An energy storage power station consistency evaluation method, system, device and medium based on an improved MK algorithm

Abstract

The application relates to an energy storage power station consistency evaluation method, system, equipment and medium based on an improved MK algorithm, and comprises the following steps: based on a pre-established energy storage power station consistency evaluation index system, obtaining original observation data of each index, and forming a standardized data matrix and a fuzzy judgment matrix; based on the standardized data matrix and the fuzzy judgment matrix, a combination weight vector is obtained through a combination weight assignment method of fuzzy analytic hierarchy process-gray correlation degree-game theory; based on the improved MK algorithm, clustering analysis is performed on a data set of an energy storage power station to be evaluated, and the consistency of the energy storage power station is evaluated in combination with the combination weight vector, so that a consistency evaluation result is obtained. The application can be widely applied to the technical field of energy storage power station safety risk evaluation.

Description

Technical Field

[0001] This invention belongs to the field of energy storage power station safety risk assessment technology, specifically involving a consistency evaluation method, system, equipment and medium for energy storage power stations based on an improved MK algorithm. Background Technology

[0002] The evaluation index system for large-scale energy storage power stations is complex, and the massive amount of consistent data generated by each energy storage power station poses a great challenge to the overall evaluation work.

[0003] Current mainstream evaluation methods for energy storage power stations mostly employ the analytic hierarchy process (AHP) and entropy weight method to determine the weights of indicators. However, the AHP relies on subjective judgment, making the evaluation process susceptible to interference from subjective factors and ambiguity. The entropy weight method, on the other hand, struggles to handle situations where indicators are highly correlated, often leading to improper weight allocation.

[0004] There is an urgent need for a comprehensive and accurate method to evaluate large-scale energy storage power stations. Summary of the Invention

[0005] To address the aforementioned issues, the present invention aims to provide a method, system, equipment, and medium for evaluating the consistency of energy storage power stations based on an improved MK algorithm. This method utilizes a combined weighting method of fuzzy hierarchical analysis, grey relational analysis, and game theory, along with the improved MK algorithm, to evaluate the consistency of energy storage power stations. This results in more accurate and reasonable evaluations, while also improving the efficiency of consistency evaluation. Furthermore, it allows for fault diagnosis of energy storage power stations based on clustering results, thereby enhancing the safety and stability of energy storage power station operation.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a consistency evaluation method for energy storage power stations based on an improved MK algorithm, comprising: acquiring the original observation data of each indicator based on a pre-established consistency evaluation index system for energy storage power stations, and forming a standardized data matrix and a fuzzy judgment matrix; obtaining a combined weight vector based on the standardized data matrix and the fuzzy judgment matrix through a combined weighting method of fuzzy hierarchical analysis-grey relational degree-game theory; performing cluster analysis on the dataset of the energy storage power station to be evaluated based on the improved MK algorithm, and evaluating the consistency of the energy storage power station in combination with the combined weight vector to obtain the consistency evaluation result.

[0007] Furthermore, the process of acquiring the original observation data of each indicator based on the pre-established consistency evaluation index system for energy storage power stations and forming a standardized data matrix and a fuzzy judgment matrix includes: establishing a consistency evaluation index system for energy storage power stations; the consistency evaluation index system includes four categories of indicators: battery voltage, battery internal resistance, battery SOC, and battery temperature, and each category of indicators includes three evaluation indicators: standard deviation, coefficient of variation, and range; acquiring the original observation data and expert judgment information of each indicator and forming a standardized data matrix and a fuzzy judgment matrix.

[0008] Furthermore, the method of obtaining a combined weight vector based on a standardized data matrix and a fuzzy judgment matrix through a combined weighting method of fuzzy hierarchical analysis, grey relational analysis, and game theory includes: calculating a subjective weight vector based on the established fuzzy judgment matrix using fuzzy hierarchical analysis; calculating an objective weight vector based on the established standardized data matrix using grey relational analysis; and fusing the obtained subjective and objective weight vectors using game theory to obtain the final combined weight vector.

[0009] Furthermore, the step of calculating the subjective weight vector based on the established fuzzy judgment matrix and the fuzzy hierarchical analysis method includes: performing a consistency check on the fuzzy judgment matrix; if the consistency check passes, proceed to the next step; otherwise, regenerate the fuzzy judgment matrix; synthesize the fuzzy judgment matrix after consistency verification using fuzzy arithmetic rules, and calculate the fuzzy weights of each indicator; after defuzzifying and normalizing the fuzzy weights of each indicator, obtain the subjective weight vector composed of the subjective weights of each indicator.

[0010] Furthermore, the objective weight vector calculated based on the established standardized data matrix and the grey relational analysis method includes: processing the established standardized data matrix to obtain the corresponding ideal scheme reference sequence; calculating the correlation coefficient of each indicator based on the ideal scheme reference sequence and the grey relational analysis theory; calculating the average correlation coefficient based on the correlation coefficient of each indicator, and normalizing it to obtain the objective weight vector composed of the objective weights of each indicator.

[0011] Furthermore, the process of fusing the obtained subjective weight vector and objective weight vector based on game theory to obtain the final combined weight vector includes: constructing an objective function and determining constraints with the goal of minimizing the sum of squared Euclidean distances between the combined weight vector and the subjective and objective weight vectors; constructing a Lagrangian function to solve the constructed objective function to obtain the initial optimal combined weight vector; and normalizing the initial optimal combined weight vector to obtain the final combined weight vector.

[0012] Furthermore, the improved MK algorithm is used to perform cluster analysis on the dataset of the energy storage power stations to be evaluated, and the consistency of the energy storage power stations is evaluated by combining the combined weight vector, resulting in a consistency evaluation result, including: ① Initialize using the K-Means++ method based on Manhattan distance to select the initial set of centroids C; ② Obtain the dataset, number of clusters K, and initial set of centroids C corresponding to the energy storage power station to be evaluated; ③ Sort each column of the dataset in ascending order; ④ Based on the initial set of centroids C, the Annoy algorithm is used to assign each non-centroid in the dataset to the cluster containing the nearest centroid; ⑤ Sort each column of the dataset in ascending order, and update the center point based on the sorting result; ⑥ Repeat steps ④ to ⑤ until the centroids no longer change, and output the initial set of centroids C and the associated cluster set P; ⑦ Based on the initial set of centroids C and the associated cluster set P, the VC algorithm is used to split centroids and merge cluster pairs to obtain the optimized set of centroids C' and the associated cluster set P'. ⑧ Based on the cluster set P' and the obtained combined weight vector, the evaluation level classification work is carried out to obtain the consistency evaluation results of the energy storage power station.

[0013] Secondly, this invention provides a consistency evaluation system for energy storage power stations based on an improved MK algorithm, comprising: an indicator data acquisition module, used to acquire the original observation data of each indicator based on a pre-established consistency evaluation indicator system for energy storage power stations, and form a standardized data matrix and a fuzzy judgment matrix; a combined weight calculation module, used to obtain a combined weight vector based on the standardized data matrix and the fuzzy judgment matrix through a combined weighting method of fuzzy hierarchical analysis-grey relational degree-game theory; and a consistency evaluation module, used to perform cluster analysis on the dataset of the energy storage power station to be evaluated based on the improved MK algorithm, and evaluate the consistency of the energy storage power station in combination with the combined weight vector to obtain the consistency evaluation result.

[0014] Thirdly, the present invention provides a computer-readable storage medium for storing one or more programs, said one or more programs including instructions that, when executed by a computing device, cause the computing device to perform any method.

[0015] Fourthly, the present invention provides a computing device comprising: one or more processors and a memory, wherein the memory stores one or more programs and is configured to be executed by the one or more processors, the one or more programs including instructions for performing any method.

[0016] The present invention has the following advantages due to the adoption of the above technical solutions: 1. This invention constructs a universal consistency evaluation index system for energy storage power stations, and uses a combined weighting method based on fuzzy hierarchical analysis, grey relational analysis, and game theory to calculate the weights of evaluation indicators by combining subjective and objective methods, making the setting of indicator weights more reasonable. 2. Based on the combined weighting, this invention uses an improved MK algorithm to perform cluster analysis on the consistency index data of energy storage power stations. The evaluation level is divided according to the clustering results to achieve the purpose of analyzing consistency data, thereby improving the evaluation efficiency of the consistency of energy storage power stations.

[0017] This invention relies on cluster analysis, comprehensive scoring, and anomaly detection technologies, with core compatibility for the operation and maintenance management of energy storage power stations, while also possessing potential for expansion into multiple fields. In the energy storage field, it can calculate the comprehensive score of cluster centers, classify energy storage power stations and operating periods into performance levels, screen low-level periods for secondary clustering, accurately locate abnormal operating periods, and support power station assessment, scheduling optimization, and fault early warning. In the power system field, it can be used for grid energy storage ancillary service assessment, wind and solar integrated energy storage operation and maintenance, and distribution area monitoring, contributing to grid stability and new energy consumption. In the power battery field, it can achieve battery pack grading and screening and on-board battery fault early warning, promoting the tiered utilization of batteries. In the industrial field, it can conduct enterprise energy efficiency rating and predictive maintenance of industrial equipment. Furthermore, this technology can also be used as a general module, embedded in big data platforms such as environmental monitoring and smart cities, to complete multi-indicator subject rating and anomaly data detection. Attached Figure Description

[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. In the drawings: Figure 1 This is a flowchart of the energy storage power station consistency evaluation method based on the improved MK algorithm provided in this embodiment of the invention; Figure 2 This is the consistency evaluation index system for energy storage power stations provided in the embodiments of the present invention; Figure 3 This is a flowchart of the FV algorithm provided in an embodiment of the present invention; Figure 4 This is a flowchart of the VC algorithm provided in an embodiment of the present invention; Figure 5 This is a flowchart of the MK algorithm provided in an embodiment of the present invention; Figure 6 This is a flowchart of the improved MK algorithm provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0020] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0021] In some embodiments of the present invention, a consistency evaluation method for energy storage power stations based on an improved MK algorithm is provided. To address the problem of unreasonable weight allocation, a combined weighting method based on fuzzy hierarchical analysis, grey relational analysis, and game theory is proposed, which reduces the subjective uncertainty of weight allocation. To address the problem of the large amount of consistency data of energy storage power stations making it difficult to conduct comprehensive evaluation, an improved MK algorithm is proposed. This algorithm reduces the computational time complexity, and using this algorithm for comprehensive evaluation improves the accuracy and efficiency of the evaluation.

[0022] Correspondingly, in other embodiments of the present invention, a consistency evaluation system, device and medium for energy storage power stations based on an improved MK algorithm are provided.

[0023] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0024] Example 1 like Figure 1 As shown in the figure, this embodiment provides a consistency evaluation method for energy storage power stations based on an improved MK algorithm. The method includes the following steps: Step 1: Based on the pre-established consistency evaluation index system for energy storage power stations, obtain the original observation data of each index and form a standardized data matrix and a fuzzy judgment matrix.

[0025] Specifically, it includes the following steps: Step 1.1: Establish a consistency evaluation index system for energy storage power stations.

[0026] like Figure 2 As shown in this embodiment, when establishing the consistency evaluation index system for energy storage battery stacks, four consistency influencing factors are comprehensively considered: battery voltage, battery internal resistance, battery SOC (State of Charge), and battery temperature. For the four consistency influencing factors, three types of evaluation indicators, namely standard deviation, coefficient of variation, and range, are selected respectively.

[0027] ① Voltage consistency: (1) (2) (3) in, For voltage standard deviation, For the first The voltage of each battery The average voltage of all batteries. This is the maximum voltage value. This is the minimum voltage value. For voltage range, is the voltage variation coefficient.

[0028] ② Internal resistance consistency: (4) (5) (6) in, The standard deviation of internal resistance, For the first The internal resistance of a battery. This represents the average internal resistance of all batteries. The internal resistance variation coefficient is... This represents the maximum internal resistance. This represents the minimum internal resistance. This represents the extreme difference in internal resistance.

[0029] ③SOC consistency: (7) (8) (9) in, SOC standard deviation For the first SOC of a battery The SOC average for all batteries. The coefficient of variation of SOC. Represents the maximum SOC. Represents the minimum SOC value. This represents the worst SOC range.

[0030] ④ Temperature consistency: (10) (11) (12) in, For temperature standard deviation, For the first The temperature of each battery, The average temperature of all batteries. The coefficient of variation is the temperature variation. Represents the maximum temperature. Represents the minimum temperature. This is due to an extreme temperature difference.

[0031] Step 1.2: Obtain the original observation data and expert judgment information for each indicator, and form a standardized data matrix and a fuzzy judgment matrix.

[0032] Specifically, based on the established consistency evaluation index system for energy storage power stations, the original observation data of each index are collected and the original observation data are preprocessed. The preprocessing methods include positive transformation (used to convert negative indicators into positive indicators, see formula (21)) and standardization (used to eliminate the influence of dimensions, see formula (22)) to obtain a standardized data matrix, which is used to calculate the objective weight of the index.

[0033] Meanwhile, experts in the field were invited to use triangular fuzzy numbers (ã1,ã2,...,ã9) to compare the importance of indicators within the same level based on the established consistency evaluation index system for energy storage power stations, forming a fuzzy judgment matrix to calculate the subjective weight of the indicators.

[0034] Step 2: Based on the standardized data matrix and fuzzy judgment matrix, obtain the combined weight vector through the combined weighting method of fuzzy hierarchical analysis-grey relational degree-game theory.

[0035] Specifically, it includes the following steps: Step 2.1: Based on the fuzzy judgment matrix established in Step 1, calculate the subjective weight vector using FAHP (fuzzy hierarchical analysis).

[0036] In this embodiment, FAHP includes the following steps: Step 2.1.1: Perform a consistency check based on the fuzzy judgment matrix. If the consistency check passes, proceed to step 2.1.2; otherwise, return to step 1 to regenerate the fuzzy judgment matrix.

[0037] Assume the fuzzy judgment matrix established in step 1.2 is as follows: (13) in, For triangular fuzzy numbers, To be the minimum value, The most likely value, It is the maximum value and satisfies ( ), , The number of indicators.

[0038] First, the fuzzy judgment matrix Deblurring yields the precise matrix. ， The deblurring formula uses the centroid method: (14) Secondly, based on the exact matrix The consistency index was calculated.

[0039] The formula for calculating the consistency index is as follows: (15) in, For matrix The largest eigenvalue.

[0040] Finally, based on the calculated consistency index, combined with the random consistency index... Perform a consistency check.

[0041] Among them, the random consistency index This can be obtained through table lookup. When performing consistency verification, the consistency ratio needs to be calculated based on the consistency index and the random consistency index. ,like If the fuzzy judgment matrix meets the consistency requirement, proceed to step 2.1.2; otherwise, feedback is given to domain experts to correct the fuzzy judgment matrix until... .

[0042] Step 2.1.2: Use fuzzy arithmetic rules (extended principle) to synthesize the fuzzy judgment matrix after consistency verification, and calculate the fuzzy weight of each indicator. .

[0043] First, calculate the product of the fuzzy elements in each row of the fuzzy judgment matrix, using the following formula: (16) in, , , ; Secondly, based on the product of the fuzzy elements in each row Calculate the sum of the products of all row fuzzy elements. The formula is: (17) in, , , .

[0044] Finally, based on the sum of the products of all row fuzzy elements The fuzzy weights of each indicator are calculated using the following formula: (18) in, , , , This is a fuzzy multiplication operation. This is a fuzzy inverse operation.

[0045] Step 2.1.3: Assigning fuzzy weights to each indicator After defuzzing and normalization, a subjective weight vector is obtained, which consists of the subjective weights of each indicator.

[0046] First, the fuzzy weights of each indicator. Defuzzification is performed (using the centroid method) to convert the fuzzy weights into precise weights. ， The formula is: (19) Then, the precise weights of each indicator. After normalization, a subjective weight vector consisting of the subjective weights of each indicator is obtained. The normalization formula is: (20) in, Total number of indicators in the indicator layer ， For the first Subjective weights of each indicator ， And satisfy ( ).

[0047] Step 2.2: Based on the standardized data matrix established in Step 1, calculate the objective weight vector using GRA (grey relational analysis method).

[0048] Specifically, it includes the following steps: Step 2.2.1: Based on the established standardized data matrix, process it to obtain the corresponding ideal scheme reference sequence.

[0049] Specifically, including: First, the acquired raw observation data matrix （ For the number of evaluation objects, For the number of indicators, For the first The first evaluation object The raw data of each indicator are preprocessed, including positiveing ​​and standardization.

[0050] Among them, positive optimization refers to directly retaining positive indicators (the larger the value, the better), that is... For negative indicators (the smaller the value, the better), the inverse method is used for conversion, and the formula is: (twenty one) in, These are the indicator data after positive transformation.

[0051] Standardization (using range standardization) to eliminate the influence of dimensions, the formula is: (twenty two) in, For the first The first evaluation object The standardized values ​​of each indicator are used to obtain a standardized data matrix. ; Then, the optimal values ​​of each index in the standardized data matrix are selected to form a reference sequence of ideal solutions: (twenty three) in, (Positive indicator, the optimal value after standardization is 1), that is (Simplified form of the reference sequence after standardization).

[0052] Step 2.2.2: Based on the ideal scheme reference sequence, calculate the correlation coefficient of each indicator according to the grey relational theory.

[0053] Specifically, including: First, based on the grey relational analysis theory, calculate the absolute difference between the standardized values ​​of each indicator and the reference sequence. The formula is: (twenty four) Secondly, calculate the minimum difference between the two levels. and the maximum difference between the two levels The formulas are as follows: (25) (26) Finally, the correlation coefficient is calculated. The calculation formula is: (27) in, The resolution coefficient ranges from 0 to 1, and is usually set to 0.5 (to balance resolution and stability). The closer the value is to 1, the higher the correlation.

[0054] Step 2.2.3: Calculate the average correlation coefficient based on the correlation coefficient of each indicator, and after normalization, obtain the objective weight vector composed of the objective weights of each indicator.

[0055] Calculate objective weights: First, calculate the average correlation coefficient of each indicator. This reflects the overall correlation between the indicator and the reference series, and the formula is: (28) The average correlation coefficient is then normalized to obtain the objective weight vector. The normalization formula is: (29) in, For the first The objective weights of each indicator, and satisfying , ( ).

[0056] Step 2.3: Based on game theory, fuse the obtained subjective weight vector and objective weight vector to obtain the final combined weight vector.

[0057] Specifically, it includes the following steps: Step 2.3.1: With the goal of minimizing the sum of squared Euclidean distances between the combined weight vector and the subjective and objective weight vectors, construct the objective function and determine the constraints.

[0058] In this embodiment, the combined weight vector to be determined is defined as follows: (Column vectors) An optimization model is constructed with the objective of minimizing the sum of squared Euclidean distances between the combined weights and the subjective and objective weights. The objective function is: (30) in , , The distance is Euclidean, which, when expanded, is: , This represents the matrix transpose operation. Introducing constraints: The combined weights must satisfy the normalization constraint (the sum of the weights must be 1), i.e. ,in It is a column vector of all 1s, and the constraints ensure the rationality and interpretability of the weights.

[0059] Step 2.3.2: Construct a Lagrangian function to solve the objective function constructed in Step 2.3.1, and obtain the optimal combined weight vector. .

[0060] Specifically, the Lagrangian function is constructed, incorporating the constraints into the objective function, and expressed as: (31) in, For Lagrange multipliers (real numbers); about Taking the partial derivatives and setting them to zero, according to the matrix differentiation rule, we have: (32) The preliminary solution with optimal weights is obtained by rearranging. : (33) Constraints Substitute into the above equation to solve for the Lagrange multiplier. : (34) in, The number of indicators.

[0061] Will Substituting the initial solution, we obtain the unconstrained optimal combination weight vector. .

[0062] Step 2.3.3: Calculate the optimal combination weights Normalization is performed (to ensure that the weight constraints are met) to obtain the final combined weight vector. , represented as: (35) The normalization formula is: (36) in, satisfy .

[0063] Step 3: Based on the improved MK algorithm, cluster analysis is performed on the dataset of the energy storage power station to be evaluated, and the consistency of the energy storage power station is evaluated by combining the combined weight vector to obtain the consistency evaluation result.

[0064] The K-Medoids algorithm mainly consists of two parts: the Fast Voronoi algorithm (FV) and the Voronoi Correction (VC). To better understand this invention, we will first introduce the theory of K-Medoids. The K-Medoids problem aims to solve problems involving a given dataset... (in Select K center points from )) This minimizes the error function. The error function is shown below: (37) in, Dissimilarity is a measure of dissimilarity used to measure the degree of difference between data points. The MK algorithm uses Manhattan distance as the dissimilarity measure.

[0065] Theorem 1: For a vector sorted in ascending order If instance Within a specific range, or ,and For vectors If the median is found, then the following equation holds: (38) Theorem 1 establishes the relationship between the 1-median and the 1-center point error function.

[0066] Data set D is based on a set of center points Divided into Clusters , express In the A 1-D array consisting of all elements of dimension 1. At this point, the error function of the K-Medoids problem... It can be rewritten as:

[0067] This means that the conclusion of Theorem 1 can be applied to any dimension of the K-Medoids error function. This allows for the rapid calculation of the center points of a given cluster.

[0068] Based on Theorem 1, Corollary 1 is derived. For a point set sorted in ascending order according to each feature... ,have (40) in, This inference provides a direct basis for the center point update step of the FV algorithm, making center point update a selection process. .

[0069] like Figure 3 As shown, the FV algorithm includes the following steps: Input: Obtain the input dataset D, the number of clusters K, and the initial set of centroids C.

[0070] Sorting phase: Sort each column of dataset D in ascending order.

[0071] Allocation Phase: Based on the current set of centroids C, allocate each point in dataset D. Assigning a node to the cluster containing the nearest centroid results in a cluster. .

[0072] Update phase: Based on the sorting results, calculate all of And select to make smallest As the new center.

[0073] Loop: Repeat the above allocation and update steps until the stopping condition is met (such as the center point no longer changing or the error decrease is negligible).

[0074] Output: The set of centroids C and the associated cluster set P.

[0075] like Figure 4 As shown, the specific steps of the VC algorithm are as follows.

[0076] Input: Obtain the set of centroids C and the associated cluster set P obtained by the FV algorithm.

[0077] Splitting phase: For each cluster The FV algorithm is used to approximate the 2-centroid problem. This process involves finding the centroid for each cluster. Generate paired center points After generating these centroid pairs, each cluster is computed. At a single center point and double center point Error difference between the two cases ,Right now .choose Largest cluster ,satisfy In this cluster Let K=2 and perform another FV iteration to add a new center point. This is because... The larger the value, the greater the error reduction that adding a center point to the cluster can bring. This also means that the cluster may have a more complex internal structure and needs to be further subdivided. This helps to break the local optimum that may be trapped and provides more possibilities for subsequent cluster optimization.

[0078] Merging phase: Selecting the two clusters that minimize the increase in error after merging. and Merging is used to identify regions in space that may be overrepresented by the current set of centroids. Specifically, this is achieved by calculating... To estimate the increase in error after merging, this value is The upper bound of the cluster pair to be merged. This means that, in selecting cluster pairs to merge... and When merging these two clusters, the algorithm will find the cluster pair that minimizes this estimate. This is because the smaller the estimate, the smaller the increase in error after merging these two clusters.

[0079] Output: The final set of centroids C and the associated cluster set P.

[0080] like Figure 5 As shown, the combination of the FV algorithm and VC algorithm described above constitutes the Manhattan K-Medoids algorithm.

[0081] To address the issues of the MK algorithm's heavy reliance on initialization and the existence of optimization space in the allocation steps, this invention improves the MK algorithm using the Annoy algorithm and K-means++ initialization based on Manhattan distance. K-means++, a commonly used intelligent initialization strategy, employs Manhattan distance within the MK algorithm framework, significantly reducing the probability of the MK algorithm getting trapped in local optima.

[0082] like Figure 6 As shown, the improved MK algorithm proposed in this invention specifically includes the following steps: Step 3.1: Initialize using K-Means++ with Manhattan distance to select the initial set of centroids C.

[0083] Specifically, including: ① Select the first cluster center.

[0084] Randomly select a point from dataset D and designate it as the first cluster center, denoted as . This is the starting point for initialization, laying the foundation for the subsequent selection of cluster centers.

[0085] ② Calculate the Manhattan distance from any data point in dataset D to the selected cluster center.

[0086] For any data point in dataset D ( i =1,2,…, n ), n (Based on dataset size), according to the Manhattan distance formula ,in, For data dimensions, and Representing data points and cluster center In the The value on the dimension calculates the distance from each data point to the selected cluster center (assuming it has been selected). Cluster centers, The distance, i.e. , .

[0087] ③ Determine the probability that each data point will be selected as the next cluster center based on the Manhattan distance from each data point to the selected cluster center.

[0088] The probability of each data point being selected as the next cluster center. The calculation formula is as follows: (41) in, Each data point The minimum distance to the selected cluster center.

[0089] ④ Update the initial set of center points C based on the probability that each data point is selected as the next cluster center.

[0090] In this embodiment, the purpose of using K-Means++ initialization is to make the initial cluster centers as dispersed as possible. Therefore, according to the probability formula (41), it can be seen that the distance from the current center point... The more distant the data point, the greater the probability that it will be selected as the next centroid C. This can avoid the clustering process from being too concentrated with random initial points, thus increasing the number of clustering iterations and improving clustering efficiency.

[0091] Step 3.2: Obtain the dataset D (i.e., the index data after positive transformation), the number of clusters K, and the initial set of centroids C corresponding to the energy storage power station to be evaluated.

[0092] Step 3.3: Sort each column of dataset D in ascending order.

[0093] Step 3.4: Based on the current set of centroids C, use the Annoy algorithm to assign each non-centroid in dataset D to the cluster containing the nearest centroid.

[0094] In the allocation phase of the FV algorithm, determining the distance of each data point to each centroid and assigning it to the cluster containing the nearest centroid is a crucial and computationally intensive process. This invention optimizes this step using the Annoy algorithm, further reducing the algorithm's computational complexity. The specific steps are as follows: ① Construct the Annoy index using all the central points.

[0095] During the construction process, starting from the central point set ( K Two center points are randomly selected from the number of center points. and The hyperplane is determined to divide the space.

[0096] ② In allocating each data point x At that time, the center point of its approximate nearest neighbor is found through the constructed Annoy index.

[0097] Step 3.5: Sort each column of dataset D in ascending order, and calculate all results based on the sorting. of And select to make smallest As the new center.

[0098] Step 3.6: Repeat steps 3.4 to 3.5 to output the initial set of centroids C and the associated cluster set P.

[0099] Step 3.7: Based on the initial set of centroids C and the associated cluster set P, the VC algorithm is used to split centroids and merge cluster pairs to obtain the optimized set of centroids C' and the associated cluster set P'.

[0100] Step 3.8: Based on the cluster set P' and combined with the combined weight vector obtained in Step 2, carry out the evaluation level classification work to obtain the consistency evaluation results of the energy storage power station.

[0101] Specifically, it includes the following steps: ① Derive the cluster center coordinates of each data cluster. These center coordinates are composed of the average scores of all samples in the data cluster on each evaluation index, which can intuitively reflect the overall average level of the corresponding data cluster.

[0102] ② Multiply the values ​​of each index of the cluster center by the corresponding weights and sum them up to obtain the comprehensive score of each cluster center. The score is positively correlated with the performance of the object represented by the cluster.

[0103] ③ Sort all data clusters by comprehensive score from high to low, and complete the evaluation level classification of all data clusters. The final evaluation level of each energy storage power station is consistent with the level of the data cluster to which it belongs.

[0104] ④ Based on the evaluation results, the data period with the lowest evaluation level is selected, and secondary clustering is performed on this part of the data. Then, for the data period corresponding to the lowest-scoring cluster, abnormal data identification and analysis are carried out.

[0105] Example 2 The above-described embodiment 1 provides a consensus evaluation method for energy storage power stations based on an improved MK algorithm. Correspondingly, this embodiment provides a consensus evaluation system for energy storage power stations based on an improved MK algorithm. The system provided in this embodiment can implement the consensus evaluation method for energy storage power stations based on the improved MK algorithm of embodiment 1. This system can be implemented through software, hardware, or a combination of both. For example, the system may include integrated or separate functional modules or units to execute the corresponding steps in the methods of embodiment 1. Since the system in this embodiment is basically similar to the method embodiment, the description process in this embodiment is relatively simple. For relevant details, please refer to the description in embodiment 1. The system embodiment provided in this embodiment is merely illustrative.

[0106] The energy storage power station consistency evaluation system based on the improved MK algorithm provided in this embodiment includes: The indicator data acquisition module is used to acquire the original observation data of each indicator based on the pre-established consistency evaluation indicator system for energy storage power stations, and to form a standardized data matrix and a fuzzy judgment matrix. The combined weight calculation module is used to obtain the combined weight vector based on the standardized data matrix and the fuzzy judgment matrix through the combined weight assignment method of fuzzy hierarchical analysis-grey relational degree-game theory. The consistency evaluation module is used to perform cluster analysis on the dataset of the energy storage power station to be evaluated based on the improved MK algorithm, and to evaluate the consistency of the energy storage power station by combining the combined weight vector, so as to obtain the consistency evaluation result.

[0107] Example 3 This embodiment provides a processing device corresponding to the energy storage power station consistency evaluation method based on the improved MK algorithm provided in Embodiment 1. The processing device can be a client-side processing device, such as a mobile phone, laptop, tablet computer, desktop computer, etc., to execute the method of Embodiment 1.

[0108] The processing device includes a processor, a memory, a communication interface, and a bus. The processor, memory, and communication interface are connected via the bus to enable communication between them. The memory stores a computer program that can run on the processor. When the processor runs the computer program, it executes the energy storage power station consistency evaluation method based on the improved MK algorithm provided in Embodiment 1.

[0109] Preferably, the memory may be high-speed random access memory (RAM), and may also include non-volatile memory, such as at least one disk storage device.

[0110] Preferably, the processor can be any type of general-purpose processor such as a central processing unit (CPU) or a digital signal processor (DSP), and there is no limitation herein.

[0111] Example 4 The energy storage power station consistency evaluation method based on the improved MK algorithm in Embodiment 1 can be specifically implemented as a computer program product. The computer program product may include a computer-readable storage medium on which computer-readable program instructions for executing the energy storage power station consistency evaluation method based on the improved MK algorithm described in Embodiment 1 are loaded.

[0112] A computer-readable storage medium can be a tangible device that holds and stores instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof.

[0113] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A consensus evaluation method for energy storage power stations based on an improved MK algorithm, characterized in that, include: Based on the pre-established consistency evaluation index system for energy storage power stations, the original observation data of each index are obtained, and a standardized data matrix and a fuzzy judgment matrix are formed. Based on the standardized data matrix and fuzzy judgment matrix, a combined weight vector is obtained through a combined weighting method of fuzzy hierarchical analysis-grey relational degree-game theory. Based on the improved MK algorithm, cluster analysis is performed on the dataset of the energy storage power station to be evaluated, and the consistency of the energy storage power station is evaluated by combining the combined weight vector, so as to obtain the consistency evaluation result.

2. The consensus evaluation method for energy storage power stations based on the improved MK algorithm as described in claim 1, characterized in that, The method, based on a pre-established consistency evaluation index system for energy storage power stations, acquires the original observation data of each index and forms a standardized data matrix and a fuzzy judgment matrix, including: Establish a consistency evaluation index system for energy storage power stations; the consistency evaluation index system includes four categories of indicators: battery voltage, battery internal resistance, battery SOC and battery temperature, and each category of indicators includes three evaluation indicators: standard deviation, coefficient of variation and range. Obtain the original observation data and expert judgment information for each indicator, and form a standardized data matrix and a fuzzy judgment matrix.

3. The consensus evaluation method for energy storage power stations based on the improved MK algorithm as described in claim 1, characterized in that, The method, based on a standardized data matrix and a fuzzy judgment matrix, uses a combined weighting method of fuzzy hierarchical analysis, grey relational analysis, and game theory to obtain a combined weight vector, including: Based on the established fuzzy judgment matrix, the subjective weight vector is calculated using the fuzzy hierarchical analysis method. Based on the established standardized data matrix, the objective weight vector is calculated using the grey relational analysis method. Based on game theory, the obtained subjective weight vector and objective weight vector are fused to obtain the final combined weight vector.

4. The consistency evaluation method for energy storage power stations based on the improved MK algorithm as described in claim 3, characterized in that, The subjective weight vector, calculated based on the established fuzzy judgment matrix and using fuzzy hierarchical analysis, includes: A consistency check is performed based on the fuzzy judgment matrix. If the consistency check passes, proceed to the next step; otherwise, the fuzzy judgment matrix is ​​regenerated. The fuzzy judgment matrix after consistency verification is synthesized using fuzzy arithmetic rules, and the fuzzy weights of each indicator are calculated. After defuzzifying and normalizing the fuzzy weights of each indicator, a subjective weight vector composed of the subjective weights of each indicator is obtained.

5. The consensus evaluation method for energy storage power stations based on the improved MK algorithm as described in claim 3, characterized in that, The objective weight vector, calculated based on the established standardized data matrix and the grey relational analysis method, includes: Based on the established standardized data matrix, the corresponding ideal solution reference sequence is obtained; Based on the ideal scheme reference sequence, the correlation coefficients of each indicator are calculated according to the grey relational theory; The average correlation coefficient is calculated based on the correlation coefficient of each indicator, and after normalization, an objective weight vector composed of the objective weights of each indicator is obtained.

6. The consistency evaluation method for energy storage power stations based on the improved MK algorithm as described in claim 3, characterized in that, The method of fusing the obtained subjective and objective weight vectors based on game theory to obtain the final combined weight vector includes: With the goal of minimizing the sum of squared Euclidean distances between the combined weight vector and the subjective and objective weight vectors, an objective function is constructed and constraints are determined. The Lagrangian function is constructed to solve the constructed objective function, and the initial optimal combination weight vector is obtained; The initial optimal combination weights are normalized to obtain the final combination weight vector.

7. The consistency evaluation method for energy storage power stations based on the improved MK algorithm as described in claim 1, characterized in that, The improved MK algorithm is used to perform cluster analysis on the dataset of the energy storage power stations to be evaluated, and the consistency of the energy storage power stations is evaluated by combining the combined weight vector, resulting in a consistency evaluation result, including: ① Initialize using the K-Means++ method based on Manhattan distance to select the initial set of centroids C; ② Obtain the dataset, number of clusters K, and initial set of centroids C corresponding to the energy storage power station to be evaluated; ③ Sort each column of the dataset in ascending order; ④ Based on the initial set of centroids C, the Annoy algorithm is used to assign each non-centroid in the dataset to the cluster containing the nearest centroid; ⑤ Sort each column of the dataset in ascending order, and update the center point based on the sorting result; ⑥ Repeat steps ④ to ⑤ until the centroids no longer change, and output the initial set of centroids C and the associated cluster set P; ⑦ Based on the initial set of centroids C and the associated cluster set P, the VC algorithm is used to split centroids and merge cluster pairs to obtain the optimized set of centroids C' and the associated cluster set P'. ⑧ Based on the cluster set P' and the obtained combined weight vector, the evaluation level classification work is carried out to obtain the consistency evaluation results of the energy storage power station.

8. A consistency evaluation system for energy storage power stations based on an improved MK algorithm, characterized in that, include: The indicator data acquisition module is used to acquire the original observation data of each indicator based on the pre-established consistency evaluation indicator system for energy storage power stations, and to form a standardized data matrix and a fuzzy judgment matrix. The combined weight calculation module is used to obtain the combined weight vector based on the standardized data matrix and the fuzzy judgment matrix through the combined weight assignment method of fuzzy hierarchical analysis-grey relational degree-game theory. The consistency evaluation module is used to perform cluster analysis on the dataset of the energy storage power station to be evaluated based on the improved MK algorithm, and to evaluate the consistency of the energy storage power station by combining the combined weight vector, so as to obtain the consistency evaluation result.

9. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods described in claims 1 to 7.

10. A computing device, characterized in that, include: One or more processors and a memory, wherein the memory stores one or more programs and is configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods described in claims 1 to 7.

Images