Vibration analysis system and vibration analysis method

The vibration analysis system predicts equipment abnormalities using Mahalanobis distances on signal intensities and centers of gravity, enabling timely maintenance planning by forecasting when issues will arise.

JP7594599B2Active Publication Date: 2024-12-04VALQUA LTD
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
JP2022551891
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2020-09-25
Filing Date
2021-09-13
Publication Date
2024-12-04
Estimated Expiration
2041-09-13

AI Technical Summary

Technical Problem

Existing methods for detecting equipment abnormalities lack the capability to predict the timing of such occurrences for predictive maintenance.

Method used

A vibration analysis system and method that utilize Mahalanobis distances to analyze vibration signals, calculating signal intensities and centers of gravity to predict abnormality occurrence times, with different thresholds for near-future and immediate predictions.

Benefits of technology

Enables accurate prediction of abnormality occurrences in equipment, allowing for planned maintenance by identifying potential issues days or hours in advance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007594599000003
    Figure 0007594599000003
  • Figure 0007594599000004
    Figure 0007594599000004
  • Figure 0007594599000005
    Figure 0007594599000005
Patent Text Reader

Abstract

This vibration analysis system (100) comprises: a signal input unit (202) which receives an input of a vibration signal detected by a sensor attached to a moving object; an intensity calculation unit (204) which calculates a plurality of signal intensities respectively corresponding to a plurality of frequency bands by analyzing a vibration signal corresponding to an object; a first distance calculation unit (206) which calculates a first Mahalanobis distance of a first signal space composed of the plurality of signal intensities with respect to a first unit space; a center-of-gravity calculation unit (208) which calculates two-dimensional center of gravity data indicating the positions of the centers of gravity of the plurality of signal intensities; a second distance calculation unit (210) which calculates a second Mahalanobis distance of a second signal space composed of the center of gravity data with respect to a second unit space; and an anomaly prediction unit (212) which predicts, on the basis of the first Mahalanobis distance and the second Mahalanobis distance, a time when an abnormality occurs in the object.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present disclosure relates to a vibration analysis system and a vibration analysis method. [Background technology]

[0002] 2. Description of the Related Art Conventionally, a method for inspecting a machine for abnormalities is known in which the presence or absence of an abnormality in the machine is determined by detecting a signal caused by abnormal vibration during operation of the machine.

[0003] For example, in JP 2019-35585 A (Patent Document 1), an anomaly determination process based on a predetermined neighborhood method that calculates an index value indicating the degree of anomaly based on the distance between data and determines whether or not there is an anomaly is applied to each data of a new analysis result data group relative to an existing representative data group that represents an existing analysis result data group. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-35585 A Summary of the Invention [Problem to be solved by the invention]

[0005] In Patent Document 1, it is considered to appropriately select representative data to be used for anomaly detection from a large amount of data that is the result of monitoring the status of the target equipment. However, there is no disclosure or suggestion of a configuration for predicting the timing of anomalies for predictive maintenance of the equipment.

[0006] An object in one aspect of the present disclosure is to provide a vibration analysis system and a vibration analysis method capable of predicting when an abnormality will occur in an object by analyzing the vibration state of the object. [Means for solving the problem]

[0007] A vibration analysis system according to one embodiment includes a signal input unit that accepts input of a vibration signal detected by a sensor attached to an object in motion, an intensity calculation unit that calculates a plurality of signal intensities corresponding to a plurality of frequency bands by analyzing the vibration signal corresponding to the object, a first distance calculation unit that calculates a first Mahalanobis distance of a first signal space composed of a plurality of signal intensities for a predetermined first unit space, a centroid calculation unit that calculates two-dimensional centroid data indicating the centroid position of the plurality of signal intensities calculated by the intensity calculation unit, a second distance calculation unit that calculates a second Mahalanobis distance of a second signal space composed of the two-dimensional centroid data for a predetermined second unit space, and an abnormality prediction unit that predicts an abnormality occurrence time when an abnormality will occur in the object based on the first Mahalanobis distance and the second Mahalanobis distance.

[0008] Preferably, the first unit space is composed of a plurality of signal intensities corresponding to a plurality of frequency bands, which are calculated by analyzing a vibration signal corresponding to the object in a normal state, and the second unit space is composed of two-dimensional center-of-gravity data indicating a center-of-gravity position of the plurality of signal intensities constituting the first unit space.

[0009] Preferably, the time of occurrence of an abnormality when a first Mahalanobis distance equal to or greater than the first threshold is calculated is predicted to be in the near future than the time of occurrence of an abnormality when a first Mahalanobis distance equal to or greater than the first threshold is not calculated.

[0010] Preferably, the abnormality prediction unit predicts that an abnormality will occur in the object several days after the first Mahalanobis distance equal to or greater than the first threshold is calculated.

[0011] Preferably, the time of occurrence of an abnormality when a first Mahalanobis distance equal to or greater than a first threshold value is calculated and a second Mahalanobis distance equal to or greater than a second threshold value is calculated is predicted to be in the near future than the time of occurrence of an abnormality when a first Mahalanobis distance equal to or greater than the first threshold value is calculated and a second Mahalanobis distance equal to or greater than the second threshold value is not calculated.

[0012] Preferably, the abnormality prediction unit predicts that an abnormality will occur in the object several hours after the second Mahalanobis distance equal to or greater than the second threshold is calculated.

[0013] Preferably, the device further includes an output control unit that outputs first warning information when a first Mahalanobis distance equal to or greater than a first threshold is calculated, and outputs second warning information having a higher warning level than the first warning information when a second Mahalanobis distance equal to or greater than a second threshold is calculated.

[0014] Preferably, the output control unit causes the time series data of the first Mahalanobis distance and the time series data of the second Mahalanobis distance to be displayed on a display.

[0015] A vibration analysis method according to another embodiment includes the steps of receiving an input of a vibration signal detected by a sensor attached to an object in motion, calculating a plurality of signal intensities corresponding to a plurality of frequency bands by analyzing the vibration signal corresponding to the object, calculating a first Mahalanobis distance of a first signal space composed of the plurality of signal intensities for a predetermined first unit space, calculating two-dimensional center of gravity data indicating the center of gravity position of the calculated plurality of signal intensities, calculating a second Mahalanobis distance of a second signal space composed of the two-dimensional center of gravity data for a predetermined second unit space, and predicting a time when an abnormality will occur in the object based on the first Mahalanobis distance and the second Mahalanobis distance. Effect of the Invention

[0016] According to the present disclosure, by analyzing the vibration state of an object, it is possible to predict when an abnormality will occur in the object. [Brief description of the drawings]

[0017] [Figure 1] FIG. 1 is a diagram for explaining an overview of a system. [Diagram 2] 1 is a block diagram showing an example of an overall configuration of a vibration analysis system. [Diagram 3] FIG. 2 is a block diagram showing an example of a hardware configuration of an analysis device. [Figure 4] 13 is a flowchart showing an example of a preparation process. [Diagram 5] FIG. 2 is a diagram showing an example of a data set of signal strength for each frequency band. [Figure 6] FIG. 13 is a diagram showing an example of a data set of center of gravity positions. [Figure 7] 13 is a flowchart showing an example of an analysis process. [Figure 8] 13 is a flowchart showing an example of a first abnormality prediction step. [Figure 9] FIG. 13 is a diagram showing time-series data of Mahalanobis distance based on multidimensional data. [Figure 10] 13 is a flowchart showing an example of a second abnormality prediction step. [Figure 11] FIG. 13 is a diagram showing time-series data of Mahalanobis distance based on two-dimensional data. [Figure 12] 13 is a flowchart illustrating an example of a trend analysis process. [Figure 13] FIG. 13 is a diagram showing an example of the layout of a user interface screen. [Figure 14] FIG. 2 is a functional block diagram of the analysis device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the following description, the same components are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated.

[0019] <System configuration> FIG. 1 is a diagram for explaining an overview of the system 1000. Referring to FIG. 1, the system 1000 is a system for predicting the occurrence of an abnormality in a maintenance object such as a pump (hereinafter, also simply referred to as "object") by analyzing a signal of vibration generated during operation of the object. In the following, the object is described as being a pump, but the object is not limited thereto, and the system 1000 can be applied to any object that generates vibration (or sound) during operation. For example, the system 1000 can also be applied to predict an abnormality in a motor, a part that vibrates due to vibration from a vibrating body, and the like.

[0020] The system 1000 includes a vibration analysis system 100, a plurality of sensors 30, a terminal device 40, a network 50, and a plurality of pumps 70. The vibration analysis system 100 performs vibration analysis of the pump 70. The vibration analysis system 100 includes an analysis device 10 and a sensor unit 20. The sensor unit 20 is electrically connected to the plurality of sensors 30. In the system 1000, two sensor units 20 are connected to the analysis device 10, but three or more sensor units 20, or one sensor unit 20 may be connected to the analysis device 10. Each sensor unit 20 may be electrically connected to one sensor 30. Each sensor unit 20 may be electrically connected to a plurality of sensors 30 attached to the plurality of pumps 70, respectively.

[0021] The sensor 30 is attached to the pump 70 and acquires a detection signal (vibration signal) detected due to vibration or sound of the pump. The analysis device 10 performs vibration analysis of the pump 70 based on the vibration signal input from the sensor 30 via the sensor unit 20. The analysis device 10 is configured to be able to communicate with the terminal device 40 via the network 50. The analysis device 10 transmits vibration analysis results and the like to the terminal device 40.

[0022] The analysis device 10 typically has a structure conforming to a general-purpose computer architecture, and realizes various processes described below by a processor executing a pre-installed program. The analysis device 10 is, for example, a laptop PC (Personal Computer). However, the analysis device 10 may be any other device (for example, a desktop PC, a tablet terminal device) as long as it is capable of executing the functions and processes described below.

[0023] The network 50 includes various networks such as the Internet, etc. The network 50 may employ a wired communication system, or may employ other wireless communication systems such as a wireless LAN (local area network).

[0024] The terminal device 40 is, for example, a portable tablet terminal device. However, the terminal device 40 is not limited to this, and may be realized by a smartphone, a desktop PC (Personal Computer), or the like. Note that the vibration analysis system 100 according to the present embodiment is configured as a separate device in which the analysis device 10 and the sensor unit 20 are separated, but may be configured as an integrated device in which the analysis device 10 and the sensor unit 20 are integrated.

[0025] 2 is a block diagram showing an example of an overall configuration of the vibration analysis system 100. With reference to FIG.

[0026] The sensor 30 connected to the sensor unit 20 is a sensor capable of detecting vibration and sound signals, and is configured, for example, by an acceleration sensor using an organic piezoelectric element. Note that the sensor 30 may be configured as an acceleration sensor of another type (for example, a servo type) as long as it is a sensor capable of detecting vibration and sound signals, and may be configured as various other sensors.

[0027] When the signal obtained by the sensor 30 is a charge signal, a charge converter is provided between the sensor 30 and the vibration analysis system 100. In this case, the charge converter converts the charge signal from the sensor 30 into a voltage signal and outputs it to the vibration analysis system 100. Note that when the sensor 30 has a function of converting a charge signal into a voltage signal, the charge converter is not necessary.

[0028] The sensor unit 20 converts the vibration signal acquired from the sensor 30 (or a charge converter) into a signal that can be processed by the analysis device 10. Specifically, the sensor unit 20 includes a filter 21, an amplifier 22, and an A / D converter .

[0029] The filter 21 is an analog filter, and removes noise components from the vibration signal output from the sensor 30. The filter 21 is configured by a low-pass filter, a high-pass filter, or the like.

[0030] The amplifier 22 amplifies the analog signal output from the filter 21 by a predetermined factor, and outputs the amplified signal to the A / D converter 23 .

[0031] The A / D converter 23 converts the signal input from the amplifier 22 from an analog signal to a digital signal at a predetermined sampling frequency. The A / D converter 23 outputs the digitally converted signal to the analysis device 10.

[0032] Fig. 3 is a block diagram showing an example of a hardware configuration of analysis device 10. With reference to Fig. 3, analysis device 10 includes processor 101, memory 103, display 105, input device 107, signal input interface (I / F) 109, and communication interface (I / F) 111. These components are connected to each other so as to be able to communicate data with each other.

[0033] Processor 101 is typically an arithmetic processing unit such as a CPU (Central Processing Unit), an MPU (Multi Processing Unit), etc. Processor 101 controls the operation of each part of analysis device 10 by reading and executing a program stored in memory 103. More specifically, processor 101 realizes each function of analysis device 10 by executing the program.

[0034] The memory 103 is realized by a random access memory (RAM), a read-only memory (ROM), a flash memory, a hard disk, etc. The memory 103 stores programs executed by the processor 101, etc.

[0035] Display 105 is, for example, a liquid crystal display, an organic EL (Electro Luminescence) display, etc. Display 105 may be configured integrally with analysis device 10, or may be configured separately from analysis device 10.

[0036] The input device 107 accepts operation inputs for the analysis device 10. The input device 107 is realized by, for example, a keyboard, a button, a mouse, etc. The input device 107 may also be realized as a touch panel.

[0037] The signal input interface 109 mediates data transmission between the processor 101 and the sensor unit 20. The signal input interface (I / F) 109 accepts an input of a vibration signal from the sensor 30 via the sensor unit 20. Specifically, the signal input interface 109 accepts an input of a digital signal from the A / D converter 23.

[0038] The communication interface 111 mediates data transmission between the processor 101 and the terminal device 40, etc. As a communication method, for example, a wireless communication method such as Bluetooth (registered trademark) or a wireless LAN (Local Area Network) is used. Note that a wired communication method such as a USB (Universal Serial Bus) may also be used as the communication method.

[0039] <Anomaly prediction method> The abnormality prediction method according to the present embodiment includes a preparation step of preparing reference data and an analysis step of analyzing the abnormal state of the pump 70.

[0040] (preparation process) In the preparation step according to this embodiment, for example, the vibration state of the pump 70 at the beginning of operation is measured. Since the pump 70 at the beginning of operation is in a brand new state, the vibration state of the pump 70 in a normal state is measured as a reference. However, instead of the pump 70, a pump of the same type as the pump 70 in a normal state may be separately prepared, and the vibration state of that pump may be measured as a reference.

[0041] 4 is a flow chart showing an example of the preparation process. Typically, the following steps are realized by processor 101 of analysis device 10 executing a program stored in memory 103.

[0042] 4, the processor 101 acquires a vibration signal output from the sensor 30 via the sensor unit 20 (step S10). Specifically, the processor 101 acquires a vibration signal corresponding to the pump 70 in a normal state (a vibration signal indicating a vibration state of the pump 70) from the sensor 30.

[0043] The processor 101 performs octave analysis on the vibration signal accumulated for a predetermined time (for example, several tens to several hundreds of milliseconds) (step S12). In the present embodiment, 1 / 3 octave analysis is used. Therefore, each vibration signal is separated into 48 bands from, for example, 0.4 Hz to 20 kHz by a 1 / 3 band-pass filter, and the signal intensity (vibration intensity) is averaged for each band (that is, frequency band). In the following description, the signal intensity averaged in the frequency band is also simply referred to as the "signal intensity of the frequency band".

[0044] For each frequency band, the processor 101 stores, in the memory 103, as reference data R, the signal intensity of the frequency band corresponding to the normal pump 70 (step S14).

[0045] FIG. 5 is a diagram showing an example of a data set of the signal intensity of each frequency band. Referring to FIG. 5, the data set 310 includes the signal intensity L of each frequency band f1 to fn (where n is a natural number, n < m) for each time T1 to Tm (where m is a natural number). When each vibration signal is separated into 48 bands, n = 48. For example, the data set 310 includes the signal intensities L1_1 to L1_n of each frequency band f1 to fn at time T1, and includes the signal intensities Lm_1 to Lm_n of each frequency band f1 to fn at time Tm. The reference data R includes, for example, the signal intensity L of each frequency band f1 to fn for each time T1 to Tn. In this case, the period of time T1 to Tn corresponds to the initial operation period.

[0046] Referring again to FIG. 4, the processor 101 sets a unit space U1 in the Mahalanobis-Taguchi method (MT method) using the reference data R (step S16). Specifically, the processor 101 sets the data of the signal intensity L of each frequency band f1 to fn at each time T1 to Tn as the unit space U1. The unit space U1 is used in the first anomaly prediction step described later.

[0047] The processor 101 calculates two-dimensional center of gravity data indicating the center of gravity of the signal strength of each frequency band (step S18). Specifically, the processor 101 generates the data set shown in FIG.

[0048] Fig. 6 is a diagram showing an example of a data set of center positions. Referring to Fig. 6, the data set 320 includes a center position Gx of a frequency band and a center position Gy of a signal intensity for times T1 to Tm. For each of the times T1 to Tm, two-dimensional center position data (center position Gx, Gy) is generated. The center position Gx is expressed by the following formula (1), and the center position Gy is expressed by the following formula (2).

[0049]

number

[0050]

number

[0051] f i represents the i-th frequency band, and L i represents the signal strength of the i-th frequency band, and S represents the sum of the signal strengths of all frequency bands. From the above (1) and (2), a data set 320 consisting of two-dimensional data (center of gravity positions Gx, Gy) is generated from a data set 310 consisting of n-dimensional (multidimensional) data (signal strength L of each frequency band f1 to fn). In the data set 320, for example, the center of gravity positions Gx, Gy at time T1 are represented by center of gravity positions Gf_1, GL_1, respectively, and the center of gravity positions Gx, Gy at time Tm are represented by center of gravity positions Gf_m, GL_m, respectively.

[0052] 4 again, the processor 101 sets a unit space U2 in the MT method using the center of gravity data in an initial period of operation (for example, times T1 to Tn) (step S20). Specifically, the processor 101 sets the two-dimensional center of gravity data (center of gravity positions Gx, Gy) for each of the times T1 to Tn as the unit space U2. The unit space U2 is used in a second abnormality prediction step described later.

[0053] (Analysis process) The analysis step includes first and second abnormality prediction steps for predicting when an abnormality will occur in the vibration state of the pump 70, and a trend analysis step for analyzing the future tendency of the vibration state of the pump 70.

[0054] Fig. 7 is a flowchart showing an example of the analysis step. Referring to Fig. 7, the processor 101 acquires a vibration signal output from the sensor 30 via the sensor unit 20 (step S30). Specifically, the processor 101 acquires a vibration signal corresponding to the pump 70 during a normal period (for example, a period after the end of an initial period after the start of operation) from the sensor 30.

[0055] The processor 101 performs octave analysis on the vibration signal accumulated for a predetermined time (step S32). The processor 101 stores, for each frequency band, the signal strength of the frequency band corresponding to the pump 70 in the memory 103 (step S34). Specifically, the signal strength of each frequency band in the pump 70 at a certain time Ts is stored in the form of a data set 310. Here, a series of signal strengths Ls_1 to Ls_n of frequency bands f1 to fn in the pump 70 at time Ts is also referred to as signal strength data Ps. Note that the period from time T1 to Tn corresponds to the initial period of operation, and therefore the normal period corresponds to the period from time Tn+1 to time Tm. Therefore, time Ts is any of times Tn+1 to Tm.

[0056] The processor 101 receives the signal strength data Ps and preparationUsing the data obtained in the steps, a first anomaly prediction step (step S40), a second anomaly prediction step (step S60), and a trend analysis step (step S70) are performed. These steps may be performed in parallel or sequentially.

[0057] [First abnormality prediction process] Fig. 8 is a flowchart showing an example of the first abnormality prediction step. In the first abnormality prediction step, a data set 310 consisting of multidimensional data is used. Referring to Fig. 8, the processor 101 calculates a Mahalanobis distance MD1 (hereinafter also simply referred to as "distance MD1") of a signal space X1s consisting of a plurality of signal intensities with respect to the unit space U1 set in step S16 of Fig. 4 (step S41). The signal space X1s is composed of a plurality of signal intensities Ls_1 to Ls_n (i.e., signal intensity data Ps) at time Ts.

[0058] The processor 101 determines whether the distance MD1 is equal to or greater than a threshold value Th1 (e.g., 5) (step S43). If the distance MD1 is less than the threshold value Th1 (NO in step S43), the processor 101 ends the first abnormality prediction process. If the distance MD1 is equal to or greater than the threshold value Th1 (YES in step S43), the processor 101 predicts that an abnormality will occur in the pump 70 in the near future, outputs an abnormality alert (step S45), and ends the first abnormality prediction process. Typically, the abnormality alert is displayed on the display 105. The abnormality alert may be configured to be output as an audio signal via a speaker.

[0059] Here, the reason why the processor 101 makes the above prediction will be described with reference to Fig. 9. Fig. 9 is a diagram showing time series data of Mahalanobis distance based on multidimensional data. Graph 410 shows time series data of Mahalanobis distance MDx1 for a reference pump of the same type as pump 70. Therefore, it can be said that the Mahalanobis distance MDx1 of graph 410 shows the same tendency as the Mahalanobis distance MD1 for pump 70.

[0060] According to the graph 410, the Mahalanobis distance MDx1 of the signal space to the unit space exceeds the threshold value Th1 (=5) for the first time at time Ta1. After that, the Mahalanobis distance MDx1 exceeds the threshold value Th1 at times Ta2, Ta3, and Ta4, and finally an abnormality occurs in the reference pump at time Ta5. The period from time Ta1 to time Ta5 is 4.5 days, the period from time Ta2 to time Ta5 is 3 days, the period from time Ta3 to time Ta5 is 1.5 days, and the period from time Ta4 to time Ta5 is 8.5 hours. From this, it can be understood that an abnormality occurs several days after the Mahalanobis distance MDx1 exceeds the threshold value Th1 for the first time.

[0061] Therefore, when a Mahalanobis distance MD1 equal to or greater than the threshold value Th1 is calculated, the processor 101 predicts that an abnormality will occur in the pump 70 in the near future. In particular, the processor 101 may predict that an abnormality will occur in the pump 70 several days after the Mahalanobis distance MD1 equal to or greater than the threshold value Th1 is first calculated. In this case, the processor 101 may output an abnormality alert with a relatively high warning level to notify that an abnormality may occur in the pump 70 in several days.

[0062] [Second abnormality prediction process] 10 is a flowchart showing an example of the second abnormality prediction step. Referring to FIG. 10, the processor 101 calculates two-dimensional centroid data indicating the centroid position of the signal strength of each frequency band (step S61). Specifically, the processor 101 calculates two-dimensional centroid data indicating the centroid position of the signal strength of each frequency band (step S61). 34 in memory From the signal intensity data Ps thus obtained, two-dimensional center of gravity data (that is, center of gravity positions Gx, Gy) are calculated using equations (1) and (2).

[0063] The processor 101 calculates the Mahalanobis distance MD2 (hereinafter also simply referred to as "distance MD2") of the signal space X2s composed of two-dimensional center of gravity data with respect to the unit space U2 set in step S20 of Fig. 4 (step S63). The signal space X2 is composed of center of gravity positions Gx, Gy (i.e., Gf_s, GL_s) at time Ts.

[0064] The processor 101 determines whether the distance MD2 is equal to or greater than a threshold Th2 (for example, 5) (step S65). If the distance MD2 is less than the threshold Th2 (step S 65 If the distance MD2 is equal to or greater than the threshold value Th2 (step S 65 In step S100, the processor 101 predicts that an abnormality will occur in the pump 70 in the very near future, and outputs an abnormality alert (YES in step S100). 67 ), and the second abnormality prediction process is terminated.

[0065] Here, the reason why the processor 101 makes the above prediction will be described with reference to Fig. 11. Fig. 11 is a diagram showing time series data of the Mahalanobis distance based on two-dimensional data. Graph 420 shows time series data of the Mahalanobis distance MDx2 in a reference pump of the same type as the pump 70. Therefore, it can be said that the Mahalanobis distance MDx2 in the graph 420 shows the same tendency as the Mahalanobis distance MD2 in the pump 70.

[0066] According to the graph 420, the Mahalanobis distance MDx2 exceeds the threshold value Th2 (=5) for the first time at time Tb1. After that, an abnormality occurs in the reference pump at time Tb2. The period from time Tb1 to time Tb2 is 5.5 hours. From this, the Mahalanobis distance MD x2 It can be said that an abnormality occurs several hours after the threshold value Th2 is first exceeded.

[0067] Therefore, when a Mahalanobis distance MD2 equal to or greater than the threshold value Th2 is calculated, the processor 101 predicts that an abnormality will occur in the pump 70 in the extremely near future. In particular, the processor 101 may predict that an abnormality will occur in the pump 70 several hours after the Mahalanobis distance MD2 equal to or greater than the threshold value Th2 is first calculated. In this case, the processor 101 may output an abnormality alert with a high warning level to notify that an abnormality may occur in several hours.

[0068] [Trend analysis process] Fig. 12 is a flow chart showing an example of the trend analysis step. Referring to Fig. 12, the processor 101 calculates a difference H by subtracting representative data of the reference data R from the signal strength data Ps (step S71).

[0069] The reference data R is composed of a data group including signal strength L of each frequency band f1 to fn at each time T1 to Tn. For example, the processor 101 extracts signal strengths Ln_1 to Ln_n of each frequency band f1 to fn at a certain time Tn from this data group as representative data. Then, the processor 101 calculates a difference H by subtracting the representative data from the signal strength data Ps. In this way, the difference H for each frequency band is calculated.

[0070] The representative data may be composed of the average value of the signal strength L of each of the frequency bands f1 to fn during the time periods T1 to Tn. In this case, for example, the signal strength L1 of the frequency band f1 included in the representative data is composed of the average value of the signal strengths L1_1 to Ln_1, and the signal strength Ln of the frequency band fn included in the representative data is composed of the average value of the signal strengths L1_n to Ln_n.

[0071] Subsequently, the processor 101 determines the abnormality level of the vibration state of the pump 70 in each frequency band by comparing the difference H with a plurality of reference values Z1, Z2, Z3 for each frequency band (step S73). For example, when the difference H is 0 or more and less than the reference value Z1 (for example, 3 dB), the abnormality level is "0" and the vibration state is "normal". When the difference H is equal to or more than the reference value Z1 and less than the reference value Z2 (for example, 6 dB), the abnormality level is "1", and it is recommended to check the state of the pump 70. When the difference H is equal to or more than the reference value Z2 and less than the reference value Z3 (for example, 10 dB), the abnormality level is "2", and maintenance of the pump 70 is recommended. When the difference H is equal to or more than the reference value Z3, the abnormality level is "3", which is a dangerous state where replacement of the pump 70 or the like is necessary.

[0072] Based on the determination result of the abnormality level of the vibration state in the pump 70, the processor 101 outputs an abnormality alert (step S75). Specifically, when the abnormality level is "3" (that is, when H ≧ Z3), the processor 101 outputs an abnormality alert with a high warning level (for example, "Danger"), and when the abnormality level is "2" (that is, when Z2 ≦ H < Z3), the processor 101 outputs an abnormality alert with a relatively high warning level (for example, "Maintenance recommended"), and when the abnormality level is "1" (that is, when Z1 ≦ H < Z2), the processor 101 outputs an abnormality alert with a low warning level (for example, "Caution"). Note that when the abnormality level is "0" (that is, when H < Z1), the processor 101 may output that the vibration state of the pump 70 is "normal". Subsequently, the processor 101 stores the determination result of the abnormality level (step S77).

[0073] Among the vibration states of the pump 70 in each frequency band, the processor 101 identifies the vibration states of a predetermined number (for example, 5) of the highest abnormality levels at a reference time point (for example, the current time point), and extracts the frequency bands corresponding to the identified vibration states (step S79). The reference time point is configured to be arbitrarily selectable by the user.

[0074] The processor 101 executes a trend prediction process for the vibration state of the extracted frequency band (hereinafter also referred to as the "extracted frequency band") (step S81). Specifically, the processor 101 predicts the trend of the future difference H based on the time series data of the existing (past) difference H in the extracted frequency band. For example, the processor 101 approximates the time series data of the difference H before the reference time point using an approximation curve (e.g., linear approximation, exponential approximation, etc.) to predict the difference H after the reference time point. The processor 101 may also obtain a regression line by performing regression analysis on the time series data of the past difference H, and predict the future difference H based on the slope and intercept of the regression line.

[0075] The processor 101 displays the result of the trend prediction process as a trend graph on the display 105 (step S83). The processor 101 stores data related to the trend graph in the memory 103 (step S85) and ends the trend analysis process. For example, the processor 101 stores various data such as the extracted frequency band, the trend graph, etc. in the memory 103.

[0076] <Screen example> Fig. 13 is a diagram showing an example of the layout of a user interface screen 500. However, the user interface screen 500 may have a layout other than that shown in Fig. 13 as long as the layout can realize the functions described below.

[0077] 13, a user interface screen 500 includes display areas 502-512, a measurement condition and setting value display area 514, various buttons 516, display areas 520-540, and graphs 550-570.

[0078] In the display area 502, the identification number (unit number) of the sensor unit 20, the identification number (sensor number) of the sensor 30, the name of the measurement target (e.g., pump), etc. are displayed. In the display area 504, a status based on the distance MD1 calculated by the MT method using multidimensional data is displayed. This status changes depending on the value of the calculated distance MD1. For example, when the distance MD1 is calculated to be equal to or greater than the threshold Th1 (e.g., 5), the status "Danger" is displayed. When the distance MD1 is calculated to be less than the threshold Th1 and equal to or greater than the threshold Th1a (e.g., 4), the status "Maintenance Recommended" is displayed. When the distance MD1 is calculated to be less than the threshold Th1a and equal to or greater than the threshold Th1b (e.g., 3), the status "Caution" is displayed. When the distance MD1 is calculated to be less than the threshold Th1b, the status "Normal" is displayed. In this way, the larger the distance MD1, the higher the warning level of the status.

[0079] A status based on the distance MD2 calculated by the MT method using two-dimensional data is displayed in the display area 506. This status changes according to the value of the distance MD2. For example, if the calculated distance MD2 is equal to or greater than a threshold Th2 (e.g., 5), the status "Danger" is displayed; if the calculated distance MD2 is less than the threshold Th2 and equal to or greater than a threshold Th2a (e.g., 4), the status "Maintenance Recommended" is displayed; if the calculated distance MD2 is less than the threshold Th2a and equal to or greater than a threshold Th2b (e.g., 3), the status "Caution" is displayed; and if the calculated distance MD2 is less than the threshold Th2b, the status "Normal" is displayed. In this way, the greater the distance MD2, the higher the warning level of the status.

[0080] A status based on trend analysis is displayed in the display area 508. This status changes according to the difference H at the reference time point. For example, if the difference H at the reference time point is equal to or greater than the reference value Z3 (e.g., 10 dB), the status "Danger" is displayed; if the difference H is less than the reference value Z3 and equal to or greater than the reference value Z2 (e.g., 6 dB), the status "Maintenance Recommended" is displayed; if the difference H is less than the reference value Z2 and equal to or greater than the reference value Z1 (e.g., 3 dB), the status "Caution" is displayed; and if the difference H is less than the reference value Z1, the status "Normal" is displayed. In this way, the larger the difference H, the higher the warning level of the status.

[0081] Display area 520 displays time-series sensor data (raw data) detected by sensor 30. Display area 530 displays the analysis results (frequency spectrum) of FFT (fast Fourier transform) analysis of the time-series sensor data. Display area 540 displays signal intensity data obtained by 1 / 3 octave analysis of the time-series sensor data in a bar graph.

[0082] Graph 550 shows time series data of distance MD1. Graph 560 shows time series data of distance MD2. Graph 570 is a trend graph obtained by the process of step S83 in FIG.

[0083] As described above, it is predicted that an abnormality will occur in the pump 70 several days after the distance MD1 equal to or greater than the threshold value Th1 is first calculated, and it is predicted that an abnormality will occur in the pump 70 several hours after the distance MD2 equal to or greater than the threshold value Th2 is first calculated. Therefore, by checking the graphs 550 and 560, the user can estimate when an abnormality will occur in the pump 70.

[0084] For example, when a distance MD1 equal to or greater than a threshold value Th1 is confirmed in graph 550, the user can proceed with preparations for maintenance while checking state changes, and when a distance MD2 equal to or greater than a threshold value Th2 is confirmed in graph 560, the user can immediately begin maintenance. This allows accurate prediction of the timing of equipment maintenance such as inspection, maintenance, and repair of pump 70, so that equipment maintenance can be carried out in a planned manner.

[0085] <Functional configuration> Fig. 14 is a functional block diagram of analysis device 10. Referring to Fig. 14, analysis device 10 includes, as main functional components, signal input unit 202, intensity calculation unit 204, first distance calculation unit 206, center of gravity calculation unit 208, second distance calculation unit 210, abnormality prediction unit 212, trend analysis unit 214, and output control unit 216. Each of these functions is realized, for example, by processor 101 of analysis device 10 executing a program stored in memory 103. Note that some or all of these functions may be configured to be realized by hardware.

[0086] The signal input unit 202 accepts input of a vibration signal detected by the sensor 30 attached to the pump 70 in operation. Specifically, the signal input unit 202 receives the vibration signal (digital signal) detected by the sensor 30 via the sensor unit 20.

[0087] The intensity calculation unit 204 calculates a plurality of signal intensities corresponding to a plurality of frequency bands by analyzing the vibration signal received by the signal input unit 202. Specifically, the intensity calculation unit 204 calculates the signal intensity of each frequency band (for example, the signal intensity L of each of the frequency bands f1 to fm) by performing octave analysis (for example, 1 / 3 octave analysis) on the vibration signal corresponding to the pump 70. Note that the intensity calculation unit 204 may be configured to calculate the signal intensity of each frequency band by Fast Fourier Transform (FFT).

[0088] The first distance calculation unit 206 calculates the Mahalanobis distance (e.g., distance MD1) of a signal space (e.g., signal space X1) composed of a plurality of signal intensities with respect to a preset unit space (e.g., unit space U1) using the MT method. Typically, the unit space U1 is composed of a plurality of signal intensities L corresponding to a plurality of frequency bands f1 to fn, which are calculated by analyzing a vibration signal corresponding to the pump 70 in a normal state. The vibration signal corresponding to the pump 70 in a normal state may be, for example, a vibration signal indicating the vibration state of the pump 70 at each of times T1 to Tn (i.e., an initial period after the start of operation), or may be a vibration signal indicating the vibration state of a pump of the same type as the pump 70 in a normal state. Time series data of the distance MD1 is generated by calculating the distance MD1 for each time (e.g., times Tn+1 to Tm).

[0089] The center of gravity calculation unit 208 calculates two-dimensional center of gravity data (for example, center of gravity positions Gx, Gy) indicating the center of gravity position of the multiple signal intensities calculated by the intensity calculation unit 204.

[0090] The second distance calculation unit 210 uses the MT method to calculate the Mahalanobis distance (e.g., distance MD2) of a signal space (e.g., signal space X2) configured with two-dimensional center of gravity data with respect to a preset unit space (e.g., unit space U2). Typically, the unit space U2 is configured with two-dimensional center of gravity data indicating the center of gravity positions of multiple signal intensities L that configure the unit space U1. By calculating the distance MD2 for each time (e.g., times Tn+1 to Tm), time-series data of the distance MD2 is generated.

[0091] The abnormality prediction unit 212 predicts the time when an abnormality will occur in the pump 70 based on the distance MD1 and the distance MD2. In a certain aspect, the time when an abnormality will occur when a distance MD1 equal to or greater than the threshold value Th1 is calculated is predicted to be in the near future compared to the time when an abnormality will occur when a distance MD1 equal to or greater than the threshold value Th1 is not calculated. Specifically, the abnormality prediction unit 212 predicts that an abnormality will occur in the pump 70 several days after a distance MD1 equal to or greater than the threshold value Th1 is calculated. Note that, when a distance MD1 equal to or greater than the threshold value Th1 has not been calculated at the present time, the abnormality prediction unit 212 may predict that the possibility of an abnormality occurring in the pump 70 within several days from the present time is low.

[0092] In another aspect, the time when an abnormality occurs when a distance MD1 equal to or greater than the threshold value Th1 is calculated and a distance MD2 equal to or greater than the threshold value Th2 is calculated is predicted to be in the near future compared to the time when an abnormality occurs when a distance MD1 equal to or greater than the threshold value Th1 is calculated and a distance MD2 equal to or greater than the threshold value Th2 is not calculated. Specifically, the abnormality prediction unit 212 predicts that an abnormality will occur in the pump 70 several hours after the distance MD2 equal to or greater than the threshold value Th2 is calculated. Note that, in a case where a distance MD1 equal to or greater than the threshold value Th1 is calculated at the present time, but a distance MD2 equal to or greater than the threshold value Th2 is not calculated, the abnormality prediction unit 212 may predict that the possibility of an abnormality occurring in the pump 70 within several hours from the present time is low.

[0093] The trend analysis unit 214 calculates, for each of a plurality of frequency bands, a difference H between the signal strength of the frequency band corresponding to the pump 70 and the signal strength of the frequency band corresponding to the reference pump. The trend analysis unit 214 determines the abnormality level of the vibration state of the pump 70 in each frequency band based on the signal strength of each frequency band corresponding to the pump 70 and a predetermined reference value. Specifically, the trend analysis unit 214 determines the abnormality level of the vibration state of the pump 70 in each of the plurality of frequency bands by comparing the difference H in that frequency band with a plurality of reference values ​​Z1 to Z3.

[0094] Furthermore, the trend analysis unit 214 identifies a predetermined number (e.g., five) of vibration states having the highest abnormality levels from among the vibration states of the pump 70 in each frequency band, and extracts the frequency bands corresponding to the identified vibration states. The trend analysis unit 214 predicts the future difference H of each frequency band based on the time series data of the difference H of each frequency band stored in the memory 103. Specifically, the trend analysis unit 214 predicts the future difference H by performing a regression analysis on the time series data of the difference H from a reference time point. Alternatively, the trend analysis unit 214 predicts the future difference H by approximating the time series data of the difference H of the past using an approximation curve.

[0095] The output control unit 216 outputs various information such as the prediction result of the abnormality prediction unit 212. In one aspect, when a distance MD1 equal to or greater than a threshold value Th1 is calculated, the output control unit 216 outputs first warning information (for example, information warning of an abnormality occurring in the pump 70 in the near future), and when a distance MD2 equal to or greater than a threshold value Th2 is calculated, the output control unit 216 outputs second warning information (for example, information warning of an abnormality occurring in the pump 70 in the extremely near future) having a higher warning level than the first warning information (highly urging the user to be vigilant).

[0096] For example, if the first warning information is a warning level indicating a medium degree of urgency, the second warning information is a warning level indicating a high degree of urgency. That is, the second warning information is information with a higher degree of urgency than the first warning information. In this case, the output control unit 216 may display the second warning information on the display 105 in an emphasized manner compared to the display manner of the first warning information (that is, so that the second warning information stands out more than the first warning information).

[0097] In another aspect, the output control unit 216 causes the display 105 to display various information shown on the user interface screen 500. Specifically, the output control unit 216 causes the display 105 to display time series data of the distance MD1 (for example, graph 550 in FIG. 13) and time series data of the distance MD2 (for example, graph 560 in FIG. 13). In still another aspect, the output control unit 216 causes the display 105 to display the predicted future difference H of each frequency band (for example, a trend graph shown in graph 570). In addition, the output control unit 216 outputs (for example, displays on the display 105) an abnormality alert based on the determination result of the abnormality level determined by the trend analysis unit 214.

[0098] <Advantages> According to this embodiment, by checking the distances MD1 and MD2 using the MT method, it is possible to predict whether an abnormality will occur in the pump 70 in the near future (e.g., in a few days) or in the very near future (e.g., in a few hours). Furthermore, by also referring to the trends in the trend graph, it is possible to predict an abnormality in the pump 70 from various angles in advance. Therefore, equipment maintenance can be carried out in a planned manner.

[0099] <Other embodiments> (1) In the above embodiment, the analysis device 10 is configured to predict that an abnormality will occur several days after the Mahalanobis distance MD1 that is equal to or greater than the threshold value Th1 is calculated for the first time. 1 After first becoming greater than or equal to threshold value Th1 at time Ta2, Ta3, and Ta4, the Mahalanobis distance MDx1 becomes greater than or equal to threshold value Th1 at time Ta5, and an abnormality occurs in the reference pump at time Ta5. Therefore, it can be said that the more times the Mahalanobis distance MDx1 becomes greater than or equal to threshold value Th1, the closer the time of abnormality occurrence. Therefore, the analysis device 10 (abnormality prediction unit 212) may predict that an abnormality will occur in the pump 70 in the near future, the more times the distance MD1 is calculated to be greater than or equal to threshold value Th1.

[0100] (2) In the above-described embodiment, a program for causing a computer to function and execute the control as described in the above-described flowchart can also be provided. Such a program can be provided as a program product by being recorded on a non-transitory computer-readable recording medium such as a flexible disk, a CD-ROM (Compact Disk Read Only Memory), a secondary storage device, a main storage device, or a memory card that is attached to the computer. Alternatively, the program can be provided by being recorded on a recording medium such as a hard disk built into the computer. The program can also be provided by downloading via a network.

[0101] The program may be one that executes processing by calling necessary modules in a predetermined sequence at a predetermined timing among program modules provided as part of a computer's operating system (OS). In that case, the program itself does not include the above modules and executes processing in cooperation with the OS. Such a program that does not include a module may also be included in the program according to the present embodiment. Furthermore, the program according to the present embodiment may be one that is provided by being incorporated into a part of another program. In that case, the program itself does not include modules included in the other program, and executes processing in cooperation with the other program. Such a program incorporated into another program may also be included in the program according to the present embodiment.

[0102] (3) The configurations exemplified as the above-mentioned embodiments are merely examples of the configurations of the present invention, and may be combined with other known technologies, or may be modified, such as by omitting some parts, without departing from the scope of the present invention. In addition, the above-mentioned embodiments may be implemented by appropriately adopting the processes and configurations described in other embodiments.

[0103] The embodiments disclosed herein should be considered to be illustrative and not restrictive in all respects. The scope of the present invention is defined by the claims, not the above description, and is intended to include all modifications within the scope and meaning equivalent to the claims. [Explanation of symbols]

[0104] 10 Analysis device, 20 Sensor unit, 21 Filter, 22 Amplifier, 23 Converter, 30 Sensor, 40 Terminal device, 50 Network, 70 Pump, 100 Vibration analysis system, 101 Processor, 103 Memory, 105 Display, 107 Input device, 109 Signal input interface, 111 Communication interface, 202 Signal input section, 204 Intensity calculation section, 206 First distance calculation section, 208 Center of gravity calculation section, 210 Second distance calculation section, 212 Anomaly prediction section, 214 Trend analysis section, 216 Output control section, 310, 320 Data set, 500 User interface screen, 1000 System.

Claims

1. a signal input unit that receives an input of a vibration signal detected by a sensor attached to an object in motion; an intensity calculation unit that calculates a plurality of signal intensities respectively corresponding to a plurality of frequency bands by analyzing a vibration signal corresponding to the object; a first distance calculation unit that calculates a first Mahalanobis distance of a first signal space formed by the plurality of signal intensities with respect to a preset first unit space; a centroid calculation unit that calculates two-dimensional centroid data indicating a centroid position of the plurality of signal intensities calculated by the intensity calculation unit; a second distance calculation unit that calculates a second Mahalanobis distance of a second signal space formed by the two-dimensional center of gravity data with respect to a second unit space that is set in advance; an abnormality prediction unit that predicts an abnormality occurrence time when an abnormality will occur in the object based on the first Mahalanobis distance and the second Mahalanobis distance.

2. the first unit space is composed of a plurality of signal intensities corresponding to the plurality of frequency bands, the signal intensities being calculated by analyzing a vibration signal corresponding to the object in a normal state; 2. The vibration analysis system according to claim 1, wherein the second unit space is formed of two-dimensional centroid data indicating a centroid position of the plurality of signal intensities that form the first unit space.

3. 3. The vibration analysis system according to claim 1, wherein the time of occurrence of an abnormality when the first Mahalanobis distance calculated is equal to or greater than a first threshold value is predicted to be in the near future compared to the time of occurrence of an abnormality when the first Mahalanobis distance calculated is not equal to or greater than the first threshold value.

4. 4. The vibration analysis system according to claim 3, wherein the abnormality prediction unit predicts that an abnormality will occur in the object several days after the first Mahalanobis distance equal to or greater than the first threshold is calculated.

5. 5. The vibration analysis system according to claim 3, wherein the time of occurrence of an abnormality when the first Mahalanobis distance calculated is equal to or greater than the first threshold value and the second Mahalanobis distance calculated is equal to or greater than the second threshold value is predicted to be in the near future compared to the time of occurrence of an abnormality when the first Mahalanobis distance calculated is equal to or greater than the first threshold value and the second Mahalanobis distance not calculated is equal to or greater than the second threshold value.

6. 6. The vibration analysis system according to claim 5, wherein the abnormality prediction unit predicts that an abnormality will occur in the object several hours after the second Mahalanobis distance equal to or greater than the second threshold is calculated.

7. 7. The vibration analysis system of claim 5, further comprising an output control unit that outputs first warning information when the first Mahalanobis distance calculated is equal to or greater than the first threshold value, and outputs second warning information having a higher warning level than the first warning information when the second Mahalanobis distance calculated is equal to or greater than the second threshold value.

8. 8. The vibration analysis system according to claim 7, wherein the output control unit causes a display to display the time series data of the first Mahalanobis distance and the time series data of the second Mahalanobis distance.

9. Receiving an input of a vibration signal detected by a sensor attached to an object in motion; calculating a plurality of signal strengths corresponding to a plurality of frequency bands by analyzing the vibration signal corresponding to the object; calculating a first Mahalanobis distance of a first signal space formed by the plurality of signal intensities with respect to a preset first unit space; calculating two-dimensional center of gravity data indicating a center of gravity position of the calculated signal intensities; calculating a second Mahalanobis distance of a second signal space formed by the two-dimensional center of gravity data with respect to a second unit space set in advance; and predicting an abnormality occurrence time when an abnormality will occur in the object based on the first Mahalanobis distance and the second Mahalanobis distance.

Citation Information

Patent Citations

  • Bearing fault diagnosis and prediction method based on extended Kalman filtering algorithm

    CN105300692A

  • Manufacturing machine and life predicting method for rotating machine

    JP2004117253A

  • Apparatus diagnosing device, refrigeration cycle device, fluid circuit diagnosing method, apparatus monitoring system and refrigeration cycle monitoring system

    JP2005207644A

  • Life prediction method, life prediction device, and life prediction system

    JP2017142153A

  • Representative data selection device, device diagnostic device, program and representative data selection method

    JP2019035585A