of the energy island meets engineering standards.

[0215]

[0216] in LPSP max The maximum allowable power shortage rate threshold (e.g., 5% or 1%, depending on the off-grid / grid-connected type).

[0217] (d) Boundary constraints on decision variables: Limit the search space of optimization variables to ensure that the calculation results are non-negative and do not exceed the carrying capacity of the sea area.

[0218]

[0219] in The upper limit is determined by the target sea area defined in step one and the power density of each device. W / m 2 ) was calculated.

[0220] Physical constraints based on frequency band characteristics: In addition to conventional power balance and investment constraints, this invention introduces characteristic constraints based on spectral analysis and diagnosis to improve optimization efficiency and physical matching degree.

[0221] (A) Based on seasonal frequency bands F 3 The upper limit constraint of installed capacity based on grey relational analysis: This avoids over-allocation of resources that do not match the total system demand over long-term (seasonal) trends, reducing the pressure on long-term energy storage. This applies to each type of renewable energy... i , i∈{ wind, solar, wave Its planned installed capacity P i :

[0222]

[0223] K i Capacity confidence coefficient based on seasonal frequency band correlation:

[0224] ,,

[0225] The maximum total installed capacity allowed in the target sea area in terms of physical space;

[0226] ξ space,i This refers to the maximum power density or space occupancy factor of the resource.

[0227] Γ i ( F 3) For resources in seasonal frequency bands F Grey relational degree within 3;

[0228] α base Basic capacity guarantee factor;

[0229] m As a penalty factor m ≥1, m The larger the value, the greater the penalty for low-relevance resources.

[0230] (B) Based on weather frequency bands F 2) Lower limit constraint of energy storage capacity (kWh) for spectral energy integration: Power fluctuations in weather bands (1-10 days) are the main cause of power shortages in energy islands. The combined power spectrum from the stability index calculated in step four is used... S mix ( f By using the integral of fluctuating energy within the weather frequency band as the physical benchmark for energy storage capacity configuration, we can ensure that energy storage is sufficient to mitigate power deficits at the weather scale.

[0231] ,

[0232] in or safe This refers to the safety margin factor. T char The characteristic period of the weather frequency band (usually the reciprocal of the center frequency of the band) is used to convert the power standard deviation into energy demand; ∫ f∈F2 S mix The integral value representing the variance of the total power of synoptic-scale fluctuations; S mix ( f;P Let P be the power spectral density function of the system's hybrid output power under configuration P; and let P be the installed capacity vector in the current optimization iteration step. P =[ P wind ,P soar ,P wave ], P ∈ X .

[0233] (C) Based on intraday frequency bands ( F 1) Lower limit constraint of energy storage power (kW) for high-frequency variance: Rated power of the energy storage converter (PCS) P bat It must be sufficient to cope with the instantaneous power surges in the intraday high-frequency band, which are determined by the spectral variance of the intraday frequency band:

[0234] ,

[0235] in l surge It is the power surge factor (usually taken as 3-4, following the 3σ principle to ensure coverage of 99.7% of instantaneous fluctuations).

[0236] (D) Based on the comprehensive complementarity index CSD mandatory constraints

[0237] The optimized solutions must meet a passing grade at the "source-to-source complementarity" level to prevent complete reliance on energy storage while ignoring the complementary potential of the resources themselves.

[0238] ,

[0239] in CSD threshold The lower limit (80% of the benchmark value) is set for the CSD index benchmark value of the region derived from the site selection phase.

[0240] (E). Formulate frequency domain response operation strategy: In the optimization solution process, the inner layer operation simulation adopts a phase mask-based approach. M Ф The adaptive control strategy is as follows: First, extract the wind-solar and wind-wave phase masks for the intraday frequency band (F1). M Ф ( t );when M Ф ( t When the value is closer to 1 to achieve a better complementary effect, the intervention threshold of the energy storage system should be lowered, and source-to-source complementarity should be prioritized to smooth fluctuations; when M Ф ( t When the energy density is close to zero and the complementary effect is poor, the output command of the energy storage system is increased to forcibly suppress the power surge.

[0241] (F). The above model is solved using the multi-objective optimization algorithm (NSGA-II) to obtain the Pareto optimal solution set, and the knee point is selected as the final recommended equipment configuration scheme. Example 1:

[0242] The simulation area has a latitude and longitude range of 105°E-135°E and 2°N-42°N. The grid adopts a rectangular grid with a grid resolution of 0.25° × 0.25°. The entire process was verified based on ERA5 reanalysis data.

[0243] This embodiment uses ERA5 reanalysis data from January 1, 2024 to December 31, 2024 as the basic data source, including 10m meridional wind speed, 10m zonal wind speed, surface solar radiation, significant wave height, and spectral peak period data. The core physical parameter is set as: air density. r a =10225 kg / m 3 Seawater density r w =1025 kg / m 3 Gravitational acceleration g=9.81 m / s 2 Photovoltaic conversion efficiency or pv =0.18, Day and Night Irradiance Threshold G threshold =15 W / m 2 Spectral analysis parameters are set to the sampling frequency. f s =1 / 3600 Hz (i.e., sampling once per hour), maximum hysteresis N lag Take 10 days (i.e., 240 points) to ensure that periodic features at the weather scale are captured.

[0244] First, the atmospheric and wave data are aligned with the grid. The nearest neighbor interpolation method is used to map the wave data to the atmospheric data grid. Then, the public sea area is clipped by ocean masking and invalid grid points on land are removed. Missing values ​​in the data are filled in by the "nearest neighbor" method. At the same time, the exver dimension of ERA5 data is merged to ensure data integrity.

[0245] The data for each grid is preprocessed to calculate the wind, solar, and wave power densities for each grid. Subsequently, the power series is subjected to mean removal, Min-Max normalization, and stationarity processing to ensure that the data meets the requirements of spectral analysis.

[0246] Blackman-Tukey spectral analysis was performed on the preprocessed data. The autocovariance and cross-covariance of the wind-solar-wave power series were calculated. After windowing using a Hanning window, a Fourier transform was performed to obtain the autospectral density and pairwise cross-spectral density of each resource, constructing a 3×3 spectral matrix. S ( f Based on the inversion of the spectral matrix, the partial coherence function is calculated, and the influence of the third variable is eliminated to obtain the wind-solar ( c ws·p ),storm( c wp·s ), light-wave ( c sp·w The true complementary relationship between them.

[0247] Phase and gain spectra are extracted between resources. The phase spectrum reflects the lead-lag relationship of resource fluctuations, while the gain spectrum characterizes the amplitude dependence between resources. A phase masking function is constructed to reward peak offsetting and penalize in-phase fluctuations, and a gain masking function is constructed to quantify the amplitude matching degree. Both masking functions adopt the form of Gaussian functions, with parameters set as follows. s Ф =𝜋 / 3、 sG =𝜋 / 3.

[0248] The spectrum is divided into three frequency bands according to engineering scale: intraday frequency band (period (T<1) days), weather frequency band (period (1≤T≤10) days), and seasonal frequency band (period (T>10) days). The arithmetic mean of the partial coherence, phase mask, and gain mask in each frequency band is used to complete the aggregation calculation.

[0249] Complementarity C ,stability S Contribution D Three-dimensional indicators: complementarity is the weighted sum of the products of partial coherence and mask in each frequency band; stability is characterized by the variance of the mixed power spectrum; and contribution is quantified by combining the energy proportion entropy balance with grey relational measurement.

[0250] Construct a global evaluation matrix and use the analytic hierarchy process (AHP) to obtain subjective weights. W sub It is believed that system stability is more important than resource contribution in offshore scenarios, and a subjective weight vector is set. W sub =[0.40,0.30,0.30], corresponding to complementarity, stability, and contribution, respectively. The objective weights are obtained using the entropy weight method. W abj Calculated W abj =[0.417, 0.120, 0.463]. The optimal combination coefficients are solved using game theory to obtain the comprehensive weight. W Finally, the comprehensive complementarity index is calculated. CSD To obtain the entire sea area CSD The exponential spatial distribution, with calculated weights for each dimension, is as follows: w C =0.408、 w S =0.210、 w D =0.382,, CSD The index has a mean of 0.419 and a range of [0.197, 1.000]. [Plotting...] CSD The index spatial distribution cloud map, with the filtering range set to 115°E-123°E and 20°N-42°N, is then... CSD The top 200 high-value points were extracted in descending order of value for regional comprehensive complementarity discrimination, as shown in Table 1.

[0251] Table 1. Top 20 high CSD values:

[0252]

[0253] An analysis was conducted on a typical area of ​​the Zhoushan Islands (target area range 121.50°E-123°E, 29°N-31°N). If the target point was an invalid area, the system automatically searched for the nearest valid sea area point within a radius of 5 grids, ultimately locking in the optimal site and ensuring its safety. CSD The values ​​are non-zero and the data is complete. A detailed spectral analysis is then performed again on the selected island area. The calculation steps are as follows:

[0254] 1) Extract wind, solar, and wave parameters from the target site, convert them into power density sequences, and perform mean removal, normalization, and ADF stationarity test.

[0255] 2) Calculate the autocovariance and crosscovariance functions of each resource power sequence, apply a Hanning window, and construct a three-variable power spectral density matrix using Fourier transform to analyze the energy distribution characteristics at different time scales. Figure 1 );

[0256] 3) Calculate the partial coherence function based on the inversion of the spectral matrix, eliminate the common driving interference of the third variable, and obtain the true complementary relationships between wind-solar, wind-wave, and solar-wave. Figure 2 );

[0257] 4) Analyze the phase spectrum and gain spectrum between resources, quantify the lead-lag relationship and amplitude dependence of fluctuations, and construct phase and gain mask functions, as shown in Table 2;

[0258] 5) Divide the entire spectrum into three frequency bands: intraday, weather, and seasonal. Aggregate and calculate the spectral indices within each frequency band to form the band-specific characteristic parameters.

[0259] 6) Draw complementary polar coordinate diagnostic maps based on zonal data ( Figure 3 ), to identify strong complementary regions and synchronous fluctuation regions in specific frequency bands (such as synoptic scales).

[0260] Table 2 Calculation of Spectral Analysis Characteristics for Each Combination

[0261]

[0262] Finally, a two-layer optimization model based on frequency band feature decoupling is used to verify micro-capacity gradation and operation strategy. The specific steps are as follows:

[0263] 1) Set equipment cost and operation and maintenance parameters: wind power unit investment cost 8,000 yuan / kW, photovoltaic 9,000 yuan / kW, wave energy 12,000 yuan / kW, energy storage 2,000 yuan / kWh, annual operation and maintenance fee rate 3%, discount rate 6%, project life cycle 20 years;

[0264] 2) The lowest levelized cost of electricity (LCOE) and the comprehensive complementarity index (SCI) are used as the benchmarks. CSDThe maximum index is the joint optimization objective. The initial capacity is set as (corresponding to the installed power of wind, solar, and wave and the energy storage capacity respectively). The multi-objective evolutionary algorithm is used to find the optimal configuration scheme that takes into account both economy and stability. The results are shown in Tables 3 and 4.

[0265] 3) Comparison curves of the output grid-connected power sequence and the original hybrid power generation sequence ( Figure 4 This directly characterizes the system's peak-shaving and valley-filling capabilities; through comparison of spectrum smoothing effects and analysis of spectrum attenuation efficiency ( Figure 5 , Figure 6 ), probability distribution shaping effect ( Figure 7 ) and battery operating phase space trajectory ( Figure 8 The effectiveness of energy storage systems in mitigating power fluctuations was quantitatively verified from both frequency domain and statistical perspectives.

[0266] 4) Extract the typical weekly operating data with the most drastic fluctuations ( Figure 9 The dynamic balance mechanism of source-side input, energy storage high-frequency / low-frequency response and grid-side output was analyzed to verify the continuous power supply stability and reliability of the system under extreme weather conditions.

[0267] Table 3 Microscopic Volume Fixation Scheme

[0268] .

[0269] Table 4 Economic and Technical Indicators

[0270] .

[0271] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for site selection and capacity determination of offshore energy islands, characterized in that, Includes the following steps: (1) Obtain ERA5 multidimensional reanalysis data of the target sea area to construct the power time series, and complete the preprocessing work of mean removal, normalization and stationarity test; (2) Construct a three-variable power spectrum matrix and use the matrix inversion method to calculate the partial coherence function in order to eliminate the coupling interference of the third variable and analyze the basic data for evaluating the true complementarity between resources; (3) A dual masking function of phase and gain is introduced to correct the evaluation of true complementarity, and the entire spectrum is divided into three characteristic frequency bands: intraday, weather and seasonal. (4) The subjective and objective weights are calculated by integrating the analytic hierarchy process and the entropy weight method. The optimal weight vector under Nash equilibrium is solved by using the game theory combined weighting model, thereby constructing a three-dimensional evaluation framework covering complementarity, stability and contribution, and forming a comprehensive complementarity index. (5) Based on the spatial distribution characteristics of the comprehensive complementarity index, high-value areas are delineated to identify priority candidate sites for the construction of the energy island; (6) Establish a two-level optimization model with the goal of minimizing the levelized cost of electricity and maximizing the comprehensive complementarity index. Introduce physical constraints and adaptive frequency domain response strategies based on the energy integral of each frequency band spectrum. Use a multi-objective evolutionary algorithm to solve the optimal capacity configuration of wind-solar-wave-storage.

2. The method for site selection and capacity determination of offshore energy islands as described in claim 1, characterized in that, In step (1), ERA5 or similar data of the target sea area are obtained. The wind data mainly includes meridional wind speed and zonal wind speed. The light data mainly includes solar irradiance. The wave data mainly includes significant wave height and spectral peak period.

3. The method for site selection and capacity determination of offshore energy islands as described in claim 1, characterized in that, In step (2), based on the preprocessed power time series, the autospectral density and cross-spectral density are calculated to construct the spectral matrix of the three variables of wind, solar and wave.

4. The method for site selection and capacity determination of offshore energy islands as described in claim 3, characterized in that, Calculate the self-spectrum and cross-spectral density to construct the spectral matrix of the three variables of wind, light, and waves, specifically as follows: a. Calculate the autocovariance function of each resource power sequence and the cross-covariance function between each pair of sequences; b. To reduce spectral leakage, the above covariance function is windowed using the Hanning window calculation method; c. By performing Fourier transforms on the windowed autocovariance and crosscovariance functions, the power spectral density estimate is obtained; d. Construct a spectral matrix of wind-light-wave three variables using the calculated autospectral density and cross-spectral density; e. Calculate the partial coherence function by inverting the spectral matrix to eliminate the influence of the third variable.

5. The method for site selection and capacity determination of offshore energy islands as described in claim 1, characterized in that, In step (3), for any two resources X and Y, their relative position spectra are... Φ XY ( f ) is its cross-spectral density S XY ( f The argument of a frequency is described by its angle of reference. f Above, resources X and Y Phase lag relationship between them; gain spectrum G XY ( f ) indicates frequency f Above, resources Y Relative to resources X The degree of linear dependence; Phase masks are constructed using Gaussian kernels. M Φ ( f ) and gain mask M G ( f The above values ​​are decomposed into frequency bands and aggregated into bands. The entire spectrum is divided into three frequency bands according to different time scales of interest in engineering. Then, the spectral indices in each frequency band are averaged to obtain representative indices for each frequency band.

6. The method for site selection and capacity determination of offshore energy islands as described in claim 1, characterized in that, Step (4) constructs a three-dimensional indicator system, including complementarity, stability and contribution. To overcome the limitations of a single weighting method, a strategy combining subjective and objective weighting is adopted. Considering that the selection of sea areas involves a large number of grid points, in order to avoid the redundancy of computing power in point-by-point calculation and to capture the resource distribution characteristics of the entire sea area, the method of "global weighting and local mapping" is adopted.

7. The method for site selection and capacity determination of offshore energy islands as described in claim 1, characterized in that, Step (5) Calculate the value of each grid point within the simulation area. CSD The index was plotted using MATLAB tools. CSD Exponential spatial distribution cloud map; based on the cloud map, delineate... CSD Areas with higher indices are potential development zones for the energy island; representative test sites are selected within these potential development zones, and their zoning diagnostic results are analyzed.

8. The method for site selection and capacity determination of offshore energy islands as described in claim 7, characterized in that, The specific details of the diagnosis are as follows: If the intraday frequency band complementarity at a certain point C (1) High indicates good short-term complementarity, which can reduce the need for power-type energy storage configuration; If the weather frequency band at a certain point is stable S (2) High values ​​indicate large weather-scale fluctuations, necessitating the allocation of sufficient energy storage capacity. If the seasonal frequency band contribution at a certain point D (3 Balanced means that resources are used in a balanced way, and equipment configuration can be more diversified.

9. The method for site selection and capacity determination of offshore energy islands as described in claim 1, characterized in that, Step (6) Within the priority candidate sites defined in step (5), based on the frequency band diagnosis results obtained in step (3), construct a two-layer optimization model that considers frequency domain feature coupling to determine the optimal configuration of wind turbines, photovoltaics, wave energy devices and energy storage systems.

10. The method for site selection and capacity determination of offshore energy islands as described in claim 9, characterized in that, The detailed steps for optimal configuration of wind turbines, photovoltaic, wave energy devices, and energy storage systems are as follows: 1) Define decision variables; 2) Construct a multi-objective optimization function; 3) Before executing the two-level optimization algorithm, the following four types of key parameters need to be preset: equipment technical parameters, economic model parameters, spectral analysis feature constraint thresholds, and optimization algorithm parameters. 4) When constructing a two-level optimization model, it is first necessary to establish a set of basic constraints to ensure the energy conservation of the system and the safe operation of the equipment. Specifically, these constraints include: real-time power balance constraints, energy storage system state and operation constraints, power supply reliability constraints, decision variable boundary constraints, and setting physical constraints based on frequency band characteristics. 5) Develop a frequency domain response operation strategy: 6) The above model is solved using a multi-objective optimization algorithm to obtain the Pareto optimal solution set, and the knee point is selected as the final recommended equipment configuration scheme.

Images