A method for configuring a hybrid electrolytic cell considering wind and light combined modeling

Abstract

The application discloses a kind of mixed electrolytic cell configuration method considering wind and light combined modeling, it is related to electrolytic cell technical field, including the following steps: based on historical wind and light power generation data, the edge distribution function of wind and photovoltaic power generation data is calculated using kernel density estimation method;The edge distribution function of wind and photovoltaic power generation data is fitted using Copula function, and the joint distribution function is obtained;The parameter in Copula function is estimated using Canonical Maximum Likelihood, combined with original data characteristics, the best Copula function is selected using rank correlation coefficient, Euclidean distance fitting, and wind and light output scene is generated;In the case where different electrolytic cell performance is considered, the optimal mixed electrolytic cell configuration is calculated using ensemble empirical mode decomposition power, with economy as target.

Description

Technical Field

[0001] This invention relates to the field of electrolytic cell technology, and more specifically to a hybrid electrolytic cell configuration method that takes into account joint modeling of wind and solar power. Background Technology

[0002] With the deepening of my country's energy transition, hydrogen energy has become a key strategic emerging industry and a priority for future development. The wind-solar hybrid electrolyzer hydrogen production technology uses hydrogen as an energy medium to store the electricity generated by wind and solar power, solving the problem of wind and solar power consumption while meeting domestic and international demand for hydrogen energy. It represents a major direction for the future development of hydrogen energy. Currently, electrolyzer hydrogen production technology has entered the commercialization stage. However, maximizing the utilization of renewable energy sources such as photovoltaic and wind power, and determining the optimal configuration of the hybrid electrolyzer based on the wind and solar power output characteristics of specific regions, remains a pressing issue. This problem becomes even more complex when considering the interdependence of wind and solar power output.

[0003] Currently, commercially available water electrolysis hydrogen production technologies include alkaline electrolyzers (AEL) and proton exchange membrane electrolyzers (PEM). AEL has a relatively low cost (3000 RMB / kW), but a long start-up time, exceeding 60 minutes for cold start-up, and low electrolysis efficiency under fluctuating power output, making efficient operation difficult. PEM, on the other hand, has a short start-up time and high hydrogen production efficiency, but its cost is higher at 9000 RMB / kW. Therefore, hybrid electrolyzers are considered. Regarding the configuration of hybrid electrolyzers, there is limited research literature. There are optimal configuration methods for hybrid electrolyzers based on the analytic hierarchy process (AHP) and entropy weight method, and methods based on non-dominated hierarchical genetic algorithms. However, these methods only consider the characteristics and economics of the two types of electrolyzers themselves, without considering their power output characteristics. Specifically, for wind-solar hybrid hydrogen production, the degree of wind and solar power integration is not considered. Regarding wind-solar joint modeling, photovoltaic and wind power generation are highly stochastic, and results based on prior experience often differ significantly from actual data.

[0004] To address the above issues, how to propose a hybrid electrolyzer configuration method that considers joint wind and solar energy modeling based on historical data, and maximize the utilization of wind and solar energy, is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a hybrid electrolyzer configuration method that considers wind and solar joint modeling, which solves the problems existing in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A hybrid electrolyzer configuration method considering joint wind and solar modeling includes the following steps:

[0008] Based on historical wind and solar power generation data, the marginal distribution function of wind and solar power generation data is calculated using the kernel density estimation method.

[0009] The marginal distribution functions of wind and solar power generation data are fitted using the Copula function to obtain the joint distribution function;

[0010] The parameters in the Copula function are estimated using Canonical Maximum Likelihood. Combined with the features of the original data, the best-fitting Copula function is selected using the rank correlation coefficient and Euclidean distance to generate a wind and solar power output scenario.

[0011] Taking into account the performance of different electrolyzers, the optimal hybrid electrolyzer configuration is calculated by using ensemble empirical mode decomposition of power with economic efficiency as the objective.

[0012] Optionally, the specific methods for obtaining historical wind and solar power generation data are as follows:

[0013] Collect solar power output data for a specific year in the same region, and sample the data every 5 minutes.

[0014] Let X be the sequence of raw photovoltaic power generation data obtained from sampling. pv =[X pv1 ,X pv2 ,…,X pvn The rated power of photovoltaic power generation is P. pv0 The raw wind power generation data sequence is Y. w =[Y w1 ,Y w2 ,…,Y wn The rated power of wind power generation is P. w0 , where n is the number of samples.

[0015] Optionally, this also includes per-unit normalization of historical wind and solar power generation data, specifically:

[0016] Let P pvi =X pvi / P pv0 P wi =Y wi / P w0 Where i∈[1,n],i∈Z, we obtain the photovoltaic power generation dataset P. pv =[p pv1 ,...,p pvn Wind power generation data set P w =[p w1 ,...,p wn ].

[0017] Optionally, the marginal distribution function of wind and solar power generation data is calculated, specifically including the following steps:

[0018] Photovoltaic power generation data set P pv =[p pv1 ,...,p pvn ] and wind power generation data set P w =[p w1 ,...,p wn Substituting these values ​​into equations (1) and (2) respectively, the marginal distribution of the wind and solar data is estimated using the kernel density estimation probability model:

[0019]

[0020]

[0021] In the formula, X represents an estimate of the probability density function, x, y ∈ R; i and Y i These are the power data for photovoltaic power generation and wind power generation, respectively; h represents the window width, n represents the number of samples, and K(x) represents the kernel function.

[0022] The photovoltaic power generation P is obtained from equations (1) and (2). pv probability density function f Ppv (p pv Wind power generation P w probability density function f Pw (p w By integrating equations (3) and (4), the marginal distribution function is obtained. and

[0023]

[0024]

[0025] Photovoltaic power generation data set P pv =[p pv1 ,...,p pvn ] and wind power generation data set P w =[p w1 ,...,p wn Each data point p in ] pvi and p wi The marginal distribution function is used to transform it into a random variable u that conforms to the distribution. i and v i :

[0026]

[0027] Equation (5) yields the empirical distribution values ​​u = [u1, u2, ..., u] for two sets of photovoltaic and wind power data uniformly distributed in [0, 1]. n ] and v = [v1, v2, ..., v n ].

[0028] Optionally, select the best-fitting Copula function, which includes the following steps:

[0029] The parameters of the Copula function are obtained by maximizing the likelihood function of equation (6):

[0030]

[0031] In the formula, c(·) is the joint density function. Let the empirical distribution functions of photovoltaic power and wind power be respectively, then the corresponding parameter estimates are:

[0032]

[0033] The empirical distribution values ​​of photovoltaic and wind power data and Substituting the Copula function yields different Copula models;

[0034] Compare the Kendall rank correlation coefficient, Spearman rank correlation coefficient, and squared Euclidean distance, and select the Copula function that best fits the data. Let Copula be the optimal Copula function. best .

[0035] Optionally, generate a landscape output scene, which specifically includes the following steps:

[0036] Based on the best-fitting Copula function Copula best Generate several daily wind and solar power output samples on a uniform distribution;

[0037] By inversely transforming the marginal distribution, the daily wind and solar power output samples are converted back to the original scale of wind and solar power, resulting in a random variable U′=[u′1,u′2,...,u′] with the same dependency structure as the original wind and solar data. n ] and V′=[v′1,v′2,...,v′ n Clustering generates multiple wind and solar power output scenarios.

[0038] Optionally, power can be decomposed using ensemble empirical mode decomposition, specifically:

[0039] Different white noise h k Added to the original power sequence Obtain the sequence

[0040]

[0041] For each power sequence P after adding noise hi Perform EMD decomposition and obtain the decomposition results:

[0042]

[0043] In the formula, IMF itk The power sequence with added noise (k) is the t-th subsequence obtained after decomposition; M is the number of subsequences; S(t) is the residual of the t-th subsequence; N h The quantity of white noise;

[0044] The components after EMD decomposition of each sequence are averaged to obtain the EEMD decomposition results of the original power sequence:

[0045]

[0046] Repeat the above steps to decompose the power of each wind and solar power output scenario into power components (IMFs). ij and residual S i .

[0047] Optionally, the optimal hybrid electrolyzer configuration is determined by the following steps:

[0048] Considering the volatility V of the AEL electrolyzer AEL and amplitude M AEL PEM electrolyzer volatility V PEM and amplitude M PEM Distribute power with different volatility and amplitude to hybrid electrolyzers with different configurations;

[0049] The costs of wind and solar curtailment penalties, electricity purchase costs, and electrolyzer costs are calculated and normalized using the max-min criterion to obtain the overall economic evaluation scalar value R under different hybrid electrolyzer configurations. The optimal hybrid electrolyzer configuration scheme is then selected.

[0050] As can be seen from the above technical solution, compared with the prior art, the present invention provides a hybrid electrolyzer configuration method considering wind and solar joint modeling. Based on wind and solar hybrid electrolyzer hydrogen production technology, with economic efficiency as the objective, it solves the model for the optimal configuration of the hybrid electrolyzer, and has the following beneficial effects:

[0051] (1) This invention uses the Copula function to generate wind and solar power output scenarios that conform to the distribution characteristics of the original data, taking into account the natural intermittency and uncertainty of wind and solar power output data as well as the special nonlinear relationship between photovoltaic power generation and wind power generation. It is of great significance for the consumption of wind and solar energy and the safe and stable operation of the power system.

[0052] (2) Based on specific data, this invention yields the optimal configuration of a hybrid electrolyzer, considering three economic indicators: wind and solar curtailment penalties, electricity purchase costs, and electrolyzer costs. best (AEL electrolytic cell: PEM electrolytic cell);

[0053] (3) Considering that the unit price of AEL is 3000 yuan / kW and the unit price of PEM is 9000 yuan / kW, the optimal hybrid electrolyzer configuration Q of the present invention is... best For k A-P (Between 3000 and 9000 yuan), giving full play to their respective advantages to achieve synergistic operation, improving the ability of electrolytic hydrogen production units to control complex power fluctuations, maximizing the use of wind and solar energy, saving costs, maximizing benefits, and providing a new idea and method for optimizing the configuration of hybrid electrolyzers. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0055] Figure 1 A flowchart of a hybrid electrolyzer configuration method considering wind and solar co-modeling provided by the present invention. Detailed Implementation

[0056] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] To develop more efficient renewable energy hydrogen production technology, this embodiment innovatively proposes a hybrid electrolyzer configuration method that considers joint wind and solar energy modeling, based on the power output characteristics of two types of electrolyzers. This maximizes the utilization of wind and solar energy and optimizes the electrolyzer configuration. Figure 1 As shown, it includes the following steps:

[0058] Based on historical wind and solar power generation data, the marginal distribution function of wind and solar power generation data is calculated using the kernel density estimation method.

[0059] The marginal distribution functions of wind and solar power generation data are fitted using the Copula function to obtain the joint distribution function;

[0060] The parameters in the Copula function are estimated using Canonical Maximum Likelihood. Combined with the features of the original data, the best-fitting Copula function is selected using the rank correlation coefficient and Euclidean distance to generate a wind and solar power output scenario.

[0061] Taking into account the performance of different electrolyzers, the optimal hybrid electrolyzer configuration is calculated by using ensemble empirical mode decomposition of power with economic efficiency as the objective.

[0062] based on Figure 1 The steps shown in this invention utilize the Copula function to more accurately capture the dependency structure between wind and solar power generation, ensuring that the generated wind and solar power output scenario is closer to reality. Compared to traditional methods, the Copula method can provide higher data accuracy and reliability, which is of great significance for the output prediction and economic dispatch of wind and solar combined power generation systems. Based on the wind and solar combined model built above, ensemble empirical mode decomposition is used to analyze the configuration of different types of hybrid electrolyzers. Considering the different costs of the two types of electrolyzers, the high penalties for wind and solar curtailment, and the consumption of electricity purchase costs, a comprehensive evaluation is conducted to select the optimal hybrid electrolyzer configuration with the goal of optimal economic efficiency. This research provides a new idea and method for the integration and optimization of wind and solar power generation and electrolysis systems, which can improve the utilization of wind and solar energy, reduce costs, and help promote the further development and application of clean energy technologies.

[0063] Next, for Figure 1 The steps shown are described in detail to provide a further understanding of the technical solution of this embodiment.

[0064] 1. Data collection and segmentation

[0065] Collect solar power output data for a specific year in the same region, and sample the data every 5 minutes.

[0066] Let X be the sequence of raw photovoltaic power generation data obtained from sampling. pv =[X pv1 ,X pv2 ,…,X pvn The rated power of photovoltaic power generation is P. pv0 The raw wind power generation data sequence is Y. w =[Y w1 ,Yw2 ,…,Y wn The rated power of wind power generation is P. w0 Where n is the sample size, which is 105120 in this embodiment. The rated power of wind power generation is P. w0 The configuration ratio of AEL and PEM electrolyzers is Q, and the fluctuation rate of the AEL electrolyzer is V. AEL The amplitude is M AEL The fluctuation rate of the PEM electrolyzer is V PEM The amplitude is M PEM The rated power of the mixed electrolytic cell is P0.

[0067] 2. Standardization

[0068] Let P pvi =X pvi / P pv0 P wi =Y wi / P w0 Where i∈[1,n],i∈Z, we obtain the photovoltaic power generation dataset P. pv =[p pv1 ,...,p pvn Wind power generation data set P w =[p w1 ,...,p wn ].

[0069] 3. Probabilistic model estimation

[0070] Photovoltaic power generation data set P pv =[p pv1 ,...,p pvn ] and wind power generation data set P w =[p w1 ,...,p wn Substituting these values ​​into equations (1) and (2) respectively, the marginal distribution of the wind and solar data is estimated using the kernel density estimation probability model:

[0071]

[0072]

[0073] In the formula, X represents an estimate of the probability density function, x, y ∈ R; i and Y i These are the power data for photovoltaic power generation and wind power generation, respectively; h represents the window width, n represents the number of samples, and K(x) represents the kernel function.

[0074] The photovoltaic power generation P is obtained from equations (1) and (2).pv probability density function Wind power generation P w probability density function By integrating equations (3) and (4), the marginal distribution function is obtained. and

[0075]

[0076]

[0077] Photovoltaic power generation data set P pv =[p pv1 ,...,p pvn ] and wind power generation data set P w =[p w1 ,...,p wn Each data point p in ] pvi and p wi The marginal distribution function is used to transform it into a random variable u that conforms to the distribution. i and v i :

[0078]

[0079] Equation (5) yields the empirical distribution values ​​u = [u1, u2, ..., u] for two sets of photovoltaic and wind power data uniformly distributed in [0, 1]. n ] and v = [v1, v2, ..., v n ].

[0080] 4. Select the optimal Copula function

[0081] The empirical distribution values ​​u = [u1, u2, ..., u] of photovoltaic and wind power data are used. n ] and v = [v1, v2, ..., v n Substituting the five Copula functions—Gaussian Copula, t-Copula, Clayton Copula, Gumbel Copula, and Frank Copula—the formula is as follows:

[0082] (1) Gaussian Copula function

[0083] C(u,v;Σ)=Φ2(Φ -1 (u),Φ -1 (v);Σ);

[0084] In the formula, Φ2 is the bivariate normal distribution function; Φ -1Let P be the inverse function of a bivariate normal distribution; u and v are random variables, respectively, representing photovoltaic power generation P. pv And wind power generation P w The cumulative distribution function; Σ is the correlation coefficient matrix.

[0085] (2) t-Copula function

[0086]

[0087] In the formula, t λ It is a multivariate t-distribution function; It is t λ The inverse function; u and v are random variables, photovoltaic power generation P pv And wind power generation P w The cumulative distribution function; λ is the degree of freedom; Σ is the correlation coefficient matrix.

[0088] (3) Clayton Copula function

[0089] C(u,v;θ)=max(u -θ +v -θ -1,0) -1 / θ ;

[0090] In the formula, u and v are random variables, and photovoltaic power generation P is the power generation. pv And wind power generation P w The marginal distribution function value; θ is P pv and P w The degree of dependence between them.

[0091] (4) Gumbel Copula function

[0092] C(u,v;θ)=exp(-[(-ln(u)) θ +(-ln(v)) θ ] 1 / θ );

[0093] In the formula, u and v are random variables, and photovoltaic power generation P is the power generation. pv And wind power generation P w The marginal cumulative distribution function; θ is P pv and P w The degree of dependence between them.

[0094] (5) Frank Copula function

[0095]

[0096] In the formula, u and v are random variables, and photovoltaic power generation P is the power generation. pv And wind power generation P w The cumulative distribution function; θ is the cumulative distribution function of P.pv and P w The degree of dependence between them.

[0097] The parameters of the Copula function are obtained by maximizing the likelihood function of equation (6):

[0098]

[0099] In the formula, c(·) is the joint density function. Let the empirical distribution functions of photovoltaic power and wind power be respectively, then the corresponding parameter estimates are:

[0100]

[0101] The empirical distribution values ​​of photovoltaic and wind power data and Substituting the values ​​into the Copula function yields different Copula models.

[0102] To test the fitting effect of different Copula functions, the rank correlation coefficient and Euclidean distance were calculated as follows:

[0103] (1) Kendall's rank correlation coefficient

[0104]

[0105] In the formula, C(·) is the Copula function, and u and v are the empirical distribution values ​​of the variable photovoltaic and wind power data.

[0106] (2) Spearman rank correlation coefficient

[0107]

[0108] In the formula, C(·) is the Copula function, and u and v are the empirical distribution values ​​of the variable photovoltaic and wind power data.

[0109] (3) Square Euclidean distance

[0110]

[0111] In the formula, n is the total sample size, u i ,v i p pvi ,p wi Empirical distribution values ​​of (samples of photovoltaic and wind power output data) The empirical Copula function has the following formula:

[0112]

[0113] In the formula, I[·] is the indicator function. and Let be the empirical distribution functions of photovoltaic and wind power. When I[·] = 1, otherwise I[·] = 0.

[0114] Compare the rank correlation coefficient and the Euclidean distance, and select the Copula function that best fits the data. Let Copula be the optimal Copula function. best .

[0115] 5. Generate scenes and cluster them

[0116] Based on the best-fitting Copula function Copula best 1000 daily wind and solar power output samples on a uniform distribution are generated. Through inverse transformation of the marginal distribution, these 1000 samples are transformed back to the original scale of wind and solar power, resulting in a random variable U′=[u′1,u′2,...,u′] with the same dependency structure as the original wind and solar data. n ] and V′=[v′1,v′2,...,v′ n This generates 1000 scenes. The main steps are as follows:

[0117] (1) Draw two sets of random numbers independently from a standard uniform distribution: {x′1, x′2, ..., x′ n} and {y′1,y′2,...,y′ n};

[0118] (2) Let the marginal distribution function value of the first random variable be u′1=x′1, and obtain the marginal distribution function value of the second random variable according to the formula of the Gaussian Copula function, where C is the selected Copula function. best function:

[0119]

[0120] For the marginal distribution function value u′ of the nth random variable n Find the solution to the following equation:

[0121]

[0122] Repeat the above steps to obtain a random variable U′=[u′1,u′2,...,u′] with the same dependency structure as the original landscape data. n ] and V′=[v′1,v′2,...,v′ n ].

[0123] The simulated random variables are transformed back into the original actual data space of wind power and photovoltaic power generation using the inverse cumulative distribution function, and the calculation is as follows:

[0124]

[0125]

[0126] In the formula, and yes and The inverse function, p′ pvi and p′ wi It generates photovoltaic power generation data and wind power generation data with the same distribution as the original data.

[0127] The photovoltaic power generation data column P′ is obtained from the above steps. pv =[p′ pv1 ,p′ pv2 ,...,p′ pvn ] and wind power generation data column P′ pw =[p′ w1 ,p′ w2 ,...,p′ wn After dimensionality reduction, k-means clustering was performed, ultimately yielding six typical daily photovoltaic power generation scenarios. Six classic scenarios of wind power generation in Japan

[0128] In summary, considering the correlation between wind and solar power generation, we derive the main characteristics and trends of scenarios incorporating wind and solar power output. The calculation is as follows:

[0129]

[0130] In the formula, 1≤i≤6, i∈Z.

[0131] 6. Empirical Mode Decomposition of Sets

[0132] Scenario 1 with variables For example, the steps for performing ensemble empirical mode decomposition (EEMD) are as follows:

[0133] (1) Different white noise h k Added to the original power sequence Obtain the sequence

[0134]

[0135] (2) For each power sequence after adding noise Perform EMD decomposition and obtain the decomposition results:

[0136]

[0137] In the formula, IMF 1tkThe power sequence with added noise (k) is the t-th subsequence obtained after decomposition; M is the number of subsequences; S(t) is the residual of the t-th subsequence; N h This represents the amount of white noise.

[0138] (3) Averaging the components after EMD decomposition of each sequence yields the EEMD decomposition results of the original power sequence:

[0139]

[0140] In the formula, N h =288, M=6; The scene is obtained from the above steps. The M IMF power components and the residual S1.

[0141] Repeat the above steps to generate power for the 6 landscape scenes. Power components (IMFs) were obtained by performing power decomposition separately. ij and residual S i , where 1≤i≤6, 1≤j≤6, i,j∈Z.

[0142] 7. Select the optimal electrolytic cell configuration

[0143] Considering the volatility V of the AEL electrolyzer AEL and amplitude M AEL PEM electrolyzer volatility V PEM and amplitude M PEM Power with different volatility and amplitude is allocated to hybrid electrolyzers with different configurations. Let the AEL electrolyzer component be P. AEL The PEM electrolytic cell component is P PEM And the remaining power is P c Let Q take different values, such as Q = 1:0 (configuration Q1), 1:1 (configuration Q2), 1:2 (configuration Q3), 1:3 (configuration Q4), 3:1 (configuration Q5), 2:1 (configuration Q6), 0:1 (configuration Q7), etc. Taking scenario 1 as an example, power is distributed to hybrid electrolyzers with different configurations.

[0144]

[0145] If the remaining power P c If the power is greater than 0, then it is discarded power; if the remaining power P... c If <0, then the required power is the power to be purchased. Repeat the above steps to assign different scenarios to hybrid electrolyzers with different configurations. With economic efficiency as the goal, calculate the cost of 7 hybrid electrolyzer configurations and select the optimal hybrid electrolyzer configuration scheme.

[0146] (1) The penalty fee for wind and solar power curtailment within one day is calculated as follows:

[0147]

[0148] In the formula, C p The penalty fee for abandoning wind and solar power (in yuan), k p P represents the cost coefficient for the light-wasting penalty. pv.c (t), P w.c (t) represents the discarded power (kW) of wind power and photovoltaic power at time t, respectively, and Δt is the sampling time, which is 5 min in this embodiment, i.e. 1 / 12 h.

[0149] (2) The formula for calculating electricity purchase costs is as follows:

[0150]

[0151] In the formula, C F For electricity purchase cost (yuan), k F For the electricity purchase price, P G (t) represents the power purchased at time t (kW), and Δt is the sampling time, which is 5 min in this embodiment, i.e. 1 / 12 h.

[0152] (3) Equipment cost is calculated as follows:

[0153] C s =k ael ·P ael.E +k pem ·P pem.E ;

[0154] In the formula, C s k represents the total cost of the equipment (in yuan). ael For the unit price of an AEL electrolyzer, k pem P is the unit price of a PEM electrolyzer. ael.E P pem.E These represent the installed capacity of AEL electrolytic cells and PEM electrolytic cells, respectively.

[0155] Combining the above three indicators, we use max-mincriterion for normalization to calculate the economics of each configuration scheme. The normalized calculations for each indicator are as follows:

[0156]

[0157]

[0158]

[0159] Where I Cp I CF I Cs These are the normalized results for wind and solar curtailment penalties, electricity purchase costs, and equipment costs, respectively; C piC Fi C si , respectively, represent the wind and solar curtailment penalty cost, electricity purchase cost, and equipment cost for the i-th configuration scheme; max and min represent the maximum and minimum values ​​of the desired indicators.

[0160] Different configurations Q i Substituting the power allocated to the hybrid electrolyzer into the aforementioned cost calculation formula, we can obtain scalar values ​​for three economic indicators under different configurations of the hybrid electrolyzer. Then, summing these three scalar values ​​yields the overall economic evaluation scalar value R for different configurations of the hybrid electrolyzer:

[0161]

[0162] Sort the R values ​​of the hybrid electrolyzer configurations and select the optimal hybrid electrolyzer configuration scheme, denoted as Q. best Taking economic factors into account, Q best This is the optimal hybrid electrolyzer configuration scheme for this region.

[0163] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0164] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for configuring a hybrid electrolyzer considering joint wind and solar modeling, characterized in that, Includes the following steps: Based on historical wind and solar power generation data, the marginal distribution function of wind and solar power generation data is calculated using the kernel density estimation method. The marginal distribution functions of wind and solar power generation data are fitted using the Copula function to obtain the joint distribution function; The parameters in the Copula function are estimated using Canonical Maximum Likelihood. Combined with the features of the original data, the best-fitting Copula function is selected using the rank correlation coefficient and Euclidean distance to generate a wind and solar power output scenario. Taking into account the performance of different electrolyzers, the optimal hybrid electrolyzer configuration is calculated by using ensemble empirical mode decomposition power with economic efficiency as the objective. The process of generating a landscape power output scene includes the following steps: Based on the best-fitting Copula function Generate several daily wind and solar power output samples on a uniform distribution; By inversely transforming the marginal distribution, the daily wind and solar power output samples are converted back to the original scale of wind and solar power, resulting in random variables with the same dependency structure as the original wind and solar data. and Clustering generates multiple landscape power output scenarios; The power is decomposed using ensemble empirical mode decomposition, specifically as follows: Different white noise Added to the original power sequence , to obtain the sequence : （8）； For each power sequence with added noise Perform EMD decomposition and obtain the decomposition results: （9）； In the formula, For the first k The power sequence after adding noise is decomposed to obtain the first... t Subsequences; M The number of subsequences; for t The residuals of the sequence; The quantity of white noise; The components after EMD decomposition of each sequence are averaged to obtain the EEMD decomposition results of the original power sequence: （10）； Repeat the above steps to decompose the power of each wind and solar power output scenario into power components. and residuals ; The optimal configuration of the hybrid electrolyzer is determined by the following steps: Considering the volatility of AEL electrolyzers and amplitude PEM electrolyzer volatility and amplitude Distribute power with different volatility and amplitude to hybrid electrolyzers with different configurations; The costs of wind and solar power curtailment penalties, electricity purchase costs, and electrolyzer costs were calculated and normalized using the max-min criterion to obtain the overall economic scalar values ​​for different configurations of hybrid electrolyzers. R Select the optimal hybrid electrolyzer configuration scheme.

2. The hybrid electrolyzer configuration method considering wind and solar joint modeling according to claim 1, characterized in that, The specific methods for obtaining historical wind and solar power generation data are as follows: Collect solar power output data for a specific year in the same region, and sample the data every 5 minutes. Let the sampled raw photovoltaic power generation data sequence be... The rated power of photovoltaic power generation is The original wind power generation data sequence is as follows: The rated power of wind power generation is ,in n This represents the number of samples.

3. The hybrid electrolyzer configuration method considering joint wind and solar modeling according to claim 2, characterized in that, This also includes standardizing historical wind and solar power generation data, specifically: make , ,in Obtain photovoltaic power generation dataset Wind power generation data set .

4. The hybrid electrolyzer configuration method considering joint wind and solar modeling according to claim 3, characterized in that, The calculation of the marginal distribution function for wind and solar power generation data includes the following steps: Photovoltaic power generation data set Wind power generation data set Substituting into equations (1) and (2) respectively, the marginal distribution of the wind and solar data is estimated using the kernel density estimation probability model: （1）； （2）； In the formula, , This represents an estimate of the probability density function. ; and These are the power data for photovoltaic power generation and wind power generation, respectively; h Indicates window width. n Indicates the number of samples. Represents the kernel function. ; The photovoltaic power generation is obtained from equations (1) and (2). probability density function Wind power generation capacity probability density function By integrating using formulas (3) and (4), the marginal distribution function is obtained. and : （3）； （4）； Photovoltaic power generation data set Wind power generation data set Each data point and Transform it into a random variable that conforms to a distribution using a marginal distribution function. and : （5）； Equation (5) yields the empirical distribution values ​​of two sets of photovoltaic and wind power data uniformly distributed in [0,1]. and .

5. A hybrid electrolyzer configuration method considering joint wind and solar modeling according to claim 4, characterized in that, Selecting the best-fitting Copula function involves the following steps: The parameters of the Copula function are obtained by maximizing the likelihood function of equation (6): （6）； In the formula, For the joint density function, Let the empirical distribution functions of photovoltaic power and wind power be respectively, then the corresponding parameter estimates are: （7）； The empirical distribution values ​​of photovoltaic and wind power data and Substituting the Copula function yields different Copula models; Compare the Kendall rank correlation coefficient, Spearman rank correlation coefficient, and squared Euclidean distance, and select the Copula function that best fits the data. .

Images