Information processing device, information processing method, and computer program

Abstract

To detect sections in which a measurement object is in a second state in measured waveform data relating to the measurement object.SOLUTION: An information processing device, on the basis of measured waveform data including multiple first data points relating to a measurement object that can be in both a first state and a second state and reference waveform data including multiple second data points related to the first state, generates first distance information including distances between the first and second data points, generates first correspondence data by determining one or more second data points corresponding to each first data point, generates second distance information by replacing the distances between a target data point selected from the multiple second data points and one or more first data points with set values corresponding to the target data points, generates second correspondence data by determining one or more second data points corresponding to each first data point on the basis of the second distance information, and evaluates the measured waveform data on the basis of the first and second correspondence data to detect sections in which a measurement object is in a second state.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present embodiment relates to an information processing device, an information processing method, and a computer program. [Background technology]

[0002] In time-series waveform data in which a certain waveform pattern is repeated at a predetermined cycle, such as a patient's electrocardiogram data or pressure data from a spacecraft's propulsion system, there exists a type of anomaly in which a portion of the waveform pattern is distorted (partial anomaly). By monitoring the time-series waveform data, it is possible to detect in which cycle of the waveform pattern an anomaly has occurred. In this case, there is a demand to detect the section from the start to the end of the anomaly in the waveform pattern in which the anomaly has occurred. In other words, there is a demand to detect the state (e.g., normal or abnormal) of a certain section in the waveform pattern. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2013-120128 [Non-patent literature]

[0004] [Non-Patent Document 1] Seiichi Uchida, "Special Lecture: Overview of DP Matching - Fundamentals and Various Extensions," Technical Report of the Institute of Electronics, Information and Communication Engineers, December 2006, Vol. 106, No. 428, pp. 31-36 Summary of the Invention [Problem to be solved by the invention]

[0005] The present embodiment provides an information processing device, an information processing method, and a computer program that are capable of detecting the state of a partial section in measured waveform data relating to an object to be measured. [Means for solving the problem]

[0006] The information processing device of this embodiment includes a processing unit that generates first distance information including distances between the first data points and the second data points based on measured waveform data including a plurality of first data points related to a measurement object that can be in a first state and a second state, and reference waveform data including a plurality of second data points related to the first state, generates first correspondence data that associates the first data points with the second data points by determining one or more corresponding second data points for each of the first data points based on the first distance information, generates second distance information by replacing the distance between a target data point, which is at least one second data point selected from the plurality of second data points, and one or more of the first data points with a set value corresponding to the target data point, generates second correspondence data that associates the first data points with the second data points by determining one or more corresponding second data points for each of the first data points based on the second distance information, and generates second correspondence data that associates the first data points with the second data points, and evaluates the measured waveform data based on the first correspondence data and the second correspondence data. [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 2 is a diagram illustrating an outline of the operation of the information processing device according to the first embodiment. [Figure 2] 1 is a block diagram showing a configuration of an information processing device according to a first embodiment. [Figure 3] 10A and 10B are diagrams illustrating an example of normal measured waveform data and reference waveform data. [Figure 4] 10A and 10B are diagrams illustrating an example of measured waveform data including a partial abnormality and reference waveform data. [Figure 5] FIG. 10 is a diagram illustrating an example of a first cost matrix. [Figure 6] FIG. 10 is a diagram illustrating an example of a first DTW matrix. [Figure 7] FIG. 10 is a diagram illustrating a method for setting a setting value. [Figure 8] FIG. 10 is a diagram illustrating an example of a second cost matrix. [Figure 9] FIG. 10 is a diagram illustrating an example of a second DTW matrix. [Figure 10] FIG. 4 is a diagram illustrating the relationship between a first similarity and a second similarity. [Figure 11] 4 is a flowchart illustrating the operation of the information processing device according to the first embodiment. [Figure 12] FIG. 10 is a block diagram showing a configuration of an information processing device according to a second embodiment. [Figure 13] FIG. 4 is a block diagram showing a detailed configuration of a second similarity calculation section. [Figure 14] FIG. 10 is a diagram illustrating an example of an inverse DTW matrix. [Figure 15] FIG. 10 is a diagram illustrating an example of a forward DTW matrix. [Figure 16] FIG. 10 is a diagram illustrating a method for calculating an evaluation vector. [Figure 17] FIG. 10 is a diagram illustrating a configuration of an information processing system according to a third embodiment. [Figure 18] FIG. 2 is a diagram illustrating a hardware configuration of an information processing device according to each embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0008] Hereinafter, the present embodiment will be described with reference to the drawings. In the drawings, the same or corresponding elements are designated by the same reference numerals, and detailed descriptions thereof will be omitted as appropriate.

[0009] (Embodiment 1) FIG. 1 is a diagram illustrating an outline of the operation of an information processing device 101 according to the first embodiment. For example, periodic time-series data measured from a measurement object is input to the information processing device 101. In the periodic time-series data, waveforms of approximately the same pattern appear repeatedly. Examples of such time-series data include electrocardiogram data of a patient, pressure data of a spacecraft propulsion device, and sensing data of equipment that repeatedly performs multiple processes at regular intervals.

[0010] In the case of electrocardiogram data, the length of the waveform in one cycle in the time direction expands or contracts depending on the patient's heart rate, but if the length of each waveform in the time direction is appropriately adjusted, the waveforms will generally match.In the case of pressure data from a propulsion device, the length of the waveform in one cycle in the time direction expands or contracts depending on the output of the propulsion device, but if the length of each waveform in the time direction is appropriately adjusted, the waveforms will generally match.

[0011] In the case of electrocardiogram data, if a patient develops arrhythmia, a portion of the periodic waveform pattern is deformed. In the case of pressure data from a propulsion device, if the propulsion device knocks, a portion of the periodic waveform pattern is deformed. Similarly, if an abnormality occurs in one process of equipment that performs multiple processes, a portion of the waveform pattern corresponding to that process is deformed. The information processing device 101 detects such a partial deformation of the waveform pattern as a partial abnormality and detects a section in the time-series data where that partial abnormality exists. In other words, it detects a section in which the measurement object is in a second state (e.g., an abnormal state) that is different from a first state (e.g., a normal state or a normal state). However, the state (second state) for which the section is detected is not limited to an abnormal state, and may be any state that is different from the normal state. For example, it may be a state that is a precursor to an abnormality or a state in which deterioration has progressed significantly.

[0012] 2 is a block diagram showing the configuration of an information processing device 101 according to embodiment 1. The information processing device 101 includes a data acquiring unit 10, a data storage unit 11, a first cost calculating unit 12, a first similarity calculating unit 13, a second cost calculating unit 14, a second similarity calculating unit 15, an anomaly detecting unit 16, and an output unit 17.

[0013] Some or all of the data acquisition unit 10, data storage unit 11, first cost calculation unit 12, first similarity calculation unit 13, second cost calculation unit 14, second similarity calculation unit 15, anomaly detection unit 16, and output unit 17 constitute a processing unit that performs processing related to this embodiment.

[0014] The data acquisition unit 10 receives periodic time series data measured by a sensor that senses a measurement object that can be in both a first state and a second state. The time series data is sampled in the time direction at a predetermined sampling frequency and quantized in the amplitude direction with a predetermined number of bits. The data acquisition unit 10 sequentially extracts measured waveform data corresponding to each period of the periodic time series data. Each measured waveform data includes M data points (first data points) from the first to the Mth. The first first data point corresponds to the first first data point, and the Mth first data point corresponds to the last first data point.

[0015] Data storage unit 11 pre-stores reference waveform data including M data points (second data points) from 1st to Mth related to a first state (e.g., a normal state), and predetermined setting values ​​(described in detail below) corresponding to each of the M data points. The first second data point in the reference waveform data corresponds to the first second data point, and the Mth second data point corresponds to the last second data point. Note that, in the first embodiment, a case will be described in which the measured waveform data and the reference waveform data have the same number of data points, but the numbers of data points may differ between the two.

[0016] Although there is no particular limitation on the method for creating the reference waveform data, for example, test waveform data corresponding to one period of periodic time series data measured from a measurement object in a normal state are sequentially extracted, and the barycenter waveform of these multiple test waveform data is obtained, thereby creating the reference waveform data. The barycenter waveform can be obtained, for example, by the soft-DTW method or the DBA method.

[0017] Figure 3 shows an example of normal measured waveform data and an example of reference waveform data when M=10. The upper part of Figure 3 shows the normal measured waveform data, and the lower part of Figure 3 shows the reference waveform data. Comparing the two waveform data, they are roughly consistent.

[0018] Figure 4 shows an example of measured waveform data containing a partial anomaly when M=10, and an example of reference waveform data. The upper part of Figure 4 shows the measured waveform data containing a partial anomaly, and the lower part of Figure 4 shows the reference waveform data. Comparing the two waveform data, a large discrepancy occurs between the sixth and seventh data points, which correspond to the partial anomaly contained in the measured waveform data.

[0019] The first cost calculation unit 12 calculates the distances between M data points (first data points) included in the measured waveform data and M data points (second data points) included in the reference waveform data. More specifically, for each of the M first data points included in the measured waveform data, the distance between each of the M second data points included in the reference waveform data is calculated. Based on the calculated distances, the first cost calculation unit 12 generates distance information (first distance information) including the distances between the first data points and the second data points.

[0020] Here, a matrix (first distance matrix) of size M×M is generated in which the calculated distance is stored in the corresponding element for each pair of the first data point and the second data point as the first distance information. More specifically, the distance between the first data point and the second data point is defined as the cost, and a matrix (first cost matrix D1) is calculated in which the cost is stored in the corresponding element for each pair of the first data point and the second data point.

[0021] In general, the distance δ(x,y) between two data points x and y, i.e., the cost, can be defined as follows, where γ is a real parameter:

[0022]

number

[0023] In the first embodiment, the case where γ=1 in the above equation, that is, the Euclidean distance is considered as the cost. In this case, if the reference waveform data is {x1, x2,... xM} and the measured waveform data is {y1, y2,... yM}, the first cost matrix D1 is calculated as follows:

[0024]

number

[0025] FIG. 5 shows two examples of the first cost matrix D1 (first distance information or first distance matrix). The left side of FIG. 5 shows the first cost matrix D1 calculated from normal measured waveform data and reference waveform data. The right side of FIG. 5 shows the first cost matrix D1 calculated from measured waveform data including a partial abnormality and reference waveform data. The element in the i-th row and j-th column in the first cost matrix D1 is represented as element (i, j). For example, the element in the 6th row and 6th column of the first cost matrix D1 on the right side of FIG. 5 is element (6, 6), and the value of element (6, 6) is 1.5. The value of element (6, 6) being 1.5 indicates that the distance (cost) between the sixth data point x1 of the reference waveform data and the sixth data point y6 of the measured waveform data including a partial abnormality is 1.5.

[0026] The first similarity calculation unit 13 determines one or more corresponding second data points for each first data point based on the first cost matrix D1, and generates first correspondence data indicating the correspondence between each first data point and the determined second data point. For example, the first correspondence data is generated by sequentially searching for one or more corresponding second data points for each first data point so that the sum of the distances between the corresponding second data points is minimized and all second data points are associated with at least one first data point. The first correspondence data is then generated by calculating a first evaluation value based on the sum of the distances between corresponding first and second data points included in the first correspondence data. In this embodiment, the first evaluation value corresponds to the first similarity S1 between the measured waveform data and the reference waveform data, but may be a value other than similarity. The processing of the first similarity calculation unit 13 is described in more detail below.

[0027] In the first embodiment, the first correspondence data is generated and the first similarity S1 is calculated from the first cost matrix D1 using DTW (Dynamic Time Warping). DTW is characterized by its robustness against expansion / contraction in the time direction and phase shifts. By using DTW, even when the lengths of the measured waveform data and the reference waveform data are different or when the phases of the two data are shifted, the similarity between the two waveform data can be calculated taking these factors into consideration.

[0028] In detail, the first similarity calculation unit 13 calculates the first DTW matrix V1 from the first cost matrix D1 according to the following equation.

[0029]

number

[0030] That is, starting from an element corresponding to the first pair of a first data point and a first second data point in the first cost matrix D1 (first distance matrix), each adjacent element is sequentially traced from the starting point. A route from the starting point to an element other than the starting point that minimizes the sum of the distances of the elements included in the route is identified, and the sum of the distances of the elements included in the identified route is calculated as a DTW value (first intermediate evaluation value). The DTW values ​​calculated for each element are then stored in the corresponding elements to generate a first DTW matrix V1 (first evaluation matrix). The value of the element in the first DTW matrix V1 corresponding to the first pair of a first data point and a first second data point may be the same as the value of the starting point of the first cost matrix D1.

[0031] The value of the end point (M, M) of the first DTW matrix V1 (the element corresponding to the pair of the last first data point and the last second data point) is the first similarity S1 (first evaluation value). When generating the first DTW matrix V1, the correspondence between the first data point and the second data point indicated by each element included in the path traced from the start point to the end point in the first cost matrix D1 corresponds to the first correspondence data described above. The sum of the distances indicated by each correspondence included in the first correspondence data is the first similarity S1 (first evaluation value). The smaller the difference between the measured waveform data and the reference waveform data, the smaller the first similarity S1; if the two data completely match, the first similarity S1 is zero. Conversely, the greater the difference between the measured waveform data and the reference waveform data, the greater the first similarity S1.

[0032] In this embodiment, the similarity (evaluation value) is defined so that the greater the similarity (evaluation value), the greater the difference between the measured waveform data and the reference waveform data. However, the similarity (evaluation value) may also be defined so that the greater the similarity (evaluation value), the closer the measured waveform data and the reference waveform data are to each other.

[0033] Fig. 6 shows two examples of the first DTW matrix V1 (first evaluation matrix). The left side of Fig. 6 is the first DTW matrix V1 calculated from normal measured waveform data, and the value of the end point (10,10) is 0.6, so the first similarity S1 is 0.6. The right side of Fig. 6 is the first DTW matrix V1 calculated from measured waveform data including a partial abnormality, and the value of the end point (10,10) is 1.3, so the first similarity S1 is 1.3.

[0034] In each first DTW matrix V1 in FIG. 6, elements surrounded by solid-line circles represent a route (referred to as an optimal route) that minimizes the sum of costs (distances) for reaching the end point (10,10) from the starting point (1,1). This optimal route corresponds to the first correspondence data described above. More specifically, the correspondence between the first data point and the second data point indicated by each element included in the optimal route corresponds to the first correspondence data. The first similarity S1 corresponds to a first evaluation value calculated from the first correspondence data.

[0035] The optimal path obtained for the first DTW matrix V1 represents the maximum likelihood correspondence between each data point included in the reference waveform data and each data point included in the measured waveform data. For example, in the case of the first DTW matrix V1 on the right side of FIG. 6, the first to fourth data points (first data points) of the reference waveform data correspond to the first to fourth data points (second data points) of the measured waveform data, respectively. The fifth data point of the reference waveform data corresponds to two data points, namely, the fifth and sixth data points of the measured waveform data. The sixth to seventh data points of the reference waveform data correspond to the seventh to eighth data points of the measured waveform data, respectively. The eighth to tenth data points of the reference waveform data correspond to the eighth to tenth data points of the measured waveform data, respectively.

[0036] The second cost calculation unit 14 calculates the second cost matrix D2 by replacing some of the elements in the first cost matrix D1 with set values ​​corresponding to the second data points corresponding to the elements. More specifically, the second cost calculation unit 14 selects at least one second data point from multiple second data points included in the reference waveform data as a target data point, and replaces the distances (costs) between the target data point and one or more first data points in the first cost matrix D1 with set values ​​corresponding to the target data point. In the first embodiment, the second cost calculation unit 14 selects all elements from the pth row to the qth row of the first cost matrix D1 and replaces the values ​​of each element with set values ​​corresponding to the second data points corresponding to each row to calculate the second cost matrix D2. The second cost matrix D2 corresponds to second distance information or a second distance matrix in which the distances (costs) between the target data point and one or more first data points in the first cost matrix D1 are replaced with set values ​​corresponding to the target data points.

[0037] The data storage unit 11 previously stores M set values ​​{r1, r2, . . . rM} corresponding to M data points (second data points) included in the reference waveform data. The set values ​​have values ​​corresponding to a range expected as a distance between the second data point and the first data point associated with the second data point in the processing of the first similarity calculation unit 13 in measured waveform data acquired when the measurement target is in a normal state. For example, the set value has an upper limit value of the range or a value greater than the upper limit value.

[0038] FIG. 7 is a diagram illustrating a method for generating or determining M setting values.

[0039] First, T test waveform data sets including M data points (third data points) related to the first state (e.g., normal state) are prepared. The method for creating the test waveform data is not particularly limited, but for example, the test waveform data used to create the above-mentioned reference waveform data can be used as is.

[0040] Next, for each test waveform data, the maximum likelihood correspondence between M data points (second data points) included in the reference waveform data and M data points (third data points) included in the test waveform data is determined by DTW. In the example of FIG. 7, the data point x1 of the reference waveform data corresponds to the data point y of the test waveform data y(k). (k) 1, the data point x2 of the reference waveform data corresponds to the data point y of the test waveform data y(k). (k) 2, y (k) 3, and the data point x3 of the reference waveform data corresponds to the data point y of the test data set y(k). (k) 4, the data point x4 of the reference waveform data corresponds to the data point y of the test waveform data y(k). (k) 4, and so on until the data point xM of the reference waveform data corresponds to the data point y of the test waveform data y(k). (k) Corresponds to M.

[0041] Next, the mean(i) and standard deviation std(i) of the distance δ(xi, y(k)) between the i-th data point xi of the reference waveform data and one or more data points corresponding to the i-th data point of each test waveform data y(k) are calculated.

[0042]

number

number

[0043] Finally, for M setpoints {r1, r2, rM}, the ith setpoint ri is calculated by adding a certain margin, for example, α times the standard deviation, to the average distance (average cost) between the ith data point of the reference waveform data and one or more data points corresponding to the ith data point of each test waveform data. The value of the parameter α is preferably, for example, about 3.0.

[0044]

number

[0045] The method for generating the M setting values ​​{r1, r2, rM} is not limited to the above method, and various modifications and adjustments may be made to the above method. Alternatively, a method completely different from the above method may be used.

[0046] The second cost calculation unit 14 sets an interval [p, q] defined by positive integers p and q that satisfy the relationship 1≦p≦q≦M, and replaces the value of each element in the i-th row of the first cost matrix D1 with the i-th setting value r i for all positive integers i∈[p, q]. In this way, the second cost matrix D2 (second distance information or second distance matrix) is calculated for each combination of p and q.

[0047] FIG. 8 is a diagram showing two examples of the second cost matrix D2 when p = 6 and q = 7. In this case, each element in the 6th row of the first cost matrix D1 is replaced with the 6th set value r6, and each element in the 7th row of the first cost matrix D1 is replaced with the 7th set value r7. In this embodiment, the values (costs) of all elements are replaced in row units, but only the values (costs) of one or more partial elements may be replaced. For example, the values of the leftmost element and the rightmost element among each element in a certain row may not be replaced.

[0048] In the first embodiment, since M = 10, as patterns of the interval [p, q], when p < q 10 There are 45 patterns for C2, 10 patterns when p = q, and a total of 55 patterns can be considered. The second cost calculation unit 14 calculates 55 second cost matrices D2 corresponding to these 55 intervals [p, q]. That is, a plurality of second cost matrices D2 are calculated for one first cost matrix D1. Hereinafter, the second cost matrix D2 obtained by replacing the values of the elements related to the interval [p, q] in the first cost matrix D1 with the set values will be denoted as "D2" [p,q] " as needed. Note that this notation does not represent the element (p, q) of the second cost matrix D2.

[0049] The second similarity calculation unit 15 calculates a plurality of second similarities S2 (second evaluation values) corresponding to the plurality of second cost matrices D2 calculated for a plurality of intervals [p, q] that satisfy the relationship 1 ≤ p ≤ q ≤ M. The method of calculating the second similarity S2 may be performed using DTW from the second cost matrix D2 in the same manner as in the case of the first similarity S1. Although the details have been described above, they will be described again.

[0050] Based on the second cost matrix D2, one or more second data points corresponding to each first data point are determined, and second correspondence data (optimal routes) indicating the correspondence between each first data point and the determined second data points are generated. For example, the second correspondence data is generated by searching for one or more second data points corresponding to each first data point so that the sum of the distances to the corresponding second data points is minimized and so that all second data points are associated with at least one first data point. Then, a second evaluation value is calculated as second similarity S2 based on the sum of the distances between corresponding first data points and second data points included in the second correspondence data.

[0051] That is, starting from the element corresponding to the first pair of first data point and first second data point in the second cost matrix D2, each adjacent element is sequentially traced from the starting point. A route from the starting point to elements other than the starting point that minimizes the sum of the distances of the elements included in the route is identified, and the sum of the distances of the elements included in the identified route is calculated as a DTW value (second intermediate evaluation value). Then, the DTW values ​​calculated for each element are stored in the corresponding elements to generate a second DTW matrix V2 (second evaluation matrix). The value of the element corresponding to the first pair of first data point and first second data point in the second DTW matrix V2 may be the same as the value of the starting point of the second cost matrix D2. The value of the end point (M, M) of the second DTW matrix V2 (the element corresponding to the last pair of first data point and last second data point) is the second similarity S2 (second evaluation value). When generating the second DTW matrix V2, the correspondence between the first data point and the second data point indicated by each element included in the route (optimal route) traced from the start point to the end point in the second cost matrix D2 corresponds to the second correspondence data. The sum of the distances related to each correspondence included in the second correspondence data becomes the second similarity S2 (second evaluation value).

[0052] FIG. 9 shows the second cost matrix D2 for the interval [6, 7]. [6,7] The second DTW matrix V2 is calculated corresponding to [6,7] and the second similarity S2 [6,7]9 shows an example of a normal waveform (left side) and an example of a waveform including a partial abnormality (right side). The left side of FIG. 9 shows the second DTW matrix V2 calculated from normal measured waveform data. [6,7] In this example, the value of the end point (10,10) is 1.6, so the second similarity S2 [6,7] The second evaluation value is 1.6. The right side of Figure 9 shows the second DTW matrix V2 calculated from the measured waveform data containing partial abnormalities. [6,7] In this example, the value of the end point (10,10) is 1.0, so the second similarity S2 [6,7] (Second evaluation value) is 1.0.

[0053] The abnormality detection unit 16 evaluates the measured waveform data based on the optimal path (first corresponding data) identified in the first DTW matrix V1 and the optimal path (second corresponding data) identified in the second DTW matrix V2, thereby detecting a section in the measured waveform data where the measurement object is in a second state (here, an abnormal state). Based on the first similarity S1 (first evaluation value) and a plurality of second similarities S2 (second evaluation values), it determines whether or not an abnormal state section exists in the measured waveform data. If an abnormal state section exists, the second similarity S2 [P,Q] The section of the measured waveform data corresponding to the section [P, Q] relating to the second similarity S2 is determined as the section in the abnormal state. [P,Q] The section of the measured waveform data corresponding to the section [P, Q] is a section of the measured waveform data that includes a first data point that corresponds to the target data point (the second data point included in the section [P, Q]) in the two corresponding data (optimal route). The processing of the abnormality detection unit 16 will be described in detail below.

[0054] The anomaly detection unit 16 selects the second similarity S2 with the highest evaluation (best), i.e., the smallest value, from among the multiple second similarities S2, and checks whether the smallest second similarity S2 is smaller than the first similarity S1 by a threshold or more. In other words, it checks whether the difference between the first similarity S1 and the smallest second similarity S2 is a threshold or more (whether the best second similarity S2 is improved over the first similarity S1 by a threshold or more).

[0055] For example, the smallest second similarity is S2 [P,Q] As described above, the smallest second similarity S2 [P,Q] The second cost matrix D2 corresponding to [P,Q] is calculated by replacing the value of each element in the i-th row of the first cost matrix D1 with the i-th setting value ri for all positive integers i∈[P,Q]. This calculation is equivalent to replacing each distance (cost) between the i-th data point of the reference waveform data and the 1st to Mth data points of the measured waveform data with a distance (cost) obtained by adding a certain margin to the average distance (average cost) between the i-th data point of the reference waveform data and the corresponding data points of the multiple test waveform data.

[0056] Therefore, when the measured waveform data is normal and the above replacement is performed, the smallest second similarity S2 [P,Q] Even the best second similarity S2 [P,Q] Even the similarity S1 should be worse than the first similarity S1.

[0057] On the other hand, as shown on the right side of Figure 10, the second similarity S2 [P,Q] is smaller than the first similarity S1 by a threshold or more, in other words, the best second similarity S2 [P,Q]If the first similarity S1 is improved by more than the threshold, it means that each distance (cost) between the ith data point of the reference waveform data and the 1st to Mth data points of the measured waveform data is greater than the average distance (average cost) between the ith data point of the reference waveform data and the corresponding data point of the test waveform data plus a certain margin. This means that an abnormality exists in the section of the measured waveform data corresponding to the section [P,Q] of the reference waveform data (the measurement target is in an abnormal state). For example, when P=6 and Q=7, by referring to the second DTW matrix V2(i,j) on the right side of FIG. 9, it is determined that an abnormality exists in the section [6,7] of the measured waveform data corresponding to the section [6,7] of the reference waveform data.

[0058] In summary, the smallest second similarity S2 [P,Q] is smaller than the first similarity S1 by a threshold value or more (the first similarity S1 and the smallest second similarity S2 [P,Q] (When the difference between the two is equal to or greater than a threshold value), in other words, the best second similarity S2 [P,Q] is improved by a threshold or more compared to the first similarity S1, the abnormality detection unit 16 determines whether the second similarity S2 of the reference waveform data is improved by a threshold or more. [P,Q] It is determined that an abnormality exists in the section of the measured waveform data corresponding to the section [P, Q] relating to

[0059] On the other hand, the second similarity S2 [P,Q] is not smaller than the first similarity S1 by a threshold value or more (when the first similarity S1 and the smallest second similarity S2 [P,Q] (When the difference between the two is less than the threshold value), in other words, the best second similarity S2 [P,Q] If the first similarity S1 has not improved by more than the threshold value, the abnormality detection unit 16 determines that the section of the measured waveform data corresponding to the above section [P, Q] does not contain an abnormality (the object to be measured is not in an abnormal state), that is, the object to be measured is in a normal state.

[0060] When the abnormality detection unit 16 determines that the above-mentioned section of the measured waveform data contains an abnormality, the output unit 17 displays the section in which the abnormality exists. The displayed information may be, for example, information indicating the section in which the abnormality exists in association with the waveform of the measured waveform data.

[0061] FIG. 11 is a flowchart illustrating the operation of the information processing device 101 according to the first embodiment.

[0062] In step S11, the data acquiring section 10 sequentially extracts a plurality of pieces of measured waveform data corresponding to one cycle of the periodic time series data from the time series data, and starts processing each piece in turn.

[0063] In step S12, the first cost calculation unit 12 calculates a first cost matrix D1 (first distance information or first distance matrix) based on the distances (costs) between M data points included in the measured waveform data currently being processed and M data points included in the reference waveform data.

[0064] In step S13, the first similarity calculation section 13 calculates a first similarity S1 (first evaluation value) between the measured waveform data currently being processed and the reference waveform data based on the first cost matrix D1.

[0065] In step S14, the second cost calculation unit 14 calculates multiple second cost matrices D2 by replacing some of the elements of the first cost matrix D1 with predetermined set values. That is, considering an interval [p, q] defined by positive integers p and q satisfying the relationship 1≦p≦q≦M, the second cost calculation unit 14 calculates the second cost matrix D2 by replacing the value of each element in the i-th row of the first cost matrix D1 with the i-th set value r i for all positive integers i∈[p, q]. By performing this calculation for all intervals [p, q] satisfying the relationship 1≦p≦q≦M, multiple second cost matrices D2 (second distance information or second distance matrices) are obtained.

[0066] In step S15, the second similarity calculation unit 15 calculates a plurality of second similarities S2 corresponding to a plurality of second cost matrices D2.

[0067] In step S16, the anomaly detection unit 16 checks whether the measured waveform data currently being processed contains an anomaly based on the first similarity S1 and the multiple second similarities S2. For example, the second similarity S2 with the highest evaluation is selected from the multiple second similarities S2, and if the difference between the first similarity S1 and the second similarity S2 is equal to or greater than a threshold, it determines that the measured waveform data contains an anomaly, and if it is less than the threshold, it determines that the measured waveform data does not contain an anomaly.

[0068] If the measured waveform data currently being processed contains an abnormality (S16=YES), the output unit 17 displays the section in which the abnormality exists (step S17), and the process returns to step S11. This section is the section of measured waveform data corresponding to the section [p, q] for which the selected second similarity S2 was calculated. More specifically, this is the section of measured waveform data that includes a first data point corresponding to a second data point in the section [p, q] in the optimal route (second corresponding data) corresponding to the section [p, q].

[0069] On the other hand, if the measured waveform data currently being processed does not contain any abnormality (S16=NO), the output unit 17 does not display anything and returns to the processing of step S11.

[0070] As described above, the information processing device 101 according to the first embodiment calculates a first cost matrix D1 based on the distances (costs) between multiple data points included in the measured waveform data and multiple data points included in the reference waveform data, and calculates a first similarity S1 between the measured waveform data and the reference waveform data from the first cost matrix D1. The information processing device 101 replaces the element values ​​(costs) corresponding to each section [p, q] in the first cost matrix D1 with predetermined set values ​​to calculate multiple second cost matrices D2, and calculates multiple second similarities S2 from the multiple second cost matrices D2. The information processing device 101 detects anomalies included in the measured waveform data and the sections in which the anomalies exist, based on the first similarities S1 and the multiple second similarities S2.

[0071] With the above features, the information processing device 101 according to the first embodiment can detect an abnormality contained in the measured waveform data and a section in which the abnormality exists.

[0072] Furthermore, the information processing device 101 replaces each distance (cost) between one or more data points included in the section [p, q] of the reference waveform data and multiple data points included in the measured waveform data with a predetermined set value. The information processing device 101 identifies that an abnormality exists in the section of the measured waveform data corresponding to the section [p, q] of the reference waveform data.

[0073] With the above-described features, when the measured waveform data contains a partial abnormality, the section in which the partial abnormality exists can be identified.

[0074] Furthermore, in the first embodiment, the first similarity S1 is calculated from the first cost matrix D1 using the DTW. Similarly, the second similarity S2 is calculated from the second cost matrix D2 using the DTW. Due to these characteristics, even when the lengths of the measured waveform data and the reference waveform data are different or when the phases of the two are shifted, it is possible to identify the section containing the abnormality contained in the measured waveform data.

[0075] The information processing device 101 also includes a data acquisition unit 10 that sequentially extracts multiple pieces of measurement waveform data from periodic time-series data. With this feature, it is possible to sequentially detect anomalies contained in periodic time-series data and sections in which anomalies exist.

[0076] If the section where an abnormality exists can be predicted or if it is desired to detect only an abnormality existing in a specific section, the above section [p, q] may be limited to a predetermined range of the reference waveform data. This makes it possible to reduce the number of patterns of the second cost matrix D2 and the second similarity S2. As a result, it is possible to reduce the amount of calculation by the second cost calculation unit 14 and the second similarity calculation unit 15.

[0077] (Another configuration example in the first embodiment) In the first embodiment described above, the presence or absence of an abnormality (the state of the object to be measured) in a corresponding section of the measured waveform data is detected based on the difference between the first similarity S1 and the second similarity S2 (for example, the second similarity S2 with the highest evaluation), but other methods may be used. For example, the first correspondence data and the second correspondence data may be input to a pre-trained machine model or artificial intelligence, and the presence or absence of an abnormality may be obtained as an output. For example, a model for determining the presence or absence of an abnormality may be constructed by performing machine learning or the like on the difference between the first correspondence data and the second correspondence data, i.e., the difference in the pattern (shape) of the corresponding optimal route, for each of the normal state and the abnormal state of the object to be measured.

[0078] (Embodiment 2) 12 is a block diagram showing the configuration of an information processing device 201 according to the second embodiment. The information processing device 201 includes a second similarity calculation unit 215 instead of the second similarity calculation unit 15 of the information processing device 101 according to the first embodiment. The second similarity calculation unit 215 calculates the second similarity S2 [p,q] has already been calculated, the second similarity S2 for the interval [p, q+1] is calculated using this. [p,q+1] This calculates the second similarity S2 [p,q+1] Since the amount of calculation required to calculate the above can be reduced, efficient or short-time processing is possible.

[0079] 13 is a block diagram showing a detailed configuration of the second similarity calculation unit 215. The second similarity calculation unit 215 includes an inverse DTW value calculation unit 215a, a forward DTW value calculation unit 215b, and a synthesis unit 215c.

[0080] The inverse DTW value calculation unit 215a calculates the following DTW matrix from the first cost matrix D1 according to the following equation, starting from (M,M) and ending at (1,1): This DTW matrix is ​​called the inverse DTW matrix B (third evaluation matrix).

[0081]

number

[0082] That is, starting from the element corresponding to the pair of the last first data point and the last second data point, each adjacent element is sequentially traced from the starting point. A route from the starting point to an element other than the starting point that minimizes the sum of the distances of the elements included in the route is identified, and the sum of the distances of the elements included in the identified route is calculated as a DTW value (third intermediate evaluation value). Then, the DTW values ​​calculated for each element are stored in the corresponding elements to generate an inverse DTW matrix B (third evaluation matrix). Note that the value (M, M) of the inverse DTW matrix B may be the same as the value (M, M) of the first cost matrix D1.

[0083] Furthermore, since there is only one first cost matrix D1, the process of calculating the inverse DTW matrix B only needs to be performed once.

[0084] The inverse DTW value calculation unit 215a identifies the inverse DTW values ​​from the end point (M, M) of the first cost matrix D1 to each element in the q+2th row, based on the inverse DTW matrix B. Specifically, each element in the q+2th row of the inverse DTW matrix B represents the sum of the costs (third intermediate evaluation value) of reaching each element in the q+2th row from the end point (M, M) of the first cost matrix D1. Therefore, the inverse DTW value calculation unit 215a extracts the value of each element in the q+2th row of the inverse DTW matrix B. That is, based on the inverse DTW matrix B, the value of each element corresponding to the q+2th second data point (third intermediate evaluation value) is obtained.

[0085] 14 is a diagram showing an example of an inverse DTW matrix B for M=10, and an example of extracting the value of each element in the q+2-th row from the inverse DTW matrix B. For example, when q=6, the inverse DTW value calculation unit 215a extracts the value of each element in the 6+2=8-th row of the inverse DTW matrix B.

[0086] The forward DTW value calculation unit 215b calculates the second cost matrix D2 [p,q] The forward DTW value from the starting point (1,1) to each element of the qth row and the second cost matrix D2 [p,q+1] and the second cost matrix D2 [p,q+1]Calculate the forward DTW value from the starting point (1,1) to each element in the q+2th row of the second cost matrix D2 [p,q] corresponds to the second distance matrix, and the second cost matrix D2 [p,q+1] corresponds to the third distance matrix.

[0087] In detail, the forward DTW value calculation unit 215b calculates the second cost matrix D2 [p,q+1] and the DTW matrix already calculated for the interval [p,q] (forward DTW matrix A [p,q] ) based on the DTW matrix (forward DTW matrix A [p,q+1] The forward DTW matrix A is calculated according to the following formula: [p,q] corresponds to a fourth evaluation matrix in which DTW values ​​(fourth intermediate evaluation values) are stored in elements corresponding to at least the first to qth second data points.

[0088]

number

[0089] However, the forward DTW matrix A [p,q+1] Each element from the 1st row to the qth row of [p,q] Therefore, in reality, the forward DTW matrix A [p,q+1] Only the elements in the q+1th and q+2th rows of

[0090] Figure 15 shows the forward DTW matrix A for the case of p=6 and q=6. [6,6] The forward DTW matrix A is calculated based on [6,7] 1 is a diagram showing an example of a forward DTW matrix A. [6,7] Each element in the first to sixth rows of the [6,6] Therefore, the forward DTW matrix A [6,7] Only the elements in the 6+1=7th line and the 6+2=8th line need to be calculated. The value of each element in the 8th line corresponds to the fifth intermediate evaluation value.

[0091] The forward DTW value calculation unit 215b calculates the forward DTW matrix A[p,q+1] The second cost matrix D2 is created by extracting the value of each element in the q+2th row of [p,q+1] The forward DTW value (fifth intermediate evaluation value) of the optimal path (path with the smallest total cost) from the starting point (1,1) to each element in the q+2th row of the forward DTW matrix A is obtained. [6,7] Each element in the 6+2=8th row is extracted.

[0092] The synthesizing unit 315c synthesizes the inverse DTW value (third intermediate evaluation value) of the optimal path from the end point (M, M) of the first cost matrix D1 calculated by the inverse DTW value calculation unit 215a to each element of the q+2th row, and the second cost matrix D2 calculated by the forward DTW value calculation unit 215b. [p,q+1] Based on the forward DTW value (fifth intermediate evaluation value) of the optimal path from the starting point (1,1) to each element in the q+2th row, the second similarity S2 [p,q+1] In other words, the route obtained by combining (combining) these two optimal routes corresponds to the optimal route for the section [p, q+1], and the evaluation value (second evaluation value) of this optimal route is the second similarity S2 [p,q+1] The data including the correspondence between the first data point and the second data point associated with each element included in the optimum route becomes the second correspondence data.

[0093] In detail, as shown in FIG. 16, the synthesis unit 315c synthesizes B(q+2,j) corresponding to the inverse DTW value from the end point (M,M) to each element in the q+2th row, where j=1 to M, and A(q+2,j) corresponding to the forward DTW value from the start point (1,1) to each element in the q+2th row. [p,q+1] Based on (q+2,j) and D1(q+2,j), which corresponds to the cost (distance) of each element in the q+2th row, the evaluation vector cost(j) is calculated according to the following formula: This formula means that the cost (distance) of each element is subtracted from the sum of the third and fifth intermediate evaluation values ​​of each element, and a vector is generated in which the calculated value is stored in the component corresponding to each element.

[0094]

number

[0095] The value of each component of the evaluation vector cost(j) calculated in this way corresponds to the total cost from the starting point (1,1) via the element (q+2,j) to the end point (M,M). Then, the component value with the highest evaluation (the smallest component value in this embodiment) is selected. The selected component value is added to the second cost matrix D2 [p,q+1] The minimum total cost of reaching the end point (M,M) from the starting point (1,1) through the elements corresponding to the selected components, that is, the second cost matrix D2 [p,q+1] The second similarity S2 to [p,q+1] (second evaluation value).

[0096] In the example of FIG. 16, the smallest component of the 10 components of the evaluation vector cost(j) obtained as a result of synthesis is 1.0, which is the second cost matrix D2 [6,7] Therefore, the synthesis unit 215c calculates the second similarity S2 [6,7] is calculated as 1.0.

[0097] As described above, the information processing device 201 according to the second embodiment calculates the second similarity S2 [p,q+1] When calculating the inverse DTW value from the end point (M, M) of the first cost matrix D1 to each element of the q+2th row, the inverse DTW value is calculated. [p,q] Based on the forward DTW values ​​from the starting point (1,1) to the qth row of [p,q+1] The forward DTW values ​​are calculated from the starting point (1,1) to each element in the q+2th row, and these forward DTW values ​​and inverse DTW values ​​are combined to obtain the second similarity S2 [p,q+1] Calculate.

[0098] Due to the above characteristics, the information processing device 201 according to the second embodiment calculates the second similarity S2 [p,q] If the second similarity S2 has already been calculated, [p,q+1] This makes it possible to reduce the amount of calculation required to calculate the second similarity S2.

[0099] (Embodiment 3) FIG. 17 is a diagram showing the configuration of an information processing system 300 according to the third embodiment. The information processing system 300 includes a control target device 302 that is the measurement target, an information processing device 301, and a control device 303. The control target device 302 is not particularly limited, but may be, for example, a propulsion device of a spacecraft. Here, the description will be given assuming the case of a propulsion device. A pressure sensor (not shown) is mounted inside the control target device 302, and pressure data acquired by the pressure sensor is input to the information processing device 301 as time-series sensor data (hereinafter, time-series data). The control device 303 may be part of the information processing device 301 as a control unit.

[0100] The information processing device 301 is the same as the information processing device 101 or 201 according to the first or second embodiment. The information processing device 301 detects an anomaly contained in the time-series data and the section in which the anomaly exists. The control device 303 adjusts a control parameter contained in a control signal to be transmitted to the control target device 302 based on the detection result of the information processing device 301. For example, the control device 303 maintains the current value of the control parameter while the time-series data is normal, and when an anomaly is detected in the time-series data, changes the value of the control parameter according to the section in which the anomaly exists. Furthermore, if the anomaly contained in the time-series data is a fatal anomaly that suggests a failure of the propulsion device 302, the control device 303 transmits a control signal to stop the propulsion device 302.

[0101] (Hardware configuration) 18 is a diagram showing the hardware configuration of the information processing devices 101 to 301 according to each embodiment. The information processing devices 101 to 301 are configured by a computer device 600. The computer device 600 includes a CPU 601, an input interface 602, a display device 603, a communication device 604, a main storage device 605, and an external storage device 606, which are interconnected by a bus 607.

[0102] The CPU (Central Processing Unit) 601 executes a signal processing program, which is a computer program, on the main storage device 605. The signal processing program is a program that realizes each of the above-mentioned functional configurations of the information processing devices 101 to 301. The signal processing program may be realized not as a single program, but as a combination of multiple programs and scripts. Each functional configuration is realized by the CPU 601 executing the signal processing program.

[0103] The input interface 602 is a circuit for inputting operation signals from input devices such as a keyboard, a mouse, and a touch panel to the information processing devices 101 to 301.

[0104] The display device 603 displays data output from the information processing devices 101 to 301. The display device 603 is, for example, but not limited to, an LCD (liquid crystal display), an organic electroluminescence display, a CRT (cathode ray tube), or a PDP (plasma display). Data output from the computer device 600 can be displayed on the display device 603. The display device 603 corresponds to the output unit 17 of the information processing devices 101 to 301 according to each embodiment.

[0105] The communication device 604 is a circuit that enables the information processing devices 101 to 301 to communicate with an external device wirelessly or via a cable. Data can be input from the external device via the communication device 604. The data input from the external device can be stored in the main memory device 605 or the external memory device 606. The communication device 604 corresponds to the data acquisition unit 10 of the information processing devices 101 to 301 according to each embodiment.

[0106] The main memory device 605 stores a signal processing program, data required for executing the signal processing program, data generated by executing the signal processing program, etc. The signal processing program is expanded and executed on the main memory device 605. The main memory device 605 is, for example, a RAM, a DRAM, or an SRAM, but is not limited to these. The data storage unit 102 of the information processing devices 101 to 301 according to each embodiment may be constructed on the main memory device 605.

[0107] The external storage device 606 stores the signal processing program, data required for executing the signal processing program, and data generated by executing the signal processing program. These signal processing programs and data are read into the main storage device 605 when the signal processing program is executed. The external storage device 606 is, for example, a hard disk, an optical disk, a flash memory, or a magnetic tape, but is not limited to these. The data storage unit 102 of the information processing devices 101 to 301 according to each embodiment may be constructed on the external storage device 606.

[0108] The signal processing program may be pre-installed in the computer device 600, or may be stored in a storage medium such as a CD-ROM. The signal processing program may also be uploaded onto the Internet.

[0109] Furthermore, the information processing devices 101 to 301 according to each embodiment may be configured as a single computer device 600, or may be configured as a system made up of a plurality of computer devices 600 connected to each other.

[0110] Although several embodiments have been described, these embodiments are presented as examples and are not intended to limit the scope of the embodiments. These embodiments can be implemented in various other forms, and various omissions, substitutions, modifications, combinations, etc. can be made without departing from the spirit of the embodiments. These embodiments and their modifications are included in the scope of the claims and their equivalents, as well as the scope and spirit of the embodiments.

[0111] This embodiment can also be configured as follows. [Item 1] generating first distance information including distances between the first data points and the second data points based on measured waveform data including a plurality of first data points relating to a measurement object that is in a first state and also in a second state, and reference waveform data including a plurality of second data points relating to the first state; determining one or more corresponding second data points for each of the first data points based on the first distance information, thereby generating first correspondence data that associates the first data points with the second data points; generating second distance information by replacing a distance between a target data point, which is at least one second data point selected from the plurality of second data points, and one or more of the first data points with a set value corresponding to the target data point; determining one or more corresponding second data points for each of the first data points based on the second distance information, thereby generating second correspondence data that associates the first data points with the second data points; a processing unit that detects a section in the measured waveform data in which the object to be measured is in the second state by evaluating the measured waveform data based on the first correspondence data and the second correspondence data; An information processing device comprising: [Item 2] the set value has a value corresponding to an expected range of distance between the target data point and the first data point associated with the target data point when the measurement target is in the first state. Item 1. An information processing device according to item 1. [Item 3] the processing unit calculates a first evaluation value by the sum of the distances between the first data points and the second data points based on the first correspondence data; calculating a second evaluation value based on the second correspondence data as a sum of distances between the first data points and the second data points; determining whether or not there is a section in the measured waveform data in which the object to be measured is in the second state based on the first evaluation value and the second evaluation value; Item 3. The information processing device according to item 1 or 2. [Item 4] the processing unit determines that there is a section in the measured waveform data in which the object to be measured is in the second state when a difference between the first evaluation value and the second evaluation value is equal to or greater than a threshold value. Item 3. An information processing device according to item 3. [Item 5] When the processing unit determines that there is a section in which the object to be measured is in the second state, it detects a section of the measured waveform data including the first data point associated with the object data point in the second correspondence data as a section in which the object to be measured is in the second state. Item 3. The information processing device according to item 3 or 4. [Item 6] the processing unit calculates the first correspondence data by searching for one or more second data points corresponding to the first data points based on the first distance information so as to minimize a sum of the distances to the corresponding second data points and so as to associate all the second data points; calculating the second correspondence data by searching for one or more second data points corresponding to the first data points based on the second distance information so as to minimize the sum of the distances to the corresponding second data points and so as to associate all the second data points; 6. The information processing device according to any one of items 1 to 5. [Item 7] the processing unit calculates the first correspondence data using a DTW, and calculates the second correspondence data using the DTW; Item 7. An information processing device according to item 6. [Item 8] the processing unit generates the second distance information by replacing the distances to all of the first data points corresponding to the target data point in the first distance information with the set value. 8. The information processing device according to any one of items 1 to 7. [Item 9] the processing unit selects p-th to q-th second data points from the plurality of second data points as target data points for a plurality of combinations of p and q, and calculates a plurality of second evaluation values ​​for each combination of values ​​of p and q; selecting one second evaluation value based on the plurality of second evaluation values, and evaluating the measured waveform data based on the second corresponding data used in calculating the selected second evaluation value; 6. The information processing device according to any one of items 3 to 5. [Item 10] the processing unit selects the second evaluation value with the highest evaluation from the plurality of second evaluation values. Item 10. The information processing device according to item 9. [Item 11] the processing unit generates the second distance information by replacing, for each of a plurality of combinations of p and q, the distances to all of the first data points corresponding to the target data point with the set value. Item 11. The information processing device according to item 9 or 10. [Item 12] the processing unit generates a first distance matrix in which distances between the first data point and the second data point are stored in elements corresponding to pairs of the first data point and the second data point, as the first distance information; generating a third evaluation matrix in which, when an element corresponding to a pair of the last first data point and the last second data point in the first distance matrix is ​​used as a first starting point and adjacent elements are sequentially traced to reach another element, the sum of distances of elements included in a path that minimizes the sum of distances is stored in the other element as a third intermediate evaluation value; As the second distance information, each of the second data points from the pth to the qth in the first distance matrix is ​​set as a target data point, and a second distance matrix is ​​generated in which the value of each element corresponding to the target data point is replaced with the set value; generating a fourth evaluation matrix in which a sum of distances of elements included in a path that minimizes the sum of distances when elements corresponding to the first first data point and the first second data point in the second distance matrix are traced sequentially from a second starting point to elements corresponding to the first to qth second data points as a fourth intermediate evaluation value, As the second distance information, each of the second data points from the pth to the q+1th in the first distance matrix is ​​set as a target data point, and a third distance matrix is ​​generated in which the value of each element corresponding to the target data point is replaced with the set value; using the fourth evaluation matrix, setting an element corresponding to the first first data point and the first second data point in the third distance matrix as a third starting point, tracing adjacent elements from the third starting point to each element corresponding to the q+2th second data point, calculating a fifth intermediate evaluation value as a sum of distances of the elements included in a path that minimizes the sum of distances; Obtaining the third intermediate evaluation value of each element corresponding to the q+2-th second data point based on the third evaluation matrix; selecting an element from among the elements corresponding to the q+2-th second data point based on the fifth intermediate evaluation value and the third intermediate evaluation value, and calculating the second evaluation value for the third distance matrix based on the fifth intermediate evaluation value and the third intermediate evaluation value of the selected element; The information processing device according to any one of Items 9 to 11. [Item 13] the processing unit calculates a value by subtracting the distance for each element in the second distance information from the sum of the fifth intermediate evaluation value and the third intermediate evaluation value for each element, and sets the highest evaluation value among the calculated values ​​as the second evaluation value for the third distance matrix. Item 13. An information processing device according to item 12. [Item 14] the processing unit generates a route by combining a first route, which is the route from the third starting point to the selected element in the third distance matrix, and a second route, which is the route from the first starting point to the selected element in the third evaluation matrix, and defines the correspondence between the first data point and the second data point, which are associated by each element included in the generated route, as the second correspondence data. Item 14. The information processing device according to item 12 or 13. [Item 15] The reference waveform data is created by calculating a centroid waveform of a plurality of test waveform data associated with the first state. 15. The information processing device according to any one of items 1 to 14. [Item 16] when determining third data points corresponding to each second data point of the reference waveform data for each of a plurality of test waveform data including a plurality of third data points associated with the first state, the set value is set based on an average and a standard deviation of distances between each of the second data points and the corresponding third data point; 16. The information processing device according to any one of items 1 to 15. [Item 17] the first state is a normal state, The second state is an abnormal state. 17. The information processing device according to any one of items 1 to 16. [Item 18] the processing unit acquires the measured waveform data from a sensor that senses the measurement object; 18. The information processing device according to any one of items 1 to 17. [Item 19] the processing unit sequentially extracts data for each period from the time series data having periodicity detected by the sensor as the measured waveform data; Item 19. An information processing device according to item 18. [Item 20] a control unit that controls the object to be measured based on the state of the object to be measured in the section determined by the processing unit; 20. The information processing device according to item 18 or 19, comprising: [Item 21] generating first distance information including distances between the first data points and the second data points based on measured waveform data including a plurality of first data points relating to a measurement object that is in a first state and also in a second state, and reference waveform data including a plurality of second data points relating to the measurement object having the first state; determining one or more corresponding second data points for each of the first data points based on the first distance information, thereby generating first correspondence data that associates the first data points with the second data points; generating second distance information by replacing a distance between a target data point, which is at least one second data point selected from the plurality of second data points, and one or more of the first data points with a set value corresponding to the target data point; determining one or more corresponding second data points for each of the first data points based on the second distance information, thereby generating second correspondence data that associates the first data points with the second data points; detecting a section in the measured waveform data in which the object to be measured is in the second state by evaluating the measured waveform data based on the first correspondence data and the second correspondence data; A computer-implemented information processing method. [Item 22] generating first distance information including distances between the first data points and the second data points based on measured waveform data including a plurality of first data points relating to a measurement object that is in a first state and also in a second state, and reference waveform data including a plurality of second data points relating to the first state; determining one or more corresponding second data points for each of the first data points based on the first distance information, thereby generating first correspondence data that associates the first data points with the second data points; generating second distance information by replacing a distance between a target data point, which is at least one second data point selected from the plurality of second data points, and one or more of the first data points with a set value corresponding to the target data point; determining one or more corresponding second data points for each of the first data points based on the second distance information, thereby generating second correspondence data that associates the first data points with the second data points; detecting a section in the measured waveform data in which the object to be measured is in the second state by evaluating the measured waveform data based on the first correspondence data and the second correspondence data; A computer program that is executed by a computer. [Explanation of symbols]

[0112] 10 Data Acquisition Section 11 Data storage unit 12 First Cost Calculation Section 13 First similarity calculation unit 14 Second Cost Calculation Section 15 Second similarity calculation unit 16 Abnormality detection unit 17 Output section 101 Information processing equipment 201 Information processing equipment 215 Second similarity calculation unit 215a Inverse DTW value calculation unit 215b Forward DTW value calculation unit 215c Synthesis Department 300 Information Processing Systems 301 Information processing equipment 302 Controlled Device 303 Control Device 600 Computer equipment 601 CPU 602 Input Interface 603 Display device 604 Communication equipment 605 Main storage 606 External storage device 607 Bus A ordered DTW matrix B inverse DTW matrix D1 First cost matrix D2 Second cost matrix S1 1st similarity S2 Second Similarity V1, 1st DTW row V2 2DTW row

Claims

1. generating first distance information including distances between the first data points and the second data points based on measured waveform data including a plurality of first data points relating to a measurement object that is in a first state and also in a second state, and reference waveform data including a plurality of second data points relating to the first state; determining one or more corresponding second data points for each of the first data points based on the first distance information, thereby generating first correspondence data that associates the first data points with the second data points; generating second distance information by replacing a distance between a target data point, which is at least one second data point selected from the plurality of second data points, and one or more of the first data points with a set value corresponding to the target data point; determining one or more corresponding second data points for each of the first data points based on the second distance information, thereby generating second correspondence data that associates the first data points with the second data points; a processing unit that detects a section in the measured waveform data in which the object to be measured is in the second state by evaluating the measured waveform data based on the first correspondence data and the second correspondence data; An information processing device comprising:

2. the set value has a value corresponding to an expected range of distance between the target data point and the first data point associated with the target data point when the measurement target is in the first state. The information processing device according to claim 1 .

3. the processing unit calculates a first evaluation value by the sum of the distances between the first data points and the second data points based on the first correspondence data; calculating a second evaluation value based on the second correspondence data as a sum of distances between the first data points and the second data points; determining whether or not there is a section in the measured waveform data in which the object to be measured is in the second state based on the first evaluation value and the second evaluation value; The information processing device according to claim 1 .

4. the processing unit determines that there is a section in the measured waveform data in which the object to be measured is in the second state when a difference between the first evaluation value and the second evaluation value is equal to or greater than a threshold value. The information processing device according to claim 3 .

5. When the processing unit determines that there is a section in which the object to be measured is in the second state, it detects a section of the measured waveform data that includes the first data point that is associated with the object data point in the second correspondence data as a section in which the object to be measured is in the second state. The information processing device according to claim 3 .

6. the processing unit calculates the first correspondence data by searching for one or more second data points corresponding to the first data points based on the first distance information so as to minimize a sum of the distances to the corresponding second data points and so as to associate all the second data points; calculating the second correspondence data by searching for one or more second data points corresponding to the first data points based on the second distance information so as to minimize the sum of the distances to the corresponding second data points and so as to associate all the second data points; The information processing device according to claim 1 .

7. the processing unit calculates the first correspondence data using DTW, and calculates the second correspondence data using DTW; The information processing device according to claim 6 .

8. the processing unit generates the second distance information by replacing the distances to all of the first data points corresponding to the target data point in the first distance information with the set value. The information processing device according to claim 1 .

9. the processing unit selects p-th to q-th second data points from the plurality of second data points as target data points for a plurality of combinations of p and q, and calculates a plurality of second evaluation values ​​for each combination of values ​​of p and q; selecting one second evaluation value based on the plurality of second evaluation values, and evaluating the measured waveform data based on the second correspondence data used in calculating the selected second evaluation value; The information processing device according to claim 3 .

10. the processing unit selects the second evaluation value with the highest evaluation from the plurality of second evaluation values. The information processing device according to claim 9 .

11. the processing unit generates the second distance information by replacing, for each of a plurality of combinations of p and q, the distances to all of the first data points corresponding to the target data point with the set value. The information processing device according to claim 9 .

12. the processing unit generates a first distance matrix in which distances between the first data point and the second data point are stored in elements corresponding to pairs of the first data point and the second data point, as the first distance information; generating a third evaluation matrix in which, when an element corresponding to a pair of the last first data point and the last second data point in the first distance matrix is ​​used as a first starting point and adjacent elements are sequentially traced to reach another element, the sum of distances of elements included in a path that minimizes the sum of distances is stored in the other element as a third intermediate evaluation value; As the second distance information, each of the second data points from the pth to the qth in the first distance matrix is ​​set as a target data point, and a second distance matrix is ​​generated in which the value of each element corresponding to the target data point is replaced with the set value; generating a fourth evaluation matrix in which a sum of distances of elements included in a path that minimizes the sum of distances when elements corresponding to the first first data point and the first second data point in the second distance matrix are sequentially traced from a second starting point to elements corresponding to the first to qth second data points as a fourth intermediate evaluation value, and As the second distance information, each of the second data points from the pth to the q+1th in the first distance matrix is ​​set as a target data point, and a third distance matrix is ​​generated in which the value of each element corresponding to the target data point is replaced with the set value; using the fourth evaluation matrix, setting an element corresponding to the first first data point and the first second data point in the third distance matrix as a third starting point, sequentially tracing adjacent elements from the third starting point to each element corresponding to the q+2th second data point, calculating as a fifth intermediate evaluation value the sum of distances of the elements included in a path that minimizes the sum of distances; Obtaining the third intermediate evaluation value of each element corresponding to the q+2-th second data point based on the third evaluation matrix; selecting an element from among the elements corresponding to the q+2-th second data point based on the fifth intermediate evaluation value and the third intermediate evaluation value, and calculating the second evaluation value for the third distance matrix based on the fifth intermediate evaluation value and the third intermediate evaluation value of the selected element; The information processing device according to claim 9 .

13. the processing unit calculates a value by subtracting the distance for each element in the second distance information from the sum of the fifth intermediate evaluation value and the third intermediate evaluation value for each element, and sets the highest evaluation value among the calculated values ​​as the second evaluation value for the third distance matrix. The information processing device according to claim 12.

14. the processing unit generates a route by combining a first route, which is the route from the third starting point to the selected element in the third distance matrix, and a second route, which is the route from the first starting point to the selected element in the third evaluation matrix, and defines the correspondence between the first data point and the second data point, which are associated by each element included in the generated route, as the second correspondence data. The information processing device according to claim 12.

15. The reference waveform data is created by calculating a centroid waveform of a plurality of test waveform data associated with the first state. The information processing device according to claim 1 .

16. the set value is set based on an average and a standard deviation of a distance between each of the second data points and the corresponding third data point when determining, for each of a plurality of test waveform data including a plurality of third data points associated with the first state, a third data point corresponding to each of the second data points of the reference waveform data. The information processing device according to claim 2 .

17. the first state is a normal state, the second state is an abnormal state; The information processing device according to claim 1 .

18. the processing unit acquires the measured waveform data from a sensor that senses the measurement object; The information processing device according to claim 1 .

19. the processing unit sequentially extracts data for each period from the time series data having periodicity detected by the sensor as the measured waveform data; The information processing device according to claim 18.

20. a control unit that controls the object to be measured based on the state of the object to be measured in the section determined by the processing unit; The information processing device according to claim 18, comprising:

21. generating first distance information including distances between the first data points and the second data points based on measured waveform data including a plurality of first data points relating to a measurement object that is in a first state and also in a second state, and reference waveform data including a plurality of second data points relating to the measurement object having the first state; determining one or more corresponding second data points for each of the first data points based on the first distance information, thereby generating first correspondence data that associates the first data points with the second data points; generating second distance information by replacing a distance between a target data point, which is at least one second data point selected from the plurality of second data points, and one or more of the first data points with a set value corresponding to the target data point; determining one or more corresponding second data points for each of the first data points based on the second distance information, thereby generating second correspondence data that associates the first data points with the second data points; detecting a section in the measured waveform data in which the object to be measured is in the second state by evaluating the measured waveform data based on the first correspondence data and the second correspondence data; A computer-implemented information processing method.

22. generating first distance information including distances between the first data points and the second data points based on measured waveform data including a plurality of first data points relating to a measurement object that is in a first state and also in a second state, and reference waveform data including a plurality of second data points relating to the first state; determining one or more corresponding second data points for each of the first data points based on the first distance information, thereby generating first correspondence data that associates the first data points with the second data points; generating second distance information by replacing a distance between a target data point, which is at least one second data point selected from the plurality of second data points, and one or more of the first data points with a set value corresponding to the target data point; determining one or more corresponding second data points for each of the first data points based on the second distance information, thereby generating second correspondence data that associates the first data points with the second data points; detecting a section in the measured waveform data in which the object to be measured is in the second state by evaluating the measured waveform data based on the first correspondence data and the second correspondence data; A computer program that is executed by a computer.

Images