Monitoring device, monitoring method, and monitoring program
The monitoring device enhances abnormality detection in monitored objects by using multiple features from different frequency bands to reduce false positives and improve prediction accuracy.
Patent Information
- Application Number
- JP2022134385
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-08-25
- Publication Date
- 2026-11-11
- Estimated Expiration
- 2042-08-25
AI Technical Summary
Existing technologies struggle to accurately detect signs of abnormality in monitored objects.
A monitoring device and method that acquires waveform data from monitored objects, extracts multiple feature values, and determines abnormalities using a combination of features from different frequency bands, performing operations such as addition, subtraction, multiplication, or division to enhance detection accuracy.
Improves the detection of abnormalities by reducing false positives through the use of multiple features from different frequency bands, ensuring accurate and reliable anomaly prediction.
Smart Images

Figure 0007947820000001 
Figure 0007947820000002 
Figure 0007947820000003
Abstract
Description
Technical Field
[0001] The present disclosure relates to a monitoring device, a monitoring method, and a monitoring program.
Background Art
[0002] Patent Document 1 describes an inspection apparatus that acquires waveform data measured when an inspection object is driven for each inspection object and inspects for an abnormality in the inspection object, the inspection apparatus comprising: feature extraction means for extracting a feature value from the waveform data for each inspection object; and environment change detection means for detecting that a change in the inspection environment has occurred when the feature values extracted for each inspection object continuously exceed a predetermined threshold value across a plurality of inspection objects.
Prior Art Documents
Patent Documents
[0003]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0004] In the conventional technology, there is room for improvement in terms of grasping signs of abnormality in a monitored object.
[0005] An object of the present disclosure is to improve the technology for grasping signs of abnormality in a monitored object.
Means for Solving the Problems
[0006] A monitoring device according to the present disclosure is a monitoring device including a processing unit and an acquisition unit, wherein the processing unit acquires data representing a waveform emitted from a monitored object via the acquisition unit, extracts a plurality of feature values from the data representing the waveform, and determines whether there is a sign of abnormality in the monitored object based on the plurality of feature values.
[0007] The monitoring method relating to this disclosure is a monitoring method performed by a monitoring device, and includes acquiring data showing a waveform emitted from an object to be monitored, extracting a plurality of feature quantities from the data showing the waveform, and determining whether or not there are signs of an abnormality in the object to be monitored based on the plurality of feature quantities.
[0008] The monitoring program relating to this disclosure causes a computer to perform the following actions: acquire data showing waveforms emitted from the object being monitored; extract multiple feature quantities from the data showing the waveforms; and determine whether or not there are signs of an abnormality in the object being monitored based on the multiple feature quantities. [Effects of the Invention]
[0009] According to this disclosure, it is possible to improve the technology for detecting signs of abnormalities in monitored objects. [Brief explanation of the drawing]
[0010] [Figure 1] This figure shows the configuration of a monitoring system according to one embodiment of the present disclosure. [Figure 2] This figure shows the operation of a monitoring device according to one embodiment of the present disclosure. [Figure 3] This diagram illustrates the first and second feature quantities in this disclosure. [Modes for carrying out the invention]
[0011] Hereinafter, an embodiment of this disclosure will be described with reference to the drawings.
[0012] Referring to Figure 1, the configuration of the monitoring system 1 according to this embodiment will be described.
[0013] The monitoring system 1 comprises at least one monitoring device 10, an object to be monitored 20, and a waveform measuring instrument 30.
[0014] The monitoring device 10 is, for example, a server belonging to a cloud computing system or other computing system.
[0015] The monitored object 20 is any equipment installed in a factory or similar facility. Examples of monitored objects 20 include, but are not limited to, machine tools such as machining centers, pumps used in semiconductor manufacturing equipment, bearings used in belt conveyors, or processing equipment such as cutters. While Figure 1 shows one monitored object 20, the monitoring system 1 may have multiple monitored objects 20. For example, if multiple pieces of equipment are in physical or electrical contact or in close proximity and thus affect each other's waveforms, these multiple pieces of equipment may be considered monitored objects 20.
[0016] The waveform measuring instrument 30 is any measuring instrument capable of measuring the waveform emitted from the object being monitored 20. The waveform measuring instrument 30 is appropriately connected to the monitoring device 10 via any amplifier 40. Here, the waveform can be, for example, an vibration waveform, a voltage waveform, a current waveform, or an AE (Acoustic Emission) wave. The shape and other features are shown, but are not limited to these. Figure 1 shows one waveform measuring instrument 30, but the monitoring system 1 may have multiple waveform measuring instruments 30. For example, as mentioned above, if there are multiple objects to be monitored 20, a waveform measuring instrument 30 may be attached to each object to be monitored 20. In this case, the multiple feature quantities described later may be extracted from the waveform data measured by each waveform measuring instrument 30. Also, a waveform measuring instrument 30 may be attached to each component of a single object to be monitored 20. Furthermore, a waveform measuring instrument 30 may be attached to each component of the multiple components of the object to be monitored 20 that influence each other's waveforms. In addition, Figure 1 shows one amplifier 40, but the monitoring system 1 may have multiple amplifiers 40. Also, one amplifier 40 may be connected to multiple objects to be monitored 20.
[0017] Next, the configuration of the monitoring device 10 according to this embodiment will be described.
[0018] The monitoring device 10 includes a processing unit 11, a storage unit 12, an acquisition unit 13, an input unit 14, and an output unit 15.
[0019] The processing unit 11 includes one or more processors, one or more programmable circuits, one or more dedicated circuits, or a combination thereof. The processor is, for example, a CPU (Central Processing Unit) or a general-purpose processor such as a GPU (Graphics Processing Unit), or a specific dedicated processor specialized for processing. The programmable circuit is, for example, an FPGA (Field-Programmable Gate Array). The dedicated circuit is, for example, an ASIC (Application Specific Integrated Circuit). The processing unit 11 executes various processes related to the operation of the monitoring device 10 while controlling each unit of the monitoring device 10.
[0020] The storage unit 12 includes one or more semiconductor memories, one or more magnetic memories, one or more optical memories, or a combination thereof. The semiconductor memory is, for example, a RAM (Random Access Memory) or a ROM (Read Only Memory). The storage unit 12 functions as, for example, a main storage device, an auxiliary storage device, or a cache memory. The storage unit 12 stores information used for the operation of the monitoring device 10 and information obtained by the operation of the monitoring device 10.
[0021] The acquisition unit 13 can acquire data from the waveform measuring instrument 30 or data from an external device such as an arbitrary database and includes one or more acquisition interfaces. The acquisition interface is, for example, but not limited to, an analog input interface, a digital input interface, or a LAN (Local Area Network) interface. The acquisition unit 13 acquires data used for the operation of the monitoring device 10, and converts the data into a data format readable by the processing unit 11.
[0022] The input unit 14 includes one or more input interfaces. The input interface is, for example, a physical key, a capacitive key, a pointing device, or a touch screen provided integrally with a display. The input unit 14 receives an operation for inputting information used for the operation of the monitoring device 10. Instead of being provided in the monitoring device 10, the input unit 14 may be connected to the monitoring device 10 by any method such as USB (Universal Serial Bus), HDMI (registered trademark), or Bluetooth (registered trademark ), as an external input device such as a tablet terminal.
[0023] The output unit 15 includes one or more output interfaces. The output interface is, for example, a display. The display is, for example, an LCD (Liquid Crystal Display) or an organic EL (Electro Luminescence) display. The output unit 15 outputs information obtained by the operation of the monitoring device 10. Instead of being provided in the monitoring device 10, the output unit 15 may be connected to the monitoring device 10 by any method such as USB (Universal Serial Bus), HDMI (registered trademark), or Bluetooth (registered trademark), as an external output device such as a tablet terminal.
[0024] The operation of the monitoring device 10 according to the present embodiment will be described with reference to Figure 2. This operation corresponds to the monitoring method according to the present embodiment.
[0025] Step S1: The processing unit 11 of the monitoring device 10 acquires, via the acquisition unit 13, data indicating a waveform emitted from the monitoring target 20 measured by the waveform measuring instrument 30.
[0026] The waveform emitted from the monitoring target 20 may be a waveform during the operation of the monitoring target 20, or may be a waveform when the monitoring target 20 is stationary.
[0027] Step S2: The processing unit 11 of the monitoring device 10 extracts multiple features from the waveform data acquired in step S1.
[0028] Multiple features include the effective value (RMS value) of the waveform, peak amplitude or impact (waveform kurtosis) that can be extracted from the waveform using an arbitrary bandpass filter, or the power spectrum (e.g., peak amplitude) that can be extracted from the waveform using an arbitrary FFT (Fast Fourier Transformation) method. This may include, but is not limited to, the specified band peak power (a numerical representation of the power in a frequency band specified by the user from the power spectrum). The choice of which feature to adopt may be appropriately selected based on user input, etc. In the case of user input, the processing unit 11 of the monitoring device 10 can accept the user's operation to select the desired type of feature via the input unit 14.
[0029] Step S3: The processing unit 11 of the monitoring device 10 determines whether or not there are signs of an abnormality in the monitored object 20 based on the multiple feature quantities extracted in step S2.
[0030] Here, the processing unit 11 of the monitoring device 10 may extract further features from the multiple features extracted in step S2, and determine whether or not there are signs of an abnormality in the monitored object 20 based on these further features. This case will be explained in detail below.
[0031] First, the processing unit 11 of the monitoring device 10 extracts a first feature quantity and a second feature quantity, which correspond to multiple features, from the waveform data acquired in step S1. Then, the processing unit 11 of the monitoring device 10 extracts a third feature quantity, which corresponds to further features, from the first feature quantity and the second feature quantity. Finally, the processing unit 11 of the monitoring device 10 determines that there is an indication of an abnormality in the monitored object 20 if the value of the third feature quantity is above a predetermined threshold.
[0032] The first feature may be a feature extracted from the waveform data acquired in step S1, specifically from the waveform data in a first frequency band. The second feature may be a feature extracted from the waveform data acquired in step S1, specifically from the waveform data in a second frequency band different from the first frequency band. Here, the first frequency band (e.g., low frequency range) and the second frequency band (e.g., high frequency range) can be appropriately selected based on user input, etc., according to the characteristics of the waveform emitted from the object being monitored 20, so that signs of anomalies are detected with appropriate frequency and accuracy for operating the monitoring system 1. For example, using only a single feature may result in an overly high detection frequency of signs of anomalies, which can be problematic in practice. In this case, by using a second feature corresponding to a second frequency band different from the first frequency band, in addition to a first feature corresponding to the first frequency band, the first and second frequency bands are selected so that a composite state of an anomaly corresponding to the first feature and an anomaly corresponding to the second feature is detected with appropriate frequency and accuracy for operating the monitoring system 1. This is based on the following technical knowledge. In other words, as also shown in Figure 3, if only the first feature in the first frequency band is used to determine the presence of an anomaly, even if the monitored object 20 is not actually showing any signs of an anomaly, if the first feature exceeds a predetermined threshold, vibrations that are not even signs of an anomaly may be detected as signs of an anomaly, and this becomes noise, leading to problems such as overdetection. Similarly, if only the second feature in the second frequency band is used to determine the presence of an anomaly, even if the monitored object 20 is not actually showing any signs of an anomaly, if the second feature exceeds a predetermined threshold, vibrations that are not even signs of an anomaly may be detected as signs of an anomaly, and this becomes noise, leading to problems such as overdetection. In contrast to this, by multiplying the feature quantities in two different frequency bands (AND operation), the true Only composite states that contribute to the prediction of an anomaly appear as a third feature with high probability, and noise that does not contribute to the prediction of an anomaly is suppressed. The first and second frequency bands may be identified, for example, by analyzing the frequency components corresponding to the anomaly of the object being analyzed using any method. The boundary between the low-frequency range and the high-frequency range can be appropriately set considering the characteristics (natural frequency, etc.) unique to the object being monitored 20, and may be, for example, 500 Hz or nearby, or 1000 Hz or nearby. That is, the low-frequency range may be below the boundary frequency, and the high-frequency range may be above the boundary frequency. As an example, the first frequency band corresponding to the low-frequency range may be related to the vibration of the device itself, and the second frequency band corresponding to the high-frequency range may be related to vibration caused by the rotating mechanism or vibration caused by friction. However, this disclosure is not limited to these examples, and it is possible to appropriately combine different frequency bands related to the predictions that the user wishes to detect.
[0033] The third feature may be a feature calculated by performing a predetermined operation on the first and second features.
[0034] The predetermined operation may be addition, subtraction, multiplication, or division of the first feature quantity and the second feature quantity. For example, the predetermined operation may be an operation to add the second feature quantity to the first feature quantity at each time step, an operation to subtract the second feature quantity from the first feature quantity at each time step, an operation to subtract the first feature quantity from the second feature quantity at each time step, an operation to multiply the first feature quantity by the second feature quantity at each time step, or an operation to divide the first feature quantity by the second feature quantity at each time step, or an operation to divide the second feature quantity by the first feature quantity at each time step. The user may appropriately set which of the four basic arithmetic operations to use via the input unit 14 of the monitoring device 10. Here, the processing unit 11 of the monitoring device 10 may improve user convenience by outputting the following message to the user via the output unit 15. That is, the message may include that addition or multiplication is suitable for detecting a composite state of an abnormal state corresponding to a first feature and an abnormal state corresponding to a second feature, or for detecting an abnormal state that occurs in common equipment even if the abnormal states occur at different timings. The message may also include that the appropriateness of addition or multiplication can be appropriately selected according to the scale and degree of influence of the first and second features. The message may also include that subtraction or division is suitable for detecting an abnormal state corresponding to either the first or second feature. However, this disclosure is not limited to these.
[0035] Furthermore, the predetermined operation may be addition, subtraction, multiplication, or division of the power of the first feature and the power of the second feature. For example, the predetermined operation may be an operation to add the square of the second feature to the square of the first feature at each time step, an operation to subtract the square of the second feature from the square of the first feature at each time step, an operation to subtract the square of the first feature from the square of the second feature at each time step, an operation to multiply the square of the first feature by the square of the second feature at each time step, or an operation to divide the square of the first feature by the square of the second feature at each time step, or an operation to divide the square of the second feature by the square of the first feature at each time step.
[0036] Furthermore, when performing the predetermined calculations described above, the processing unit 11 of the monitoring device 10 may assign a predetermined weight to at least one of the first feature or a power of the first feature and the second feature or a power of the second feature. The weighting may be performed by multiplying by a predetermined value if the predetermined calculation is addition or subtraction, or by dividing by a predetermined value if the predetermined calculation is addition or division, but is not limited to these. Here, the weight for the first feature or a power of the first feature and the weight for the second feature or a power of the second feature can be set appropriately according to the detection accuracy of anomaly predictions desired by the user. For example, the weight for the one that has a greater impact on detection accuracy between the first feature or a power of the first feature and the second feature or a power of the second feature may be set to be larger. Also, the weight for the first feature or a power of the first feature and the weight for the second feature or a power of the second feature can be set appropriately in order to match the scale of the first feature and the scale of the second feature. The weight can be set by the user as appropriate via the input unit 14 of the monitoring device 10.
[0037] The predetermined threshold can be set as appropriate according to the detection accuracy of anomaly predictions desired by the user, and may be determined based on actual values of the waveform data. For example, the actual values may include statistical values such as the average or standard score of the third feature over a predetermined period, or values calculated using any machine learning method, but are not limited to these. The predetermined threshold can be set as appropriate by the user via the input unit 14 of the monitoring device 10.
[0038] As an example, the processing unit 11 of the monitoring device 10 extracts a first feature quantity corresponding to abnormal noise (abnormal noise that does not correspond to vibration noise observed in the high frequency range) by using a bandpass filter having a first frequency band (low frequency range) as a passband on the peak amplitude or specified band peak power corresponding to the first feature quantity from the vibration waveform data corresponding to the waveform data. The processing unit 11 of the monitoring device 10 also extracts a second feature quantity corresponding to vibration noise by using a bandpass filter having a second frequency band (high frequency range) as a passband on the peak amplitude or specified band peak power corresponding to the second feature quantity from the vibration waveform data corresponding to the waveform data. Then, the processing unit 11 of the monitoring device 10 multiplies the first feature quantity and the second feature quantity (performs an AND operation) Then, a third feature is calculated. The processing unit 11 of the monitoring device 10 determines that there is a sign of an anomaly if the third feature is abnormal to a predetermined threshold. However, this disclosure is not limited to this example.
[0039] Step S4: The processing unit 11 of the monitoring device 10 outputs the result of the determination in step S3 to the user via the output unit 15. The output mode can be set as appropriate by the user via the input unit 14 of the monitoring device 10.
[0040] As described above, the processing unit 11 of the monitoring device 10 according to this embodiment acquires data showing waveforms emitted from the object being monitored 20 via the acquisition unit 13. The processing unit 11 of the monitoring device 10 then extracts a plurality of feature quantities from the waveform data. Based on the plurality of feature quantities, the processing unit 11 of the monitoring device 10 determines whether or not there are signs of an abnormality in the object being monitored 20.
[0041] With this configuration, by using multiple features, a composite state of an abnormal state corresponding to one of the multiple features and an abnormal state corresponding to another of the multiple features can be detected with appropriate frequency and accuracy for operating the monitoring system 1. More specifically, as described above with reference to Figure 3, by multiplying features in two different frequency bands, only composite states that truly contribute to anomaly prediction remain with a high probability, and noise that does not contribute to anomaly prediction is suppressed. This avoids a situation where the detection frequency of anomaly predictions is too high, which would be problematic in operation if a single feature were used. Therefore, the technology for identifying anomaly predictions of the monitored object 20 can be improved.
[0042] The present disclosure has been described above based on the drawings and embodiments, but it should be noted that those skilled in the art may make various modifications and alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are within the scope of the present invention. For example, the functions, etc., included in each component or step, etc., can be rearranged in a logically consistent manner, and multiple components or steps, etc., can be combined into one or divided.
[0043] For example, an embodiment is also possible in which a general-purpose computer functions as the monitoring device 10 according to the above embodiment. Specifically, a monitoring program describing the processing content that realizes each function of the monitoring device 10 according to the above embodiment is stored in the memory of the general-purpose computer, and the processor reads and executes the monitoring program. Therefore, this disclosure can also be realized as a monitoring program that can be executed by a processor, or as a non-temporary computer-readable medium that stores the monitoring program. [Explanation of Symbols]
[0044] 1. Monitoring System 10 Monitoring equipment 11 Processing Section 12 Storage section 13 Acquisition Department 14 Input section 15 Output section 20. Objects under surveillance 30 Waveform measuring instrument 40 Amplifiers
Claims
1. A monitoring device comprising a processing unit and an acquisition unit, The aforementioned processing unit, The acquisition unit acquires data showing the waveform emitted from the object being monitored. A first feature quantity is extracted from the data showing the waveform in the first frequency band among the data showing the waveform. A second feature quantity is extracted from the data showing the waveform in a second frequency band different from the first frequency band among the data showing the waveform. By performing an AND operation on the first and second feature quantities, a third feature quantity is extracted that allows for the detection of an anomaly when both the first and second feature quantities indicate an anomaly. In a monitoring device that determines whether or not there are signs of abnormality in the monitored object based on the third characteristic quantity described above, The object to be monitored is a device having a rotation mechanism, The first frequency band is a frequency band related to the vibration of the device itself, The second frequency band is a frequency band related to vibrations generated by vibration or friction caused by the rotating mechanism. monitoring equipment.
2. A monitoring method performed by a monitoring device, To acquire data showing waveforms emitted from the monitored object, Extracting a first feature quantity from the data showing the waveform in the first frequency band among the data showing the waveform, Extracting a second feature quantity from the data showing the waveform in a second frequency band different from the first frequency band among the data showing the waveform, By performing an AND operation on the first and second features, a third feature is extracted that allows for the detection of an anomaly when both the first and second features indicate an anomaly. A monitoring method that includes determining whether or not there are signs of an abnormality in the monitored object based on the third characteristic quantity, The object to be monitored is a device having a rotation mechanism, The first frequency band is a frequency band related to the vibration of the device itself, The second frequency band is a frequency band related to vibrations generated by vibration or friction caused by the rotating mechanism. Monitoring method.
3. On the computer, To acquire data showing waveforms emitted from the monitored object, Extracting a first feature quantity from the data showing the waveform in the first frequency band among the data showing the waveform, Extracting a second feature quantity from the data showing the waveform in a second frequency band different from the first frequency band among the data showing the waveform, By performing an AND operation on the first and second features, a third feature is extracted that allows for the detection of an anomaly when both the first and second features indicate an anomaly. In a monitoring program that performs the following actions based on the third characteristic quantity described above, The object to be monitored is a device having a rotation mechanism, The first frequency band is a frequency band related to the vibration of the device itself, The second frequency band is a frequency band related to vibrations generated by vibration or friction caused by the rotating mechanism. A monitoring program.
Citation Information
Patent Citations
Inspection device, inspection method, inspection device control program, and computer-readable recording medium
JP2007225525A
Analyzer and analytical method
JP2021018141A
Semiconductor manufacturing apparatus, fault prediction method of semiconductor manufacturing apparatus, and fault prediction program of semiconductor manufacturing apparatus
JP2021102817A
