Multi-dimensional time series data dynamic evolution analysis method and device, medium and electronic equipment

Abstract

A multi-dimensional time series data dynamic evolution analysis method, device, medium and electronic equipment, the method comprises the steps of: obtaining and structuring pre-processing time series data; each of the time series data is quantified and attributed in multiple dimensions, and a three-dimensional data node containing spatial coordinates, initial weight and timestamp is constructed; based on the set of three-dimensional data nodes in a specific time window, a three-dimensional data density field is constructed which can represent the data hotspot distribution and intensity in the time window; in the three-dimensional data density field, the density peak area is identified as the core data cluster, and the cluster quality is calculated according to its density and coverage, and the association strength between different core data clusters is calculated; the sequence of the state snapshots of the continuous multiple time window data fields is serialized to form a dynamic evolution view which can show the birth and death of data hotspots, the fusion and splitting process of data clusters. The abstract multi-dimensional data is converted into intuitive three-dimensional density field.

Description

Technical Field

[0001] This invention relates to the field of data processing, and in particular to a method, apparatus, medium, and electronic device for dynamic evolution analysis of multidimensional time-series data. Background Technology

[0002] With the rapid development of information technology and the Internet of Things (IoT), massive amounts of multidimensional time-series data, characterized by complex dimensions and rapid changes over time, are being generated and accumulated daily in numerous fields such as IoT, financial markets, scientific computing, bioinformatics, and social sciences. Extracting valuable information, discovering potential patterns, and understanding the evolutionary laws from these massive, high-dimensional, and dynamic data streams has become a major challenge for the field of data science.

[0003] Traditional analytical methods often process different dimensions (such as time, space, and numerical values) separately, making it difficult to provide a comprehensive and intuitive analytical perspective. Furthermore, existing methods struggle to effectively handle similarity relationships in high-dimensional data, especially when dealing with the evolution of data clustering, where effective quantification methods are often lacking. Therefore, current technologies fail to provide profound insights into the underlying patterns within the data.

[0004] Currently, the main limitations of analytical techniques for this type of data are as follows:

[0005] 1. Limitations of Static and Low-Dimensional Presentation: Traditional analytical methods, such as statistical chart analysis, mainly rely on two-dimensional charts (such as line charts, bar charts, and scatter plot matrices) to display data. These methods are difficult to intuitively show the complex relationships between three or more dimensions, and cannot effectively present the dynamic evolution of data distribution over time. The analytical perspective is fragmented and static.

[0006] 2. Loss of high-dimensional information and poor interpretability: Although some dimensionality reduction techniques (such as PCA and t-SNE) can map high-dimensional data to a low-dimensional space (two-dimensional or three-dimensional) for visualization, some original information will inevitably be lost during the dimensionality reduction process, and the coordinate axes after dimensionality reduction often lack clear physical meaning, which is not conducive to users' understanding and interpretation of the dynamic changes of data clusters.

[0007] 3. Lack of insight into inter-cluster relationships and evolutionary paths: Existing methods mostly focus on identifying data clusters (i.e., hotspots) at specific points in time, but struggle to reveal the deep relationships between data clusters. For example, the fusion (multiple small hotspots converge into a large hotspot), splitting (a hotspot differentiates into multiple sub-hotspots) and evolutionary paths of data clusters are not well understood, especially the lack of effective models to quantify and describe the intensity of mutual influence between different data clusters.

[0008] 4. Difficulty in handling the real-time nature of data streams: Many time-series data (such as financial transaction data and sensor data) are characterized by high throughput and real-time nature. Traditional batch processing analysis methods are difficult to meet the needs of real-time tracking and analysis of the dynamic evolution of data streams.

[0009] Therefore, there is an urgent need for a new technical solution that can overcome the above-mentioned shortcomings, integrate multiple core dimensions of data into a unified analytical framework, and reveal the internal structure, evolutionary laws, and interaction forces between data clusters in an intuitive, dynamic, quantifiable, and interactive way. Summary of the Invention

[0010] In view of the above situation, it is necessary to provide a method, device, medium, and electronic device for dynamic evolution analysis of multidimensional time-series data, addressing the problems in existing technologies. This method transforms abstract multidimensional data into an intuitive three-dimensional density field and introduces a similarity gravity model. By calculating and visualizing the dynamic evolution process of the data field, it ultimately achieves a comprehensive, intuitive, and quantifiable in-depth understanding of the formation, development, fusion, splitting, and mutual influence relationships of hotspots in complex data networks.

[0011] This invention provides a method for dynamic evolution analysis of multidimensional time-series data, comprising the following steps:

[0012] S1, acquire and structure preprocess time series data, the time series data including data content, timestamps and metadata;

[0013] S2, perform multi-dimensional quantization and attribute assignment on each of the aforementioned time-series data, and construct a three-dimensional data node containing spatial coordinates, initial weights and timestamps;

[0014] S3, Based on the set of three-dimensional data nodes within a specific time window, a kernel density estimation algorithm is applied to construct a three-dimensional data density field that can characterize the distribution and intensity of data hotspots within that time window;

[0015] S4, in the three-dimensional data density field, identify the density peak region as the core data cluster, calculate its cluster quality according to its density and coverage, and calculate the correlation strength between different core data clusters based on the similarity gravity model.

[0016] S5. By repeatedly executing steps S3 and S4 over multiple consecutive time windows, a series of data field state snapshots are generated, and the snapshots are serialized to form a dynamic evolution view that can show the birth and death of data hotspots, the fusion and splitting of data clusters.

[0017] Furthermore, in the above-mentioned method for dynamic evolution analysis of multidimensional time series data, the step of acquiring and structuring preprocessing time series data includes:

[0018] The acquired time-series data is cleaned;

[0019] Perform multi-dimensional processing on the cleaned time-series data;

[0020] The processed data is stored in a structured manner;

[0021] It summarizes and converts the format of all time-series data within a preset time period.

[0022] Furthermore, in the above-mentioned method for dynamic evolution analysis of multidimensional time-series data, step S2 includes:

[0023] Based on at least three feature dimensions of the time series data, each data point is mapped to a coordinate position in three-dimensional space;

[0024] Assign an initial weight value to each data point based on at least one attribute of the data.

[0025] Furthermore, in the above-mentioned multidimensional time-series data dynamic evolution analysis method, step S3 includes:

[0026] For any grid point p in three-dimensional space, calculate its density value ρ(p,t) within the time window t. The density value is calculated using the following kernel density estimation formula:

[0027] ;

[0028] Where, p i and m i Here, K represents the coordinates and initial weight of the i-th 3D data node within the time window t, where n is the total number of 3D data nodes within the time window t, and K is the total number of 3D data nodes within the time window t. h It is a kernel function with bandwidth h;

[0029] A continuous three-dimensional data density field is constructed by calculating all grid points in three-dimensional space.

[0030] Furthermore, in the above-mentioned method for dynamic evolution analysis of multidimensional time-series data, step S4 includes:

[0031] The density values ​​in the three-dimensional data density field that are greater than the threshold ρ threshold The region was identified as a high-density region;

[0032] Cluster or segment the high-density regions, and define each independent connected high-density region as a core data cluster;

[0033] Calculate the centroid position of each of the core data clusters;

[0034] The total mass M of each core data cluster is calculated by integration, using the following formula:

[0035] ,

[0036] Where V is the volume region covered by the core data cluster;

[0037] Based on the similarity gravity model, calculate the correlation strength F between any two core data clusters A and B. AB The formula is:

[0038] ;

[0039] Among them, M A and M B denoted by , r represents the mass of core data clusters A and B, r is the distance between the centroids of the two clusters, and G is a configurable gravitational coefficient.

[0040] Furthermore, in the above-mentioned multidimensional time-series data dynamic evolution analysis method, the step of serializing the snapshots to form a dynamic evolution view that can show the birth and death of data hotspots, data cluster fusion and splitting processes includes:

[0041] By comparing the sets of core data clusters in two adjacent time windows, a correspondence between core data clusters between adjacent time windows is established based on the Euclidean distance between the centroids of the core data clusters and the overlap of their coverage areas.

[0042] Based on the correspondence, the evolutionary state of each core data cluster between adjacent time windows is determined, and the evolutionary state includes: persistence, fusion, splitting, new formation, and extinction;

[0043] Based on the evolutionary state, trajectories and event markers representing the evolutionary process of data clusters are generated in three-dimensional space, thereby forming the dynamic evolutionary view.

[0044] This invention also discloses a multidimensional time-series data dynamic evolution analysis device, comprising:

[0045] The acquisition module is used to acquire and preprocess time-series data in a structured manner. The time-series data includes data content, timestamps, and metadata.

[0046] The data node construction module is used to perform multi-dimensional quantization and attribute assignment on each piece of time-series data to construct a three-dimensional data node containing spatial coordinates, initial weights, and timestamps.

[0047] The density field construction module is used to construct a three-dimensional data density field that can characterize the distribution and intensity of data hotspots within a specific time window by applying a kernel density estimation algorithm based on the set of three-dimensional data nodes within that time window.

[0048] The identification and calculation module is used to identify density peak regions as core data clusters in the three-dimensional data density field, calculate their cluster quality based on their density and coverage, and calculate the correlation strength between different core data clusters based on the similarity gravity model.

[0049] The dynamic evolution view generation module is used to serialize multiple consecutive time window snapshots of the data field state to form a dynamic evolution view that can show the birth and death of data hotspots, the fusion and splitting of data clusters.

[0050] Furthermore, in the aforementioned multidimensional time-series data dynamic evolution analysis device, the data node construction module is used for:

[0051] Based on at least three feature dimensions of the time series data, each data point is mapped to a coordinate position in three-dimensional space;

[0052] Assign an initial weight value to each data point based on at least one attribute of the data.

[0053] The present invention also discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the multidimensional time-series data dynamic evolution analysis method described in any of the above claims.

[0054] The present invention also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the multidimensional time-series data dynamic evolution analysis method described in any of the above claims.

[0055] The technical effects of the multidimensional time-series data dynamic evolution analysis method based on a three-dimensional density field in this invention are as follows:

[0056] (1) Multidimensional attribute fusion and three-dimensional presentation: Multiple key dimensions of data (e.g., by quantization mapping to X, Y, Z coordinates) are unified into a three-dimensional spatial model, realizing a three-dimensional and holistic understanding of the data situation, overcoming the shortcomings of traditional methods with fragmented perspectives.

[0057] (2) Dynamic visualization of data evolution process: By analyzing and tracking continuous time windows, the complete life cycle of data hotspots can be displayed intuitively, including their generation, extinction, and the fusion and splitting process between data clusters, which greatly enhances the insight into the laws of data development.

[0058] (3) Quantifiable analysis of data cluster interactions: The similarity gravity model is creatively introduced to transform the abstract influence relationship between data clusters into a calculable "gravity" or "association strength", providing a brand-new quantitative analysis method for judging the association strength and influence of different data focal points.

[0059] (4) Improve analysis efficiency and interactivity: By transforming complex time series data into intuitive three-dimensional density cloud maps, motion trajectories and gravitational relationships, and providing adjustable time window parameters, the cognitive load of analysts is significantly reduced, making the judgment and decision-making on the evolution of complex data faster and more accurate. Attached Figure Description

[0060] Figure 1 A flowchart of the multidimensional time-series data dynamic evolution analysis method provided in the first embodiment of the present invention;

[0061] Figure 2 Visualization results for traditional methods;

[0062] Figure 3 This is a visualization result from an embodiment of the present invention;

[0063] Figure 4 This is a structural block diagram of the multidimensional time-series data dynamic evolution analysis device in the second embodiment of the present invention;

[0064] Figure 5 This is a schematic diagram of the structure of an electronic device in an embodiment of the present invention. Detailed Implementation

[0065] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0066] Embodiments of the present invention will become clear from the following description and accompanying drawings. In these descriptions and drawings, some specific implementations of the embodiments of the invention are specifically disclosed to illustrate some ways of carrying out the principles of the embodiments of the invention; however, it should be understood that the scope of the embodiments of the invention is not limited thereto. Rather, the embodiments of the invention include all variations, modifications, and equivalents falling within the spirit and scope of the appended claims.

[0067] Please see Figure 1 The multidimensional time-series data dynamic evolution analysis method in the first embodiment of the present invention includes steps S1 to S5.

[0068] Step S1: Obtain and structure preprocessed time series data, which includes data content, timestamps, and metadata.

[0069] First, raw time-series data related to the specific monitoring topic is obtained from various data sources, such as sensor data, IoT data, financial data, and meteorological data. This data typically contains a lot of noise and is in an unstructured format, therefore, it usually requires structured preprocessing. Specifically:

[0070] S11. Clean the acquired raw time-series data. Remove HTML tags, URL links, special characters, stop words, and noise information irrelevant to the analysis objective to ensure data purity and consistency, providing high-quality data for subsequent analysis.

[0071] S12. Perform multi-dimensional processing on the cleaned time-series data. This may include: timestamp standardization, unifying the timestamps of data from different sources into a standard format for time alignment and synchronization; feature extraction, extracting various features (such as numerical values, categories, geographical locations, etc.) from the original data to form a multi-dimensional data structure; and data normalization and standardization, normalizing the data of each dimension to ensure consistent scale of different features for subsequent analysis.

[0072] S13. Store the processed data in a structured format. Transform the processed data into a structured format to facilitate subsequent analysis and calculations. In practice, a time-series data structure can be defined, with each data instance containing the following fields: data content, data collection timestamp, numerical characteristics, category, and data source name. Meta-information for each data entry (such as data source identifier, data generation timestamp, data type, and event identifier) ​​will be stored as supplementary information for easy subsequent querying and tracking.

[0073] S14. Summarize and convert the format of all time-series data within the preset time period to form an initial data set for subsequent analysis. This data set provides the foundation for constructing the three-dimensional density field and subsequent dynamic evolution analysis. The preset time period is, for example, the past 24 hours or the past week.

[0074] Transform the data into a format suitable for subsequent analysis (such as CSV, JSON, Parquet, etc.) to facilitate importing into the analysis system for visualization or modeling. Simultaneously, perform integrity checks on the stored data to ensure that each data entry has been properly cleaned and formatted, and meets the expected structural requirements.

[0075] Step S2: Perform multi-dimensional quantization and attribute assignment on each of the time-series data to construct a three-dimensional data node containing spatial coordinates, initial weights, and timestamps.

[0076] This step aims to transform each piece of structured time-series data into geometric points that can be represented and computed in three-dimensional space. The specific steps are as follows:

[0077] S21, based on at least three feature dimensions of the time series data, map each data point to a coordinate position in three-dimensional space;

[0078] S22, assign an initial weight value to each data item based on at least one attribute of the data.

[0079] Specifically, firstly, using an appropriate model (e.g., sentiment analysis model, topic model, or other specialized feature extraction algorithm), multiple key attributes of each time-series data point are calculated and normalized to a preset interval. First:

[0080] To determine the X-coordinate, based on a key feature of the data (such as numerical attributes, sentiment tendency, or other indicators), calculate the feature value using an appropriate analysis model (such as a deep learning-based sentiment analysis model), and normalize it to a predetermined interval (such as [-1, 1] or [0, 1]), which will be used as the X-coordinate of the information point.

[0081] Determine the Y-coordinate, use topic modeling or similarity calculation to evaluate the similarity between the data and a specific topic or related features, and normalize it as the Y-coordinate;

[0082] The Z-coordinate is determined by calculating and normalizing its value based on another dimension of the data (such as the data's dissemination influence, quality assessment, or other important attributes) through a comprehensive evaluation model (considering the data source, the strength of the data flow, etc.), and thus serving as the Z-coordinate. Therefore, each time-series data point corresponds to a unique coordinate (X, Y, Z) in three-dimensional space.

[0083] Secondly, initial weight values ​​are assigned by using a pre-defined mapping function (e.g., a linear or logarithmic function) to assign an initial weight value m to each 3D data node. i This mass m i This reflects the contribution of each data point to the data field in the initial stage. Data with greater influence has a greater initial weight, thus occupying a more important position in the three-dimensional density field.

[0084] Step S3: Based on the set of three-dimensional data nodes within a specific time window, apply the kernel density estimation algorithm to construct a three-dimensional data density field that can characterize the distribution and intensity of data hotspots within that time window.

[0085] The purpose of this step is to transform the discrete cloud of data nodes into a continuous field, thereby enabling the intuitive identification of data clusters. The specific steps are as follows:

[0086] S31, For any grid point p in three-dimensional space, calculate its density value ρ(p,t) within the time window t. The density value is calculated using the following kernel density estimation formula:

[0087] ;

[0088] Where, p i and m i Here, K represents the coordinates and initial weight of the i-th 3D data node within the time window t, where n is the total number of 3D data nodes within the time window t, and K is the total number of 3D data nodes within the time window t. h It is a kernel function with bandwidth h;

[0089] S32 constructs a continuous three-dimensional data density field by calculating all grid points in three-dimensional space.

[0090] In this embodiment, the user can interactively set the length of a specific time window (e.g., 1 hour) and the step size for sliding along the time axis (e.g., 10 minutes). Both the length of the time window and the step size for sliding along the time axis are user-adjustable parameters to adapt to the analysis needs of different time scales.

[0091] In addition, it is necessary to set the bandwidth parameter h and select the kernel function K in the kernel density estimation algorithm. In this embodiment, the Gaussian kernel function is selected because it has good smoothing properties.

[0092] Using the kernel density estimation (KDE) algorithm, for any grid point p in the 3D data space, its density value ρ(p,t) within the time window t is calculated by the following formula. Among them, p i and m i These are the 3D coordinates and initial weights of the i-th data node within the time window t, where n is the total number of information points within the window, and K is the weight of the i-th data node. h It is a kernel function with bandwidth h. By calculating for all grid points in three-dimensional space, a continuous three-dimensional data density field can be constructed. The higher the density of a region, the more concentrated the data nodes are, and the higher the heat of hotspot regions.

[0093] Step S4: In the three-dimensional data density field, identify the density peak region as the core data cluster, calculate its cluster quality based on its density and coverage, and calculate the correlation strength between different core data clusters based on the similarity gravity model.

[0094] This step aims to extract structured, meaningful data focus regions from a continuous density field and perform quantitative analysis on them. The specific steps are as follows:

[0095] S41, in the three-dimensional data density field, the density value is greater than the threshold ρ threshold The region was identified as a high-density region;

[0096] S42, cluster or segment the high-density region, and define each independent connected high-density region as a core data cluster;

[0097] S43, Calculate the centroid position of each of the core data clusters;

[0098] S44, the total mass M of each core data cluster is calculated by integration, using the following formula:

[0099] ,

[0100] Where V is the volume region covered by the core data cluster;

[0101] S45, Calculate the correlation strength F between any two core data clusters A and B based on the similarity gravity model. AB The formula is:

[0102] ;

[0103] Among them, M A and M B denoted by , r represents the mass of core data clusters A and B, r is the distance between the centroids of the two clusters, and G is a configurable gravitational coefficient.

[0104] In this embodiment, a density threshold ρ is first set. threshold The process involves identifying high-density regions in the density field whose density values ​​exceed a threshold. Clustering algorithms (such as DBSCAN) or image segmentation algorithms (such as the watershed algorithm) are then applied to these high-density regions, defining each independent, connected high-density region as a core data cluster. These core data clusters represent regions with strong correlation and concentration in three-dimensional space. The centroid of each core data cluster is calculated as its center in three-dimensional space. The total mass M is calculated by integrating the volume of the high-density region covered by each core data cluster. The total mass M represents the overall importance or size of the core data cluster. The formula for its calculation is... Where V is the volume region occupied by this core data cluster.

[0105] The magnitude of the gravitational force F between any two core data clusters A and B AB According to the gravitational formula The calculation yielded the gravitational value F. AB The larger the value, the stronger the correlation and mutual influence between the two core data clusters.

[0106] S5. By repeatedly executing steps S3 and S4 over multiple consecutive time windows, a series of data field state snapshots are generated, and the snapshots are serialized to form a dynamic evolution view that can show the birth and death of data hotspots, the fusion and splitting of data clusters.

[0107] This step tracks changes in data clusters by continuously analyzing multiple time windows, ultimately forming a dynamic view that analysts can understand. Specifically, the step of serializing the snapshots to form a dynamic evolution view that shows the birth and death of data hotspots, and the merging and splitting of data clusters includes:

[0108] S51. By comparing the sets of core data clusters in two adjacent time windows, and based on the Euclidean distance of the centroids of the core data clusters and the overlap of their coverage areas, a correspondence between core data clusters in adjacent time windows is established.

[0109] S52, based on the correspondence, determine the evolution state of each core data cluster between adjacent time windows, the evolution state including: persistence, fusion, splitting, new birth and extinction;

[0110] S53, based on the evolutionary state, generate trajectories and event markers representing the evolutionary process of the data cluster in three-dimensional space, thereby forming the dynamic evolution view.

[0111] Specifically, the core data cluster sets are compared between two adjacent time windows (e.g., time t and time t+1). Based on two key indicators: the Euclidean distance between the centroids of the core data clusters and the volume overlap of high-density regions, a correspondence between the core data clusters in the two time windows is established. For example, if a core data cluster A at time t has the closest centroid distance and the highest volume overlap with a core data cluster A' at time t+1, then A is considered to have evolved into A'.

[0112] Based on the established correspondence between core data clusters, the evolutionary state of each core data cluster is determined. Typical evolutionary states include:

[0113] Continuing: A single core data cluster at time t uniquely and closely corresponds to a single core data cluster at time t+1.

[0114] Merging refers to the process where multiple core data clusters at time t are collectively mapped to a single core data cluster at time t+1.

[0115] Splitting: A single core data cluster at time t corresponds to multiple core data clusters at time t+1.

[0116] Nascent refers to the core data cluster that appears at time t+1 but has no corresponding source at time t.

[0117] Ceasing: The core data cluster at time t cannot be found to have any corresponding data at time t+1.

[0118] By serializing a series of state snapshots over time windows (including the position, mass, mutual gravitational pull, and evolutionary state of core data clusters), and rendering them using a 3D visualization engine, end users can see a dynamic view. This dynamic evolutionary view uses 3D visualization technology to showcase the birth and death of data hotspots, the merging and splitting of data clusters through dynamically changing density cloud maps, data cluster trajectories, and gravitational lines. For example, the rise and fall of the density cloud map ("nebula") represents the movement, growth, and shrinking of the "stars" of the data clusters in space, as well as their interactions connected by gravitational lines, clearly demonstrating the entire data evolution process, including the aggregation, changes, and interactions of data clusters.

[0119] In summary, this invention provides a method for dynamic evolution analysis of multidimensional time-series data based on a three-dimensional density field. This method organically integrates multiple dimensional attributes of the data (such as numerical characteristics, sentiment, themes, and influence) into a three-dimensional spatial model, achieving a three-dimensional presentation of the time-series data situation. Through time window snapshots and tracking, it achieves dynamic visualization of the life cycle of data hotspots. Furthermore, it innovatively introduces a gravity model to quantify the interactions between data clusters. This method significantly improves the depth, intuitiveness, and efficiency of time-series data analysis, providing a powerful technical tool for in-depth insights and decision support for complex data events.

[0120] like Figure 2 and Figure 3 As shown, the visualization results of the traditional method ( Figure 2 The visualizations presented in the form of independent two-dimensional statistical charts suffer from information fragmentation. In contrast, the visualizations generated by the method of this invention (…) Figure 3 This is a unified, dynamic 3D view that simultaneously displays the spatial distribution, evolutionary trajectories, and merging and splitting events between multiple data clusters. It shows the changes in clusters from the left to the right image, significantly enhancing the overall understanding and in-depth insight into complex data situations.

[0121] Please see Figure 4 The multidimensional time-series data dynamic evolution analysis device provided in the second embodiment of the present invention includes:

[0122] The acquisition module 21 is used to acquire and structure preprocess time series data, wherein the time series data includes data content, timestamps and metadata;

[0123] The data node construction module 22 is used to perform multi-dimensional quantization and attribute assignment on each piece of time-series data to construct a three-dimensional data node containing spatial coordinates, initial weights and timestamps.

[0124] Density field construction module 23 is used to construct a three-dimensional data density field that can characterize the distribution and intensity of data hotspots within a specific time window by applying a kernel density estimation algorithm based on the set of three-dimensional data nodes within that time window.

[0125] The identification and calculation module 24 is used to identify density peak regions as core data clusters in the three-dimensional data density field, calculate their cluster quality based on their density and coverage, and calculate the correlation strength between different core data clusters based on the similarity gravity model.

[0126] The dynamic evolution view generation module 25 is used to serialize multiple consecutive time window data field state snapshots to form a dynamic evolution view that can show the birth and death of data hotspots, the fusion and splitting of data clusters.

[0127] Furthermore, in the aforementioned multidimensional time-series data dynamic evolution analysis device, the data node construction module is used for:

[0128] Based on at least three feature dimensions of the time series data, each data point is mapped to a coordinate position in three-dimensional space;

[0129] Assign an initial weight value to each data point based on at least one attribute of the data.

[0130] The multidimensional time-series data dynamic evolution analysis device provided in this embodiment of the invention has the same implementation principle and technical effects as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0131] In another aspect, the present invention also proposes an electronic device, please refer to [link to relevant documentation]. Figure 5 The image shows an electronic device according to an embodiment of the present invention, including a processor 10, a memory 20, and a computer program 30 stored in the memory and executable on the processor. When the processor 10 executes the computer program 30, it implements the multidimensional time-series data dynamic evolution analysis method as described above.

[0132] The electronic device may be, but is not limited to, a personal computer, a mobile phone, or other computer equipment. In some embodiments, the processor 10 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, etc.

[0133] The memory 20 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 20 can be an internal storage unit of an electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 20 can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory 20 can include both internal and external storage units of the electronic device. The memory 20 can be used not only to store application software and various types of data installed on the electronic device, but also to temporarily store data that has been output or will be output.

[0134] Optionally, the electronic device may further include a user interface, a network interface, a communication bus, etc. The user interface may include a display, an input unit such as a keyboard, and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface. The network interface may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface), typically used to establish communication connections between the device and other electronic devices. The communication bus is used to enable communication between these components.

[0135] It should be pointed out that, Figure 5 The structure shown does not constitute a limitation on the electronic device. In other embodiments, the electronic device may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0136] The present invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the multidimensional time-series data dynamic evolution analysis method described above.

[0137] Those skilled in the art will understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequential list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system or apparatus (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from or in conjunction with such an instruction execution system or apparatus). For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit programs for use by or in conjunction with an instruction execution system or apparatus.

[0138] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0139] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0140] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0141] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for dynamic evolution analysis of multidimensional time-series data, characterized in that, Including the following steps: S1, acquire and structure preprocess time series data, the time series data including data content, timestamps and metadata; S2, perform multi-dimensional quantization and attribute assignment on each of the aforementioned time-series data, and construct a three-dimensional data node containing spatial coordinates, initial weights and timestamps; S3, Based on the set of three-dimensional data nodes within the time window, a kernel density estimation algorithm is applied to construct a three-dimensional data density field that can characterize the distribution and intensity of data hotspots within the time window; S4, in the three-dimensional data density field, identify the density peak region as the core data cluster, calculate its cluster quality according to its density and coverage, and calculate the correlation strength between different core data clusters based on the similarity gravity model. S5, by repeatedly executing steps S3 and S4 over multiple consecutive time windows, a series of data field state snapshots are generated, and the snapshots are serialized to form a dynamic evolution view showing the birth and death of data hotspots, the fusion and splitting of data clusters; Step S3 includes: For any grid point p in three-dimensional space, calculate its density value ρ(p,t) within the time window t. The density value is calculated using the following kernel density estimation formula: ； Where, p i and m i Here, K represents the coordinates and initial weight of the i-th 3D data node within the time window t, where n is the total number of 3D data nodes within the time window t, and K is the total number of 3D data nodes within the time window t. h It is a kernel function with bandwidth h; A continuous three-dimensional data density field is constructed by calculating all grid points in three-dimensional space. Step S4 includes: The density values ​​in the three-dimensional data density field that are greater than the threshold ρ threshold The region was identified as a high-density region; Cluster or segment the high-density regions, and define each independent connected high-density region as a core data cluster; Calculate the centroid position of each of the core data clusters; The total mass M of each core data cluster is calculated by integration, using the following formula: ， Where V is the volume region covered by the core data cluster; Based on the similarity gravity model, calculate the correlation strength F between any two core data clusters A and B. AB The formula is: ； Among them, M A and M B Let r be the mass of core data clusters A and B, r be the distance between the centroids of the two clusters, and G be the gravitational coefficient.

2. The method for dynamic evolution analysis of multidimensional time-series data as described in claim 1, characterized in that, The steps for acquiring and structuring preprocessing time series data include: The acquired time-series data is cleaned; Perform multi-dimensional processing on the cleaned time-series data; The processed data is stored in a structured manner; It summarizes and converts the format of all time-series data within a preset time period.

3. The method for dynamic evolution analysis of multidimensional time-series data as described in claim 1, characterized in that, Step S2 includes: Based on at least three feature dimensions of the time series data, each data point is mapped to a coordinate position in three-dimensional space; Assign an initial weight value to each data point based on at least one attribute of the data.

4. The method for dynamic evolution analysis of multidimensional time-series data as described in claim 1, characterized in that, The step of serializing the snapshots to form a dynamic evolutionary view showing the generation and destruction of data hotspots, and the fusion and splitting of data clusters includes: By comparing the sets of core data clusters in two adjacent time windows, a correspondence between core data clusters between adjacent time windows is established based on the Euclidean distance between the centroids of the core data clusters and the overlap of their coverage areas. Based on the correspondence, the evolutionary state of each core data cluster between adjacent time windows is determined, and the evolutionary state includes: persistence, fusion, splitting, new formation, and extinction; Based on the evolutionary state, trajectories and event markers representing the evolutionary process of data clusters are generated in three-dimensional space, thereby forming the dynamic evolutionary view.

5. A device for dynamic evolution analysis of multidimensional time-series data, characterized in that, include: The acquisition module is used to acquire and preprocess time-series data in a structured manner. The time-series data includes data content, timestamps, and metadata. The data node construction module is used to perform multi-dimensional quantization and attribute assignment on each piece of time-series data to construct a three-dimensional data node containing spatial coordinates, initial weights, and timestamps. The density field construction module is used to construct a three-dimensional data density field that can characterize the distribution and intensity of data hotspots within the time window by applying a kernel density estimation algorithm based on the set of three-dimensional data nodes within the time window. The identification and calculation module is used to identify density peak regions as core data clusters in the three-dimensional data density field, calculate their cluster quality based on their density and coverage, and calculate the correlation strength between different core data clusters based on the similarity gravity model. The dynamic evolution view generation module is used to serialize multiple consecutive time window data field state snapshots to form a dynamic evolution view that shows the birth and death of data hotspots, the fusion and splitting of data clusters. The density field construction module is used for: For any grid point p in three-dimensional space, calculate its density value ρ(p,t) within the time window t. The density value is calculated using the following kernel density estimation formula: ； Where, p i and m i Here, K represents the coordinates and initial weight of the i-th 3D data node within the time window t, where n is the total number of 3D data nodes within the time window t, and K is the total number of 3D data nodes within the time window t. h It is a kernel function with bandwidth h; A continuous three-dimensional data density field is constructed by calculating all grid points in three-dimensional space. The identification and calculation module is used for: The density values ​​in the three-dimensional data density field that are greater than the threshold ρ threshold The region was identified as a high-density region; Cluster or segment the high-density regions, and define each independent connected high-density region as a core data cluster; Calculate the centroid position of each of the core data clusters; The total mass M of each core data cluster is calculated by integration, using the following formula: ， Where V is the volume region covered by the core data cluster; Based on the similarity gravity model, calculate the correlation strength F between any two core data clusters A and B. AB The formula is: ； Among them, M A and M B Let r be the mass of core data clusters A and B, r be the distance between the centroids of the two clusters, and G be the gravitational coefficient.

6. The multidimensional time-series data dynamic evolution analysis device as described in claim 5, characterized in that, The data node construction module is used for: Based on at least three feature dimensions of the time series data, each data point is mapped to a coordinate position in three-dimensional space; Assign an initial weight value to each data point based on at least one attribute of the data.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method for dynamic evolution analysis of multidimensional time-series data as described in any one of claims 1 to 4.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the computer program, it implements the multidimensional time-series data dynamic evolution analysis method as described in any one of claims 1 to 4.

Images