Information processing method, and information processing device

Abstract

To make it possible to achieve further accurate alignment.SOLUTION: An information processing method according to the present application includes: an acquisition step for acquiring object data being time series data containing sampling values at respective sampling timing and standard time series data serving as a reference when determining whether the object data is normal or not; a generation step for determining sampling timing corresponding to the sampling values for each of sampling values of the standard time series data as relative sampling timing being sampling timing measured from the prescribed start point, and generating distributed data indicating a distributing range of the relative sampling timing; and a determination step for determining a warping path based on the objective data and the standard time series data so as to be contained in the distribution range of the relative sampling timing to be determined on the basis of the distributed data.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an information processing method and an information processing device. [Background technology]

[0002] In the past, in production process management, a technique has been known that aligns time-series data of measured values such as temperature and pressure measured in production equipment with time-series data under normal conditions in order to determine abnormalities in the production equipment. For example, a technique is known in which, when performing alignment, the allowable range of time lag in sampling timing is flexibly set for each sampled value. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 5492622 Summary of the Invention [Problem to be solved by the invention]

[0004] However, in the above-mentioned conventional technology, in an alignment method using the elapsed time from the start of a process consisting of multiple steps, the later the process, the wider the distribution tends to be. Therefore, when it is desired to detect a deviation in the required time of a later process as an anomaly, even if the required time of that process deviates from normal data, which is time-series data under normal conditions, it may be determined to be within the normal data range. In this case, there is a risk that the deviation in the required time of that process will not be detected as an anomaly.

[0005] The present invention is intended to solve such problems and has an object to make it possible to achieve more accurate alignment. [Means for solving the problem]

[0006] The information processing method according to the present application is characterized by including an acquisition step of acquiring target data, which is time series data including sampling values for each sampling timing, and standard time series data that serves as a reference for determining whether the target data is normal; a generation step of determining, for each sampling value of the standard time series data, the sampling timing corresponding to the sampling value as a relative sampling timing, which is a sampling timing measured from a predetermined starting point, and generating distribution data that indicates the distribution range of the relative sampling timing; and a determination step of determining a warping path based on the target data and the standard time series data so that the warping path is included in the distribution range of the relative sampling timing determined based on the distribution data.

[0007] In the above information processing method, all sampled values of the target data and the standard time series data are compared with each other, and the warping path that minimizes the distance between the target data and the standard time series data is determined as the optimal warping path.

[0008] In the above information processing method, the warping path is determined by restricting the route of the warping path within the distribution range of the relative sampling timing determined based on the distribution data.

[0009] The information processing method further includes a calculation step of calculating the distance between the target data and the standard time series data by integrating values on the route of the warping path.

[0010] The information processing method further includes a determining step of determining that the target data is abnormal if the distance calculated in the calculating step is equal to or greater than a predetermined threshold.

[0011] In the above information processing method, the distribution data is a range of variation in values represented by the difference between the relative sampling timing and the sampling timing of the target data. [Effects of the Invention]

[0012] According to the information processing method described above, target data, which is time series data including sampling values for each sampling timing, and standard time series data that serves as a reference for determining whether the target data is normal are acquired, and for each sampling value of the standard time series data, the sampling timing corresponding to the sampling value is determined as a relative sampling timing, which is a sampling timing measured from a predetermined starting point, distribution data indicating the distribution range of the relative sampling timings is generated, and a warping path is determined based on the target data and the standard time series data so that the target data falls within the distribution range of the relative sampling timings determined based on the distribution data, thereby enabling more accurate alignment to be achieved. [Brief explanation of the drawings]

[0013] [Figure 1] FIG. 1 is a conceptual diagram showing an outline of the embodiment. [Figure 2] FIG. 2 is a diagram illustrating an example of the configuration of an information processing system according to the embodiment. [Figure 3] FIG. 3 is a diagram illustrating an example of a dataset information storage unit according to the embodiment. [Figure 4] FIG. 4 is a diagram illustrating an example of a threshold information storage unit according to the embodiment. [Figure 5] FIG. 5 is a conceptual diagram showing the generation process according to the embodiment. [Figure 6] FIG. 6 is a flowchart showing an example of the flow of the decision process executed by the information processing device according to the embodiment. [Figure 7] FIG. 7 is a flowchart showing an example of the flow of the determination process executed by the information processing device according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0014] Next, embodiments will be described with reference to the drawings. In the following description, components common to the embodiments will be designated by the same reference numerals, and repeated description will be omitted.

[0015] [Embodiment] 1. Overview of the embodiment In production process management, early detection of abnormalities in production equipment is necessary to ensure stable operation of the equipment. Conventionally, in production process management, a technique is known for aligning time series data of measured values such as temperature and pressure measured by the production equipment (time series data of the monitored object to monitor for abnormalities) with normal time series data in order to determine abnormalities in the equipment. For example, a technique is known for flexibly setting the tolerance for time lag in sampling timing for each sampled value when aligning. Dynamic Time Warping (DTW) is used for this alignment.

[0016] However, with conventional technology, there is a risk that abnormalities in the measurement values measured by production equipment may not be detected in processes consisting of multiple steps. Generally, plant operation data may consist of multiple steps. In this case, process data may be, for example, data from the plant's batch process, or data from various sensors in the plant.

[0017] For example, a process may consist of step 1 of feeding raw materials, step 2 of preparing the raw materials, step 3 of reacting the prepared raw materials, step 4 of terminating the process, step 5 of removing the product from the equipment, and step 6 of shutting down the plant. In this way, the process targeted in this application is assumed to have a clear start and end.

[0018] In such cases, the time required to complete the process may vary depending on the process as a whole or each step. For example, even among normal data, there may be data that is long or short along the time axis, which may cause fluctuations along the time axis. Therefore, it was necessary to perform synchronization processing along the time axis.

[0019] However, alignment methods using the elapsed time from the start of a process tended to have a wider distribution the later the process. This tendency is thought to occur because the distribution of elapsed time from the start of a process for a later process includes not only the variance in the required time for that process but also the variance in all processes that have been performed up to that point. Therefore, when attempting to detect a deviation in the required time for a later process as an anomaly, even if the required time for that process deviates from the normal data, it may be determined to be within the range of normal data in the distribution that includes the variance in the required time for the previous processes, and therefore the deviation in the required time for that process may not be detected as an anomaly.

[0020] Here, the deviation in the required time of a process will be explained using FIG. 1. FIG. 1 is a conceptual diagram showing an overview of an embodiment. In the example of FIG. 1, FIG. FI1 shows time series data in which the horizontal axis represents time and the vertical axis represents sensor values (an example of sampled values). FIG. FI1 is also a diagram in which multiple time series data are superimposed. For example, in FIG. FI1, the dotted line represents normal data and the solid line represents target data OD1. The normal data here is, for example, time series data of normal sensor values acquired from a plant.

[0021] In this case, the process consists of steps ST1, ST2, ST3, ST4, and ST5. As shown in FIG. 1, it can be seen that the variation FL1 in step ST5 is larger than in the other steps. As such, when it is desired to determine whether the required time for each of steps ST1 to ST5 is normal or abnormal, or when it is desired to determine whether the required time for the entire process is normal or abnormal, there is a problem in that the variation in the required time becomes larger in the later steps (e.g., step ST5 in the example of FIG. 1).

[0022] When applying conventional technology to such time series data, even if the sampling timing of the target data is out of sync with the variation in the time axis direction of the normal data, there is a possibility that the sampling timing of the standard time series data and the sampling timing of the target data will be synchronized when performing alignment using DTW. The sampling timing may be the actual time, the elapsed time from the start of the process, an index indicating time, or the like.

[0023] To address the above-mentioned problem, the present application introduces relative sampling timing, which is sampling timing measured from a predetermined starting point. For example, the present application provides a solution for generating distribution data based on the relative sampling timing. For example, the present application sets a point selected from any data points among the sampling timings of time-series data as the starting point of the relative sampling timing. The present application then determines the range of difference between the current sampling timing and the starting point as the synchronizable range. The present application then performs alignment based on the synchronizable range.

[0024] In the following, an example will be described in which the starting point of the relative sampling timing is a sampling timing that is a predetermined value before the current sampling timing.

[0025] [2. Information Processing System Configuration] An example of the configuration of the information processing system 1 will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the configuration of the information processing system 1 according to the embodiment. In the example of Fig. 2, the information processing system 1 includes a data server 10 and an information processing device 100. The data server 10 and the information processing device 100 are connected to each other via a network N so as to be able to communicate with each other via a wired or wireless connection. Note that the information processing system 1 shown in Fig. 2 may include a plurality of data servers 10 and a plurality of information processing devices 100.

[0026] The data server 10 is an information processing device that acquires various data and provides it to the information processing device 100, and is realized, for example, by a server device or a cloud system. For example, the data server 10 stores target data newly acquired from a plant. For example, the target data is time-series data including sampling values at each sampling timing. In this case, the data server 10 may store information about a process corresponding to the sampling timing of the target data in association with the target data.

[0027] The data server 10 also stores standard time series data. Here, the standard time series data is data generated based on normal data indicating time series data of normal sensor values acquired from a plant. For example, the standard time series data is data that serves as a reference when determining whether target data newly acquired from a plant is normal.

[0028] The standard time series data is data generated from the average values of multiple normal data. To give a more specific example, the value of the sampling timing (time axis value) of the standard time series data is the average value of the value of the sampling timing of the first normal data and the value of the sampling timing of the second normal data. Furthermore, the sensor value of the standard time series data is the average value of the sensor value of the first normal data and the sensor value of the second normal data.

[0029] In the above example, the value of the sampling timing of the standard time series data is not limited to the average value of the value of the sampling timing of the first normal data and the value of the sampling timing of the second normal data. For example, instead of the average value, a weighted average may be used.

[0030] The information processing device 100 is an information processing device capable of communicating with various devices via a network N, and is realized by, for example, a server device or a cloud system. For example, the information processing device 100 is connected to various other devices via the network N so as to be able to communicate with them.

[0031] For example, the information processing device 100 acquires target data and standard time series data from the data server 20. Next, the information processing device 100 generates distribution data indicating a distribution range of relative sampling timings, which are sampling timings measured from a predetermined starting point. Then, the information processing device 100 determines a warping path based on the target data and the standard time series data so that the target data is included in the distribution range of the relative sampling timings determined based on the generated distribution data.

[0032] That is, the information processing device 100 determines the range of difference between the current sampling timing and the starting point of the relative sampling timing as the synchronizable range, and performs alignment based on this synchronizable range. As a result, the information processing device 100 can achieve more accurate alignment.

[0033] 3. Configuration of Information Processing Device Next, a description will be given of an example of the functional configuration of the information processing device 100. As shown in FIG.

[0034] (Regarding the communication unit 110) The communication unit 110 is realized by, for example, a network interface card (NIC), etc. The communication unit 110 is connected to a network N by wire or wirelessly, and transmits and receives information to and from various other devices.

[0035] (Regarding the storage unit 120) The storage unit 120 is realized by, for example, a semiconductor memory element such as a random access memory (RAM) or a flash memory, or a storage device such as a hard disk or an optical disk. For example, the storage unit 120 includes a dataset storage unit 121 and a threshold information storage unit 122.

[0036] (Regarding the dataset storage unit 121) The dataset storage unit 121 stores various types of data. An example of the dataset storage unit 121 according to the embodiment is shown in Fig. 3. In the example shown in Fig. 3, the dataset storage unit 121 has items such as "dataset ID (identifier)", "standard time series data ID", "sampling timing", "sensor value", "target data ID", "sampling timing", "sensor value", "relative distribution data ID", and "relative distribution data".

[0037] "Dataset ID" is an identifier that identifies a dataset, which is a group of data. "Standard time series data ID" is an identifier that identifies the standard time series data associated with the "Dataset ID". "Sampling timing" is information about the sampling timing of the standard time series data associated with the "Standard time series data ID". For example, the sampling timing of the standard time series data is the time measured with the point at which the process starts as 0. "Sensor value" is information about the sensor value of the standard time series data associated with the "Standard time series data ID".

[0038] "Target data ID" is an identifier that identifies the target data associated with "Data set ID." "Sampling timing" is information about the sampling timing of the target data associated with "Target data ID." For example, the sampling timing of the target data is the time measured with the point at which the process starts as 0. "Sensor value" is information about the sensor value of the target data associated with "Target data ID."

[0039] "Distribution data ID" is an identifier that identifies the distribution data associated with the "Data set ID." "Distribution data" is information about the distribution data associated with the "Distribution data ID." For example, the distribution data is a range of variation in values represented by the difference between the relative sampling timing and the sampling timing of normal data.

[0040] For example, in Figure 3, "D1" identified by the dataset ID has a standard time series data ID of "SD1", a sampling timing of "ST1", and a sensor value of "SS1". Also, the target data ID is "OD1", a sampling timing of "DT1", and a sensor value of "DS1". Also, the distribution data ID is "AD1", and the distribution data is "ADD1".

[0041] In the example shown in Figure 3, the sampling timing, etc. is expressed by an abstract code such as "ST1", but the sampling timing, etc. may also be a specific numerical value, a specific character string, or the file format of a file containing various information indicating the sampling timing, etc.

[0042] (Regarding the threshold information storage unit 122) The threshold information storage unit 122 stores information related to a predetermined threshold. An example of the threshold information storage unit 122 according to the embodiment is shown in Fig. 4. In the example shown in Fig. 4, the threshold information storage unit 122 has items such as "threshold ID" and "threshold information."

[0043] The "threshold ID" is an identifier that identifies the threshold information. The "threshold information" is information related to the threshold associated with the "threshold ID." For example, the information related to the threshold indicates a predetermined threshold. For example, the predetermined threshold is a value estimated from the time series data at the time of an abnormality, and is an arbitrary value. The predetermined threshold may be set in consideration of the distance between the time series data at the time of an abnormality and the standard time series data. In this case, the predetermined threshold is the lower limit of the calculated distance.

[0044] For example, in Fig. 4, "T1" identified by the threshold ID has threshold information "TH1." Note that, although the threshold information, etc. is expressed by an abstract code such as "TH1" in the example shown in Fig. 4, the threshold information, etc. may be a specific numerical value, a specific character string, or the file format of a file containing various information indicating the threshold information, etc.

[0045] (Regarding the control unit 130) The control unit 130 is a controller, and is realized, for example, by a CPU (Central Processing Unit) or an MPU (Micro Processing Unit) using RAM as a work area to execute various programs stored in a storage device inside the information processing device 100. The control unit 130 is also a controller, and is realized, for example, by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0046] 2, the control unit 130 has an acquisition unit 131, a generation unit 132, a calculation unit 133, a determination unit 134, and a judgment unit 135, and realizes or executes the functions and actions of information processing described below. Note that the internal configuration of the control unit 130 is not limited to the configuration shown in FIG. 2, and may have other configurations as long as they perform the information processing described below. Furthermore, the connection relationship between the processing units included in the control unit 130 is not limited to the connection relationship shown in FIG. 2, and may be other connection relationships.

[0047] (Regarding the acquisition unit 131) The acquiring unit 131 acquires various information. Specifically, the acquiring unit 131 acquires target data, which is time-series data including sampling values at each sampling timing, and standard time-series data that serves as a reference for determining whether the time-series data is normal.

[0048] For example, the acquiring unit 131 acquires the target data and the standard time series data from the data server 10. Then, the acquiring unit 131 stores the acquired target data and the standard time series data in the data set storage unit 121.

[0049] (Regarding the generation unit 132) The generating unit 132 generates various information. Specifically, for each sampling value of the standard time series data, the generating unit 132 determines the sampling timing corresponding to the sampling value as a relative sampling timing, which is a sampling timing measured from a predetermined starting point. In this case, the generating unit 132 determines the starting point of the relative sampling timing for each sampling timing of the standard time series data.

[0050] For example, the generating unit 132 determines, as the starting point of the relative sampling timing, a sampling timing that is a predetermined value earlier than the current sampling timing.

[0051] Next, the generating unit 132 generates distribution data indicating the distribution range of the determined relative sampling timings, and stores the generated distribution data in the data set storage unit 121.

[0052] Next, the generating unit 132 determines a range to which the sampling timing is to be associated based on the distribution data. For example, the generating unit 132 determines the distribution range of the relative sampling timings indicated by the distribution data as the range to which the sampling timing is to be associated.

[0053] Here, the distribution data generation process will be described with reference to Fig. 5. Fig. 5 is a conceptual diagram showing the generation process according to the embodiment. In the example of Fig. 5, table TA1 shows the sampling timing of the standard time series data and the sampling timing of normal data 1 to N (for example, N is an arbitrary integer) that is the basis of the standard time series data.

[0054] Here, in table TA1, when the sampling timing of the standard time series data is "1", the sampling timing of normal data 1 is "1", the sampling timing of normal data 2 is "1", and the sampling timing of normal data N is "1".

[0055] Also, in table TA1, when the sampling timing of the standard time series data is "20", the sampling timing of normal data 1 is "17", the sampling timing of normal data 2 is "22", and the sampling timing of normal data N is "19".

[0056] Also, in table TA1, when the sampling timing of the standard time series data is "35", the sampling timing of normal data 1 is "28", the sampling timing of normal data 2 is "33", and the sampling timing of normal data N is "38".

[0057] Here, it is assumed that the starting point of the relative sampling timing is sampling timing "20" of the standard time series data. In this case, the relative sampling timing is represented by the difference between the current sampling timing and the starting sampling timing. In the example of Figure 5, the relative sampling timing is represented by the difference between sampling timing "35" of the standard time series data and sampling timing "20" of the standard time series data.

[0058] Next, generate distribution data of relative sampling timings corresponding to the sampling timings of the standard time series data. First, calculate the difference between each sampling timing of normal data corresponding to the relative sampling timing. In the example of Figure 5, the difference between the sampling timing of normal data 1 corresponding to the starting sampling timing, which is "17," and the sampling timing of normal data 1 corresponding to the current sampling timing, which is "28," is "+11."

[0059] Furthermore, the difference between the sampling timing of normal data 2 corresponding to the starting sampling timing, which is "22," and the sampling timing of normal data 2 corresponding to the current sampling timing, which is "33," is "+11."

[0060] Furthermore, the difference between the sampling timing of normal data N corresponding to the starting sampling timing, which is "19", and the sampling timing of normal data N corresponding to the current sampling timing, which is "38", is "+19".

[0061] Next, the generating unit 132 generates distribution data from the difference between each sampling timing in the normal data. In the example of FIG. 5, the generating unit 132 generates distribution data "+11 to +19" in the normal data 1 to N. In this way, the generating unit 132 generates distribution data "+11 to +19" of the relative sampling timing corresponding to the sampling timing "35" of the standard time series data. Then, the generating unit 132 stores the generated distribution data "+11 to +19" in the dataset storage unit 121.

[0062] The distribution data may include, for example, a minimum value and a maximum value of the sampling timing, and may also indicate a standard deviation of the sampling timing.

[0063] (Regarding the calculation unit 133) The calculation unit 133 calculates a distance matrix based on the target data and the standard time series data. In this case, in the distance matrix, the rows correspond to the standard time series data, and the columns correspond to the target data. For example, the calculation unit 133 calculates the absolute difference between the sensor values of the target data and the sensor values of the standard time series data, as an example of the absolute difference between the sensor values of the target data and the sensor values of the standard time series data. In this way, the calculation unit 133 calculates the distance matrix based on the target data and the standard time series data. Then, the calculation unit 133 stores the calculated distance matrix in the storage unit 120.

[0064] Furthermore, the calculation unit 133 calculates the distance between the target data and the standard time series data by accumulating values on the warping path. For example, in an example where the warping path is determined in advance, the calculation unit 133 calculates the distance between the target data and the standard time series data using values on the warping path. For example, the calculation unit 133 refers to the distance matrix stored in the storage unit 120 and accumulates the absolute difference of each matrix element forming the warping path to calculate the distance between the standard time series data and the target data. The matrix elements here correspond to the matrix elements of the distance matrix based on the target data and the standard time series data. The calculation unit 133 then stores the calculated distance in the storage unit 120.

[0065] (Regarding the decision unit 134) The determination unit 134 determines a warping path based on the target data and the standard time series data so that the warping path is included in the distribution range of the relative sampling timing determined based on the distribution data.

[0066] For example, the determination unit 134 compares all sampling values of the target data and the standard time series data with each other, and determines a warping path that minimizes the distance between the target data and the standard time series data. In this case, the determination unit 134 determines the warping path by restricting the route of the warping path to within the distribution range of the relative sampling timing determined based on the distribution data stored in the dataset storage unit 121. To explain using the example of FIG. 5, the determination unit 134 determines the warping path by restricting the route of the warping path to within the range of the distribution data "+11 to +19" generated by the generation unit 132 in FIG. 5.

[0067] To give a more specific example, the determination unit 134 refers to the distance matrix calculated by the calculation unit 133 and stored in the memory unit 120 and determines to exclude matrix elements included in the warping path that are located outside the distribution range of the relative sampling timing from the route of the warping path.

[0068] For example, suppose one matrix element included in a warping path is located outside the distribution range of relative sampling timings. In this case, the determination unit 134 changes the one matrix element included in the warping path to a matrix element in the same column but a different row that is included in the warping path and is located within the distribution range of relative sampling timings. In this way, the determination unit 134 determines a warping path that is limited to the distribution range of relative sampling timings.

[0069] (Regarding the determination unit 135) The determination unit 135 determines that the target data is abnormal when the distance calculated by the calculation unit 133 is equal to or greater than a predetermined threshold. For example, the determination unit 135 determines that the target data is abnormal when the distance calculated by the calculation unit 133 and stored in the storage unit 120 is equal to or greater than a predetermined threshold stored in the threshold information storage unit 122. On the other hand, the determination unit 135 determines that the target data is normal when the distance stored in the storage unit 120 is less than the predetermined threshold stored in the threshold information storage unit 122.

[0070] [4. Processing Procedure (1)] Next, the procedure of the decision process executed by the information processing device 100 according to the embodiment will be described with reference to Fig. 6. Fig. 6 is a flowchart showing an example of the flow of the decision process executed by the information processing device 100 according to the embodiment.

[0071] For example, it is assumed that the standard time series data and the target data have been previously acquired from the data server 10. It is also assumed that alignment has already been performed at all points between the standard time series data and the normal data.

[0072] 6, the information processing device 100 assigns 1 to the sampling timing t (step S101). Subsequently, the acquisition unit 131 acquires information about the sampling timing of the target data from the dataset storage unit 121 (step S102).

[0073] Then, the generating unit 132 selects a starting point of the relative sampling timing from among the sampling timings of the target data (step S103). For example, the generating unit 132 determines a sampling timing that is a predetermined value earlier than the current sampling timing as the starting point of the relative sampling timing.

[0074] Next, the acquiring unit 131 acquires the sampling timing of the standard time series data that corresponds to the starting point of the relative sampling timing (step S104). For example, the acquiring unit 131 acquires the sampling timing of the standard time series data that is associated with the relative sampling timing of the target data.

[0075] Then, the generating unit 132 generates distribution data based on the relative sampling timings (step S105). For example, the generating unit 132 generates distribution data that indicates the distribution range of the relative sampling timings.

[0076] Next, the generating unit 132 determines a range to which the sampling timing is to be associated based on the distribution data (step S106). For example, the generating unit 132 determines the distribution range of the relative sampling timings indicated by the distribution data as the range to which the sampling timing is to be associated.

[0077] Then, the determination unit 134 determines whether to associate the target data based on the association range (step S107). For example, the determination unit 134 compares all sampling values of the target data and the standard time series data with each other, and determines the warping path that minimizes the distance between the target data and the standard time series data as the optimal warping path. In this case, the determination unit 134 determines the warping path by restricting the route of the warping path to within the distribution range of the relative sampling timing determined based on the distribution data stored in the dataset storage unit 121. That is, the determination unit 134 performs DTW within the distribution range to search for the optimal point.

[0078] To give a more specific example, the determination unit 134 refers to the distance matrix stored in the memory unit 120 and determines to exclude matrix elements included in the warping path that are located outside the distribution range of the relative sampling timing from the route of the warping path.

[0079] For example, suppose one matrix element included in a warping path is located outside the distribution range of relative sampling timings. In this case, the determination unit 134 changes the one matrix element included in the warping path to a matrix element in the same column but a different row that is included in the warping path and is located within the distribution range of relative sampling timings. In this way, the determination unit 134 determines a warping path that is limited to the distribution range of relative sampling timings.

[0080] Next, the information processing device 100 assigns t+1 to the sampling timing t (step S108). Then, the information processing device 100 determines whether t is greater than a predetermined value (step S109). For example, the predetermined value is a value indicating the data length of the target data.

[0081] Specifically, if the information processing device 100 determines that t is smaller than the predetermined value (step S109; No), the process returns to step S102. On the other hand, if the information processing device 100 determines that t is larger than the predetermined value (step S109; Yes), the information processing ends.

[0082] In the above example, in step S101, an example has been described in which the information processing device 100 assigns 1 to the sampling timing t, but this is not limited to this and may be performed by various configurations of the information processing device 100 (for example, the acquisition unit 131, etc.). In addition, in step S108, an example has been described in which the information processing device 100 assigns t+1 to the sampling timing t, but this is not limited to this and may be performed by various configurations of the information processing device 100. In addition, in step S109, an example has been described in which the information processing device 100 determines whether t is greater than a predetermined value, but this is not limited to this and may be performed by various configurations of the information processing device 100.

[0083] In the above example, the information processing is repeated until t reaches a predetermined value, but the present invention is not limited to this. For example, the information processing may be performed only for some of the sampling timings of the target data.

[0084] [5. Processing Procedure (2)] Next, the procedure of the determination process executed by the information processing device 100 according to the embodiment will be described with reference to Fig. 7. Fig. 7 is a flowchart showing an example of the flow of the determination process executed by the information processing device 100 according to the embodiment.

[0085] 7, the calculation unit 133 calculates the distance between the target data and the standard time series data (step S201). For example, the calculation unit 133 calculates the distance between the target data and the standard time series data by integrating the absolute differences on the warping paths between the target data and the standard time series data.

[0086] Next, the determination unit 135 determines whether the distance calculated by the calculation unit 133 is equal to or greater than a predetermined threshold (step S202). Specifically, if the distance is equal to or greater than the predetermined threshold (step S202; Yes), the determination unit 135 determines that the target data is abnormal (step S203).

[0087] On the other hand, if the distance is less than the predetermined threshold (step S202; No), the determining unit 135 determines that the target data is normal (step S204).

[0088] [6. Effects] As described above, the information processing device 100 according to the embodiment acquires target data and standard time series data. Next, the information processing device 100 determines, for each sampled value of the standard time series data, a sampling timing corresponding to the sampled value as a relative sampling timing. The information processing device 100 then generates distribution data indicating a distribution range of the relative sampling timings. Next, the information processing device 100 determines a warping path based on the target data and the standard time series data so that the target data falls within the distribution range of the relative sampling timings determined based on the distribution data.

[0089] For example, by using relative sampling timing, the information processing device 100 can accurately evaluate the variation in the time required for each process, regardless of its position on the time axis. Furthermore, when performing alignment using DTW, the generated distribution data serves as a limiting condition. Then, when the target data is aligned with the standard time series data, the points of the target data are linked to appropriate points of the standard time series data based on the distribution data. This allows the information processing device 100 to perform appropriate synchronization processing in line with the distribution of normal data. In this way, the information processing device 100 can achieve more accurate alignment.

[0090] Furthermore, the information processing device 100 determines the sampling timing that is a predetermined value before the current sampling timing as the starting point of the relative sampling timing, and determines the relative sampling timing, thereby avoiding the influence of variations in the required time of processes prior to the process that includes the sample.

[0091] [7. Modifications] The information processing device 100 described above may be implemented in various different forms other than the above embodiment, so other embodiments of the information processing device 100 will be described below.

[0092] [7-1. Sampling Values] In the above embodiment, the sampled values are sensor values, but the sampled values are not limited to these. For example, the sampled values may be any values that indicate measurement values such as temperature and pressure measured in the production facility.

[0093] [7-2. Starting point of relative sampling timing] In the above embodiment, an example has been described in which the starting point of the relative sampling timing is a sampling timing that is a predetermined value before the current sampling timing, but the present invention is not limited to this.

[0094] For example, the starting point of the relative sampling timing may be a sampling timing corresponding to a time acquired from actual operation history data. For example, such a starting point may be a sampling timing corresponding to a time recorded as a signal of a process operation. To give a more specific example, the information processing device 100 may determine, as the starting point of the relative sampling timing, the first sampling timing having the same value as the sampling value at the sampling timing in the identification information indicating the step acquired in the process.

[0095] The starting point of the relative sampling timing may also be estimated by machine learning. For example, the information processing device 100 may identify points in the time-series data where the trajectory of the data changes significantly based on conventional machine learning techniques. The information processing device 100 may then determine the point closest to the sampling timing among the identified points as the starting point of the relative sampling timing. Note that it is desirable for the starting point of the relative sampling timing to be a point close to the starting point of the process that includes the sample.

[0096] [7-3. Information Processing Device] Furthermore, the information processing device 100 may be an information processing device that monitors target data. For example, the information processing device 100 may determine whether time-series data including sampled values of temperature, pressure, etc. acquired from the data server 10 is normal. For example, the determination unit 135 determines that the target data is abnormal when the distance calculated by the calculation unit 133 is equal to or greater than a predetermined threshold stored in the threshold information storage unit 122. On the other hand, the determination unit 135 determines that the target data is normal when the distance calculated by the calculation unit 133 is less than the predetermined threshold stored in the threshold information storage unit 122.

[0097] The determination process executed by the information processing device 100 may determine whether the target data is normal or not when the degree of match between the aligned target data and the aligned normal data is less than a predetermined threshold.

[0098] For example, the calculation unit 133 calculates the degree of coincidence between the aligned target data and the aligned normal data. Note that, for the calculation process, various conventional techniques for calculating the degree of coincidence between data can be adopted.

[0099] Then, the determination unit 135 may determine that the target data is abnormal when the degree of coincidence calculated by the calculation unit 133 is less than a predetermined threshold. On the other hand, the determination unit 135 may determine that the target data is normal when the degree of coincidence calculated by the calculation unit 133 is equal to or greater than a predetermined threshold.

[0100] [7-4. Relative Sampling Timing] Furthermore, the information processing device 100 may combine the sampling timing and the relative sampling timing, such as by taking a product set, to obtain a new relative sampling timing.

[0101] [8. Other] Furthermore, among the processes described in the above embodiments and modifications, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using known methods. In addition, the information including the processing procedures, specific names, various data, and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified. For example, the various information shown in each drawing is not limited to the information shown in the drawings.

[0102] Furthermore, the components of each device shown in the figure are conceptual functional components and do not necessarily have to be physically configured as shown in the figure. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figure, and all or part of them can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc.

[0103] Furthermore, the above-described embodiments and modifications can be combined as appropriate within the scope of not causing any contradiction in the processing content.

[0104] Furthermore, the above-mentioned "section, module, unit" can be read as "means" or "circuit," etc. For example, a decision unit can be read as a decision means or a decision circuit.

[0105] Although some of the embodiments of the present application have been described in detail above with reference to the drawings, these are merely examples, and the present invention can be implemented in other forms that include the embodiments described in the Disclosure of the Invention section and that have undergone various modifications and improvements based on the knowledge of those skilled in the art. [Explanation of symbols]

[0106] N Network 1. Information Processing Systems 10 Data Server 100 Information processing device 110 Communications Department 120 Storage section 121 Dataset Storage 122 Threshold information storage unit 130 control section 131 Acquisition Department 132 Generation part 133 Calculation Unit 134 Decision Section 135 Judgment section

Claims

1. an acquisition step of acquiring target data, which is time-series data including sampling values at each sampling timing, and standard time-series data that serves as a reference for determining whether the target data is normal; a generating step of determining, for each sampled value of the standard time series data, a sampling timing corresponding to the sampled value as a relative sampling timing, which is a sampling timing measured from a predetermined starting point, and generating distribution data indicating a distribution range of the relative sampling timing; a determination step of determining a warping path based on the target data and the standard time series data so that the target data is included in a distribution range of the relative sampling timing determined based on the distribution data; An information processing method comprising:

2. The determining step All the sampled values of the target data and the standard time series data are compared with each other, and a warping path that minimizes the distance between the target data and the standard time series data is determined as the warping path.

2. The information processing method according to claim 1,

3. The determining step The warping path is determined by restricting the route of the warping path within a distribution range of the relative sampling timing determined based on the distribution data.

3. The information processing method according to claim 2.

4. The method further includes a calculation step of calculating the distance between the target data and the standard time series data by integrating values on the route of the warping path.

3. The information processing method according to claim 2.

5. The method further includes a determination step of determining that the target data is abnormal when the distance calculated by the calculation step is equal to or greater than a predetermined threshold.

5. The information processing method according to claim 4.

6. The distribution data is a range of variation in values represented by the difference between the relative sampling timing and the sampling timing of normal data, which is time-series data under normal conditions.

2. The information processing method according to claim 1,

7. an acquisition unit that acquires target data, which is time-series data including sampling values at each sampling timing, and standard time-series data that serves as a reference for determining whether the target data is normal; a generation unit that determines, for each of the sampled values of the standard time series data, a sampling timing corresponding to the sampled value as a relative sampling timing, which is a sampling timing measured from a predetermined starting point, and generates distribution data that indicates a distribution range of the relative sampling timing; a determination unit that determines a warping path based on the target data and the standard time series data so that the target data is included in a distribution range of the relative sampling timing determined based on the distribution data; An information processing device comprising:

Images