Offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency

Abstract

The invention discloses an offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency, and relates to the field of energy storage optimization configuration, and the offshore wind storage configuration method comprises the steps: carrying out the clustering of an offshore wind power output data set through employing a firefly forest clustering algorithm of a Pearson's correlation coefficient, obtaining an annual operation scene of offshore wind power; establishing a line impedance complex affine model based on submarine cable line impedance uncertainty by using an affine mathematical technology; carrying out load flow calculation on the optimized line impedance complex affine model by utilizing a convergence discrimination technology, and establishing a submarine cable transmission efficiency model of line parameter uncertainty according to a load flow calculation result; and based on the annual operation scene of the offshore wind power, establishing a double-stage optimization configuration model, and solving the double-stage optimization configuration model. By means of the load flow calculation and judgment method of the line impedance complex affine model, the defect that a traditional comparative complex model cannot be adopted for voltage complex affine numbers is overcome.

Description

Technical Field

[0001] The present invention relates to the field of energy storage optimization configuration, and in particular to an offshore wind storage configuration method that takes into account wind power output uncertainty and submarine cable transmission efficiency. Background Art

[0002] In recent years, with the deepening of global energy transformation, offshore wind power, as a clean and renewable energy source, has received widespread attention and rapid development. Due to the characteristics of offshore wind power such as strong output volatility and high uncertainty, direct grid connection will have an adverse impact on the stability of the power grid and the quality of power. To alleviate this problem, energy storage technology has been widely used in wind power systems to improve the stability and economy of the system by shaving peaks and filling valleys and adjusting power output.

[0003] However, in practical applications, the optimal configuration of offshore wind storage systems still faces many challenges. First, the operating scenarios of offshore wind power are complex and changeable. Traditional clustering algorithms require pre-defining the number of clusters or setting a fixed number of neighbors, which limits the accuracy and robustness of clustering results. At the same time, these methods generally rely on computationally intensive iterative processes with low computational efficiency. In addition, as an important transmission channel for offshore wind storage systems, the impedance parameters of submarine cables are easily affected by environmental factors and show certain uncertainties, which brings additional complexity to the accurate calculation of transmission losses. If the uncertainty of submarine cable parameters is not fully considered, it may lead to deviations in system planning results, thereby affecting the overall economy and stability.

[0004] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0005] In response to the problems in the related technology, the present invention proposes an offshore wind storage configuration method that takes into account the uncertainty of wind power output and the transmission efficiency of submarine cables, so as to overcome the above-mentioned technical problems existing in the existing related technology.

[0006] To this end, the specific technical solution adopted by the present invention is as follows:

[0007] An offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency, the offshore wind storage configuration method comprising:

[0008] S1. Obtain an offshore wind power output data set, and cluster the offshore wind power output data set using the firefly forest clustering algorithm of the Pearson correlation coefficient to obtain the annual operation scenario of offshore wind power;

[0009] S2. Establish a line impedance complex affine model based on the uncertainty of submarine cable line impedance using affine mathematical technology, and optimize the line impedance complex affine model based on polynomial fitting technology;

[0010] S3. Use convergence judgment technology to perform power flow calculation on the optimized line impedance complex affine model, and establish a submarine cable transmission efficiency model with line parameter uncertainty based on the power flow calculation results;

[0011] S4. Based on the annual operation scenario of offshore wind power, a two-stage optimization configuration model is established and solved to realize offshore wind storage configuration.

[0012] Preferably, an offshore wind power output data set is obtained, and the offshore wind power output data set is clustered using a firefly forest clustering algorithm with a Pearson correlation coefficient, and the annual operation scenarios of offshore wind power are obtained, including:

[0013] S11, obtaining an offshore wind power output data set containing a number of independently distributed data points, and using the Pearson coefficient to characterize the dissimilarity between the data points;

[0014] S12, based on the dissimilarity between the data points, iterate each data point in the offshore wind power output data set, and perform maximum likelihood estimation on the number of neighbors of the data point in each iteration process to obtain the likelihood ratio of the comprehensive model;

[0015] S13, judging whether the likelihood ratio of the comprehensive model meets the preset conditions, if so, updating the number of neighbors of the current maximum likelihood estimate and executing step S14, otherwise, repeating the iterative process of the data points until the likelihood ratio of the comprehensive model meets the preset conditions;

[0016] S14. The data points are regarded as fireflies with adaptive brightness, and the fireflies are clustered to form a tree structure. The tree structure is merged to generate a firefly forest, and the annual operation scenario of offshore wind power is obtained.

[0017] Preferably, the data points are used as fireflies with adaptive brightness, and the fireflies are clustered to form a tree structure, and the tree structures are merged to generate a firefly forest. The annual operation scenarios of offshore wind power include:

[0018] S141, taking the data points as fireflies, initializing firefly parameters based on the distribution parameters of a predefined offshore wind power output data set, and calculating the firefly brightness according to the firefly parameters;

[0019] S142, arranging the fireflies in descending order based on their brightness, each firefly selects a brighter firefly closest to it as a guide, obtaining a preliminary clustering result, and using a tree structure to store the preliminary clustering result to form a firefly tree;

[0020] S143. Based on the luminous interaction signals of the firefly population, the firefly trees are merged into a firefly forest to obtain the final clustering result, and the final clustering result is used as the annual operation scenario of offshore wind power.

[0021] Preferably, taking the data points as fireflies, initializing firefly parameters based on the distribution parameters of a predefined offshore wind power output data set, and calculating the firefly brightness according to the firefly parameters includes:

[0022] S1411, treating the data points as fireflies with adaptive brightness, and initializing firefly parameters according to the distribution parameters of the offshore wind power data set, and the firefly parameters include brightness and the number of neighbors of maximum likelihood estimation;

[0023] S1412, calculating the interquartile range of the number of neighbors of the maximum likelihood estimate using the interquartile range calculation formula, and calculating the upper bound and the lower bound of the number of neighbors based on the interquartile range of the number of neighbors;

[0024] S1413, performing an interquartile range test on the number of neighbors of the maximum likelihood estimate based on the upper and lower bounds of the number of neighbors, identifying the proximity of outliers in the number of neighbors of the maximum likelihood estimate, and obtaining the error of the maximum likelihood estimate density;

[0025] S1414. Based on the error of the maximum likelihood estimation density, the firefly brightness is calculated using the calculation formula of the firefly brightness.

[0026] Preferably, the fireflies are arranged in descending order based on their brightness, each firefly selects a brighter firefly closest to it as a guide, a preliminary clustering result is obtained, and a tree structure is used to store the preliminary clustering result to form a firefly tree, including:

[0027] S1421, sorting the fireflies in descending order according to their brightness values, iterating each firefly and the neighboring maximum likelihood estimate of each firefly;

[0028] S1422, comparing the brightness value of the neighboring fireflies with the brightness value of the current firefly. If the brightness value of the neighboring firefly is greater than that of the current firefly, the neighboring firefly becomes the middle firefly of the current firefly, and the current firefly is added to the corresponding firefly tree;

[0029] S1423. If the brightness values ​​of all neighboring fireflies are smaller than that of the current firefly, it indicates that the current firefly is a root firefly, and the root firefly obtains a new cluster label, and generates a new firefly tree, and uses the new firefly tree as the local cluster center;

[0030] S1424, traverse the remaining fireflies, if the root firefly becomes the middle firefly of the remaining fireflies, assign its cluster label to the remaining fireflies to obtain a preliminary clustering result, and use a tree structure to store the preliminary clustering result to form a firefly tree.

[0031] Preferably, the expression of the line impedance complex affine model based on the uncertainty of the submarine cable line impedance is:

[0032]

[0033] In the formula, represents the complex affine model of the line impedance from node i to node j; represents the affine form of the parameters; R i-j and X i-j Respectively represent the line resistance and reactance value from node i to node j; δ r,1 Indicates a parameter used to measure line resistance fluctuation.

[0034] Preferably, using convergence discrimination technology to perform power flow calculation on the optimized line impedance complex affine model, and establishing a submarine cable transmission efficiency model with line parameter uncertainty according to the power flow calculation results includes:

[0035] S31, based on the known load and initial state of node j, obtain the current between node i and node j, and calculate the injected power of each node according to the current and line impedance complex affine model;

[0036] S32, calculating the current and voltage drop between each node and other nodes based on the injected power of each node;

[0037] S33, iteratively calculating the voltage difference of each node voltage relative to the voltage of the previous iteration, using the affine number fluctuation domain overlap area comparison convergence judgment technology to judge the similarity of the two iteration results, and obtaining the power flow calculation result of the line impedance complex affine model;

[0038] S34. According to the power flow calculation results of the line impedance complex affine model, the relationship between the energy storage output and the voltage at the head end of the submarine cable is analyzed, and based on the relationship between the energy storage output and the voltage at the head end of the submarine cable, a submarine cable transmission efficiency model with line parameter uncertainty is established.

[0039] Preferably, based on the annual operation scenario of offshore wind power, a two-stage optimization configuration model is established, and the two-stage optimization configuration model is solved to achieve offshore wind storage configuration including:

[0040] S41. Taking the lowest annual comprehensive cost of offshore wind storage, the smallest net load fluctuation and the smallest voltage fluctuation as the optimization objectives, an upper capacity configuration model for offshore wind storage is constructed;

[0041] S42, taking the upper capacity configuration model as input and taking minimizing the daily loss of the submarine cable transmission line as the optimization goal, constructing the lower optimization operation model of the offshore wind storage;

[0042] S43. An improved geyser algorithm is used to solve the upper capacity configuration model, and a solver is used to solve the lower optimization operation model.

[0043] Preferably, using the improved geyser algorithm to solve the upper capacity configuration model includes:

[0044] Initialize the parent population P op , calculate the fitness function of the upper capacity configuration model, and ensure that the fitness function of the upper capacity configuration model satisfies the constraint conditions;

[0045] Generate offspring population R using the eruption mechanism T , the parent population P op and the offspring population R T Merge to form population S T , and based on the cooling process, the population S T Sort and evaluate the population S T The quality of each solution in

[0046] Based on the advantage level and reference distance of each solution, from the population S T Select the best solution and generate the next generation parent population P op,T+1 , repeat the eruption mechanism and cooling process, and continue to generate a new generation of parent populations until the convergence condition is met;

[0047] The final parent population P op As the Pareto optimal solution of the upper capacity configuration model, the final solution of the upper capacity configuration model is selected by using the grey target decision based on the entropy weight method.

[0048] Preferably, the expression of the lower layer optimization operation model is:

[0049]

[0050] Where P bg,t represents the active power injected into the head end of the submarine cable at time t; η hl,t K represents the transmission efficiency of the submarine cable considering the uncertainty of line parameters at time t; d represents the number of annual operation scenarios of offshore wind power; T k Indicates the number of days corresponding to the kth scenario.

[0051] The beneficial effects of the present invention are:

[0052] 1. The firefly forest clustering algorithm based on the Pearson correlation coefficient of the present invention realizes dynamic adjustment through adaptive neighborhood estimation. The brightness of each data point is calculated according to its distribution characteristics. There is no need to manually set the number of neighbors, which significantly simplifies parameter tuning and improves flexibility. It shows excellent scalability when processing large-scale data sets, and filters outliers through the interquartile range method to ensure the stability of brightness calculation and the accuracy of clustering results. It has significant advantages in processing complex data distribution and large-scale data, and is particularly suitable for practical application scenarios requiring high efficiency and high precision. At the same time, the present invention applies the improved firefly forest clustering algorithm to scene clustering, which has higher robustness and stability than common clustering algorithms.

[0053] 2. The present invention uses affine mathematics to model the uncertainty of line impedance to obtain a complex affine model of line impedance based on the uncertainty of submarine cable line impedance, and uses a polynomial fitting method to improve the traditional affine division, so that the conservatism of the original affine number division is overcome with a slight loss of authenticity during subsequent affine operations. Then, through the power flow calculation and discrimination method of the complex affine model of line impedance, the defect that the complex affine number of voltage cannot use the traditional comparative complex module is solved.

[0054] 3. The present invention derives a submarine cable line transmission efficiency formula that takes into account the uncertainty of line parameters. Compared with the traditional derivation method that assumes that the line impedance is fixed, it has better conservatism and can more accurately characterize the transmission efficiency of the submarine cable line under actual conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1 is a flow chart of an offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to an embodiment of the present invention;

[0057] Figure 2 It is a flowchart of a firefly forest clustering algorithm in an offshore wind storage configuration method that takes into account wind power output uncertainty and submarine cable transmission efficiency according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments. They can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention.

[0059] According to an embodiment of the present invention, a method for configuring offshore wind storage taking into account the uncertainty of wind power output and the transmission efficiency of submarine cables is provided.

[0060] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to an offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to an embodiment of the present invention, the offshore wind storage configuration method includes:

[0061] S1. Obtain the offshore wind power output data set, and use the firefly forest clustering algorithm of the Pearson correlation coefficient to cluster the offshore wind power output data set to obtain the annual operation scenario of offshore wind power.

[0062] Among them, the offshore wind power output data set is obtained, and the firefly forest clustering algorithm of the Pearson correlation coefficient is used to cluster the offshore wind power output data set, and the annual operation scenarios of offshore wind power are obtained, including:

[0063] S11, obtaining an offshore wind power output data set containing a number of independently distributed data points, and using the Pearson coefficient to characterize the dissimilarity between the data points;

[0064] S12, based on the dissimilarity between the data points, iterate each data point in the offshore wind power output data set, and perform maximum likelihood estimation on the number of neighbors of the data point in each iteration process to obtain the likelihood ratio of the comprehensive model;

[0065] S13, judging whether the likelihood ratio of the comprehensive model meets the preset conditions, if so, updating the number of neighbors of the current maximum likelihood estimate and executing step S14, otherwise, repeating the iterative process of the data points until the likelihood ratio of the comprehensive model meets the preset conditions;

[0066] S14. The data points are regarded as fireflies with adaptive brightness, and the fireflies are clustered to form a tree structure. The tree structure is merged to generate a firefly forest, and the annual operation scenario of offshore wind power is obtained.

[0067] The data points are used as fireflies with adaptive brightness, and the fireflies are clustered to form a tree structure. The tree structures are merged to generate a firefly forest. The annual operation scenarios of offshore wind power include:

[0068] S141, taking the data points as fireflies, initializing firefly parameters based on the distribution parameters of a predefined offshore wind power output data set, and calculating the firefly brightness according to the firefly parameters.

[0069] Among them, the data points are taken as fireflies, the firefly parameters are initialized based on the distribution parameters of the predefined offshore wind power output data set, and the firefly brightness is calculated according to the firefly parameters, including:

[0070] S1411, treating the data points as fireflies with adaptive brightness, and initializing firefly parameters according to the distribution parameters of the offshore wind power data set, and the firefly parameters include brightness and the number of neighbors of maximum likelihood estimation;

[0071] S1412, calculating the interquartile range of the number of neighbors of the maximum likelihood estimate using the interquartile range calculation formula, and calculating the upper bound and the lower bound of the number of neighbors based on the interquartile range of the number of neighbors;

[0072] S1413, performing an interquartile range test on the number of neighbors of the maximum likelihood estimate based on the upper and lower bounds of the number of neighbors, identifying the proximity of outliers in the number of neighbors of the maximum likelihood estimate, and obtaining the error of the maximum likelihood estimate density;

[0073] S1414. Based on the error of the maximum likelihood estimation density, the firefly brightness is calculated using the calculation formula of the firefly brightness.

[0074] S142, arranging the fireflies in descending order based on their brightness, each firefly selects the brighter firefly closest to it as a guide, obtaining a preliminary clustering result, and using a tree structure to store the preliminary clustering result to form a firefly tree.

[0075] Among them, fireflies are arranged in descending order based on their brightness, and each firefly selects the brighter firefly closest to it as a guide to obtain preliminary clustering results, and the preliminary clustering results are stored in a tree structure to form a firefly tree including:

[0076] S1421, sorting the fireflies in descending order according to their brightness values, iterating each firefly and the neighboring maximum likelihood estimate of each firefly;

[0077] S1422, comparing the brightness value of the neighboring fireflies with the brightness value of the current firefly. If the brightness value of the neighboring firefly is greater than that of the current firefly, the neighboring firefly becomes the middle firefly of the current firefly, and the current firefly is added to the corresponding firefly tree;

[0078] S1423. If the brightness values ​​of all neighboring fireflies are smaller than that of the current firefly, it indicates that the current firefly is a root firefly, and the root firefly obtains a new cluster label, and generates a new firefly tree, and uses the new firefly tree as the local cluster center;

[0079] S1424, traverse the remaining fireflies, if the root firefly becomes the middle firefly of the remaining fireflies, assign its cluster label to the remaining fireflies to obtain a preliminary clustering result, and use a tree structure to store the preliminary clustering result to form a firefly tree.

[0080] S143. Based on the luminous interaction signals of the firefly population, the firefly trees are merged into a firefly forest to obtain the final clustering result, and the final clustering result is used as the annual operation scenario of offshore wind power.

[0081] In order to facilitate understanding of the above technical solutions of the present invention, the following is a detailed description of the method of obtaining an offshore wind power output data set in the actual process of the present invention, and clustering the offshore wind power output data set using the firefly forest clustering algorithm of the Pearson correlation coefficient to obtain the annual operation scenario of offshore wind power:

[0082] Step 1: Similarity representation based on Pearson coefficient:

[0083] Given a wind power output data set, there is a series of independent and identically distributed data points {x1, ..., x n}, each point can be regarded as a firefly, and the Pearson coefficient is used to characterize the data point x i and x j The dissimilarity between them is shown in formula (1):

[0084]

[0085] Where: r i,j Represents data point x i and x j The dissimilarity between them is larger, i,j Value represents x i and x j The difference between them is greater; n represents the number of data points; Represents x i The mean of Represents x j The mean of .

[0086] Step 2: Maximum likelihood estimation based on comprehensive model:

[0087] The volume v of the hypersphere is defined as follows:

[0088] v i,l =σ·r i,l -r i,l-1 ; (2)

[0089] Where: r i,l Represents point x i The dissimilarity between it and its lth nearest neighbor; r i,l-1Represents point x i The dissimilarity between it and its l-1th nearest neighbor; σ represents the scale coefficient of the hypersphere.

[0090] The volume v of the hypersphere obeys an exponential distribution with parameter ρ, as shown in formula (3):

[0091] f(v)=ρ·e -ρ·v ; (3)

[0092] Where: f(v) means that the volume v of the hypersphere follows an exponential distribution with parameter ρ; e means the exponential function (i.e., e x ).

[0093] Data point x i The density likelihood estimation function of its k nearest neighbors is as shown in formula (4):

[0094]

[0095] Where: L i (ρ i |{r i,l} i≤k ) represents the data point x i and the density likelihood estimation function of its k nearest neighbors; V i,k Represents the data point x i The total volume of the hypersphere centered on its k nearest neighbors; ρ i represents the likelihood density of the k nearest neighbors; L i Represents data point x i The density likelihood estimation function, that is, L i (x).

[0096] To obtain the maximum likelihood estimate, we need to modify the parameter ρ in equation (4) i To perform the derivation:

[0097]

[0098] Where: Set L' i =0, calculate parameter ρ i The maximum likelihood estimate of i represents the derivative of the likelihood estimation function.

[0099] The likelihood ratio method was used to obtain the comprehensive model D k , as shown in formula (6):

[0100] D k = -2k[log(V i,j )+log(V j,k )-2log(V i,k +V j,k)+log(4)]; (6)

[0101] Where: V i,k Represents the data point x i The total volume of the hypersphere centered on its k nearest neighbors; V j,k Represents the data point x j The total volume of the hypersphere centered on its k nearest neighbors; V i,j Represents the data point x i The total volume of the hypersphere centered on its j nearest neighbors, where the comprehensive model D k It follows a chi-square distribution with 1 degree of freedom. As k increases, the larger the D k A higher value indicates a higher confidence in the maximum likelihood estimate of k.

[0102] In order to adaptively obtain the maximum likelihood estimate of the number of neighbors k for each data point, a threshold D is set thr , so that formula (6) satisfies:

[0103]

[0104] Where: The algorithm starts from the first neighbor and updates the number of neighbors of the current maximum likelihood estimate to k. Assume that there is a value k≤k such that D k <D thr And D k+1 >D thr , then it can be considered that, in thr Under the corresponding confidence level, the number of neighbors of the maximum likelihood estimate of the data point is k, and the density ρ of the point is defined as formula (8):

[0105]

[0106] At this time, the corresponding estimation error is defined as formula (9):

[0107]

[0108] Where: ε i represents the error corresponding to the maximum likelihood estimate density; k represents the number of neighbors of the maximum likelihood estimate of density ρ; Represents data point x i The number of neighbors for the maximum likelihood estimate of .

[0109] The present invention avoids the need for preset parameters and reduces the time complexity typically associated with biomimetic algorithms by allowing the algorithm to dynamically adjust nearest neighbor relationships.

[0110] Step 3: Define the submodules of the Firefly Forest Algorithm:

[0111] The firefly forest algorithm consists of three main modules: outlier filtering brightness calculation, firefly tree generation, and merging firefly trees into firefly forests. Initially, data points are treated as fireflies with adaptive brightness; these fireflies are then clustered based on brightness and proximity to form a tree structure; finally, clusters are merged into firefly forests to obtain the final clustering result.

[0112] 1) Abnormal value filtering brightness calculation:

[0113] Each data point is considered as a firefly, and the brightness and number of neighbors are initialized adaptively according to the offshore wind dataset distribution and defined parameters.

[0114] 2) Firefly tree generation:

[0115] Fireflies are arranged in descending order of brightness, and each firefly selects the closest brighter firefly as its guide, thereby joining the cluster of the guide. Fireflies that guide other fireflies but are not guided by other fireflies are called "root fireflies", and fireflies that guide other fireflies and are guided by brighter fireflies are called "intermediate fireflies". This process completes the preliminary clustering and uses a tree data structure to store these fireflies. Members of the same cluster are located on the same firefly tree.

[0116] 3) Firefly trees merged into Firefly forest:

[0117] Certain populations of fireflies gather and flash synchronously at night. They communicate through light signals and achieve collective synchronization by controlling the timing and frequency of light emission. Fireflies that do not guide other fireflies are called "leaf fireflies". Each firefly cluster represents a tree, and the interaction and merging between clusters are carried out through leaf fireflies. The merging of clusters is similar to the merging of trees, eventually forming a unified firefly forest.

[0118] Among them, the outlier filtering brightness calculation includes:

[0119] The number of nearest neighbors estimated by maximum likelihood can be used to calculate the corresponding point density. The number of nearest neighbors obtained by maximum likelihood estimation is usually related to the confidence level. The higher the confidence level, the more accurate the result, but more neighbors will be obtained, which will lead to an overestimation of the actual number of neighbors, resulting in errors in subsequent density calculations. Therefore, the over-estimated number of neighbors is corrected.

[0120] After the current number of neighbors satisfies formula (7), the estimated number of neighbors is tested for interquartile range (IQR). Define data point x i The volume of the hypersphere is V i,list , store each adjacent corresponding v in ascending order i,l , the calculation formula of IQR is:

[0121]

[0122] Where: IQR means interquartile range; Represents the volume list V of the hypersphere i,list The third quartile of Represents the volume list V of the hypersphere shell i,list The first quartile of .

[0123] After obtaining the IQR, calculate the lower and upper bounds according to formulas (11) and (12):

[0124]

[0125] Where: L ower Bound and U pper Bound Respectively represent the upper and lower bounds of the volume of the hypersphere; I thr Indicates the threshold of the IQR test, the higher the I thr A value of 0 indicates a looser test, making it easier for the estimated number of neighbors to pass the test, resulting in a slightly larger estimate and a smaller I thr A value of represents a stricter test, making it more difficult for the estimated number of neighbors to pass the test, resulting in a slightly smaller estimate.

[0126] After obtaining the lower and upper bounds, in order to identify the outlier neighbors, the estimated number of neighbors is subjected to an IQR test, as shown in equations (13)-(15):

[0127]

[0128] V i,final =V i,list -V i,koutliers ; (14)

[0129] k final =number(V i,final ); (15)

[0130] Where: Represents the volume list V of the hypersphere shell i,list The part of the V that is larger than the upper limit or smaller than the lower limit; i,final Indicates to remove all The remaining hypersphere shell volume list V i,list Part; k outliers Indicates that the outlier is adjacent; k final Represents the final estimated number of neighbors, whose value is equal to the set V i,final The number of elements in .

[0131] Considering the influence of prediction error, a larger prediction error indicates greater instability of the corresponding result. Therefore, the brightness of the firefly is inversely proportional to its prediction error. The calculation formula of the firefly brightness is as follows:

[0132]

[0133] Where: B right,i Indicates the brightness of fireflies; ε i represents the error corresponding to the maximum likelihood estimation density; k final represents the maximum likelihood neighbor number; sum(V i,final ) represents the data point x i and its surroundings final The sum of the IQR inspection volumes of the hyperspherical shells between the neighbors.

[0134] Construct a KD tree (K-Dimensional Tree, i.e., a partitioning structure for high-dimensional spatial data) to create the nearest neighbor matrix and distance matrix of the wind power output data set; initialize the nearest neighbor and threshold judgment information of each point; iterate each point in the data set until all points are traversed; in each iteration, perform maximum likelihood estimation on the corresponding ρ value according to formula (6) to obtain the comprehensive model D k and D k+1 The likelihood ratio is calculated and the relationship between the comprehensive model and the threshold is determined. If the maximum likelihood proximity is reached, the algorithm will start IQR filtering and use formula (16) to calculate the brightness and save the relevant information.

[0135] Among them, the firefly tree generation includes:

[0136] After obtaining the brightness and maximum likelihood estimated neighborhood of each point, the firefly tree generation module is executed, which performs a preliminary clustering of the dataset according to the following definition:

[0137] 1) Definition 1:

[0138] If all k maximum likelihood estimates of the neighboring brightness values ​​are lower than x i Brightness value B right,i , then we can assume that x i With a peak brightness, the firefly x i For the root firefly.

[0139] 2) Definition 2:

[0140] If around the firefly x i Among the neighbors of the k maximum likelihood estimates, there is a brightness value B right,j Greater than x i Brightness value B right,i Firefly x j , then firefly xi and Firefly x j into the same cluster, and firefly x j It's the middle firefly.

[0141] 3) Definition 3:

[0142] If Firefly x i An intermediate firefly or root firefly that does not become any other firefly is considered a leaf firefly.

[0143] like Figure 2 As shown, based on the above definition, the process of generating a firefly tree is as follows:

[0144] 1) Sort the fireflies in descending order according to their brightness values, traverse each firefly, and iterate the neighbors of the maximum likelihood estimate of each firefly. For example, when iterating the i-th firefly, it is recorded as firefly i ;

[0145] 2) For each neighbor, compare its brightness value with the brightness value of the current firefly. If the brightness value of the neighboring firefly is greater than that of the current firefly, the neighboring firefly becomes the middle firefly of the current firefly, the cluster label of the current firefly is the same as that of the neighboring firefly, and the current firefly is added to the corresponding firefly tree;

[0146] 3) If the brightness values ​​of all neighboring fireflies are smaller than that of the current firefly, it means that the current firefly is the root firefly. At this time, the root firefly obtains a new cluster label and generates a new firefly tree, which can be regarded as the local cluster center;

[0147] 4) When traversing other fireflies subsequently, the root firefly in step 3) is likely to become the intermediate firefly of other fireflies and assign its cluster label to other fireflies, thereby achieving local clustering. This step provides the basis for the preliminary clustering of the entire algorithm.

[0148] Among them, merging firefly trees into firefly forests includes:

[0149] Through the firefly tree generation module, preliminary clusters and firefly trees can be obtained. In the firefly tree merge to firefly forest module, the firefly trees will be merged to obtain the final clusters. The process of merging firefly trees into firefly forests is as follows:

[0150] 1) Confirm the leaf nodes of each cluster and guide the fireflies to traverse each cluster. Since the data points of each cluster are stored in the firefly tree, the corresponding leaf nodes can be obtained through the tree structure. The construction of the firefly tree starts from the center of the cluster and propagates to the data points with lower brightness. Therefore, the farther away from the root node of the firefly tree, the lower the brightness of the corresponding node.

[0151] 2) Traverse the nearest neighbors of each leaf node and determine whether the nearest neighbor's label is the same as its own label, and find the neighbors with different labels.

[0152] 3) Calculate the brightness difference between the leaf firefly and the fireflies in different clusters, as shown in formula (17):

[0153]

[0154] Where: BG represents the brightness difference between the leaf firefly and the fireflies in different clusters; B right,NC Indicates the brightness value of the adjacent cluster; B right,leaf represents the brightness value of the leaf node; NC represents the adjacent cluster.

[0155] In this module, first traverse the root firefly to find the corresponding leaf firefly, then traverse each leaf firefly. For each traversed leaf firefly, it is necessary to traverse its k nearest neighbors to determine whether the label of the neighboring firefly is the same as its own label. If they are the same, no operation is performed. If they are different, the brightness difference is calculated and updated until all traversals are completed.

[0156] S2. A complex affine model of line impedance based on the uncertainty of submarine cable line impedance is established using affine mathematical techniques, and the complex affine model of line impedance is optimized based on polynomial fitting technology.

[0157] In order to facilitate understanding of the above technical solution of the present invention, the following is a detailed description of the use of affine mathematical technology in the actual process of the present invention to establish a line impedance complex affine model based on the uncertainty of the submarine cable line impedance, and to optimize the line impedance complex affine model based on the polynomial fitting technology:

[0158] Affine mathematics is used to establish a complex affine model of line impedance based on the uncertainty of submarine cable line impedance:

[0159] Assume that the line resistance and reactance have fluctuation ranges in, R , They represent the lower and upper bounds of line resistance fluctuations, respectively. X , They represent the lower and upper bounds of line reactance fluctuations, respectively. The line impedance complex affine model is described as follows:

[0160]

[0161] R i-j =R0+R1δ r,1 ; (19)

[0162] X i-j =X0+X1δ x,1; (20)

[0163]

[0164] Where: R0 and X0 represent the theoretical values ​​of line resistance and reactance respectively, δ r,1 and δ x,1 They are used to measure the fluctuation of line resistance and reactance respectively, and their value range is [-1, 1]; R1 and X1 are used to indicate the fluctuation amplitude of uncertain variables; R i-j and X i-j Respectively represent the line resistance and reactance values ​​from node i to node j (the nodes here represent the nodes in the power flow calculation in the power system); Represents the complex affine model of the line impedance from node i to node j; represents the affine form of the parameters.

[0165] Line impedance usually fluctuates by 25-30%, so the upper and lower limits of line impedance are:

[0166]

[0167] In summary, the complex affine model of the line impedance from node i to node j can be obtained:

[0168]

[0169] Assume there are two correlated variables α and β, whose affine forms are as follows:

[0170] α=4+3θ1; (25)

[0171] β=2+θ1; (26)

[0172] Where: θ1 represents the noise element, and its value range is [-1, 1].

[0173] Affine numbers are essentially linear functions of noise elements. The quotient of affine numbers does not have the linear characteristics of affine numbers, so it is not an affine form and cannot participate in subsequent affine operations. The result of affine number division is converted into a polynomial function with only existing noise elements by using curve polynomial fitting. Setting the number of terms to 9 gives:

[0174]

[0175] Where: α, β represent the affine forms of two related variables respectively; p represents the coefficient of the i-th term.

[0176] S3. Use convergence judgment technology to perform power flow calculation on the optimized line impedance complex affine model, and establish a submarine cable transmission efficiency model with line parameter uncertainty based on the power flow calculation results.

[0177] Among them, the convergence judgment technology is used to calculate the power flow of the optimized line impedance complex affine model, and the submarine cable transmission efficiency model with line parameter uncertainty is established based on the power flow calculation results, including:

[0178] S31, based on the known load and initial state of node j, obtain the current between node i and node j, and calculate the injected power of each node according to the current and line impedance complex affine model;

[0179] S32, calculating the current and voltage drop between each node and other nodes based on the injected power of each node;

[0180] S33, iteratively calculating the voltage difference of each node voltage relative to the voltage of the previous iteration, using the affine number fluctuation domain overlap area comparison convergence judgment technology to judge the similarity of the two iteration results, and obtaining the power flow calculation result of the line impedance complex affine model;

[0181] S34. According to the power flow calculation results of the line impedance complex affine model, the relationship between the energy storage output and the voltage at the head end of the submarine cable is analyzed, and based on the relationship between the energy storage output and the voltage at the head end of the submarine cable, a submarine cable transmission efficiency model with line parameter uncertainty is established.

[0182] In order to facilitate understanding of the above technical solution of the present invention, the present invention uses the convergence judgment technology in the actual process to calculate the power flow of the optimized line impedance complex affine model, and establishes a submarine cable transmission efficiency model with line parameter uncertainty based on the power flow calculation results.

[0183] The power flow calculation of the line impedance complex affine model based on the traditional forward and reverse power flow calculation mainly includes the following three steps:

[0184] 1) Back generation:

[0185] Use the affine load S of node j j,l and initial state, and assume that it is equal to the rated voltage U j ;, get the current between nodes i and j:

[0186]

[0187] According to the composite affine model of current and line impedance, the power loss of the segment and the injected power at node i are calculated as follows:

[0188]

[0189] Where: S j,l represents the affine load of node j; S i,l represents the affine load of node i; U j Indicates rated voltage; R i-j and Xi-j Respectively represent the line resistance and reactance values ​​from node i to node j; Represents the complex affine model of the line impedance from node i to node j; represents the current between nodes i and j.

[0190] 2) Push forward:

[0191] Using the injected power and the first-segment voltage of each node obtained by the back-substitution process, the current and voltage drop between each node and other nodes are calculated. The calculation process is as follows:

[0192]

[0193] U j =U i -ΔU i-j ; (32)

[0194] Where: and U i-j denote the current and voltage between node i and node j respectively, represents the load of node i; U j represents the voltage at node j; U i represents the voltage of node i; represents the complex power of node i.

[0195] 3) Convergence judgment:

[0196] The iteration termination criterion of the traditional power flow algorithm is that the numerical deviation of each node voltage relative to the previous iteration is less than the allowable value. However, for the line loss calculation considering the uncertainty of line impedance, the voltage difference between the previous and next iterations is:

[0197]

[0198] Where: All represent complex affine numbers; Represents the reference term of voltage during iteration; represents the additional term of voltage during iteration; Indicates the voltage correction factor.

[0199] It should be noted that the voltage difference is not only a complex number, but also an affine number with a fluctuation range, and cannot be simply judged based on the amplitude of the voltage difference before and after the iteration. In order to solve the defect that the voltage complex affine number cannot use the traditional comparison of complex modulus, the present invention proposes a complex affine number fluctuation domain overlap area comparison convergence judgment method.

[0200] Assume that the rectangular areas S and S' are complex affine numbers and The area of ​​the overlapping part represents the convergence of the iterative process. The larger the overlapping area, the better the convergence. The idea of ​​comparing the convergence of the overlapping area of ​​the complex affine number fluctuation domain is to determine whether the ratio of the area of ​​the overlapping area S to the sum of the areas of areas S1 and S2 is less than the set threshold δ. The calculation formula is as follows:

[0201] S / (S1+S2)<δ; (34)

[0202] By comparing the overlapping areas, the similarity of the two iteration results can be obtained, thereby determining whether the iteration termination condition has been met.

[0203] Derivation of the submarine cable transmission efficiency formula based on line parameter uncertainty:

[0204] The present invention stipulates that the location of the offshore wind storage is the initial section of the submarine cable, and the grid-connected side is the end of the submarine cable. The voltage at the head end of the submarine cable can be adjusted by energy storage. The relationship between the energy storage output and the voltage at the head end of the submarine cable is as follows:

[0205]

[0206] Where: U1 represents the voltage at the head end of the submarine cable; P1 and Q1 represent the active power and reactive power of the first section of the submarine cable respectively; Y 1-j It represents the admittance between the head end of the submarine cable and other nodes; U j represents the voltage of node j; P BESS,t Represents the output power of energy storage at time t; P ref and Q ref P represents the power at the head end of the submarine cable without wind power and energy storage and without taking into account transmission loss; WT,t and Q WT,t They represent the active and reactive power output of offshore wind power at time t respectively; P bg Indicates the active power at the head end of the submarine cable; Q bg Indicates the reactive power at the head end of the submarine cable.

[0207] Assume that the voltage at the grid-connected end of the submarine cable is the reference voltage, that is:

[0208]

[0209] Assume that the voltage relationship between the first end of the submarine cable and the grid-connected end satisfies:

[0210]

[0211] Where: represents the ratio of the voltage amplitude between the head end of the submarine cable and the grid-connected end; θ represents the voltage phase difference between the head end of the submarine cable and the grid-connected end; τ represents the discount rate; Indicates the voltage at the grid-connected end of the submarine cable; Indicates the voltage at the beginning of the submarine cable.

[0212] The affine model of submarine cable impedance and admittance based on considering the uncertainty of line parameters is as follows:

[0213] Z=R+jωL; (39)

[0214] Y=G+jωC; (40)

[0215] Where: represents the affine form of the parameter; Z and Y represent the impedance and admittance of the submarine cable respectively; R and G represent the resistance and conductance of the submarine cable respectively; L and C represent the inductance and capacitance of the submarine cable respectively; ω represents the angular frequency; j represents a complex number.

[0216] The affine model of the characteristic impedance and propagation coefficient of the submarine cable is obtained as follows:

[0217]

[0218] Where: Z c represents the characteristic impedance of the submarine cable; γ represents the propagation coefficient of the submarine cable.

[0219] The voltage and current relationship between the two ends of the equivalent circuit of a long line considering the uncertainty of line parameters is:

[0220]

[0221] Thus, the Y parameter matrix of the submarine cable considering the uncertainty of line parameters can be constructed:

[0222]

[0223] Where: and Respectively represent the current at the beginning and end of the submarine cable; and Represent the voltage at the beginning and end of the submarine cable respectively.

[0224] This gives the admittance matrix Y hl :

[0225]

[0226] Where: Y hl represents the admittance matrix; Y L and Y R They represent the parameters of the admittance matrix respectively.

[0227] Based on the admittance matrix considering the uncertainty of line parameters, the calculation model of submarine cable complex affine transmission loss is obtained:

[0228]

[0229] Where: P lossRepresents the transmission loss of the submarine cable; P bg and P grid Represent the power at the head end and the end end (i.e., the grid-connected end) of the submarine cable respectively; * represents conjugation; Re represents the real part.

[0230] Based on the above affine transmission loss calculation model, the calculation formula for submarine cable transmission efficiency is obtained:

[0231]

[0232] Where: η hl Indicates the transmission efficiency of the submarine cable.

[0233] Based on the voltage relationship between the two ends of the submarine cable considering the energy storage output, we can get:

[0234]

[0235] In summary, the formula for submarine cable transmission efficiency considering the uncertainty of line parameters can be obtained:

[0236]

[0237] S4. Based on the annual operation scenario of offshore wind power, a two-stage optimization configuration model is established and solved to realize offshore wind storage configuration.

[0238] Among them, based on the annual operation scenario of offshore wind power, a two-stage optimization configuration model is established, and the two-stage optimization configuration model is solved to achieve offshore wind storage configuration including:

[0239] S41. Taking the lowest annual comprehensive cost of offshore wind storage, the smallest net load fluctuation and the smallest voltage fluctuation as the optimization goals, an upper-level capacity configuration model for offshore wind storage is constructed.

[0240] S42. Taking the upper capacity configuration model as input and minimizing the daily loss of submarine cable transmission lines as the optimization goal, a lower optimization operation model of offshore wind storage is constructed.

[0241] S43. An improved geyser algorithm is used to solve the upper capacity configuration model, and a solver is used to solve the lower optimization operation model.

[0242] Among them, the improved geyser algorithm is used to solve the upper capacity configuration model including:

[0243] Initialize the parent population P op , calculate the fitness function of the upper capacity configuration model, and ensure that the fitness function of the upper capacity configuration model satisfies the constraint conditions;

[0244] Generate offspring population R using the eruption mechanism T , the parent population Pop and the offspring population R T Merge to form population S T , and based on the cooling process, the population S T Sort and evaluate the population S T The quality of each solution in

[0245] Based on the advantage level and reference distance of each solution, from the population S T Select the best solution and generate the next generation parent population P op,T+1 , repeat the eruption mechanism and cooling process, and continue to generate a new generation of parent populations until the convergence condition is met;

[0246] The final parent population P op As the Pareto optimal solution of the upper capacity configuration model, the final solution of the upper capacity configuration model is selected by using the grey target decision based on the entropy weight method.

[0247] In order to facilitate understanding of the above technical solutions of the present invention, the following is a two-stage optimization configuration model based on the annual operation scenario of offshore wind power in the actual process of the present invention (the model is a two-stage optimization configuration model based on the uncertainty of submarine cable line impedance and submarine cable transmission efficiency), and the two-stage optimization configuration model is solved to achieve offshore wind storage configuration. A detailed description is given below:

[0248] In order to achieve the best economy of offshore wind storage, system stability and the lowest submarine cable transmission loss, the present invention constructs a two-layer optimization configuration model for offshore wind storage capacity configuration. The upper capacity configuration model takes the lowest annual comprehensive cost of offshore wind storage, the smallest net load fluctuation and the smallest voltage fluctuation as the optimization goals, determines the configuration capacity of offshore wind storage, and serves as the input of the second-stage optimization operation model. The lower optimization operation model takes minimizing the daily loss of submarine cable transmission lines as the optimization goal:

[0249] minF1=C inv +C OM -B FLC -B loss -B cur ; (51)

[0250] Upper capacity configuration model for offshore wind storage:

[0251] minF1=C inv,WT +C inv,BESS +C OM,WT +C OM,BESS -I I-P -I WT ; (52)

[0252] Where: C inv,WT and C inv,BESSRespectively represent the investment costs of wind power and energy storage; C OM,WT and C OM,BESS Represent the operation and maintenance costs of wind power and energy storage respectively; I I-P Represents the benefits of energy storage; I WT Represents the revenue from wind power sales.

[0253] The expression of investment cost of energy storage is:

[0254]

[0255] Where: τ represents the discount rate; y BESS Indicates the life of energy storage; F P and F E Respectively represent the unit power and unit capacity cost of energy storage; P BAT and E BAT They represent the configured power and configured capacity of the energy storage respectively.

[0256] The investment cost of wind power is expressed as:

[0257]

[0258] Where: y WT Indicates the investment period of wind power; F WT represents the unit investment cost of wind power; P WT Represents the configured capacity of wind power.

[0259] The expression of the operation and maintenance cost of energy storage is:

[0260]

[0261] Where: OM,P and ρ OM,E Represent the operation and maintenance coefficient of unit power and unit capacity respectively; P BESS and E BESS They represent the configured power and configured capacity of the energy storage respectively.

[0262] The investment cost of wind power is expressed as:

[0263]

[0264] Where: OM,WT Represents the operation and maintenance coefficient of wind power.

[0265] The expression of energy storage benefits is:

[0266]

[0267] Where: sell,t and ω pur,t Respectively represent the electricity sales and purchase prices at time t; Kd represents the number of annual operation scenarios of offshore wind power; T k represents the number of days corresponding to the kth scenario; P dis,t and P cha,t They represent the discharge power and charging power of the energy storage at time t, respectively, where:

[0268]

[0269] Where: WT,t P represents the electricity price of offshore wind power at time t; WT,t Represents the configured capacity of wind power at time t.

[0270]

[0271] Where: L(t) represents the net load at time t; L(t-1) represents the net load at time t-1.

[0272]

[0273] Where: minF1, minF2 and minF3 represent the minimum optimization objectives of the upper capacity configuration model, which are annual comprehensive cost, net load fluctuation and voltage fluctuation respectively; U i,j,t represents the voltage per unit value of the ith node on the kth typical day; represents the voltage mean of the ith node on the kth typical day; N nodes Indicates the number of nodes.

[0274] Constraints:

[0275] Node voltage constraints:

[0276] U i,min ≤U i,t ≤U i,max ; (61)

[0277] Where: U i,max and U i,min are the upper and lower limits of the voltage at node i; U i,t represents the voltage of node i at time t.

[0278] Power and capacity constraints of energy storage:

[0279]

[0280] Where: E BESS,max 、E BESS,min , P BESS,max , P BESS,min They represent the upper and lower limits of energy storage configuration capacity and the upper and lower limits of configuration power respectively.

[0281] Node power balance constraints:

[0282]

[0283] Where: U j,t represents the voltage of node j at time t; θ ij,t represents the voltage phase difference between node i and node j at time t; G ij and B ij represents the equivalent conductance and susceptance of the line between node i and node j; P i (t) represents the active power of node i at time t; Q i (t) represents the reactive power of node i at time t.

[0284] Offshore wind power resource endowment constraints:

[0285] 0≤P WT ≤P WT,max ; (64)

[0286] Where: P WT,max Represents the upper limit of physical resources for offshore wind power.

[0287] Lower-level optimization operation model:

[0288]

[0289] Where: P bg,t represents the active power injected into the head end of the submarine cable at time t; η hl,t K represents the transmission efficiency of the submarine cable considering the uncertainty of line parameters at time t; d Represents the annual number of operating scenarios for offshore wind power.

[0290] Constraints:

[0291] Energy storage charging and discharging constraints:

[0292]

[0293] Where: γ cha and γ dis Respectively represent the charging efficiency and discharging efficiency of energy storage; P BESS Represents the configuration power of energy storage; P cha,t and P dis,t They represent the charging power and discharging power of the energy storage at time t respectively.

[0294] SOC constraints:

[0295] SOC min ≤SOC i (t)≤SOC max ; (67)

[0296] Where: SOC max and SOC min Respectively represent the upper and lower limits of the SOC of energy storage; SOC i (t) represents the SOC of energy storage at time t.

[0297] SOC of energy storage at time t:

[0298]

[0299] Where: Δt represents the charge and discharge time; SOC i (t-1) represents the SOC at the energy storage moment t-1.

[0300] Solve the two-stage optimization configuration model:

[0301] The improved Multi-Objective Geyser Algorithm (MOGA) is used to solve the upper model (upper capacity configuration model), and the solver is used to solve the lower model (lower optimization operation model), which specifically includes:

[0302] Input the original data of each node’s load and branch parameters, offshore wind power parameters, etc.

[0303] Set parameters and variable constraints of MOGA;

[0304] Initialize a parent population P of size m op , calculate the fitness functions of equations (52), (59), and (60), and ensure that the constraints of equations (61)-(64) are satisfied;

[0305] The eruption mechanism is used to generate the offspring population R T ;

[0306] The population P op and R T Merge into a population S of size 2m T , and sort them using a cooling process;

[0307] According to the dominance level and reference distance, the best m solutions are selected from the 2m solutions to generate the next generation parent population P op,T+1 ;

[0308] Continue the eruption and cooling process until the convergence condition is met and return to the final parent population P op As the Pareto solution of the upper model, the final solution is selected by grey target decision-making based on entropy weight method.

[0309] Inputting the results of the upper model into the lower model and using the solver to solve the lower model includes:

[0310] The process of passing the results of the upper model as input to the lower model usually refers to the problem of hierarchical optimization. The solver of the lower model (such as CPLEX solver, Gurobi solver, etc.) is a mathematical solution tool used to solve this hierarchical optimization problem. The solver is used to solve the lower model, and the results of the lower model are fed back to the upper model. The upper model is modified according to the results of the lower model, and it is iterated repeatedly until convergence.

[0311] The present invention clusters the annual operation scenarios of offshore wind power based on the firefly forest clustering algorithm of the Pearson correlation coefficient; in view of the uncertainty of the impedance of the submarine cable line and to ensure that the calculation result does not fluctuate too much, a submarine cable transmission loss model considering the uncertainty of submarine cable parameters is proposed based on the improved affine division method; a two-stage optimization configuration model for offshore wind storage is established. In the first stage, the economy of energy storage, the net load fluctuation of the system and the voltage fluctuation are comprehensively considered; in the second stage, the transmission loss of the submarine cable is considered, and the improved geyser algorithm is used to solve the model.

[0312] The firefly forest algorithm based on the Pearson correlation coefficient proposed in the present invention solves the problems of the need to set a fixed number of neighbors, predefine the number of clusters, and rely on a computationally intensive group iteration process in traditional clustering technology, thereby improving the accuracy and robustness of the clustering results, eliminating the dependence on the predefined number of clusters, and reducing the computational complexity; it not only ensures the economy of offshore wind storage and the power quality of the system, but also considers the uncertainty of line parameters to make the calculation of submarine cable transmission efficiency more accurate. At the same time, it takes the transmission loss of the submarine cable as the objective function of the lower-level model, which can ensure the economy of offshore wind storage and the stability of the system while taking into account the transmission loss of the submarine cable, and further ensures the economy of the planning scheme.

[0313] In summary, with the help of the above technical scheme of the present invention, the firefly forest clustering algorithm based on the Pearson correlation coefficient of the present invention realizes dynamic adjustment through adaptive neighborhood estimation, and the brightness of each data point is calculated according to its distribution characteristics, without the need to manually set the number of neighbors, which significantly simplifies parameter tuning and improves flexibility. It shows excellent scalability when processing large-scale data sets, and filters outliers through the interquartile range method to ensure the stability of brightness calculation and the accuracy of clustering results, thereby having significant advantages in processing complex data distribution and large-scale data, and is particularly suitable for practical application scenarios requiring high efficiency and high precision. At the same time, the present invention applies the improved firefly forest clustering algorithm to scene clustering, which has higher robustness and stability; the present invention adopts affine mathematics to model the uncertainty of line impedance to obtain a complex affine model of line impedance based on the uncertainty of submarine cable line impedance, and improves the traditional affine division by using the method of polynomial fitting, so that the conservatism of the original affine number division is overcome with a slight loss of authenticity in subsequent affine operations, and then solves the defect that the complex affine number of voltage cannot use the traditional comparison complex module through the power flow calculation and discrimination method of the complex affine model of line impedance; the present invention derives the transmission efficiency formula of the submarine cable line considering the uncertainty of line parameters, which has better conservatism than the traditional derivation method that assumes that the line impedance is fixed, and thus more accurately characterizes the transmission efficiency of the submarine cable line under actual conditions.

[0314] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for configuring offshore wind storage considering the uncertainty of wind power output and the transmission efficiency of submarine cables, characterized in that: The offshore wind storage configuration method includes: S1. Obtain an offshore wind power output data set, and cluster the offshore wind power output data set using the firefly forest clustering algorithm with the Pearson correlation coefficient to obtain the annual operation scenario of offshore wind power; S2. Establish a complex affine model of line impedance based on the uncertainty of submarine cable line impedance using affine mathematical techniques, and optimize the complex affine model of line impedance based on polynomial fitting techniques; S3. Use convergence judgment technology to perform power flow calculation on the optimized line impedance complex affine model, and establish a submarine cable transmission efficiency model with line parameter uncertainty based on the power flow calculation results; S4. Based on the annual operation scenario of offshore wind power, a two-stage optimization configuration model is established and solved to realize offshore wind storage configuration.

2. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 1 is characterized in that: The offshore wind power output data set is obtained, and the firefly forest clustering algorithm of the Pearson correlation coefficient is used to cluster the offshore wind power output data set to obtain the annual operation scenarios of offshore wind power, including: S11, obtaining an offshore wind power output data set containing a number of independently distributed data points, and using the Pearson coefficient to characterize the dissimilarity between the data points; S12, based on the dissimilarity between the data points, iterate each data point in the offshore wind power output data set, and perform maximum likelihood estimation on the number of neighbors of the data point in each iteration process to obtain the likelihood ratio of the comprehensive model; S13, judging whether the likelihood ratio of the comprehensive model meets the preset conditions, if so, updating the number of neighbors of the current maximum likelihood estimate and executing step S14, otherwise, repeating the iterative process of the data points until the likelihood ratio of the comprehensive model meets the preset conditions; S14. The data points are regarded as fireflies with adaptive brightness, and the fireflies are clustered to form a tree structure. The tree structure is merged to generate a firefly forest, and the annual operation scenario of offshore wind power is obtained.

3. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 2 is characterized in that: The data points are used as fireflies with adaptive brightness, and the fireflies are clustered to form a tree structure. The tree structures are merged to generate a firefly forest, and the annual operation scenarios of offshore wind power are obtained, including: S141, taking the data points as fireflies, initializing firefly parameters based on the distribution parameters of a predefined offshore wind power output data set, and calculating the firefly brightness according to the firefly parameters; S142, arranging the fireflies in descending order based on their brightness, each firefly selects a brighter firefly closest to it as a guide, obtaining a preliminary clustering result, and using a tree structure to store the preliminary clustering result to form a firefly tree; S143. Based on the luminous interaction signals of the firefly population, the firefly trees are merged into a firefly forest to obtain the final clustering result, and the final clustering result is used as the annual operation scenario of offshore wind power.

4. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 3 is characterized in that: The method of using the data points as fireflies, initializing firefly parameters based on the distribution parameters of the predefined offshore wind power output data set, and calculating the firefly brightness according to the firefly parameters includes: S1411, treating the data points as fireflies with adaptive brightness, and initializing firefly parameters according to the distribution parameters of the offshore wind power data set, and the firefly parameters include brightness and the number of neighbors of maximum likelihood estimation; S1412, calculating the interquartile range of the number of neighbors of the maximum likelihood estimate using the interquartile range calculation formula, and calculating the upper bound and the lower bound of the number of neighbors based on the interquartile range of the number of neighbors; S1413, performing an interquartile range test on the number of neighbors of the maximum likelihood estimate based on the upper and lower bounds of the number of neighbors, identifying the proximity of outliers in the number of neighbors of the maximum likelihood estimate, and obtaining the error of the maximum likelihood estimate density; S1414. Based on the error of the maximum likelihood estimation density, the firefly brightness is calculated using the calculation formula of the firefly brightness.

5. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 4 is characterized in that: The method of arranging the fireflies in descending order based on their brightness, each firefly selects a brighter firefly closest to it as a guide, obtaining a preliminary clustering result, and using a tree structure to store the preliminary clustering result to form a firefly tree includes: S1421, sorting the fireflies in descending order according to their brightness values, iterating each firefly and the neighboring maximum likelihood estimate of each firefly; S1422, comparing the brightness value of the neighboring fireflies with the brightness value of the current firefly. If the brightness value of the neighboring firefly is greater than that of the current firefly, the neighboring firefly becomes the middle firefly of the current firefly, and the current firefly is added to the corresponding firefly tree; S1423. If the brightness values ​​of all neighboring fireflies are smaller than that of the current firefly, it indicates that the current firefly is a root firefly, and the root firefly obtains a new cluster label, and generates a new firefly tree, and uses the new firefly tree as the local cluster center; S1424, traverse the remaining fireflies, if the root firefly becomes the middle firefly of the remaining fireflies, assign its cluster label to the remaining fireflies to obtain a preliminary clustering result, and use a tree structure to store the preliminary clustering result to form a firefly tree.

6. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 5, characterized in that: The expression of the line impedance complex affine model based on the uncertainty of submarine cable line impedance is: In the formula, Represents the complex affine model of the line impedance from node i to node j; ^ Represents the affine form of the parameters; R i-j and X i-j Respectively represent the line resistance and reactance values ​​from node i to node j; δ r,1 Indicates a parameter used to measure line resistance fluctuation.

7. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 6, characterized in that: The method of using the convergence judgment technology to perform power flow calculation on the optimized line impedance complex affine model and establishing a submarine cable transmission efficiency model with line parameter uncertainty according to the power flow calculation results includes: S31, based on the known load and initial state of node j, obtain the current between node i and node j, and calculate the injected power of each node according to the current and line impedance complex affine model; S32, calculating the current and voltage drop between each node and other nodes based on the injected power of each node; S33, iteratively calculating the voltage difference of each node voltage relative to the voltage of the previous iteration, using the affine number fluctuation domain overlap area comparison convergence judgment technology to judge the similarity of the two iteration results, and obtaining the power flow calculation result of the line impedance complex affine model; S34. According to the power flow calculation results of the line impedance complex affine model, the relationship between the energy storage output and the voltage at the head end of the submarine cable is analyzed, and based on the relationship between the energy storage output and the voltage at the head end of the submarine cable, a submarine cable transmission efficiency model with line parameter uncertainty is established.

8. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 7, characterized in that: The two-stage optimization configuration model is established based on the annual operation scenario of offshore wind power, and the two-stage optimization configuration model is solved to realize the offshore wind storage configuration, including: S41. Taking the lowest annual comprehensive cost of offshore wind storage, the smallest net load fluctuation and the smallest voltage fluctuation as the optimization objectives, an upper capacity configuration model for offshore wind storage is constructed; S42, taking the upper capacity configuration model as input and taking minimizing the daily loss of the submarine cable transmission line as the optimization goal, constructing the lower optimization operation model of the offshore wind storage; S43. An improved geyser algorithm is used to solve the upper capacity configuration model, and a solver is used to solve the lower optimization operation model.

9. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 8, characterized in that: The method of solving the upper layer capacity configuration model by using the improved geyser algorithm includes: Initialize the parent population P op , calculate the fitness function of the upper capacity configuration model, and ensure that the fitness function of the upper capacity configuration model satisfies the constraint conditions; Generate offspring population R using the eruption mechanism T , the parent population P op and the offspring population R T Merge to form population S T , and based on the cooling process, the population S T Sort and evaluate the population S T The quality of each solution in Based on the advantage level and reference distance of each solution, from the population S T Select the best solution and generate the next generation parent population P op,T+1 , repeat the eruption mechanism and cooling process, and continue to generate a new generation of parent populations until the convergence condition is met; The final parent population P op As the Pareto optimal solution of the upper capacity configuration model, the final solution of the upper capacity configuration model is selected by using the grey target decision based on the entropy weight method.

10. The offshore wind storage configuration method considering wind power output uncertainty and submarine cable transmission efficiency according to claim 9, characterized in that: The expression of the lower layer optimization operation model is: Where P bg,t represents the active power injected into the head end of the submarine cable at time t; ηhl,t represents the transmission efficiency of the submarine cable considering the uncertainty of line parameters at time t; K d represents the number of annual operation scenarios of offshore wind power; T k Indicates the number of days corresponding to the kth scenario.

Images