Data processing method and related device

Abstract

The invention provides a data processing method and a related device. The method comprises the following steps: determining K reconstructed phase spaces of a target time sequence under S time scales and D embedded dimensions; constructing K graph networks corresponding to the K reconstructed phase spaces; determining K graph network features corresponding to the K graph networks; under the same embedding dimension, determining time scale change trend characteristics of the target time sequence according to the change trend of the K graph network characteristics for the time scale; under the same time scale, according to the K graph network features for the change trend of the embedded dimension, determining the change trend feature of the embedded dimension of the target time sequence; and according to the combined change trend of the K graph network features for the time scale and the embedded dimension, determining the spatio-temporal scale change trend features of the target time sequence. Through the method, when the time scale and the embedded dimension change, the corresponding change of the time sequence can be evaluated, and the comprehensiveness and the accuracy of time sequence processing are ensured.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular, to a data processing method and related devices. Background Art

[0002] In today's information society, massive amounts of time series data are pouring out, such as stock prices, meteorological data, biomedical signals, etc., which have become important research objects in related industries. Since the time series output by the evolution of the complex world usually has the characteristics of high dimension, nonlinearity and high dynamics, the quantitative analysis method of time series in related technologies is relatively simple, especially in capturing the evolution of the intrinsic dynamic characteristics of time series. Summary of the invention

[0003] The embodiments of the present application at least provide a data processing method and related device, which can evaluate the corresponding changes in the time series when the time scale and embedding dimension change, that is, it can effectively capture the intrinsic dynamic evolution characteristics of the time series, thereby ensuring the comprehensiveness and accuracy of the time series processing.

[0004] In a first aspect, the present application provides a data processing method, comprising: Determine K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions, the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions correspond to the k-th reconstructed phase space in the K reconstructed phase spaces, where S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1; Construct K graph networks with one-to-one correspondence between K reconstructed phase spaces; Determine K graph network features corresponding one-to-one to the K graph networks; Under the same embedding dimension, the time scale change trend characteristics corresponding to the target time series are determined according to the change trend of the time scale of the K graph network features; under the same time scale, the embedding dimension change trend characteristics corresponding to the target time series are determined according to the change trend of the embedding dimension of the K graph network features; according to the joint change trend of the time scale and embedding dimension of the K graph network features, the spatiotemporal scale change trend characteristics corresponding to the target time series are determined.

[0005] In a second aspect, the present application further provides a data processing device, comprising: The first determination unit is used to determine K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions, the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions correspond to the k-th reconstructed phase space in the K reconstructed phase spaces, wherein S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1; A construction unit, used to construct K graph networks corresponding to the K reconstructed phase spaces one by one; A second determining unit is used to determine K graph network features corresponding one-to-one to the K graph networks; The third determination unit is used to determine the time scale change trend characteristics corresponding to the target time series according to the change trend of the time scale of the K graph network features under the same embedding dimension; determine the embedding dimension change trend characteristics corresponding to the target time series according to the change trend of the embedding dimension of the K graph network features under the same time scale; determine the spatiotemporal scale change trend characteristics corresponding to the target time series according to the joint change trend of the time scale and the embedding dimension of the K graph network features.

[0006] In a third aspect, the present application also provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and when the machine-readable instructions are executed by the processor, the data processing method provided in the present application is executed.

[0007] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the data processing method provided by the present application is executed.

[0008] In a fifth aspect, the present application also provides a computer program product, including a computer program, which executes the data processing method provided by the present application when executed by a processor.

[0009] In summary, the present application provides a data processing method and related devices, the method comprising: determining K reconstructed phase spaces of a target time series under S time scales and D embedding dimensions, the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions corresponding to the k-th reconstructed phase space in the K reconstructed phase spaces, wherein S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1; constructing K reconstructed phase spaces; K graph networks corresponding to each other in space; determine K graph network features corresponding to each of the K graph networks; under the same embedding dimension, determine the time scale change trend features corresponding to the target time series according to the change trend of the K graph network features for the time scale; under the same time scale, determine the embedding dimension change trend features corresponding to the target time series according to the change trend of the K graph network features for the embedding dimension; determine the space-time scale change trend features corresponding to the target time series according to the joint change trend of the K graph network features for the time scale and the embedding dimension. Through the above method, the corresponding changes of the time series can be evaluated when the time scale and the embedding dimension change, that is, the intrinsic dynamic evolution characteristics of the time series can be effectively captured, thereby ensuring the comprehensiveness and accuracy of the time series processing.

[0010] Other advantages of the present application will be explained in more detail in conjunction with the following description and drawings.

[0011] It should be understood that the above description is only an overview of the technical solution of the present application, so that the technical means of the present application can be generally understood and then implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically described below by way of example. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments that conform to the present application and are used together with the specification to illustrate the technical solutions of the present application. It should be understood that the drawings only illustrate certain embodiments of the present application and should not be regarded as limiting the scope of protection. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work. Moreover, the same reference numerals are used to represent the same components throughout the drawings. In the drawings: Figure 1 A method flow chart of a data processing method provided in an embodiment of the present application; Figure 2 A schematic diagram of a data processing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0013] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0014] In the description of the embodiments of the present application, it should be understood that terms such as "including" or "having" are intended to indicate the presence of disclosed features, numbers, steps, behaviors, components, parts, or a combination thereof in the present specification, and do not exclude the possibility of the presence of one or more other features, numbers, steps, behaviors, components, parts, or a combination thereof.

[0015] Unless otherwise specified, “ / ” means or. For example, A / B can mean A or B. The “and / or” in this article is merely a way to describe the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0016] The terms "first", "second", etc. are used only to distinguish the same or similar technical features for the convenience of description, and should not be understood as indicating or implying the relative importance or quantity of these technical features. Thus, the features defined by "first", "second", etc. may explicitly or implicitly include one or more of such features. In the description of the embodiments of the present application, unless otherwise specified, the term "plurality" means two or more than two.

[0017] Since the time series output by the evolution of the complex world are usually high-dimensional, nonlinear and highly dynamic, the quantitative analysis methods of time series in related technologies are relatively simple, especially in capturing the intrinsic dynamic evolution characteristics of time series. There are serious limitations, and the evolution laws of the dynamic characteristics of time series at different time scales and different embedded dimensions are ignored.

[0018] In view of this, the present application provides a data processing method and related devices, which can evaluate the corresponding changes in the time series when the time scale and embedding dimension change, that is, can effectively capture the intrinsic dynamic evolution characteristics of the time series, thereby ensuring the comprehensiveness and accuracy of the time series processing.

[0019] The data processing method provided in the embodiments of the present application can be implemented by a computer device, which can be a terminal device or a server, wherein the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. Terminal devices include but are not limited to mobile phones, computers, intelligent voice interaction devices, smart home appliances, vehicle-mounted terminals, aircraft, etc. Terminal devices and servers can be directly or indirectly connected via wired or wireless communication, and this application does not limit this.

[0020] The data processing method provided by the present application is described below through a method embodiment. Figure 1 As shown, Figure 1 A method flow chart of a data processing method provided in an embodiment of the present application, wherein the aforementioned computer device may be a server, and the method includes: S101. Determine K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions, the i-th time scale among the S time scales and the j-th embedding dimension among the D embedding dimensions correspond to the k-th reconstructed phase space among the K reconstructed phase spaces, where S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1.

[0021] A time series refers to data arranged in chronological order. In this embodiment, the target time series refers to the time series obtained for the target object. For example, when the target object is stock, the target time series may be the stock price; when the target object is weather, the target time series may be meteorological data; when the target object is the human body, the target time series may be a biomedical signal.

[0022] It should be noted that, since the target time series is data obtained for the target object and arranged in time order, the target time series is a one-dimensional sequence, and x can be used to represent the target time series.

[0023] In order to analyze the target time series x in terms of both time scale and embedding dimension, the server can determine K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions. For the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions, the i-th time scale represents any time scale in the S time scales, the j-th embedding dimension represents any embedding dimension in the D embedding dimensions, and the k-th reconstructed phase space in the K reconstructed phase spaces corresponds to the i-th time scale and the j-th embedding dimension, that is, it can be used to represent the K reconstructed phase spaces. In practical applications, the preferred maximum number of time scales is 10-15, and the preferred maximum number of embedding dimensions is 3-10.

[0024] That is to say, by determining the K reconstructed phase spaces corresponding to the target time series and S time scales and D embedding dimensions respectively, the one-dimensional target time series is expanded to the high-dimensional phase space, so that the information contained in the target time series can be extracted more effectively, laying the foundation for the subsequent steps to analyze the changing trends of the target time series relative to the two dimensions of time scale and embedding dimension.

[0025] In a possible implementation, determining K reconstructed phase spaces of the target time series at S time scales and D embedding dimensions in S101 includes: By performing multi-scale coarse-graining processing on the target time series at S time scales, S coarse-grained time series corresponding one-to-one between the target time series and the S time scales are determined; By reconstructing S coarse-grained time series under preset delay times and D embedding dimensions, K reconstructed phase spaces are obtained.

[0026] Specifically, the server can first perform multi-scale coarse-graining processing on the target time series at S time scales to determine S coarse-grained time series that correspond one-to-one to the target time series and the S time scales. Multi-scale coarse-graining processing is a way of dividing the time series on the time axis into multiple time windows of unequal lengths. This processing can reflect the complexity of the time series at different time scales. When x is used to represent the target time series, it can be used to represent S coarse-grained time series. In practical applications, the preferred multi-scale coarse-graining processing can be mean coarse-graining or fine composite mean coarse-graining, and the preferred maximum number of time scales is 10-15.

[0027] After obtaining S coarse-grained time series, the server can reconstruct the S coarse-grained time series under preset delay time and D embedding dimensions to obtain K reconstructed phase spaces, wherein the embedding dimension refers to the number of dimensions used to represent the system state in the reconstructed phase space, the delay time refers to the time interval between different state variables when constructing the phase space vector, and the reconstruction process refers to mapping the time series to a high-dimensional phase space. When used to represent S coarse-grained time series, it can be used to represent K reconstructed phase spaces. In practical applications, the preferred reconstruction process can be a delay time reconstruction method, the preferred delay time calculation method can be an autocorrelation method, the preferred delay time parameter range is [1,20], and the preferred maximum number of embedding dimensions is 3-10. The use of delay time and autocorrelation method to reconstruct the phase space has the beneficial effects of high computational efficiency and flexible operation.

[0028] In a possible implementation, the method further includes: Get the initial time series to be processed; The initial time series is preprocessed to obtain the target time series.

[0029] Specifically, the initial time series refers to the initialized time series obtained for the target object. The server can obtain the initial time series to be processed. In order to ensure subsequent processing, after obtaining the initial time series, the server can preprocess the initial time series. The preprocessing may include steps such as denoising, so as to obtain the target time series.

[0030] S102. Construct K graph networks corresponding to the K reconstructed phase spaces one by one.

[0031] Graph network refers to a deep learning model used to process graph structured data. Graph network can capture the dependencies of the graph through information transmission between nodes.

[0032] In order to better process the changes of the K reconstructed phase spaces determined in S101 relative to the two dimensions of time scale and embedding dimension, the server can construct K graph networks that correspond one to one to the K reconstructed phase spaces. On the premise that they can be used to represent the K reconstructed phase spaces, they can be used to represent the K graph networks.

[0033] In other words, the server can use the graph network to process and analyze the reconstructed phase space data. The reconstructed phase space is obtained by mapping the target time series into a high-dimensional space. By using the graph network, the server can mine the complex patterns and structures in the reconstructed phase space.

[0034] In a possible implementation, constructing K graph networks in one-to-one correspondence with K reconstructed phase spaces in S102 includes: For the k-th reconstructed phase space, determine N nodes corresponding one-to-one to the N phase points included in the k-th reconstructed phase space, where N is an integer and N>1; According to the relationship between the N phase points, the weight values ​​of the node edges of the N nodes are determined; Based on the N nodes and the weight values ​​of the node edges between the N nodes, a graph network corresponding to the k-th reconstructed phase space is constructed.

[0035] Specifically, when the server constructs K graph networks that correspond one-to-one to the K reconstructed phase spaces, the nodes in the graph networks can correspond one-to-one to the phase points in the reconstructed phase space.

[0036] The following uses the k-th reconstructed phase space as an example to explain how to construct the reconstructed phase space. The k-th reconstructed phase space refers to any space among the K reconstructed phase spaces.

[0037] When the kth reconstructed phase space includes N phase points, the server can first determine the N nodes corresponding to the N phase points one by one; then determine the weight values ​​of the node edges of the N nodes based on the relationship between the N phase points, that is, the weight values ​​of the node edges can be determined by the relationship between the corresponding phase points in the reconstructed phase space, which can be used to represent the connection between the th node and the th node in the graph network, where the th node and the th node are any one of the N nodes. In practical applications, the preferred calculation method can be linear and nonlinear mapping based on the distance in the phase point space.

[0038] After determining the weight values ​​of the N nodes and the edges connecting the N nodes, the server may construct a graph network corresponding to the kth reconstructed phase space based on the weight values ​​of the N nodes and the edges connecting the N nodes.

[0039] In a possible implementation, constructing K graph networks corresponding to K reconstructed phase spaces in one-to-one correspondence in S102 includes: For the k-th reconstructed phase space, determine M nodes corresponding one-to-one to the M state degenerate modes included in the k-th reconstructed phase space, where M is an integer, M>1; According to the transition probabilities between the M state degenerate modes, the weight values ​​of the node links of the M nodes are determined; Based on the M nodes and the weight values ​​of the node edges between the M nodes, a graph network corresponding to the k-th reconstructed phase space is constructed.

[0040] Specifically, when the server constructs K graph networks that correspond one-to-one to K reconstructed phase spaces, the nodes in the graph networks can correspond one-to-one to the state degenerate modes in the reconstructed phase space. The state degenerate modes refer to different physical states with the same energy or propagation constant in the reconstructed phase space.

[0041] The following uses the k-th reconstructed phase space as an example to explain how to construct the reconstructed phase space. The k-th reconstructed phase space refers to any space among the K reconstructed phase spaces.

[0042] When the kth reconstructed phase space includes M state degenerate modes, the server can first determine the M nodes corresponding to the M state degenerate modes included in the kth reconstructed phase space; then the weight values ​​of the node edges of the M nodes can be determined according to the transition probabilities between the M state degenerate modes, that is, the weight values ​​of the node edges can be determined by the transition probabilities between the corresponding state degenerate modes in the reconstructed phase space, which can be used to represent the weight values ​​of the node edges corresponding to the th node and the th node in the graph network, where the th node and the th node are any one of the M nodes. In practical applications, the weight values ​​of the node edges can also be determined by the phase point sorting mode of the reconstructed phase space. Based on the reconstructed phase space, by comparing the amplitudes of the components in the state vector, each state vector is encoded in an ordinal mode, and the weight values ​​of the node edges are obtained by the distances encoded by the ordinal modes of different nodes.

[0043] In a possible implementation, the distances encoded by different node ordinal patterns may be Hamming distance, Kullback-Leibler divergence, Jaccard similarity coefficient, or Levenshtein distance, and the relationship between the distances encoded by different node ordinal patterns and the weight values ​​of the node links may be inversely proportional.

[0044] After determining the weight values ​​of the M nodes and the edges connecting the M nodes, the server may construct a graph network corresponding to the kth reconstructed phase space based on the weight values ​​of the M nodes and the edges connecting the M nodes.

[0045] S103. Determine K graph network features corresponding one-to-one to the K graph networks.

[0046] After determining the K graph networks in S102, the server may determine K graph network features corresponding one-to-one to the K graph networks. The graph network features refer to a series of information reflected by the corresponding graph networks. On the premise of being used to represent the K graph networks, they can be used to represent the K graph network features.

[0047] After obtaining K graph network features, since the K graph network features can reflect the information in the K graph networks, the K graph networks correspond to K reconstructed phase spaces, and the K reconstructed phase spaces are determined under S time scales and D embedding dimensions of the target time series, the K graph network features can reflect the relevant information of the target time series at different time scales and different embedding dimensions. In this regard, the server can use the K graph network features as a type of output for processing time series based on the two dimensions of time scale and embedding dimension.

[0048] In a possible implementation, the graph network features may include network density, global network efficiency, global clustering coefficient, network entropy, recursion ratio, average diagonal length, and average vertical line length of the corresponding graph network.

[0049] Specifically, the network density of a graph network refers to the density of the edges between nodes in the network, the global network efficiency refers to the average efficiency of information transmission between any two nodes in the graph network, the global clustering coefficient is an indicator used to evaluate the degree of node aggregation in the graph network, the recursive ratio refers to the ratio of the distance between any two nodes less than a given threshold, the network entropy is an indicator used to measure the degree of randomness of the network node degree distribution, the average diagonal length refers to the average shortest path length between any two nodes in the graph network, and the average vertical line length refers to the average distance between a node in the graph network and its direct neighbors. The above network characteristics quantify the complexity and randomness of the network from the perspective of nodes, edges, and the global perspective.

[0050] S104. Under the same embedding dimension, according to the changing trend of the time scale of the K graph network features, determine the time scale changing trend characteristics corresponding to the target time series; under the same time scale, according to the changing trend of the embedding dimension of the K graph network features, determine the embedding dimension changing trend characteristics corresponding to the target time series; according to the joint changing trend of the time scale and the embedding dimension of the K graph network features, determine the spatiotemporal scale changing trend characteristics corresponding to the target time series.

[0051] After K graph network features are determined in S103, for the time scale, under the same embedding dimension, the server can determine the time scale change trend features corresponding to the target time series based on the change trend of the time scale of the K graph network features. For example, under the premise that the embedding dimension is determined to be m, when the time scale changes in S time scales, the server can determine the corresponding time scale change trend features based on the S graph network features.

[0052] For the embedding dimension, at the same time scale, the server can determine the embedding dimension change trend characteristics corresponding to the target time series based on the change trend of the embedding dimension according to the K graph network features. For example, under the premise that the time scale is determined to be t, when the embedding dimension changes in D embedding dimensions, the server can determine the corresponding embedding dimension change trend characteristics based on these D graph network features.

[0053] For the time scale and embedding dimension, the server can determine the spatiotemporal scale change trend characteristics corresponding to the target time series based on the joint change trend of the time scale and the embedding dimension according to the K graph network features. For example, when the time scale changes jointly in S time scales and the embedding dimension changes jointly in D embedding dimensions, the server can determine the corresponding embedding dimension change trend characteristics based on these K graph network features.

[0054] In other words, the graph network features are processed not only from a single time scale and embedding dimension, but also from the joint changes of the two dimensions of time scale and embedding maintenance. This makes it possible to evaluate the corresponding changes in the time series when the time scale and embedding dimension change, that is, to effectively capture the intrinsic dynamic evolution characteristics of the time series, thereby ensuring the comprehensiveness and accuracy of time series processing.

[0055] In a possible implementation, in S104, under the same embedding dimension, according to the K graph network features for the time scale change trend, the time scale change trend feature corresponding to the target time series is determined, including: Under the same embedding dimension, determine the first difference or first ratio of K graph network features for different time scales, and determine the characteristic parameters of the time scale fitting curve corresponding to the K graph network features, and the characteristic parameters of the time scale fitting curve are used to identify the overall change trend of the graph network features for the time scale; Determine the time scale change trend characteristics according to the first difference and the time scale fitting curve characteristic parameters, or determine the time scale change trend characteristics according to the first ratio and the time scale fitting curve characteristic parameters; That is to say, the time scale change trend characteristics are determined by calculating the difference or ratio of the corresponding results of different time scales and the characteristic parameters of the time scale fitting curve that can reflect the overall change trend of the time scale. In practical applications, the method for determining the characteristic parameters of the time scale fitting curve can be curve fitting based on the least squares method, and the preferred fitting function type can be a linear function, a polynomial function, a hyperbolic function, a power function, a logarithmic function or an exponential function.

[0056] In S104, at the same time scale, according to the K graph network features, the embedding dimension change trend characteristics corresponding to the target time series are determined based on the embedding dimension change trend, including: Under the same time scale, determining the second difference or second ratio of K graph network features for different embedding dimensions, and determining the characteristic parameters of the embedding dimension fitting curve corresponding to the K graph network features, the characteristic parameters of the embedding dimension fitting curve are used to identify the overall change trend of the graph network features for the embedding dimension; Determine the embedding dimension change trend characteristic according to the second difference and the embedding dimension fitting curve characteristic parameter, or determine the embedding dimension change trend characteristic according to the second ratio and the embedding dimension fitting curve characteristic parameter; That is to say, the changing trend characteristics of the embedding dimension are determined by calculating the difference or ratio of the corresponding results of different embedding dimensions and the characteristic parameters of the embedding dimension fitting curve that can reflect the overall changing trend of the embedding dimension. In practical applications, the method for determining the characteristic parameters of the time scale fitting curve can be curve fitting based on the least squares method, and the preferred fitting function type can be a linear function, a polynomial function, a hyperbolic function, a power function, a logarithmic function or an exponential function.

[0057] In S104, according to the joint change trend of the time scale and the embedding dimension of the K graph network features, the temporal and spatial scale change trend characteristics corresponding to the target time series are determined, including: Determine the spatiotemporal fitting surface feature parameters corresponding to K graph network features, and the spatiotemporal fitting surface feature parameters are used to identify the joint overall change trend of the graph network features with respect to the time scale and the embedding dimension; According to the characteristic parameters of the spatiotemporal fitting surface, the trend characteristics of spatiotemporal scale changes are determined.

[0058] That is to say, the characteristics of the spatiotemporal scale change trend are determined by the spatiotemporal fitting surface characteristic parameters that can reflect the joint overall change trend of the time scale and the embedding dimension. In practical applications, the method for determining the spatiotemporal fitting surface characteristic parameters can be curve fitting based on the least squares method, and the preferred fitting surface type can be a polynomial surface.

[0059] In summary, the present application provides a data processing method, which includes: determining K reconstructed phase spaces of a target time series under S time scales and D embedding dimensions, the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions correspond to the k-th reconstructed phase space in the K reconstructed phase spaces, wherein S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1; constructing K reconstructed phase spaces K graph networks corresponding to each other one by one; determine K graph network features corresponding to each of the K graph networks one by one; under the same embedding dimension, determine the time scale change trend features corresponding to the target time series according to the change trend of the K graph network features for the time scale; under the same time scale, determine the embedding dimension change trend features corresponding to the target time series according to the change trend of the K graph network features for the embedding dimension; determine the spatiotemporal scale change trend features corresponding to the target time series according to the joint change trend of the K graph network features for the time scale and the embedding dimension. Through the above method, the corresponding changes of the time series can be evaluated when the time scale and the embedding dimension change, that is, the intrinsic dynamic evolution characteristics of the time series can be effectively captured, thereby ensuring the comprehensiveness and accuracy of the time series processing.

[0060] In the description of this specification, the description with reference to the terms "some possible embodiments", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application, and the above terms do not necessarily represent the same embodiment or example. Moreover, the described specific features, structures, materials or characteristics may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0061] About the method flow chart of the present application embodiment, some operations are described as different steps performed in a certain order. Such flow chart belongs to illustrative and non-restrictive. Some steps described in this article can be grouped together and performed in a single operation, or some steps can be divided into multiple sub-steps and can be performed in an order different from that shown in this article. Each step shown in the flow chart can be realized in any way by any circuit structure and / or tangible mechanism (for example, by software, hardware (for example, the logical function realized by processor or chip) etc. running on computer equipment and / or any combination thereof).

[0062] Those skilled in the art will appreciate that, in the method described in the above specific implementation, the writing order of each step does not mean a strict execution order, and the specific execution order of each step should be determined by its function and possible internal logic.

[0063] Based on the aforementioned Figure 1 The data processing device provided by the present application is described below through a device embodiment. Figure 2 As shown, the data processing device 200 includes: The first determining unit 201 is used to determine K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions, where the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions correspond to the k-th reconstructed phase space in the K reconstructed phase spaces, wherein S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1; A construction unit 202, configured to construct K graph networks corresponding one to one to the K reconstructed phase spaces; A second determining unit 203 is used to determine K graph network features corresponding one-to-one to the K graph networks; The third determination unit 204 is used to determine the time scale change trend characteristics corresponding to the target time series according to the change trend of the K graph network features with respect to the time scale under the same embedding dimension; determine the embedding dimension change trend characteristics corresponding to the target time series according to the change trend of the K graph network features with respect to the embedding dimension under the same time scale; determine the spatiotemporal scale change trend characteristics corresponding to the target time series according to the joint change trend of the K graph network features with respect to the time scale and the embedding dimension.

[0064] In a possible implementation manner, the first determining unit 201 is configured to: By performing multi-scale coarse-graining processing on the target time series at S time scales, S coarse-grained time series corresponding one-to-one between the target time series and the S time scales are determined; By reconstructing S coarse-grained time series under preset delay times and D embedding dimensions, K reconstructed phase spaces are obtained.

[0065] In a possible implementation, the construction unit 202 is configured to: For the k-th reconstructed phase space, determine N nodes corresponding one-to-one to the N phase points included in the k-th reconstructed phase space, where N is an integer and N>1; According to the relationship between the N phase points, the weight values ​​of the node edges of the N nodes are determined; Based on the N nodes and the weight values ​​of the node edges between the N nodes, a graph network corresponding to the k-th reconstructed phase space is constructed.

[0066] In a possible implementation, the construction unit 202 is configured to: For the k-th reconstructed phase space, determine M nodes corresponding one-to-one to the M state degenerate modes included in the k-th reconstructed phase space, where M is an integer, M>1; According to the transition probabilities between the M state degenerate modes, the weight values ​​of the node links of the M nodes are determined; Based on the M nodes and the weight values ​​of the node edges between the M nodes, a graph network corresponding to the k-th reconstructed phase space is constructed.

[0067] In a possible implementation, the graph network features include network density, global network efficiency, global clustering coefficient, network entropy, recursion ratio, average diagonal length, and average vertical line length of the corresponding graph network.

[0068] In a possible implementation, the third determining unit 204 is configured to: Under the same embedding dimension, determine the first difference or first ratio of K graph network features for different time scales, and determine the characteristic parameters of the time scale fitting curve corresponding to the K graph network features, and the characteristic parameters of the time scale fitting curve are used to identify the overall change trend of the graph network features for the time scale; Determine the time scale change trend characteristics according to the first difference and the time scale fitting curve characteristic parameters, or determine the time scale change trend characteristics according to the first ratio and the time scale fitting curve characteristic parameters; Under the same time scale, determining the second difference or second ratio of K graph network features for different embedding dimensions, and determining the characteristic parameters of the embedding dimension fitting curve corresponding to the K graph network features, the characteristic parameters of the embedding dimension fitting curve are used to identify the overall change trend of the graph network features for the embedding dimension; Determine the embedding dimension change trend characteristic according to the second difference and the embedding dimension fitting curve characteristic parameter, or determine the embedding dimension change trend characteristic according to the second ratio and the embedding dimension fitting curve characteristic parameter; Determine the spatiotemporal fitting surface feature parameters corresponding to K graph network features, and the spatiotemporal fitting surface feature parameters are used to identify the joint overall change trend of the graph network features with respect to the time scale and the embedding dimension; According to the characteristic parameters of the spatiotemporal fitting surface, the trend characteristics of spatiotemporal scale changes are determined.

[0069] In a possible implementation, the data processing device 200 further includes an acquisition unit, configured to: Get the initial time series to be processed; The initial time series is preprocessed to obtain the target time series.

[0070] It should be noted that the device in the implementation mode of the present application can implement each process of the implementation mode of the aforementioned method and achieve the same effects and functions, which will not be repeated here.

[0071] The embodiment of the present application also provides an electronic device, including: a processor, a memory and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the memory communicate through the bus. When the machine-readable instructions are executed by the processor, the following processing is performed: Determine K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions, the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions correspond to the k-th reconstructed phase space in the K reconstructed phase spaces, where S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1; Construct K graph networks with one-to-one correspondence between K reconstructed phase spaces; Determine K graph network features corresponding one-to-one to the K graph networks; Under the same embedding dimension, the time scale change trend characteristics corresponding to the target time series are determined according to the change trend of the time scale of the K graph network features; under the same time scale, the embedding dimension change trend characteristics corresponding to the target time series are determined according to the change trend of the embedding dimension of the K graph network features; according to the joint change trend of the time scale and embedding dimension of the K graph network features, the spatiotemporal scale change trend characteristics corresponding to the target time series are determined.

[0072] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the data processing method described in the above method embodiment are executed. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0073] An embodiment of the present application also provides a computer program product, including a computer program. The computer program product carries a program code. The instructions included in the program code can be used to execute the steps of the data processing method described in the above method embodiment. For details, please refer to the above method embodiment, which will not be repeated here.

[0074] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is implemented as a computer storage medium. In another optional embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0075] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and computer-readable storage medium embodiments, since they are basically similar to the method embodiments, their descriptions are simplified, and the relevant parts can be referred to the partial description of the method embodiments.

[0076] The apparatus, equipment and computer-readable storage medium provided in the embodiments of the present application correspond one-to-one to the method. Therefore, the apparatus, equipment and computer-readable storage medium also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the apparatus, equipment and computer-readable storage medium will not be repeated here.

[0077] Although the spirit and principle of the present application have been described above with reference to several specific embodiments, it should be understood that the present application is not limited to the disclosed specific embodiments, and the division of various aspects does not mean that the features in these aspects cannot be combined. The present application is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the attached claims.

Claims

1. A data processing method, characterized in that: include: Determine K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions, the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions correspond to the k-th reconstructed phase space in the K reconstructed phase spaces, wherein S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1; Constructing K graph networks corresponding to the K reconstructed phase spaces one by one; Determine K graph network features corresponding one-to-one to the K graph networks; Under the same embedding dimension, the time scale change trend characteristics corresponding to the target time series are determined according to the change trend of the K graph network features with respect to the time scale; under the same time scale, the embedding dimension change trend characteristics corresponding to the target time series are determined according to the change trend of the K graph network features with respect to the embedding dimension; the spatiotemporal scale change trend characteristics corresponding to the target time series are determined according to the joint change trend of the K graph network features with respect to the time scale and the embedding dimension.

2. The method according to claim 1, characterized in that The step of determining K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions includes: By performing multi-scale coarse-graining processing on the target time series at the S time scales, S coarse-grained time series corresponding one-to-one between the target time series and the S time scales are determined; The K reconstructed phase spaces are obtained by reconstructing the S coarse-grained time series under a preset delay time and the D embedding dimensions.

3. The method according to claim 1, characterized in that The step of constructing K graph networks corresponding to the K reconstructed phase spaces in one-to-one correspondence includes: For the k-th reconstructed phase space, determine N nodes corresponding one-to-one to the N phase points included in the k-th reconstructed phase space, where N is an integer and N>1; Determine the weight value of the node edge of the N nodes according to the relationship between the N phase points; Based on the N nodes and the weight values ​​of the node edges between the N nodes, a graph network corresponding to the k-th reconstructed phase space is constructed.

4. The method according to claim 1, characterized in that: The step of constructing K graph networks corresponding to the K reconstructed phase spaces in one-to-one correspondence includes: For the k-th reconstructed phase space, determine M nodes corresponding one-to-one to the M state degenerate modes included in the k-th reconstructed phase space, where M is an integer and M>1; Determining weight values ​​of node edges of the M nodes according to transition probabilities between the M state degenerate modes; Based on the M nodes and the weight values ​​of the node edges between the M nodes, a graph network corresponding to the k-th reconstructed phase space is constructed.

5. The method according to claim 1, characterized in that The graph network features include network density, global network efficiency, global clustering coefficient, network entropy, recursion ratio, average diagonal length and average vertical line length of the corresponding graph network.

6. The method according to claim 1, characterized in that The determining of the time scale change trend feature corresponding to the target time series according to the change trend of the K graph network features with respect to the time scale under the same embedding dimension includes: Under the same embedding dimension, determining the first difference or first ratio of the K graph network features for different time scales, and determining characteristic parameters of the time scale fitting curve corresponding to the K graph network features, wherein the characteristic parameters of the time scale fitting curve are used to identify the overall change trend of the graph network features for the time scale; Determining the time scale change trend characteristics according to the first difference and the time scale fitting curve characteristic parameters, or determining the time scale change trend characteristics according to the first ratio and the time scale fitting curve characteristic parameters; The step of determining the embedding dimension change trend feature corresponding to the target time series according to the K graph network features with respect to the embedding dimension change trend under the same time scale includes: Under the same time scale, determining the second difference or the second ratio of the K graph network features for different embedding dimensions, and determining the embedding dimension fitting curve characteristic parameters corresponding to the K graph network features, wherein the embedding dimension fitting curve characteristic parameters are used to identify the overall change trend of the graph network features for the embedding dimension; Determine the embedding dimension change trend characteristic according to the second difference and the embedding dimension fitting curve characteristic parameter, or determine the embedding dimension change trend characteristic according to the second ratio and the embedding dimension fitting curve characteristic parameter; The step of determining the temporal and spatial scale change trend characteristics corresponding to the target time series according to the joint change trend of the time scale and the embedding dimension of the K graph network features includes: Determine the spatiotemporal fitting surface feature parameters corresponding to the K graph network features, wherein the spatiotemporal fitting surface feature parameters are used to identify the joint overall change trend of the graph network features with respect to the time scale and the embedding dimension; The temporal and spatial scale change trend characteristics are determined according to the temporal and spatial fitting surface characteristic parameters.

7. The method according to claim 1, characterized in that The method further comprises: Get the initial time series to be processed; The initial time series is preprocessed to obtain the target time series.

8. A data processing device, characterized in that: include: A first determination unit is used to determine K reconstructed phase spaces of the target time series under S time scales and D embedding dimensions, wherein the i-th time scale in the S time scales and the j-th embedding dimension in the D embedding dimensions correspond to the k-th reconstructed phase space in the K reconstructed phase spaces, wherein S is an integer, S>1, i is an integer, S≥i≥1, D is an integer, D>1, j is an integer, D≥j≥1, K is an integer, K=S×D, k is an integer, K≥k≥1; A construction unit, used to construct K graph networks corresponding to the K reconstructed phase spaces one by one; A second determining unit, configured to determine K graph network features corresponding one-to-one to the K graph networks; The third determination unit is used to determine the time scale change trend characteristics corresponding to the target time series according to the change trend of the K graph network features with respect to the time scale under the same embedding dimension; determine the embedding dimension change trend characteristics corresponding to the target time series according to the change trend of the K graph network features with respect to the embedding dimension under the same time scale; determine the spatiotemporal scale change trend characteristics corresponding to the target time series according to the joint change trend of the K graph network features with respect to the time scale and the embedding dimension.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the data processing method as described in any one of claims 1 to 7 is performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the data processing method according to any one of claims 1 to 7 is executed.

11. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, performs the data processing method according to any one of claims 1 to 7.

Images