Data processing device, data processing method, and data processing program

Abstract

A time-series data acquisition part (101) acquires time-series data in which a data value for each unit time is described over a plurality of unit times, and in which a cycle corresponding to a multiple value of the unit time is present in the transition of the data value for each unit time. A shift width data generation part (103) sets a plurality of shift widths that are time intervals corresponding to the multiple value of the unit time, shifts the data value for each unit time of the time-series data by the plurality of shift widths, and generates a plurality of shift width data items in which the data value for each unit time of the time-series data is shifted by each of the plurality of shift widths. An association degree calculation part (104) calculates the association degree between the transition of the data value of the time-series data and the transition of the data value of each of the plurality of shift width data items. A cycle estimation part (105) estimates, on the basis of the calculated association degree, the cycle present in the time-series data.

Description

Data processing device, data processing method and data processing program

[0001] The present disclosure relates to techniques for processing data.

[0002] Time-series data (hereinafter also referred to as collected data) collected from a monitoring target is converted into feature quantities for anomaly detection or data analysis. Generally, collected data is divided at regular intervals in a process called windowing. Then, statistical processing or the like is performed on the divided collected data for each window to calculate feature quantities. Anomaly detection or collected data analysis is then performed based on changes in the feature quantities of each window. If the collected data behaves cyclically, it is desirable to set the window size when windowing so that one window contains one period of collected data.

[0003] In the technology of Patent Document 1, autocorrelation coefficients at various lags are calculated for each variable included in collected data, and the lag of the variable with the largest autocorrelation coefficient is selected as the period of the collected data, and the window size is set.

[0004] Japanese Patent Application Laid-Open No. 2023-51402

[0005] In the technology of Patent Document 1, when collected data includes category data as variables, the category data must be converted into numerical data using some method before estimating the period. Category data is data in which non-numeric values ​​are described. In other words, category data is data in which values ​​representing categories, such as IDs (identifiers) and error codes, are described, rather than numerical values ​​like sensor values. When category data is quantified, if the category data includes many categories, the number of variables in the quantified data increases. Therefore, the technology of Patent Document 1 has a problem in that if the category data includes many categories, the amount of memory required for the process of estimating the period is large.

[0006] A primary object of the present disclosure is to solve such problems. More specifically, an object of the present disclosure is to reduce memory consumption when estimating the period of time-series data.

[0007] A data processing device according to the present disclosure includes: a time series data acquiring unit that acquires time series data in which data values ​​for each unit time are described over a plurality of unit times, and in which a period corresponding to any multiple of the unit time exists in the transition of the data values ​​for each unit time; a shift width data generating unit that sets a plurality of shift widths, each of which is a time corresponding to a multiple of the unit time, by changing the multiple of the unit time, and shifts the data values ​​for each unit time of the time series data by each of the plurality of shift widths on a time axis, thereby generating a plurality of shift width data that are a plurality of time series data in which the data values ​​for each unit time of the time series data are shifted by each of the plurality of shift widths on the time axis; a relevance calculating unit that calculates a degree of relevance between the transition of the data values ​​for each unit time of the time series data and the transition of the data values ​​for each unit time of each of the plurality of shift width data; and a period estimating unit that estimates a period existing in the time series data based on the calculated degree of relevance.

[0008] According to the present disclosure, it is possible to reduce memory consumption when estimating the period of time-series data.

[0009] 1 is a diagram showing an example of the functional configuration of a data processing device according to embodiment 1. FIG. 2 is a diagram showing examples of original data, shift width data with shift width 1, and shift width data with shift width 2 according to embodiment 1. FIG. 3 is a diagram showing an example of a plot line of original data and shift width data with shift width 1 according to embodiment 1. FIG. 4 is a diagram showing an example of a plot line of original data and shift width data with shift width 5 according to embodiment 1. FIG. 5 is a diagram showing an example of a correlation coefficient for each shift width according to embodiment 1. FIG. 6 is a diagram showing examples of original data, shift width data with shift width 1, and shift width data with shift width 2 according to embodiment 1. FIG. 7 is a diagram showing an example of a Cramer's correlation coefficient for each shift width according to embodiment 1. A flowchart showing an example of the operation of a data processing device according to embodiment 1. A diagram showing an example of the hardware configuration of a data processing device according to embodiment 1.

[0010] Hereinafter, embodiments will be described with reference to the drawings. In the following description of the embodiments and the drawings, the same reference numerals denote the same or corresponding parts.

[0011] Embodiment 1. ***Description of Configuration*** Fig. 1 shows an example of the functional configuration of a data processing device 100 according to this embodiment. Fig. 9 shows an example of the hardware configuration of the data processing device 100 according to this embodiment. The operation procedure of the data processing device 100 corresponds to a data processing method. Furthermore, a program that realizes the operation of the data processing device 100 corresponds to a data processing program. First, an example of the hardware configuration of the data processing device 100 will be described with reference to Fig. 9.

[0012] The data processing device 100 according to this embodiment is a computer. The data processing device 100 includes, as hardware components, a processor 901, a main storage device 902, an auxiliary storage device 903, and a communication device 904. As shown in FIG. 1 , the data processing device 100 also includes, as functional components, a time-series data acquisition unit 101, a data determination unit 102, a shift width data generation unit 103, a degree of association calculation unit 104, a period estimation unit 105, and a period synthesis unit 106. The functions of the time-series data acquisition unit 101, the data determination unit 102, the shift width data generation unit 103, the degree of association calculation unit 104, the period estimation unit 105, and the period synthesis unit 106 are realized, for example, by programs. The auxiliary storage device 903 stores programs that realize these functional components. These programs are loaded from the auxiliary storage device 903 to the main storage device 902. The processor 901 then executes these programs to perform the operations of these functional components, which will be described later. 9 shows a schematic diagram of a state in which a processor 901 is executing a program that implements the functional components. A communication device 904 receives time-series data from a plurality of devices that are to be monitored.

[0013] Next, an example of the functional configuration of the data processing device 100 according to the present embodiment will be described with reference to FIG.

[0014] (Time-series data acquisition unit 101) The time-series data acquisition unit 101 acquires time-series data from multiple monitored devices. The time-series data describes data values ​​for each unit time over multiple unit times. In this embodiment, the unit time is one second. Furthermore, the time-series data has a period in the transition of data values ​​for each unit time that corresponds to a multiple of the unit time. The time-series data acquisition unit 101 acquires at least either numerical data or categorical data as the time-series data. The time-series data acquisition unit 101 may acquire only numerical data as the time-series data. Alternatively, the time-series data acquisition unit 101 may acquire only categorical data as the time-series data. Alternatively, the time-series data acquisition unit 101 may acquire both numerical data and categorical data as the time-series data. The numerical data is data in which numerical values ​​for each unit time are described as data values ​​for each unit time over multiple unit times. The numerical values ​​described in the numerical data are, for example, sensor values ​​of sound, vibration, etc. generated by the monitored devices. The monitored devices are devices installed in factories, power plants, etc. Specifically, the equipment to be monitored includes motors, engines, generators, turbines, transformers, etc. The sensors include microphones, vibration sensors, etc. Category data is data in which non-numeric values ​​per unit time are described over multiple unit times as data values ​​per unit time. The non-numeric values ​​described in category data are, for example, metadata. The metadata includes an ID, an identifier representing the status of the equipment, numerical attributes of numerical data, error codes, etc. The time series data acquisition unit 101 acquires time series data using, for example, the communication device 904. The processing performed by the time series data acquisition unit 101 corresponds to time series data acquisition processing.

[0015] (Data Determination Unit 102) The data determination unit 102 determines whether the time-series data acquired by the time-series data acquisition unit 101 is numerical data or categorical data.

[0016] (Shift Width Data Generator 103) The shift width data generator 103 generates multiple shift width data by shifting the data values ​​of the time series data by multiple shift widths on the time axis. The multiple shift width data are multiple pieces of time series data in which the data values ​​of the time series data per unit time are shifted by multiple shift widths on the time axis. More specifically, the shift width data generator 103 sets multiple shift widths, which are times corresponding to multiple multiples of the unit time, by changing the multiple multiples of the unit time. Specifically, the shift width data generator 103 sets shift width 1 corresponding to 1 time the unit time (1 second), shift width 2 corresponding to twice the unit time (2 seconds), etc. Then, the shift width data generator 103 shifts the data values ​​of the time series data per unit time by multiple shift widths on the time axis to generate multiple shift width data. A specific example of a method for generating shift width data by the shift width data generator 103 will be described later. The process performed by the shift width data generating unit 103 corresponds to a shift width data generating process.

[0017] (Relevance calculation unit 104) The relevance calculation unit 104 calculates the relevance between the transition of data values ​​per unit time of the time-series data and the transition of data values ​​per unit time of each of the plurality of shift range data. The relevance is an index that represents the relevance between the transition of data values ​​per unit time of the time-series data and the transition of data values ​​per unit time of each of the shift range data. The relevance calculation unit 104 calculates different relevance when the data determination unit 102 determines that the time-series data is numerical data and when the data determination unit 102 determines that the time-series data is categorical data. A specific example of a method for calculating the relevance by the relevance calculation unit 104 will be described later. The processing performed by the relevance calculation unit 104 corresponds to a relevance calculation process.

[0018] (Period Estimation Unit 105) The period estimation unit 105 estimates a period present in the time-series data based on the degree of relevance of each shift width data calculated by the degree-of-relevance calculation unit 104. A specific example of a method for estimating a period by the period estimation unit 105 will be described later. The processing performed by the period estimation unit 105 corresponds to period estimation processing.

[0019] (Period Synthesis Unit 106) The period synthesis unit 106 synthesizes multiple periods. In this embodiment, the time-series data acquisition unit 101 acquires multiple time-series data from multiple monitored devices, and the shift width data generation unit 103 generates multiple shift width data for each of the multiple time-series data acquired by the time-series data acquisition unit 101. Then, the relevance calculation unit 104 calculates, for each of the multiple time-series data acquired by the time-series data acquisition unit 101, the degree of relevance between the transition of data values ​​per unit time of each of the multiple time-series data and the transition of data values ​​per unit time of each of the corresponding multiple shift width data. Furthermore, the period estimation unit 105 estimates a period present in each of the time-series data based on the calculated degree of relevance. The period synthesis unit 106 synthesizes the multiple periods estimated by the period estimation unit 105. The period obtained by synthesizing multiple periods by the period synthesis unit 106 is called a synthesized period. The synthesized period is a period that is commonly applied to multiple time-series data from multiple monitored devices. A specific example of the period synthesis method used by the period synthesis unit 106 will be described later.

[0020] (Specific example of a method for generating shift width data: numerical data) Next, a specific example of a method for generating shift width data when the time series data (hereinafter also referred to as original data) acquired by the time series data acquisition unit 101 is numerical data will be described with reference to Fig. 2. Fig. 2 shows an example of shift width data when the original data is numerical data.

[0021] The "Time Stamp" column in FIG. 2 shows the time for each unit of time. In FIG. 2, the unit of time is 1 second. The "Original Data" column in FIG. 2 shows the numerical values ​​for each unit of time of the original data. The "Shift Width 1" column in FIG. 2 shows numerical values ​​that are shifted from the original data by shift width 1 on the time axis. The group of numerical values ​​shown in the "Shift Width 1" column is shift width data corresponding to "Shift Width 1." "Shift Width 1" is 1 time unit time (i.e., 1 second). Therefore, the numerical value "10" of timestamp: 1 in the "Original Data" becomes the numerical value of timestamp 2 in "Shift Width 1." Similarly, the numerical value "0.42" of timestamp: 2 in the "Original Data" becomes the numerical value of timestamp 3 in "Shift Width 1." The "Shift Width 2" column in FIG. 2 shows numerical values ​​that are shifted by shift width 2 from the original data on the time axis. The group of values ​​shown in the "Shift Width 2" column is shift width data corresponding to "Shift Width 2." "Shift Width 2" is twice the unit time (i.e., 2 seconds). Therefore, the numerical value "10" of timestamp: 1 in the "original data" becomes the numerical value of timestamp: 3 in "Shift Width 2." Similarly, the numerical value "0.42" of timestamp: 2 in the "original data" becomes the numerical value of timestamp: 4 in "Shift Width 2." Although FIG. 2 only illustrates "Shift Width 1" and "Shift Width 2," the shift width data generation unit 103 generates similar shift width data for "Shift Width 3," "Shift Width 4," "Shift Width 5," and so on.

[0022] (Specific example of a method for calculating relevance: numerical data) Next, a specific example of a method for calculating relevance when the original data is numerical data will be described with reference to Figures 3, 4, and 5. Figure 3 shows a graph obtained by plotting the numerical values ​​of the original data and shift width data with a shift width of 1 at each time. Figure 4 shows a graph obtained by plotting the numerical values ​​of the original data and shift width data with a shift width of 5 at each time. Figure 5 shows the correlation coefficient between the original data and each of a plurality of shift width data.

[0023] In FIG. 3 , the plot line of the original data and the plot line of the shift width data with a shift width of 1 are not similar to each other. However, when the shift width approaches the period of the original data, as shown in FIG. 4 , the plot line of the original data and the plot line of the shift width data become similar to each other. The relevance calculation unit 104 calculates the relevance to determine the shift width corresponding to such a period of the original data. That is, the relevance calculation unit 104 calculates the strength of the relevance between the transition of the data value of the original data per unit time and the transition of the data value of each shift width data per unit time as the relevance. When the original data is numerical data, the relevance calculation unit 104 calculates, for example, a correlation coefficient as the relevance. The relevance calculation unit 104 can also use a known statistical method to calculate the relevance. Here, it is assumed that the correlation coefficients shown in FIG. 5 are obtained for each shift width.

[0024] (Specific Example of a Method for Generating Shift Amount Data: Category Data) When the original data is category data, the shift amount data generator 103 generates shift amount data in the same manner as when the original data is numerical data. Fig. 6 shows an example of shift amount data when the original data is category data.

[0025] In FIG. 6, the meanings of "timestamp," "original data," "shift width 1," and "shift width 2" are the same as those shown in FIG. 2. However, the "original data," "shift width 1," and "shift width 2" columns contain identifiers such as "STATE 0" and "STATE 1" that represent the state of the device. The "shift width 1" column in FIG. 6 shows values ​​that are shifted from the original data by shift width 1 on the time axis. The group of values ​​shown in the "shift width 1" column is the shift width data corresponding to "shift width 1." "Shift width 1" is one time unit time (i.e., one second). Therefore, the numerical value "STATE 1" of timestamp: 1 in the "original data" becomes the numerical value of timestamp 2 in "shift width 1." Similarly, the numerical value "STATE 0" of timestamp: 2 in the "original data" becomes the numerical value of timestamp 3 in "shift width 1." The "Shift Width 2" column in FIG. 6 shows numerical values ​​that are shifted by shift width 2 from the original data on the time axis. The group of values ​​shown in the "Shift Width 2" column is shift width data corresponding to "Shift Width 2." "Shift Width 2" is twice the unit time (i.e., 2 seconds). Therefore, the numerical value "STATE 1" of timestamp: 1 in the "original data" becomes the numerical value of timestamp 3 in "Shift Width 2." Similarly, the numerical value "STATE 0" of timestamp: 2 in the "original data" becomes the numerical value of timestamp 4 in "Shift Width 2." Although FIG. 6 only shows "Shift Width 1" and "Shift Width 2," the shift width data generation unit 103 generates similar shift width data for "Shift Width 3," "Shift Width 4," "Shift Width 5," and so on.

[0026] (Specific Example of a Method for Calculating the Degree of Relevance: Categorical Data) When the original data is categorical data, the relevance calculation unit 104 generates the degree of relevance in the same manner as when the original data is numerical data. Here, the relevance calculation unit 104 calculates the Cramer's correlation coefficient as the degree of relevance. The Cramer's correlation coefficient is generally used as an index representing the strength of the correlation between two categorical variables. In this embodiment, the relevance calculation unit 104 treats the original data and the shift range data as two categorical variables and calculates the Cramer's correlation coefficient as the degree of relevance. FIG. 7 shows the Cramer's correlation coefficients calculated for each of multiple shift ranges. FIG. 7 corresponds to FIG. 5 . Note that when the original data is categorical data, the relevance calculation unit 104 may calculate a phi (φ) coefficient instead of the Cramer's correlation coefficient as the degree of relevance. Furthermore, the relevance calculation unit 104 may also use a known statistical method to calculate the degree of relevance.

[0027] (Specific Example of Period Estimation Method) The period estimation unit 105 estimates that the original data contains a period corresponding to a multiple of the unit time for the shift width used to generate the shift width data with the highest degree of association among multiple shift width data. For example, when the degrees of association shown in Figure 5 or 7 are obtained, the degree of association for shift width 5 is the highest. Therefore, the period estimation unit 105 estimates that the period of the original data is 5 seconds, which is the time corresponding to shift width 5.

[0028] (Specific Example of Period Synthesis Method) The period synthesis unit 106, for example, calculates the least common multiple of the multiple periods estimated by the period estimation unit 105 for multiple pieces of time series data. The period synthesis unit 106 then uses the value of the least common multiple as the synthesized period. Furthermore, the period synthesis unit 106 selects, from the multiple periods estimated by the period estimation unit 105, the period with the highest degree of association used in estimating each period as the synthesized period. For example, suppose that the period of time series data A is estimated to be 5 seconds (shift width 5) and the degree of association for shift width 5 is 1.0. Furthermore, suppose that the period of time series data B is estimated to be 4 seconds (shift width 4) and the degree of association for shift width 4 is 0.8. Furthermore, suppose that the period of time series data C is estimated to be 3 seconds (shift width 3) and the degree of association for shift width 3 is 0.9. In this case, the period synthesis unit 106 selects, as the synthesis period, 5 seconds, which is the period of the time series data A having the highest degree of association.

[0029] ***Explanation of Operation*** Next, an example of operation of the data processing device 100 according to this embodiment will be described. FIG.

[0030] (Step S1) First, in step S1, the time-series data acquisition unit 101 acquires time-series data from the device to be monitored.

[0031] (Step S2) Next, in step S2, the data determination unit 102 determines the type of time-series data. That is, the data determination unit 102 determines whether the time-series data is numerical data or categorical data. If the time-series data is numerical data, the process proceeds to step S3-1. If the time-series data is categorical data, the process proceeds to step S3-2.

[0032] (Step S3-1) In step S3-1, the shift amount data generating unit 103 generates a plurality of shift amount data for the time series data, which is numerical data.

[0033] (Step S4-1) Next, in step S4-1, the relevance calculation unit 104 calculates the relevance for the numerical data for the multiple shift width data generated in step S3-1. Specifically, the relevance calculation unit 104 calculates a correlation coefficient as the relevance.

[0034] (Step S3-2) In step S3-2, the shift width data generating unit 103 generates a plurality of shift width data for the time series data, which is category data.

[0035] (Step S4-2) Next, in step S4-2, the relevance calculation unit 104 calculates the relevance for category data for the multiple shift width data generated in step S3-2. Specifically, the relevance calculation unit 104 calculates the Cramer's correlation coefficient or the Phi coefficient as the relevance.

[0036] (Step S5) In step S5, the period estimation unit 105 estimates the period of the time-series data based on the calculated relevance.

[0037] (Step S6) In step S6, the period synthesis unit 106 determines whether a predetermined number of time-series data required for synthesizing periods has been acquired, i.e., whether a predetermined number of periods required for synthesizing periods has been estimated. If the predetermined number of periods has been estimated, the process proceeds to step S7. On the other hand, if the predetermined number of periods has not been estimated, the process returns to step S1.

[0038] (Step 7) Finally, in step S7, the period synthesis unit 106 synthesizes a plurality of periods to obtain a synthesized period.

[0039] ***Description of Effects of the Embodiment*** According to this embodiment, it is possible to reduce memory consumption when estimating a period, regardless of whether only numeric data, only categorical data, or both numeric and categorical data are acquired as time-series data. This embodiment can calculate an appropriate degree of association depending on the type of time-series data (numeric data / categorical data). In other words, even if the time-series data includes categorical data, it is possible to calculate the degree of association for the categorical data using the same process as for numerical data, without converting the categorical data to numerical data. Therefore, according to this embodiment, it is possible to reduce memory consumption and estimate a period, even if the time-series data includes categorical data.

[0040] The procedure described in this embodiment is an example. Therefore, only a part of the procedure described in this embodiment may be implemented. Also, at least a part of the procedure described in this embodiment may be implemented in combination with a procedure not described in this embodiment. Also, the configuration and procedure described in this embodiment may be changed as necessary.

[0041] ***Supplementary Explanation of Hardware Configuration*** Finally, a supplementary explanation of the hardware configuration of the data processing device 100 will be provided. The processor 901 shown in FIG. 9 is an IC (Integrated Circuit) that performs processing. The processor 901 is a CPU (Central Processing Unit), a DSP (Digital Signal Processor), or the like. The main memory device 902 shown in FIG. 9 is a RAM (Random Access Memory). The auxiliary memory device 903 shown in FIG. 9 is a ROM (Read Only Memory), a flash memory, an HDD (Hard Disk Drive), or the like. The communication device 904 shown in FIG. 9 is an electronic circuit that executes data communication processing. The communication device 904 is, for example, a communication chip or a NIC (Network Interface Card).

[0042] The auxiliary storage device 903 also stores an OS (Operating System). At least a portion of the OS is executed by the processor 901. While executing at least a portion of the OS, the processor 901 executes programs that implement the functional components of the data processing device 100 (the time-series data acquisition unit 101, the data determination unit 102, the shift width data generation unit 103, the relevance calculation unit 104, the period estimation unit 105, and the period synthesis unit 106). The processor 901 executes the OS to perform tasks, memory management, file management, communication control, and the like. At least one of information, data, signal values, and variable values ​​indicating the results of processing by the functional components of the data processing device 100 is stored in at least one of the main storage device 902, the auxiliary storage device 903, and a register and cache memory within the processor 901. The programs that implement the functional components of the data processing device 100 may be stored on a portable recording medium such as a magnetic disk, a flexible disk, an optical disk, a compact disk, a Blu-ray (registered trademark) disk, or a DVD. A portable recording medium storing a program that realizes the functional components of the data processing device 100 may be distributed.

[0043] Furthermore, the term "unit" of at least one of the functional components of the data processing device 100 may be interpreted as a "circuit," a "step," a "procedure," a "process," or a "circuitry." The data processing device 100 may also be implemented by a processing circuit. The processing circuit may be, for example, a logic IC (Integrated Circuit), a GA (Gate Array), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field-Programmable Gate Array). In this case, the functional components of the data processing device 100 are each implemented as part of the processing circuit. In this specification, the generic concept of a processor and a processing circuit is referred to as "processing circuitry." In other words, a processor and a processing circuit are each specific examples of "processing circuitry."

[0044] 100 Data processing device, 101 Time series data acquisition unit, 102 Data determination unit, 103 Shift width data generation unit, 104 Association degree calculation unit, 105 Period estimation unit, 106 Period synthesis unit, 901 Processor, 902 Main storage device, 903 Auxiliary storage device, 904 Communication device

Claims

1. A data processing device comprising: a time series data acquisition unit that acquires time series data in which data values ​​for each unit time are described over a plurality of unit times, and in which the transition of the data values ​​for each unit time has a period corresponding to any multiple of the unit time; a shift width data generation unit that sets a plurality of shift widths, which are times corresponding to multiples of the unit time, by changing the multiple values ​​of the unit time, and shifts the data values ​​for each unit time of the time series data by each of the plurality of shift widths on the time axis, to generate a plurality of shift width data that are a plurality of time series data in which the data values ​​for each unit time of the time series data are shifted by each of the plurality of shift widths on the time axis; a relevance calculation unit that calculates a degree of relevance between the transition of the data values ​​for each unit time of the time series data and the transition of the data values ​​for each unit time of each of the plurality of shift width data; and a period estimation unit that estimates a period existing in the time series data based on the calculated degree of relevance.

2. The data processing device according to claim 1, wherein the time series data acquisition unit acquires, as the time series data, at least one of numerical data in which a numerical value for each unit time is described over the plurality of unit times as the data value for each unit time and category data in which a value other than a numerical value for each unit time is described over the plurality of unit times as the data value for each unit time, and the data processing device further has a data determination unit that determines whether the time series data acquired by the time series data acquisition unit is the numerical data or the category data, and the relevance calculation unit calculates different relevance degrees when the data determination unit determines that the time series data is the numerical data and when the data determination unit determines that the time series data is the category data.

3. The data processing device according to claim 2, wherein the relevance calculation unit calculates a correlation coefficient as the relevance when the data determination unit determines that the time series data is the numerical data, and calculates at least one of a Cramer's correlation coefficient and a Phi coefficient as the relevance when the data determination unit determines that the time series data is the categorical data.

4. A data processing device as described in claim 1, wherein the period estimation unit estimates that the time series data has a period corresponding to a multiple value of the unit time in the shift width used to generate the shift width data with the highest degree of association among the multiple shift width data.

5. The data processing device according to claim 1, further comprising: a period synthesis unit that synthesizes the periods estimated by the period synthesis unit; the time series data acquisition unit acquires a plurality of time series data; the shift width data generation unit generates a plurality of shift width data for each of the plurality of time series data acquired by the time series data acquisition unit; the relevance calculation unit calculates, for each of the plurality of time series data acquired by the time series data acquisition unit, a degree of relevance between the transition of data values ​​per unit time of each of the time series data and the transition of data values ​​per unit time of each of the plurality of corresponding shift width data; the period estimation unit estimates a period present in each of the time series data based on the calculated relevance; and the data processing device further comprising: a period synthesis unit that synthesizes the plurality of periods estimated by the period estimation unit.

6. The data processing device according to claim 5, wherein the period synthesis unit calculates the least common multiple of the plurality of periods as a synthesis of the plurality of periods.

7. A data processing device according to claim 5, wherein the period synthesis unit selects, from the plurality of periods, the period having the highest degree of association used in estimating each of the periods as the synthesis of the plurality of periods.

8. A data processing method in which a computer acquires time series data in which data values ​​for each unit time are described over a plurality of unit times, and the transition of the data values ​​for each unit time has a period corresponding to any multiple of the unit time; the computer sets a plurality of shift widths, which are times corresponding to multiples of the unit time, by changing the multiple values ​​of the unit time, and shifts the data values ​​for each unit time of the time series data by each of the plurality of shift widths on the time axis to generate a plurality of shift width data, which are a plurality of time series data in which the data values ​​for each unit time of the time series data are shifted by each of the plurality of shift widths on the time axis; the computer calculates a degree of association between the transition of data values ​​for each unit time of the time series data and the transition of data values ​​for each unit time of each of the plurality of shift width data; and estimates the period existing in the time series data based on the calculated degree of association.

9. A data processing program that causes a computer to execute the following steps: a time series data acquisition process that acquires time series data in which data values ​​for each unit time are described over a plurality of unit times, and in which the transition of the data values ​​for each unit time has a period corresponding to any multiple of the unit time; a shift width data generation process that sets a plurality of shift widths, which are times corresponding to multiples of the unit time, by changing the multiple values ​​of the unit time, and shifts the data values ​​for each unit time of the time series data by each of the plurality of shift widths on the time axis, to generate a plurality of shift width data that is a plurality of time series data in which the data values ​​for each unit time of the time series data are shifted by each of the plurality of shift widths on the time axis; a relevance calculation unit that calculates the degree of relevance between the transition of the data values ​​for each unit time of the time series data and the transition of the data values ​​for each unit time of each of the plurality of shift width data; and a period estimation process that estimates the period existing in the time series data based on the calculated degree of relevance.

Images