Electronic device, recording medium, and method for clustering data related to battery charging mode

Abstract

An electronic device is provided for clustering data related to battery charging modes. The device can be configured to: acquire multiple battery charging data points from multiple vehicles; generate multiple features related to battery charging modes based on the multiple battery charging data points; classify the multiple features into a predetermined number of datasets by sampling the multiple features; determine the number of clusters corresponding to each dataset in the predetermined number of datasets based on a clustering algorithm and an index used to evaluate the clustering results; set a reference number of clusters corresponding to the multiple features based on an optimal number of clusters for each dataset in the predetermined number of datasets; and cluster the multiple features into the reference number of clusters.

Description

Technical Field

[0001] This disclosure relates to electronic devices, recording media, and methods for clustering data about battery charging patterns, and more specifically to techniques for clustering battery charging pattern data for defining types of charging patterns for electric vehicle users.

[0002] This application claims priority and benefit to Korean Patent Application No. 2024-0066357, filed on May 22, 2024, and Korean Patent Application No. 2024-0098789, filed on July 25, 2024, the entire contents of which are incorporated herein by reference. Background Technology

[0003] Clustering algorithms, used as unsupervised learning algorithms, have been proposed for standardizing and classifying time series data. However, when performing clustering on a large amount of time series data, too few clusters may be used because a series of data may exist on the boundaries of different types. This can lead to the problem of data with different characteristics being classified into one type. Considering that the purpose of clustering is to identify the features of each target data and discover meaningful data types, clustering performance can be considered significantly low when the data included in each type exhibit different characteristics due to the mixing of data.

[0004] Therefore, traditionally, problems caused by data mixing are prevented by using a sufficient number of clusters during clustering, and data types are discovered and defined by superimposing data using the properties of probability distributions and manually merging data. However, this approach involves subjective factors and requires a long processing time.

[0005] Furthermore, while clustering is typically an unsupervised learning algorithm, an evaluation index can be used to select the optimal number of clusters to assess clustering performance. However, because the index used exhibits a monotonically decreasing or monotonically increasing trend rather than a convex form, it is difficult to select an appropriate inflection point (the point where the rate of decrease of the predetermined cost function used for clustering performance evaluation shows a sharp drop). Therefore, a method is needed to derive an objective index for determining a more appropriate number of clusters to perform clustering for pattern discovery. Summary of the Invention

[0006] Technical goals

[0007] The present disclosure aims to solve at least the aforementioned problems and provides an electronic device and a method for clustering data on battery charging patterns.

[0008] The objectives to be achieved by the exemplary embodiments of this disclosure are not limited to the objectives described above, and other objectives can be inferred from the following exemplary embodiments.

[0009] Technical solution

[0010] According to one aspect, an electronic device is provided, the electronic device including a transceiver, a processor, and one or more memories configured to store one or more instructions, wherein the one or more instructions are configured to, when executed, allow the processor to: acquire multiple battery charging data of multiple vehicles; generate multiple features of battery charging patterns based on the multiple battery charging data; classify the multiple features into a predetermined number of datasets by sampling the multiple features; determine the number of clusters corresponding to each dataset in the predetermined number of datasets based on a clustering algorithm and an index for evaluating the clustering results; set a reference number of clusters corresponding to the multiple features based on an optimal number of clusters for each dataset in the predetermined number of datasets; and perform clustering of the multiple features into clusters of the reference number of clusters.

[0011] Multiple battery charging data can include the state of charge (SoC) of each of the multiple batteries included in multiple vehicles over time as data collected during a predetermined time period.

[0012] One or more instructions may be configured, when executed, to allow the processor to identify one or more charging segments based on first battery charging data of a first vehicle among multiple battery charging data of multiple vehicles, to identify the battery SoC at the start of charging and the battery SoC at the end of charging for each of the one or more charging segments, and to generate a first feature indicating the frequency of the one or more charging segments corresponding to the range of the battery SoC at the start of charging and the range of the battery SoC at the end of charging.

[0013] Each dataset in a predetermined number of datasets may include at least one feature, and the number of at least one feature may be determined in proportion to the number of multiple features.

[0014] One or more instructions can be configured to allow the processor to perform clustering multiple times on a first dataset from a predetermined number of datasets when executed, such that the first dataset has different numbers of clusters according to the clustering algorithm, and the optimal number of clusters for the first dataset is determined based on the results of the multiple clustering operations and an index used to evaluate the results of the clustering.

[0015] Clustering algorithms may include at least one of hierarchical clustering algorithms that stratify at least one feature included in each dataset based on similarity and K-means clustering algorithms that cluster at least one feature included in each dataset based on centroid.

[0016] The index may include a first index, a second index, and a third index, and one or more instructions may be configured, when executed, to allow the processor to calculate each of the first index, the second index, and the third index based on the number of clusters, based on the results of multiple executions of clustering, and to determine the optimal number of clusters for the first dataset based on the first number of clusters corresponding to the maximum or minimum value of the first index of the calculated number of clusters, the second number of clusters corresponding to the maximum or minimum value of the second index of the calculated number of clusters, and the third number of clusters corresponding to the maximum or minimum value of the third index of the calculated number of clusters.

[0017] Each of the first, second, and third indices may include at least one of the Silhouette index technique, Calinski-Harabasz index technique, and Davies-Bouldin index technique, which evaluate the clustering results based on the cohesion of each cluster and the separation between clusters for each result of clustering performed according to the clustering algorithm.

[0018] One or more instructions can be configured to, when executed, allow the processor to set the average of the optimal number of clusters for each dataset in a predetermined number of datasets as the reference number of clusters.

[0019] A method for clustering data about battery charging patterns may include: acquiring multiple battery charging data from multiple vehicles; generating multiple features of battery charging patterns based on the multiple battery charging data; classifying the multiple features into a predetermined number of datasets by sampling the multiple features; determining the number of clusters corresponding to each dataset in the predetermined number of datasets based on a clustering algorithm and an index used to evaluate the clustering results; setting a reference number of clusters corresponding to the multiple features based on an optimal number of clusters for each dataset in the predetermined number of datasets; and performing clustering of the multiple features into clusters equal to the reference number of clusters.

[0020] As a non-transitory computer-readable recording medium, storing a program for performing the method on a computer, the method for clustering data about battery charging patterns may include: acquiring multiple battery charging data for multiple vehicles; generating multiple features of battery charging patterns based on the multiple battery charging data; classifying the multiple features into a predetermined number of datasets by sampling the multiple features; determining the number of clusters corresponding to each dataset in the predetermined number of datasets based on a clustering algorithm and an index used to evaluate the clustering results; setting a reference number of clusters corresponding to the multiple features based on the optimal number of clusters for each dataset in the predetermined number of datasets; and performing clustering of the multiple features into clusters of the reference number of clusters.

[0021] Effects of the present invention

[0022] According to the example embodiment, when multiple types of battery charging data are discovered, the type of charging pattern that allows for more effective identification of the characteristics of the data can be discovered by selecting an appropriate number of clusters.

[0023] According to the example implementation, the optimal type can be defined by minimizing the influence of data present on the boundary lines between clusters and preventing overfitting that may occur in a particular cluster.

[0024] The effects of this disclosure are not limited to those described above, and other effects will become apparent to those skilled in the art from the following description. Attached Figure Description

[0025] Figure 1 This is a block diagram of an electronic device according to an example embodiment.

[0026] Figure 2 This is a flowchart illustrating a method for clustering data about battery charging patterns.

[0027] Figure 3 This is a diagram illustrating the process of generating features according to an example embodiment.

[0028] Figure 4 This is a diagram illustrating the process of classifying features according to an example embodiment.

[0029] Figure 5 This is a diagram illustrating the process of determining the optimal number of clusters corresponding to a dataset according to an example embodiment.

[0030] Figure 6a This is a graph showing the first index based on the number of clusters according to an example embodiment.

[0031] Figure 6b This is a graph showing the second index based on the number of clusters according to an example embodiment.

[0032] Figure 6c This is a graph showing the third index based on the number of clusters according to an example embodiment.

[0033] Figure 7 This is a diagram illustrating the process of setting the number of reference clusters according to an example embodiment of the present disclosure. Detailed Implementation

[0034] While taking into account the functionality obtained under this disclosure, the terms used in the exemplary embodiments are selected from the most widely used general terminology available today; however, these terms may be replaced by other terms based on the intent of those skilled in the art, convention, the emergence of new technologies, etc. Furthermore, in specific circumstances, terms arbitrarily chosen by the applicant of this disclosure may be used. Therefore, it should be noted that the terms used herein should be interpreted based on their actual meaning and the entirety of this specification, rather than simply on their names.

[0035] Throughout this specification, when an element is referred to as "comprising" or "including" another element, unless otherwise specified in the description, that element should not be construed as excluding other elements, and that element may include at least one other element. Furthermore, the terms "unit" and "module" may, for example, refer to a component that performs at least one function or operation and may be implemented in hardware or software, or by a combination of hardware and software.

[0036] Throughout the specification, the phrase "at least one of a, b, and c" may include "only a", "only b", "only c", "a and b", "a and c", "b and c", or "all of a, b, and c".

[0037] In the following description, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings, enabling those skilled in the art to readily practice the present disclosure. However, the present disclosure may be embodied in many different forms and is not limited to the exemplary embodiments described herein.

[0038] In the following description, exemplary embodiments of an electronic device for clustering data about battery charging patterns will be described with reference to the accompanying drawings.

[0039] Figure 1 This is a block diagram of an electronic device according to an example embodiment.

[0040] refer to Figure 1 According to an example embodiment, electronic device 100 may include transceiver 110, processor 120, and memory 130. Figure 1 Only components related to electronic device 100 are shown. Therefore, those skilled in the art will understand that, in addition to Figure 1 In addition to the components shown, other general-purpose components may also be included.

[0041] For example, electronic device 100 may include a communication device comprising one or more transceivers 110, input components, and output components. As a device for wired / wireless communication, the communication device can communicate with external electronic devices. These external electronic devices may be terminals or servers. Furthermore, the communication technologies supported by the communication device may include Global System for Mobile Communications (GSM), Code Division Multiple Access (CDMA), Long Term Evolution (LTE), Wi-Fi, Bluetooth, Radio Frequency Identification (RFID), Infrared Data Association (IrDA), ZigBee, Near Field Communication (NFC), etc. Input components may include, for example, conventional keypads and keyboards, mice, microphones for receiving voice signals, cameras, and various other types of input devices that detect or receive various forms of user input. Output components may include, for example, display panels that output images, speakers that output sound, haptic devices that generate vibrations, and various other output devices.

[0042] According to an example embodiment, electronic device 100 may be a server for acquiring and processing data from multiple vehicles. More specifically, the data from the multiple vehicles may include multiple battery charging data from the multiple vehicles. For example, multiple battery charging data can be acquired via transceiver 110 of electronic device 100 from at least one of an on-board diagnostic (OBD), a battery management system (BMS), and a device (e.g., a database) in each vehicle that pre-stores battery charging data. The method by which electronic device 100 acquires multiple battery charging data is not limited to the examples described above, and those skilled in the art will understand that electronic device 100 can acquire data from different devices with communication capabilities via transceiver 110. The type of electronic device 100 is not limited thereto, and the example embodiments of this disclosure can be applied to various devices capable of acquiring and processing data on vehicles.

[0043] Processor 120 can control the overall operation of electronic device 100 and process data and signals. Processor 120 can be configured as at least one hardware unit. Furthermore, processor 120 can be operated by one or more software modules generated by executing program code stored in memory 130. Since processor 120 may include memory, processor 120 can control the overall operation of electronic device 100 and process data and signals by executing program code stored in memory.

[0044] Processor 120 may be embodied as hardware, software, or a combination thereof, such as a computer or similar device. As hardware, processor 120 may be embodied as electronic circuitry for performing control functions by processing electrical signals, and as software, processor 120 may be embodied as a program in hardware form for operating processor 120. However, it should be understood that, unless otherwise mentioned in the following description, the operation of the electronic device is controlled and executed by processor 120. That is, when modules implementing a data clustering system regarding battery charging modes are executed, it can be understood that these modules control processor 120 to perform the following operations of electronic device 100.

[0045] The memory 130 can store various types of information. The memory 130 can store data temporarily or semi-permanently. For example, data about the operating system (OS) used to operate the electronic device 100 can be stored in the memory 130 of the electronic device 100. The memory 130 can be a hard disk drive (HDD), a solid-state drive (SSD), flash memory, read-only memory (ROM), random access memory (RAM), etc. The memory 130 can be provided in an embedded or removable type.

[0046] In summary, the exemplary embodiments can be implemented using different means. For example, the exemplary embodiments can be embodied in hardware, firmware, software, or a combination thereof.

[0047] When embodied in hardware, the method according to the example embodiment can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, etc.

[0048] When embodied as firmware or software, the method according to the example embodiments can be implemented as a module, process, function, etc., to perform the above-described functions or operations. For example, software code can be stored in memory and executed by a processor. The memory can be located inside or outside the processor and can exchange data with the processor through various pre-notified means.

[0049] Figure 2 This is a flowchart illustrating a method for clustering data about battery charging patterns.

[0050] In operation S210, the electronic device 100 can acquire multiple battery charging data from multiple vehicles. Here, the multiple battery charging data can include the state of charge (SoC) of each of the multiple batteries included in the multiple vehicles over time as data collected during a predetermined time period. The predetermined time period can be a period in which battery charging and discharging are performed multiple times, and, for example, the predetermined time period can be one month. Therefore, the multiple battery charging data can be time-series data including information about the change of SoC over time (during which charging and discharging of each of the multiple batteries included in the multiple vehicles is repeated). Reference will be made below. Figure 3 A more specific example embodiment describing battery charging data in detail.

[0051] In operation S220, the electronic device 100 can generate multiple features of the battery charging mode based on multiple battery charging data. (Refer to the following...) Figure 3 A more detailed description of the process by which the electronic device 100 generates features.

[0052] In operation S230, the electronic device 100 can classify multiple features into a predetermined number of datasets by sampling multiple features. For example, feature sampling can be performed randomly. Furthermore, according to an example embodiment, the electronic device 100 can perform sampling such that the composition of each dataset does not give too much importance to the features of a particular vehicle. Reference will be made below. Figure 3 A more specific example implementation of classifying multiple features into a predetermined number of datasets is described in detail.

[0053] In operation S240, electronic device 100 can determine the number of clusters corresponding to each dataset in a predetermined number of datasets based on a clustering algorithm and an index used to evaluate the clustering results. The clustering algorithm can be an algorithm used to divide target data into clusters with similar characteristics, and the index can be a metric used to quantitatively evaluate the performance of clustering based on cohesion (a criterion for seeing how tightly data within the same cluster is grouped together), separation (a criterion for seeing how distinct clusters are separated from each other), etc. References will follow below. Figure 4 A more specific example embodiment describes in detail how the electronic device 100 determines the number of clusters corresponding to each dataset in a predetermined number of datasets based on a clustering algorithm and an index.

[0054] In operation S250, the electronic device 100 can set a reference number of clusters corresponding to multiple features based on the optimal number of clusters for each dataset in a predetermined number of datasets. That is, the electronic device 100 can set a final reference number of clusters that allows for more efficient clustering of multiple features based on the optimal number of clusters for each dataset. A more specific example embodiment of the electronic device 100 using the optimal number of clusters for each dataset to set the reference number of clusters will be described in detail below with reference to FIG6.

[0055] In operation S260, the electronic device 100 can perform clustering of multiple features into a number equal to the number of reference clusters. Through clustering, battery charging patterns can be further subdivided, and different types that allow for objective classification can be discovered.

[0056] Figure 3 This is a diagram illustrating the process of generating features according to an example embodiment.

[0057] According to an example embodiment, electronic device 100 can identify one or more charging segments based on first battery charging data of a first vehicle from among multiple battery charging data of multiple vehicles, and identify the battery SoC at the start of charging and the battery SoC at the end of charging for each of the one or more charging segments. Electronic device 100 can then generate a first feature indicating the frequency of the one or more charging segments corresponding to the range of the battery SoC at the start of charging and the range of the battery SoC at the end of charging as a feature related to the charging mode of the first battery charging data. More specifically, electronic device 100 can analyze the frequency of the charging segments and estimate a probability distribution function of the frequency of each charging segment based on the first battery charging data. Electronic device 100 can generate features in the form of an image showing the frequency of the battery charging segments of the corresponding vehicle based on the estimated probability distribution function. Electronic device 100 can generate multiple features of the battery charging mode by performing this process for each of the multiple battery charging data.

[0058] Each of the multiple features presented in image form can show a specific battery charging pattern extracted from battery charging data of a particular vehicle. That is, each feature can show the charging pattern for each user of the corresponding vehicle, depending on the user from whom the battery charging data was collected.

[0059] refer to Figure 3The first battery charging data of the first vehicle can be in the form of a graph 300, where time is on the x-axis and SoC is on the y-axis. Specifically, the graph 300 may include information about the SoC of the first battery of the first vehicle during a predetermined time period, and more specifically, may include information about the charging segments identified each time the first battery is charged. Thus, the graph 300 can show at what SoC level the charging of the first battery begins and at what SoC level it ends. The electronic device 100 can estimate a probability distribution function of the frequency of the charging segments by analyzing the frequencies of multiple charging segments identified based on the graph 300 of the first battery charging data, and can generate features 310 of the first battery charging data based on the estimated probability distribution function.

[0060] Feature 310, generated in response to first battery charging data, can be in the form of a two-dimensional image, where the SoC at the start of charging is on the x-axis and the SoC at the end of charging is on the y-axis, and the shading of feature 310 can indicate the frequency of charging segments. That is, the darker the shading of a segment within feature 310, the more frequently the segment may be charged. Therefore, feature 310 can indicate charging habits or tendencies regarding at what SoC level charging begins and ends, and as a more specific example, feature 310 can indicate that the user of the first vehicle has a habit of primarily starting and ending charging the vehicle at a low SoC level.

[0061] Figure 4 This is a diagram illustrating the process of classifying features according to an example embodiment.

[0062] refer to Figure 4 Based on multiple features 410 of the battery charging mode, a predetermined number n datasets 420 can be generated, including dataset 1 indicated by reference numeral 422, dataset 2 indicated by reference numeral 424, and dataset n indicated by reference numeral 426. Each dataset in the predetermined number of datasets may include, for example, at least one feature, and the number of at least one feature included in each dataset can be determined proportionally to the number of multiple features. As a specific example, when 10,000 features are generated and the value of n is set to 10, the number of at least one feature included in each dataset is 800, which is 8000% of 10,000, calculated by dividing 8,000 by 10. According to the example embodiment, considering the total number of features generated, the number of at least one feature included in each dataset can be determined to a more appropriate value to obtain a dataset in which the generated features are sampled equally.

[0063] Figure 5This is a diagram illustrating the process of determining the optimal number of clusters corresponding to a dataset according to an example embodiment.

[0064] According to an example embodiment, electronic device 100 can perform clustering multiple times on a first dataset from a predetermined number of datasets according to a clustering algorithm, resulting in the first dataset having different numbers of clusters. The optimal number of clusters for the first dataset is determined based on the results of the multiple clustering operations and an index used to evaluate the clustering results. Specifically, electronic device 100 can perform clustering multiple times according to a clustering algorithm, resulting in the first dataset having k clusters (k∈{2, 3, ..., k}). Electronic device 100 can perform clustering multiple times on the first dataset by allowing the number of clusters to change each time clustering is performed.

[0065] According to the example embodiment, the clustering algorithm performed on the first dataset can be one of various clustering algorithms where the number of clusters can be set as a parameter (i.e., can be hyperparameterized). As an example, the clustering algorithm could be a hierarchical clustering algorithm that stratifies each dataset based on at least one feature included in it, according to similarity. Since the hierarchical clustering algorithm computes an initial distance matrix and repeats the process of merging clusters within closer distances or separating them into sub-clusters, this algorithm may require very little computation. As another example, the clustering algorithm could be a K-means clustering algorithm that clusters each dataset based on centroids, according to at least one feature included in it. The K-means clustering algorithm assigns data points to the nearest clusters based on centroids selected in the initial stage, and repeats the clustering process by computing the centroids of the clusters until they have the same centroids. Because the repeated computations can be processed in parallel, the algorithm can perform faster computations on large-scale datasets.

[0066] Then, the electronic device 100 can determine the optimal number of clusters for the first dataset based on the results of multiple clustering operations and an index used to evaluate the results of the clustering. The index may include, for example, a first index, a second index, and a third index, and the electronic device 100 can calculate each of the first index, the second index, and the third index based on the number of clusters based on the results of multiple clustering operations. The index based on the number of clusters may refer to an evaluation index that assesses the results of clustering operations performed with different k values, and may be presented, for example, as a graph. However, the type of index calculated based on the results of multiple clustering operations performed by the electronic device 100 is not limited to three types of indices, and those skilled in the art will understand, based on the example embodiment, that more types of indices can be calculated.

[0067] According to an example embodiment, each of the first, second, and third indices may include at least one of the Silhouette index technique, the Calinski-Harabasz index technique, and the Davies-Bouldin index technique, which evaluate the clustering results based on the cohesion of each cluster and the separation between clusters for each result of clustering performed according to the clustering algorithm. The aforementioned techniques have different procedures for calculating cohesion and separation. For example, according to the Silhouette index technique, cohesion can be calculated based on the average distance between each data point and all other data points within the same cluster, and separation can be calculated based on the average distance between each cluster and all data points within the next cluster closest to each cluster. On the other hand, according to the Calinski-Harabasz index technique, cohesion can be calculated based on the square of the average distance between each data point and the center of its corresponding cluster, and separation can be calculated based on the square of the average distance between the center of each cluster and the center of the entire dataset. Furthermore, according to the Davies-Bouldin index technique, cohesion is calculated based on the average distance between each data point and the center of its corresponding cluster, and separation can be calculated based on the average distance between the centers of each cluster.

[0068] Then, the electronic device 100 can determine the optimal number of clusters for the first dataset based on a first number of clusters corresponding to the maximum or minimum value of a first exponent of the calculated number of clusters, a second number of clusters corresponding to the maximum or minimum value of a second exponent of the calculated number of clusters, and a third number of clusters corresponding to the maximum or minimum value of a third exponent of the calculated number of clusters. Here, each of the maximum or minimum values ​​of the first, second, and third exponents can correspond to the optimal value among the maximum and minimum values ​​set based on the characteristics of the exponent. For example, when the exponent is the Silhouette exponent technique or the Calinski-Harabasz exponent technique, the maximum value can correspond to the optimal value, and when the exponent is the Davies-Bouldin exponent technique, the minimum value can correspond to the optimal value. However, the optimal value for each exponent is not limited to the maximum or minimum value and can be set to other appropriate values ​​according to the characteristics of the exponent.

[0069] According to an example embodiment, the optimal number of clusters for the first dataset can be set as the overlap value between the first number of clusters, the second number of clusters, and the third number of clusters. That is, as a result of evaluation using three indices, the number of clusters where the maximum or minimum values ​​intersect can be set as the optimal number of clusters. For example, when the first and second number of clusters are 9 and the third number of clusters is 8, the optimal number of clusters can be set to 9. Therefore, high-performance clustering can be performed by setting the number of clusters that are evaluated as having optimal values ​​according to various index techniques as the optimal number of clusters for the first dataset. According to other example embodiments, when the first number of clusters, the second number of clusters, and the third number of clusters have different values, their average can be set as the optimal number of clusters for the first dataset.

[0070] refer to Figure 5 This illustrates a process 500 for setting the optimal number of clusters for dataset 1, indicated by reference numeral 422. Electronic device 100 can perform clustering multiple times for dataset 1, indicated by reference numeral 422, which has k clusters. As described above, clustering can be performed multiple times with different numbers of clusters having different k values. Electronic device 100 can then calculate each of a first exponent 522, a second exponent 524, and a third exponent 526 based on the k value, and can determine the optimal number of clusters for dataset 1, indicated by reference numeral 422, based on this.

[0071] The electronic device 100 can repeat the same process for each dataset in a predetermined number of datasets, and according to the example embodiment, the repetition can be performed sequentially or simultaneously in parallel. Traditionally, operator intervention is required to reduce the number of clusters by identifying and manually merging content considered similar data. However, according to this disclosure, since the reference number of clusters can be set to a value collected through a repetitive process (such as the process described above using clustering algorithms and indices) without operator intervention, more efficient and objective classification of data can be achieved.

[0072] Figure 6a This is a graph showing the first index based on the number of clusters according to an example embodiment.

[0073] Here, the first index can be the Silhouette index. Graph 600 shows the Silhouette index evaluated at each k value, where k is the number of clusters ranging from 2 to 30, with the number of clusters k on the x-axis and the Silhouette index on the y-axis. SI JSThe graph shows the Silhouette index calculated using Jensen-Shannon divergence, which measures the similarity between data points based on a probability distribution, and SI EUC The graph illustrates the Silhouette index calculated using Euclidean distance, which is used to calculate the distance between data points based on the straight-line distance between them. In the case of the Silhouette index technique, considering that the maximum value is the optimal value, for SI... JS Curve graph and SI EUC For both curves, the number of clusters corresponding to the optimal value of the first index can be 7.

[0074] Figure 6b This is a graph showing the second index based on the number of clusters according to an example embodiment.

[0075] Here, the second index can be the Calinski-Harabasz index. Figure 610 shows the Calinski-Harabasz index (CHI) evaluated at each k value, where k is the number of clusters ranging from 2 to 30, with the number of clusters k on the x-axis and the Calinski-Harabasz index on the y-axis. In the case of the Calinski-Harabasz index technique, considering the maximum value as the optimal value, the number of clusters corresponding to the optimal value of the second index can be 7, and... Figure 6a The same as shown.

[0076] Figure 6c This is a graph showing the third index based on the number of clusters according to an example embodiment.

[0077] Here, the third index can be the Davies-Bouldin index. Figure 620 shows the Davies-Bouldin index (DBI) evaluated at each k value, where k is the number of clusters ranging from 2 to 30, with the number of clusters k on the x-axis and the Davies-Bouldin index on the y-axis. In the case of the Davies-Bouldin index technique, considering the minimum value as the optimal value, the number of clusters corresponding to the optimal value of the third index can be 7, which is... Figure 6a and Figure 6b The same as shown.

[0078] In other words, Figures 6a to 6c In the example shown, the optimal number of clusters for the target data, determined based on the calculated index, can be 7.

[0079] Figure 7 This is a diagram illustrating the process of setting the number of reference clusters according to an example embodiment.

[0080] According to an example embodiment, electronic device 100 can set the average of the optimal number of clusters for each dataset in a predetermined dataset as the reference number of clusters. Figure 7 The electronic device 100 can reference each of the datasets 1 (indicated by reference numeral 422), 2 (indicated by reference numeral 424), and n (indicated by reference numeral 426) repeatedly. Figure 4 and Figure 5 The process described in detail is to determine the optimal number of clusters. When K1 is the optimal number of clusters corresponding to dataset 1 indicated by reference numeral 422, and K2 is the optimal number of clusters corresponding to dataset 2 indicated by reference numeral 424, and K... n When the optimal number of clusters corresponds to the dataset n indicated by reference numeral 426, the reference number of clusters set for performing the final clustering on multiple features can be K1, K2, and K. n The average value. In other words, the electronic device 100 can perform clustering on multiple features with a more appropriate number of features by setting an optimal reference number of clusters corresponding to the optimal number of clusters for each dataset generated by sampling multiple features.

[0081] The electronic device according to the above example embodiments may include a processor, a memory for storing program data and executing the stored program data, a permanent memory such as a disk drive, a communication port for communicating with external devices, and a user interface device such as a touch panel, buttons, and keypads. Methods implemented by software modules or algorithms can be stored as computer-readable code or program instructions executable by a processor in a computer-readable recording medium. Here, the computer-readable recording medium may be a magnetic storage device (such as a read-only memory (ROM), random access memory (RAM), floppy disk, and hard disk) or an optical storage device (such as a CD-ROM and a digital versatile optical disc (DVD)). The computer-readable recording medium can be distributed across a computer system connected to a network, allowing the computer-readable code to be stored and executed in a distributed manner. The medium can be read by a computer, stored in memory, and executed by a processor.

[0082] The example embodiments can be described in terms of functional block components and various processing operations. Such functional blocks can be implemented by any number of hardware and / or software components configured to perform specified functions. For example, the example embodiments may employ direct circuit configurations, such as memory, processors, logic circuits, and lookup tables, which can perform various functions by controlling one or more microprocessors or other control devices. Similar elements can be implemented by software programming or software elements, and the example embodiments can be implemented by programming or scripting languages ​​such as C, C++, Java, and assemblers, including various algorithms implemented by combinations of data structures, procedures, routines, or other programming configurations. Functional aspects can be implemented by algorithms executed by one or more processors. Furthermore, the example embodiments may employ related techniques for, for example, electronic environment setup, signal processing, and / or data processing. The terms “mechanism,” “element,” “device,” and “configuration” can be used broadly and are not limited to mechanical and physical components. For example, these terms may include the meaning of a series of software routines associated with a processor.

[0083] The above example embodiments are merely examples, and other embodiments may be implemented within the scope of the appended claims.

Claims

1. An electronic device comprising: transceiver; processor; as well as One or more memories, configured to store one or more instructions. Wherein, one or more instructions are configured to, when executed, allow the processor to: Acquire multiple battery charging data from multiple vehicles; Multiple features of the battery charging mode are generated based on the multiple battery charging data. The multiple features are sampled and classified into a predetermined number of datasets; The number of clusters corresponding to each dataset in the predetermined number of datasets is determined based on a clustering algorithm and an index used to evaluate the clustering results. The reference cluster number corresponding to the multiple features is set based on the optimal cluster number for each dataset in the predetermined number of datasets; and The process involves clustering the multiple features into clusters with the same number of reference clusters.

2. The electronic device according to claim 1, wherein, The multiple battery charging data includes the SoC of each of the multiple batteries included in the multiple vehicles as data collected during a predetermined time period.

3. The electronic device according to claim 2, wherein, The one or more instructions are configured to, when executed, enable the processor to: One or more charging segments are identified based on the first battery charging data of the first vehicle among the multiple battery charging data of the multiple vehicles. For each of the one or more charging segments, identify the battery SoC at the start of charging and the battery SoC at the end of charging; as well as Generate a first feature indicating the frequency of one or more charging segments corresponding to the range of the battery SoC at the start of charging and the range of the battery SoC at the end of charging.

4. The electronic device according to claim 1, wherein, Each of the predetermined number of datasets includes at least one feature, and The number of the at least one feature is determined in proportion to the number of the plurality of features.

5. The electronic device according to claim 1, wherein, The one or more instructions are configured to, when executed, enable the processor to: Clustering is performed multiple times on a first dataset within the predetermined number of datasets, such that the first dataset has different numbers of clusters according to the clustering algorithm; as well as The optimal number of clusters for the first dataset is determined based on the results of multiple clustering operations and an index used to evaluate the results of the clustering.

6. The electronic device according to claim 5, wherein, The clustering algorithm includes at least one of a hierarchical clustering algorithm that stratifies at least one feature included in each dataset based on similarity and a K-means clustering algorithm that clusters the at least one feature included in each dataset based on centroid.

7. The electronic device according to claim 5, wherein, The indices include a first index, a second index, and a third index, and The one or more instructions are configured to, when executed, enable the processor to: Based on the results of multiple clustering operations, each of the first index, the second index, and the third index based on the number of clusters is calculated; as well as The optimal number of clusters for the first dataset is determined based on the first number of clusters corresponding to the maximum or minimum value of the first index of the calculated number of clusters, the second number of clusters corresponding to the maximum or minimum value of the second index of the calculated number of clusters, and the third number of clusters corresponding to the maximum or minimum value of the third index of the calculated number of clusters.

8. The electronic device according to claim 7, wherein, Each of the first, second, and third indices includes at least one of the Silhouette index technique, Calinski-Harabasz index technique, and Davies-Bouldin index technique, which evaluate the clustering results based on the cohesion of each cluster and the separation between clusters for each result of clustering performed according to the clustering algorithm.

9. The electronic device according to claim 1, wherein, The one or more instructions are configured, when executed, to allow the processor to set the average of the optimal cluster number for each dataset in the predetermined number of datasets as the reference cluster number.

10. A method for clustering data about battery charging patterns, the method being performed by an electronic device and comprising: Acquire multiple battery charging data from multiple vehicles; Multiple features of the battery charging mode are generated based on the multiple battery charging data. The multiple features are sampled and classified into a predetermined number of datasets; The number of clusters corresponding to each dataset in the predetermined number of datasets is determined based on a clustering algorithm and an index used to evaluate the clustering results. The reference cluster number corresponding to the multiple features is set based on the optimal cluster number for each dataset in the predetermined number of datasets; as well as The process involves clustering the multiple features into clusters with the same number of reference clusters.

11. A non-transitory computer-readable recording medium storing a program for performing the method of claim 10 on a computer.

Images