Computer system, processing method, and processing program

Abstract

To identify one period of signal data even when periodicity of the signal data is temporarily lost.SOLUTION: A computer system has one or more processors and one or more memory resources. The one or more processors each acquire partial data of a predetermined period from signal data, divide the partial data into a plurality of pieces of division data for each of a plurality of time zones, identify a peak time of a cross-correlation function between the partial data and the signal data for every division data, calculate an evaluation value for evaluating if a temporal first order of the plurality of time zones is the same as a second order of the plurality of peak times for the plurality of pieces of division data belonging respectively to the plurality of time zones, identify the period of the signal data on the basis of the evaluation value, and acquire reference data of the identified period from the signal data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] This invention relates to a computer system, a processing method, and a processing program. [Background technology]

[0002] In manufacturing lines at factories and other facilities, products are manufactured using manufacturing equipment such as robots. However, if the manufacturing equipment deviates from normal operation, it may cause defects in the products. To prevent this, it is effective to monitor whether the manufacturing equipment is operating correctly. For example, abnormalities in the operation of the manufacturing equipment can be detected by using signal data such as current values ​​and vibration frequencies output from the manufacturing equipment. In this case, a portion of the signal data corresponding to one cycle, which is the period from when the manufacturing equipment starts working on one product until when it starts working on the next product, can be extracted from the signal data and compared with normal data for one cycle under normal conditions to detect abnormalities in the operation of the manufacturing equipment.

[0003] However, if sensors for acquiring signal data are retrofitted to the manufacturing equipment, it is difficult to obtain information from the manufacturing equipment at the start of the work, making it difficult to acquire partial data for one cycle. For this reason, a method has been proposed to identify one cycle of signal data, for example, by calculating the autocorrelation of the signal data (Patent Document 1). [Prior art documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 11-311646 [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] However, since the method of Patent Document 1 using autocorrelation assumes periodic signal data, it is impossible to identify one period based on autocorrelation for signal data whose periodicity is temporarily lost. For example, when a plurality of manufacturing apparatuses are arranged along a production line, if the operation of a certain manufacturing apparatus is delayed, a waiting time will occur in the subsequent manufacturing apparatus. In this case, a stop period during which the manufacturing apparatus stops will occur, and the periodicity of the signal data output from the sensor attached to the manufacturing apparatus will be temporarily lost.

[0006] The present invention has been made in view of such a situation, and an object thereof is to be able to identify one period of signal data even when the periodicity of the signal data is temporarily lost.

Means for Solving the Problem

[0007] This application includes a plurality of means for solving at least a part of the above problems. If an example is given, it is as follows.

[0008] In order to solve the above problems, a computer system according to an aspect of the present invention is a computer system having one or more processors and one or more memory resources, wherein the one or more processors acquire partial data for a predetermined period from signal data, divide the partial data into a plurality of divided data for each of a plurality of time zones, identify the peak time of the cross-correlation function between the divided data and the signal data for each of the divided data, calculate an evaluation value for evaluating whether a first order in time of the plurality of time zones is the same as a second order of the plurality of peak times for the plurality of divided data belonging to each of the plurality of time zones, identify the period of the signal data based on the evaluation value, and acquire reference data of the identified period from the signal data.

Effect of the Invention

[0009] According to the present invention, even when the periodicity of signal data is temporarily lost, one period of the signal data can be identified.

[0010] Other issues, configurations, and effects not mentioned above will be clarified by the following description of the embodiments. [Brief explanation of the drawing]

[0011] [Figure 1] Figure 1 is a configuration diagram of an example of a computer system according to the first embodiment. [Figure 2] Figure 2 is a schematic diagram of an example of signal data. [Figure 3] Figure 3(a) is a schematic diagram of an example of the cross-correlation function between segmented data and signal data, and Figure 3(b) is a schematic diagram showing an example of the definition of evaluation value. [Figure 4] Figure 4(a) is a schematic diagram of an example of the cross-correlation function between divided data and signal data when the predetermined interval is not equal to the period of the signal data, and Figure 4(b) is a schematic diagram showing an example of how to calculate the evaluation value in this case. [Figure 5] Figure 5(a) shows an example of measured partial data obtained from signal data, and Figure 5(b) shows an example of measured cross-correlation function between the segmented data obtained from this partial data and the original signal data. [Figure 6] Figure 6(a) shows another example of measured partial data, and Figure 6(b) shows an example of measured cross-correlation function between the segmented data obtained from this partial data and the original signal data. [Figure 7] Figure 7 is a schematic diagram illustrating an example of a method for identifying the period of signal data from the evaluation values ​​described in Figures 3(b) and 4(b). [Figure 8] Figure 8 is an example of a graph showing how evaluation values ​​were calculated while varying a predetermined period. [Figure 9] Figure 9 is a schematic diagram illustrating an example of a method for obtaining reference data. [Figure 10] Figure 10 is an example of a flowchart of the processing method according to the first embodiment. [Figure 11] Figure 11 is a schematic diagram illustrating an example of a method for obtaining inspection data to be used for anomaly detection from signal data by using reference data. [Figure 12] Figure 12 is a schematic diagram illustrating an example of a method for detecting abnormalities in test data. [Figure 13] Figure 13 is an example of a flowchart of the anomaly detection method according to the first embodiment. [Figure 14] Figure 14 is an example of a flowchart of the processing method according to the second embodiment. [Figure 15] Figure 15 is a graph showing an example of the relationship between a predetermined period and the evaluation value when the number of divisions is changed. [Figure 16] Figure 16 is an example of a flowchart of the processing method according to the third embodiment. [Figure 17] Figure 17 is a schematic diagram showing an example of signal data in the fourth embodiment. [Figure 18] Figure 18 is a schematic diagram of an example of the cross-correlation function between segmented data and signal data. [Figure 19] Figure 19 is an example of a flowchart of the processing method according to the fourth embodiment. [Modes for carrying out the invention]

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In all drawings used to describe the embodiment, the same reference numerals will be used for identical components, and repeated descriptions will be omitted as appropriate. Furthermore, in the following embodiment, it goes without saying that the components (including element steps, etc.) are not necessarily essential unless specifically stated or considered to be clearly essential in principle. Also, when we say "consisting of A," "made of A," "having A," or "containing A," it goes without saying that other elements are not excluded unless specifically stated that only that element is included. Similarly, in the following embodiment, when referring to the shape, positional relationship, etc. of components, etc., it will include those that are substantially similar or approximate to their shape, etc., unless specifically stated or considered to be clearly not the case in principle.

[0013] <First Embodiment> Figure 1 is a configuration diagram of an example of a computer system according to the first embodiment.

[0014] Computer system 100 performs data generation, transmission, reception, and various other processing tasks. The processor reads processing programs stored in memory resources, and the processor then executes the processing according to the program. Computer system 100 is a computer such as a personal computer, tablet terminal (computer), smartphone, server computer, blade server, or cloud server, and is a system that includes at least one of these computers. That is, computer system 100 also includes a system that includes, for example, a cloud server and a display computer (for example, a tablet terminal or smartphone). Furthermore, a controller that controls or manages some kind of device, including a processor and memory resources, is also an example of computer system 100.

[0015] Specifically, as shown in Figure 1, the computer system 100 includes one or more processors 101, one or more memory resources 104, one or more UI devices 102, and one or more NI (Network Interface) devices 103. The computer system 100 may also include other components. Furthermore, the processors 101, UI (User Interface) devices 102, NI devices 103, and memory resources 104 are interconnected via a bus 106.

[0016] The processor 101 is an arithmetic unit that reads various programs stored in the memory resource 104 and executes the processing corresponding to each program. Examples of processors 101 include microprocessors, CPUs (Central Processing Units), GPUs (Graphics Processing Units), FPGAs (Field Programmable Gate Arrays), quantum processors, or other arithmetic semiconductor devices.

[0017] The memory resource 104 is a storage device that stores the processing program 105 and various information, and examples include non-volatile memory and / or volatile memory. Examples of volatile memory include RAM (Random Access Memory) and ROM (Read Only Memory). Examples of non-volatile memory may be rewritable storage media such as flash memory, hard disks, or SSDs (Solid State Drives), and may also be USB (Universal Serial Bus) memory, memory cards, and hard disks. In addition, RAM such as MRAM (Magnetoresistive RAM), PRAM (Phase Change RAM), and ReRAM (Resistive RAM) may be considered non-volatile memory. The processor 101 may also provide a service to distribute the processing program 105 stored in the memory resource 104 to other computers.

[0018] The UI device 102 is an input device that inputs user (or operator) instructions to the computer system 100, and an output device that outputs information generated by the computer system 100. Input devices include, for example, keyboards, touch panels, pointing devices such as mice, and voice input devices such as microphones. Output devices include, for example, displays, printers, and speech synthesizers. Unless otherwise specified below, it is assumed that information input and output between the computer system 100 and the user is performed via the UI device 102. The UI device 102 may consist solely of an input device or solely of an output device.

[0019] NI device 103 is a communication device that communicates information with external devices. NI device 103 communicates information with external devices via a predetermined communication network 110, such as the Internet or a LAN (Local Area Network). Unless otherwise specified below, information communication between the computer system 100 (or processor 101) and external devices is assumed to be performed via NI device 103.

[0020] This computer system 100, by executing the processing program 105, identifies one cycle of the signal data output from the manufacturing device 111, as described later. For example, the manufacturing device 111 is a product manufacturing device on a production line in a factory or the like.

[0021] The manufacturing apparatus 111 is equipped with sensors 112, such as current sensors and vibration sensors. The sensors 112 may be built into the manufacturing apparatus 111 or may be added to the manufacturing apparatus 111 afterwards.

[0022] Sensor 112 outputs signal data 200 indicating changes over time in the current value, vibration intensity, etc., of the manufacturing apparatus 111. This signal data 200 is converted into a digital value by the data acquisition unit 113, which then stores it in the storage 115. The data acquisition unit 113 can perform the described processing as long as it has at least an AD converter and an interface device (a specific example similar to NI device 103) or connecting line that can communicate with the storage 115. The data acquisition unit 113 may also include components of the same type as the computer system 100 (e.g., a processor).

[0023] The storage 115 may be a secondary storage device of the computer system 100, or it may be cloud storage connected to the communication network 110.

[0024] Next, we will describe an example of the processing performed by processor 101.

[0025] Figure 2 is a schematic diagram of an example of signal data 200.

[0026] The signal data 200 contains waveforms corresponding to operations V, W, and X performed by the manufacturing apparatus 111. The period T of the signal data 200 is the time interval between the start time of one operation and the start time of the next operation, and is determined by the processor 101 as follows.

[0027] First, the processor 101 selects a predetermined interval T from the signal data 200 stored in the storage 115. in Retrieve partial data 200a.

[0028] Next, the processor 101 divides the partial data 200a into multiple partitioned data 201 to 203 for each of the multiple time zones Q1 to Q3. In this example, the number of partitions N is set to 3, but the number of partitions N may be 2 or 4 or more. Furthermore, the length of each time zone Q1 to Q3 is equal, and all are from the original predetermined interval T in It is 1 / N of that.

[0029] Next, the processor 101 identifies the peak time of the cross-correlation function between the divided data 201-203 and the signal data 200 for each of the divided data 201-203. The cross-correlation function h of the two data f and g is defined by the following equation (1).

[0030]

number

[0031] The cross-correlation function F1 is the cross-correlation function between the divided data 201 and the signal data 200. Similarly, the cross-correlation functions F2 and F3 are the cross-correlation functions between the divided data 202 and 203 and the signal data 200, respectively.

[0032] As shown in Figure 3(a), for example, multiple peaks P1 appear in the cross-correlation function F1. The peak times t1a, t1b, and t1c, when peaks P1 appear, are equal to the times when the divided data 201 and the signal data 200 overlap, and the interval between adjacent peak times is equal to the period T of the signal data 200.

[0033] Since the peak times t1a, t1b, and t1c appear at each period T, the symbols "a," "b," and "c" are used in this example as labels to identify which period each peak time belongs to.

[0034] Similarly, the cross-correlation function F2 shows peaks P2 at peak times t2a, t2b, and t2c, and the cross-correlation function F3 shows peaks P3 at peak times t3a, t3b, and t3c.

[0035] Here, as shown in Figure 2, the time zones Q1 to Q3 of the divided data 201 to 203 are arranged in this order chronologically. Therefore, the predetermined interval T in If the period T of the signal data 200 is equal to the period of "a", then, for example, in period "a", the peak P1 of segmented data 201 belonging to time zone Q1 appears first, followed by the peak P2 of segmented data 202 belonging to time zone Q2, and then the peak P3 of segmented data 203 belonging to time zone Q3. The same applies to periods "b" and "c".

[0036] Therefore, in each period of "a", "b", and "c", if the peaks P1 to P3 appear in the same order as the time periods Q1 to Q3, then the predetermined interval T in It can be determined that this is equal to the period T of signal data 200, and if not, a predetermined interval T in It can be determined that this is not equal to the period T of signal data 200.

[0037] Therefore, in this embodiment, an evaluation value is defined as shown in Figure 3(b) to evaluate whether the order of peak times for peaks P1 to P3 is the same as the order of time zones Q1 to Q3.

[0038] Figure 3(b) is a schematic diagram showing an example of the definition of evaluation values. First, the difference between peak time t2a and peak time t1a belonging to the period identified by "a" (t2a-t1a) is defined as dt21a. Also, the difference between peak time t3a and peak time t2a belonging to the period identified by "a" (t3a-t2a) is defined as dt32a. Similarly, dt21b, dt32b, dt21c, and dt32c are defined for each peak time belonging to the periods identified by "b" and "c" respectively.

[0039] In the period identified by "a", if the order of the peak times of peaks P1 to P3 is the same as the order of the time zones Q1 to Q3, then both dt21a and dt32a will be positive. Therefore, if both dt21a and dt32a are positive, the result is "○", and if either of them is negative, the result is "×".

[0040] Similarly, in the period identified by "b", if both dt21b and dt32b are positive, the result is "○", and if either of them is negative, the result is "×". The same applies to the period identified by "c".

[0041] Then, the proportion of periods judged as "○" within the number of periods "3" for each of "a", "b", and "c" is used as an evaluation value to assess whether the order of peak times for peaks P1 to P3 is the same as the order of time zones Q1 to Q3.

[0042] According to this definition, the evaluation value is a value that assesses whether the first temporal order of multiple different time zones Q1 to Q3 is the same as the second order of peak times P1 to P3. Furthermore, the evaluation value is maximized when the first order and the second order match in all of each period of "a," "b," and "c." Therefore, depending on the magnitude of the evaluation value, it is possible to evaluate whether the first temporal order of time zones Q1 to Q3 is the same as the second order of peak times P1 to P3. Note that instead of determining the sign of dt21b and dt32b in all periods of "a," "b," and "c," the sign of dt21b and dt32b may be determined in one or two periods.

[0043] In the examples in Figures 3(a) and 3(b), the order of peaks P1 to P3 and the order of time zones Q1 to Q3 are the same in all periods of "a," "b," and "c," so the percentage of the evaluation value is the maximum of 100%. Therefore, in this example, the given interval T in It can be estimated that this is equal to the period T of signal data 200.

[0044] Figure 4(a) shows the predetermined interval T inFIG. 4(a) is a schematic diagram of an example of the cross-correlation functions F1 to F3 between the divided data 201 to 203 and the signal data 200 when it is not equal to the period T of the signal data 200. Further, FIG. 4(b) is a schematic diagram showing an example of calculation of the evaluation value in this case.

[0045] In the examples of FIGS. 4(a) and 4(b), the peak P3 appears before the peak P2 in all the periods of "a", "b", and "c". Therefore, dt32 becomes negative in all the periods of "a", "b", and "c", and the percentage of the evaluation value becomes 0%.

[0046] FIG. 5(a) is a diagram showing an actual measurement example of the partial data 200a obtained from the signal data 200, and the total width of the horizontal axis corresponds to a predetermined period T in corresponding thereto. Further, FIG. 5(b) is a diagram showing an actual measurement example of the cross-correlation function between the divided data obtained from this partial data 200a and the original signal data 200. In FIG. 5(b), the number of divisions N of the divided data is set to 4. The same applies to FIG. 6(b) described later.

[0047] As shown in FIG. 5(b), in this example, the peak times appear in the order of peak P1, peak P2, peak P3, and peak P4. Therefore, it can be determined that the predetermined period T in is equal to the period T of the signal data 200.

[0048] On the other hand, FIG. 6(a) is a diagram showing another actual measurement example of the partial data 200a, and FIG. 6(b) is a diagram showing an actual measurement example of the cross-correlation function between the divided data obtained from this partial data 200a and the original signal data 200.

[0049] As shown in FIG. 6(b), in this example, the peak P1 appears before the peak P4, and the peaks P1 to P4 do not appear in order. Therefore, it can be determined that the predetermined period T in is not equal to the period T of the signal data 200.

[0050] FIG. 7 is a schematic diagram showing an example of a method for specifying the period T of the signal data 200 from the evaluation values described in FIGS. 3(b) and 4(b).

[0051] In this example, the processor 101 performs a predetermined period T in The evaluation value is calculated while changing the value. Then, the processor 101 detects a change point D where the evaluation value has changed significantly, and a predetermined period T at that change point D is calculated. in This is identified as the period T of the signal data 200. As an example, the processor 101 considers the point where the evaluation value crosses a pre-defined threshold Th as a change point D, and determines a predetermined period T at that change point D. in This can be identified as the period T of signal data 200.

[0052] Figure 8 shows the actual situation over a predetermined period T. in This is an example of a graph showing the calculation of evaluation values ​​while varying a given value T. The horizontal axis of the graph represents a predetermined period T. in Not that thing, but a predetermined period T in This refers to the number of data points included. In addition, in Figure 8, the threshold Th for the evaluation value is set to 60%.

[0053] In this case, the processor 101 processes data for a predetermined period T, which includes approximately 65,000 data points. in This is identified as the period T of signal data 200.

[0054] Once the period T can be identified in this way, reference data can be obtained from the signal data 200 to detect whether there is an anomaly in the signal data 200.

[0055] Figure 9 is a schematic diagram illustrating an example of how to obtain the reference data. In the example in Figure 9, the processor 101 obtains the reference data for the specified period T from the signal data 200.

[0056] This concludes the general explanation of the processing method performed by processor 101.

[0057] Next, we will explain a specific example of the processing method performed by processor 101.

[0058] Figure 10 is an example of a flowchart of the processing method according to this embodiment.

[0059] First, the processor 101 acquires signal data 200 from the storage 115 (see Figure 1) (step S1001), and then, for a predetermined period T in Initialize the length to 0 (step S1002).

[0060] Next, the processor 101 performs a predetermined period T in A positive increment ΔT is added to (step S1003), and the predetermined period T in A portion of data 200a (see Figure 2) having a length is obtained from the signal data 200 (step S1004).

[0061] Next, the processor 101 performs a predetermined period T in The partial data 200a is divided into multiple partitioned data for each of the N time zones obtained by dividing the time zone into N equal parts (step S1005). The following explanation will use the case where the number of partitions N is 3 as an example. In this case, the processor 101 will divide the partial data 200a into partitioned data 201 to 203 as shown in Figure 2.

[0062] Next, as shown in Figure 3(a), the processor 101 calculates the cross-correlation functions F1 to F3 between the divided data 201 to 203 and the signal data 200 (step S1006), and detects the peaks P1 to P3 of the cross-correlation functions F1 to F3 (step S1007). Then, the processor 101 identifies the peak times t1a, t1b, t1c, t2a, t2b, t2c, t3a, t3b, and t3c of the detected peaks P1 to P3 for each of the divided data 201 to 203.

[0063] Next, the processor 101 calculates the evaluation value shown in Figure 3(b) from the identified peak time (step S1008).

[0064] Then, the processor 101 determines whether there is a change point D where the evaluation value has changed significantly (step S1009). For example, as shown in Figure 7, the processor 101 determines that there is a change point D when the evaluation value crosses the threshold Th, and determines that there is no change point D when the evaluation value does not cross the threshold Th.

[0065] If it is determined that there is no change (NO), the process returns to step S1003. On the other hand, if it is determined that there is a change (YES), the process proceeds to step S1010.

[0066] In step S1010, the processor 101 identifies the period T of the signal data 200 based on the evaluation value. For example, the processor 101 identifies a predetermined period T at the change point D where the evaluation value crosses the threshold Th, as shown in Figure 7. in We identify this as the period T.

[0067] Subsequently, the processor 101 acquires reference data 205 with period T from the signal data 200, as shown in Figure 9 (step S1011).

[0068] This concludes the basic processing of the processing method according to this embodiment.

[0069] According to this processing method, as shown in Figures 3(a), (b) and 4(a), (b), the processor 101 determines the period T of the signal data 200 based on an evaluation value that assesses whether the order of the peak times of the peaks P1 to P3 of the cross-correlation function F1 to F3 is the same as the order of the time zones Q1 to Q3. Since this method does not use the autocorrelation function of the signal data 200, the period T of the signal data 200 can be determined even if the periodicity of the signal data 200 is temporarily lost.

[0070] Furthermore, the processor 101 repeats step S1003 until it is determined that there is a change point (YES) in step S1009, thereby completing the process for a predetermined period T. in This increases the predetermined period T when it is determined in step S1009 that there is a change point (YES). in This can be identified as the period T of signal data 200.

[0071] <<Anomaly Detection Method>> Next, we will explain how to detect anomalies in signal data 200 using reference data 205 (see Figure 9).

[0072] Figure 11 is a schematic diagram illustrating an example of a method for obtaining inspection data 301, which is the target of anomaly detection, from signal data 200 by using reference data 205.

[0073] In the example shown in Figure 11, the processor 101 calculates the cross-correlation function F between the signal data 200 and the reference data 205, and calculates the peak times t0a, t0b, and t0c in which the peak P appears in the cross-correlation function F.

[0074] Then, the processor 101 acquires the signal data 200 as inspection data 301-303 for a period whose length is the period T identified in step S1010 (see Figure 10), and which is centered around the respective peak times t0a, t0b, and t0c. These inspection data 301-303 become the data to be inspected.

[0075] Figure 12 is a schematic diagram illustrating an example of a method for detecting abnormalities in the inspection data 301.

[0076] The processor 101 determines whether there is an abnormality for each of the inspection data 301 to 303 by comparing it with normal data 304. Normal data 304 is data whose duration is the same as the duration of each of the inspection data 301 to 303, and represents data from when the manufacturing equipment 111 (see Figure 1) is operating normally. For example, normal data 304 is pre-stored in storage 115 by the user.

[0077] Furthermore, the algorithm for determining whether or not there is an abnormality is not particularly limited. For example, the processor 101 may determine that there is an abnormality in the test data 301 when the difference in amplitude between the normal data 304 and the test data 301 exceeds a threshold. Alternatively, the processor 101 may extract the respective frequency components of the test data 301 and the normal data 304 using the Fast Fourier Transform, and determine that there is an abnormality in the test data 301 when the difference between them exceeds a threshold. The same applies to the test data 302 and 303.

[0078] Figure 13 is an example of a flowchart of the anomaly detection method according to this embodiment.

[0079] First, the processor 101 acquires signal data 200 from the storage 115 (see Figure 1) (step S1301).

[0080] Next, the processor 101 calculates the cross-correlation function F between the reference data 205 and the signal data 200, as shown in Figure 11 (step S1302).

[0081] Next, the processor 101 detects the peak P of the cross-correlation function F and calculates the peak times t0a, t0b, and t0c in which the peak P appears (step S1303).

[0082] Next, the processor 101 extracts the test data 301-303 from the signal data 200, as shown in Figure 11 (step S1304).

[0083] Subsequently, the processor 101 stores the inspection data 301-303 in the storage 115 (see Figure 1) (step S1305).

[0084] Next, the processor 101 reads the inspection data 301 to 303 from the storage 115 and detects whether there is an abnormality in the inspection data 301 to 303 by comparing each inspection data 301 to 303 with the normal data 304, as shown in Figure 12 (step S1306).

[0085] This concludes the basic processing of the anomaly detection method according to this embodiment.

[0086] In this anomaly detection method, the reference data 205 acquired in step S1011 of Figure 10 is used. Even if the periodicity of the signal data 200 is temporarily lost, the period T of the reference data 205 can be identified as explained with reference to Figure 10, so it is possible to detect whether there is an anomaly in the signal data 200.

[0087] <Second Embodiment> In the first embodiment, as described with reference to Figure 10, step S1003 is repeated until it is determined that there is a change point (YES) in step S1009, thereby extending the predetermined period T in The initial value of 0 was increased. In contrast, in this embodiment, a predetermined period T is used as follows. in Reduce.

[0088] Figure 14 is an example of a flowchart of the processing method according to this embodiment.

[0089] First, the processor 101 acquires signal data 200 from the storage 115 (see Figure 1) (step S1401).

[0090] Next, processor 101, tmp The input is received from the user (step S1402). (Imposed period T) tmp This represents the operating cycle of the manufacturing equipment 111 (see Figure 1) as perceived intuitively by users performing daily tasks on the production line, etc.

[0091] Next, the processor 101 performs a predetermined period T in As a hypothetical period T tmp Set a value greater than (step S1403). (False period T) tmp This period relies on the user's subjective perception and may be smaller than the actual period. In that case, the hypothetical period T tmp for a predetermined period T in If we decrease that value as the initial value, over a predetermined period T in This will never be equal to the actual period. Therefore, in step S1403, a predetermined period T is set to allow for safety. in The assumed period T tmp Set it to be larger than that. For example, processor 101 sets the assumed period T tmp A value approximately 1.5 to 2 times that value over a predetermined period T in Set it as follows.

[0092] Next, steps S1004 to S1011 of Figure 10 in the first embodiment are performed. However, in this embodiment, if it is determined that there is no change (NO) in step S1009, the process moves to step S1404, and the processor 101 runs for a predetermined period T in By subtracting a positive value ΔT from this, a predetermined period T is reached until a change point is detected. in We will reduce it.

[0093] This concludes the basic processing of the processing method according to this embodiment.

[0094] In this processing method, in step S1403, the false period T tmp Based on a predetermined period T in An initial value is set, and the conversion point is detected in step S1009 while decreasing this initial value. (Provisional period T) tmp Since this is a cycle that users perceive intuitively, this value is used for a predetermined period T. in Setting this as the initial value increases the likelihood of quickly detecting change points.

[0095] <Third Embodiment> In the first embodiment, as shown in Figure 2, the number of divisions N for dividing the partial data 200a into divided data 201 to 203 was fixed to one. If the number of divisions N is made too large, for example, divided data 201 may contain only data with a value of zero. As a result, many intervals similar to divided data 201 may be generated in the original signal data 200, and as shown in Figure 3, each peak P1 to P3 may not be arranged in the order of time period Q1 to Q3. Therefore, if the predetermined period T of the partial data 200a is in Even if the period is equal to the actual period T, the evaluation value may fall below 100%.

[0096] Figure 15 shows the results for a predetermined period T when the number of divisions N is changed. in This graph shows an example of the relationship between the evaluation value and the evaluation value.

[0097] As shown in Figure 15, a predetermined period T inEven if the period T is equal to the actual period, the evaluation value can change significantly depending on the number of divisions N, potentially reducing the accuracy of detecting changes in the evaluation value. To minimize such possibilities, this embodiment uses multiple divisions N as follows.

[0098] Figure 16 is an example of a flowchart of the processing method according to this embodiment.

[0099] First, the processor 101 performs steps S1401 to S1403 and S1004 to S1008 in the order shown in Figure 14. However, in this embodiment, the processor 101 performs steps S1005 to S1008 for each division number N by changing the number of divisions N.

[0100] Then, the processor 101 calculates the average value of the evaluation values ​​across multiple divisions (step S1601).

[0101] Next, the processor 101 determines whether there is a change point where the average value of the evaluation values ​​has changed significantly (step S1602). For example, the processor 101 determines that there is a change point when the average value of the evaluation values ​​crosses the threshold Th, and determines that there is no change point when the average value of the evaluation values ​​does not cross the threshold Th.

[0102] If it is determined that there is no change here (NO), then, as in Figure 14, the processor 101 will run for a predetermined period T in step S1404. in Subtract ΔT from it.

[0103] On the other hand, if it is determined that there is a change point (YES), the process proceeds to step S1010. In step S1010, the processor 101 identifies the period T of the signal data 200 based on the average value of the evaluation values. For example, the processor 101 identifies a predetermined period T at the change point where the average value of the evaluation values ​​crosses the threshold Th. in We identify this as the period T.

[0104] Subsequently, the processor 101 obtains reference data 205 of period T from the signal data 200 (step S1011).

[0105] This concludes the basic processing of the processing method according to this embodiment.

[0106] According to the embodiment described above, when the partial data 200a is divided into multiple divided data, an evaluation value is calculated for each division N (step S1008), and the period is identified based on the average value of the evaluation values ​​(step S1010). Therefore, it is possible to suppress a decrease in the accuracy of detecting change points in the evaluation value when the evaluation value changes significantly depending on the number of divisions N.

[0107] In this example, a predetermined period T is required until a change point is detected in step S1602. in The initial value is repeatedly subtracted by ΔT, but a predetermined period T is used until a change point is detected, as shown in Figure 10. in You may also repeatedly add ΔT to the initial value of .

[0108] <Fourth Embodiment> In the first embodiment, although the period T of the signal data 200 was identified, the point in the signal data 200 at which the manufacturing apparatus 111 starts operating was not identified. In this embodiment, the processor 101 identifies the start time of the manufacturing apparatus 111 as follows.

[0109] Figure 17 is a schematic diagram showing an example of signal data 200 in this embodiment.

[0110] In this example, the waveforms corresponding to operations V, W, and X of the manufacturing apparatus 111 appear in the signal data 200. Furthermore, the period T from the start of operation V to the start of the next operation W is equal to the period T from the start of operation W to the start of the next operation X. x It is longer than that.

[0111] Here, a predetermined period T in The start time t1 is shifted in the positive direction from the start time t0 of operation V, and as a result, the predetermined period T inLet's consider the case where the data spans both operation V and operation W. In this case, as shown by arrow 201x, the same waveform as the divided data 201 of operation V appears in the waveform of the next operation W, so the period of the divided data 201 is T. Similarly, as shown by arrow 202x, the period of the divided data 202 is also T.

[0112] On the other hand, as shown by arrow 203x, the same waveform as the divided data 203 of operation W appears when the next operation X occurs, so the period of the divided data 203 is Tx.

[0113] Figure 18 is a schematic diagram of an example of the cross-correlation functions F1-F3 between the divided data 201-203 and the signal data 200 in this case.

[0114] In Figure 18, similar to Figures 3(a) and 4(a), dt21b represents the time interval between peaks P1 and P2 of segmented data 201 and 202 in adjacent time zones Q1 and Q2, respectively. Similarly, dt32b represents the time interval between peaks P2 and P3 of segmented data 202 and 203 in adjacent time zones Q2 and Q3, respectively.

[0115] As shown in Figure 18, the aforementioned periods T, T x Due to the difference in length, time interval dt32b is longer than time interval dt21b.

[0116] Thus, for a predetermined period T in If the start time t1 is different from the start time t0 of operation V, the time intervals dt21b and dt32b of the peak times will not match. In this embodiment, this is used to determine the start time of operation of the manufacturing apparatus 111.

[0117] Figure 19 is an example of a flowchart of the processing method according to this embodiment.

[0118] In this example, step S1010 is performed by executing one of the processes shown in Figures 10, 14, and 16, and it is assumed that the period T has already been determined in step S1010.

[0119] First, the processor 101 performs a predetermined period T in The period T identified in step S1010 is set as the value, and a predetermined period T in Initialize the start time t1 with an appropriate value (step S1901).

[0120] Next, the processor 101 processes data that starts from a start time t1 and has a predetermined period T in A portion of data 200a (see Figure 2) having a length is obtained from the signal data 200 (step S1902).

[0121] Next, as shown in Figure 17, the processor 101 divides the partial data 200a into divided data 201 to 203 for each of several equally spaced time periods (step S1903). In this example, the number of divisions N is set to 3, but the number of divisions N may be 2 or more than 3.

[0122] Next, as shown in Figure 18, the processor 101 calculates the cross-correlation functions F1 to F3 between the divided data 201 to 203 and the signal data 200 (step S1904), and detects the peaks P1 to P3 of the cross-correlation functions F1 to F3 (step S1905).

[0123] Next, the processor 101 calculates the time interval dt21b in adjacent time zones Q1 and Q2, and the time interval dt32b in adjacent time zones Q2 and Q3, from the time intervals of the peak times of the detected peaks P1 to P3 (step S1906).

[0124] Next, the processor 101 determines whether the respective time intervals dt21b and dt32b are the same (step S1907).

[0125] If it is determined that there is no match (NO), the process moves to step S1909, where the processor 101 reduces the start time t1 by Δt, and then returns to step S1902.

[0126] On the other hand, if it is determined that there is a match (YES), the process moves to step S1908. In step S1908, the processor 101 identifies the start time of period T as time t1. Then, the processor 101 obtains reference data 205 for period T starting at time t1 from the signal data 200.

[0127] This concludes the basic processing of the processing method according to this embodiment.

[0128] According to the embodiment described above, the processor 101 detects a time t1 such that the time intervals dt21b and dt32 of the peak times coincide by executing step S1907. Since this time t1 is equal to the actual start time of operation of the manufacturing apparatus 111, reference data 205 starting from the actual start time of operation can be obtained.

[0129] The effects described herein are merely illustrative and not limited to those described herein; other effects may also occur.

[0130] The present invention is not limited to the embodiments described above, and various modifications are included. For example, each of the embodiments described above is described in detail for the purpose of clearly illustrating the present invention, and the present invention is not necessarily limited to having all of the described components. Furthermore, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. In addition, it is possible to add, delete, or replace parts of the configuration of each embodiment with other configurations.

[0131] Furthermore, each of the above configurations, functions, processing units, and processing means may be implemented in hardware, either partially or entirely, by designing them as integrated circuits, for example. Alternatively, each of the above configurations and functions may be implemented in software by a processor interpreting and executing programs that implement each function. Information such as programs, decision tables, and files that implement each function can be stored in memory, storage devices such as HDDs and SSDs, or recording media such as IC (Integrated Circuit) cards, SD (Secure Digital) cards, and DVDs (Digital Versatile Discs). Also, control lines and information lines are shown only if deemed necessary for explanation, and not all control lines and information lines are necessarily shown in the actual product. In practice, almost all configurations can be considered interconnected. [Explanation of Symbols]

[0132] 100...Computer system, 101...Processor, 102...UI device, 103...NI device, 104...Memory resource, 105...Processing program, 106...Bus, 110...Communication network, 111...Manufacturing equipment, 112...Sensor, 113...Data acquisition unit, 115...Storage, 200...Signal data, 200a...Partial data, 201-203...Segmented data, 205...Reference data, 301-303...Inspection data, 304...Normal data.

Claims

1. A computer system having one or more processors and one or more memory resources, The one or more processors mentioned above are: By acquiring partial data for a predetermined period from the signal data, The partial data is divided into multiple segmented data for each of the multiple time periods. The peak time of the cross-correlation function between the divided data and the signal data is identified for each of the divided data. An evaluation value is calculated to assess whether the first temporal order of the multiple time periods is the same as the second order of the multiple peak times for the multiple segmented data belonging to each of the multiple time periods. Based on the evaluation value, the period of the signal data is identified. Reference data for the specified period is obtained from the signal data. Computer system.

2. The computer system according to claim 1, The one or more processors mentioned above are: The signal data is acquired as test data during a period whose duration is the specified period, centered on the peak time of the cross-correlation function between the reference data and the signal data. Based on the normal data and the inspection data during the aforementioned period, it is detected whether there is an abnormality in the inspection data. Computer system.

3. The computer system according to claim 1, The aforementioned evaluation value is maximized when the first order and the second order are the same. Computer system.

4. The computer system according to claim 1, The one or more processors mentioned above are: The evaluation value is calculated while changing the predetermined period. The predetermined period during which the evaluation value crosses a threshold is identified as the period. Computer system.

5. The computer system according to claim 4, The one or more processors mentioned above are: Accepting input for a provisional period, A value larger than the aforementioned provisional period is set as the predetermined period. The evaluation value is calculated while decreasing the predetermined period. Computer system.

6. The computer system according to claim 4, The one or more processors mentioned above are: The evaluation value is calculated while increasing the predetermined period. Computer system.

7. The computer system according to claim 1, The one or more processors mentioned above are: When the aforementioned partial data is divided into multiple divided data, the evaluation value is calculated for each division. The average value of the evaluation value across multiple divisions is calculated, The period is determined based on the above average value. Computer system.

8. The computer system according to claim 1, The one or more processors mentioned above are: For each adjacent time period, calculate the time interval between the peak times. The start time of the predetermined period such that the multiple calculated time intervals coincide is identified as the start time of the cycle. Computer system.

9. A processing method to be performed by a computer system having one or more processors and one or more memory resources, Steps include: acquiring partial data for a predetermined period from signal data, The steps include dividing the aforementioned partial data into multiple segmented data for each of the multiple time periods, The steps include: identifying the peak time of the cross-correlation function between the divided data and the signal data for each of the divided data; A step of calculating an evaluation value to evaluate whether the first temporal order of the multiple time periods and the second order of the multiple peak times for the multiple divided data belonging to each of the multiple time periods are the same, A step of identifying the period of the signal data based on the evaluation value, A step of obtaining reference data for the specified period from the signal data, A processing method that includes this.

10. The processing method according to claim 9, A step of acquiring the signal data as test data for a period of time whose duration is the specified period, and which is centered on the peak time of the cross-correlation function between the reference data and the signal data. A step of detecting whether there is an abnormality in the inspection data based on the normal data and the inspection data during the aforementioned period, A processing method that includes this.

11. The processing method according to claim 9, The aforementioned evaluation value is maximized when the first order and the second order are the same. Processing method.

12. The processing method according to claim 9, The step of identifying the period is, A step of calculating the evaluation value while changing the predetermined period, A step of identifying the predetermined period when the evaluation value crosses a threshold as the period, A processing method that includes this.

13. The processing method according to claim 12, A step to accept input for a provisional period, The step includes setting a value larger than the provisional period as the predetermined period, The step of calculating the evaluation value while changing the predetermined period is performed by calculating the evaluation value while decreasing the predetermined period. Processing method.

14. The processing method according to claim 12, The step of calculating the evaluation value while changing the predetermined period is performed by calculating the evaluation value while increasing the predetermined period. Processing method.

15. The processing method according to claim 9, The steps include: calculating the evaluation value for each number of divisions when the partial data is divided into multiple divided data; The step includes calculating the average value of the evaluation value across a plurality of divisions, The step of identifying the period was carried out by identifying the period based on the average value. Processing method.

16. The processing method according to claim 9, A step of calculating the time interval of the peak time for each adjacent time period, The step includes identifying the start time of the predetermined period such that the multiple calculated time intervals coincide as the start time of the cycle, Processing method.

17. A processing program that causes a computer system to execute the processing method described in any one of claims 9 to 16.

Images