WO1997030329A1 - Procede de detection d'anomalies et systeme de detection d'anomalies - Google Patents
Procede de detection d'anomalies et systeme de detection d'anomalies Download PDFInfo
- Publication number
- WO1997030329A1 WO1997030329A1 PCT/JP1997/000367 JP9700367W WO9730329A1 WO 1997030329 A1 WO1997030329 A1 WO 1997030329A1 JP 9700367 W JP9700367 W JP 9700367W WO 9730329 A1 WO9730329 A1 WO 9730329A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- vector
- time
- calculated
- generating
- abnormality
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D1/00—Measuring arrangements giving results other than momentary value of variable, of general application
- G01D1/18—Measuring arrangements giving results other than momentary value of variable, of general application with arrangements for signalling that a predetermined value of an unspecified parameter has been exceeded
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D1/00—Measuring arrangements giving results other than momentary value of variable, of general application
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M99/00—Subject matter not provided for in other groups of this subclass
Definitions
- the present invention relates to a method for detecting an abnormality in a manufacturing apparatus, a manufacturing plant, a product, or the like, and a system used for implementing the method.
- a main object of the present invention is to provide an abnormality detection method capable of detecting various abnormalities with high accuracy by using a time-series data and a prediction model thereof, and a system used for implementing the method. is there. Disclosure of the invention
- the present inventor has constructed a prediction model based on a plurality of past time-series signals, provided the obtained time-series signals to the prediction model, predicted the trends of the time-series signals, and obtained the prediction results.
- linear autoregression for example, GEP ⁇ . ⁇ and GM Jenkins, Time Series Analysis: Forecasting and Control Holden-Day Inc., 1976
- neural network for example, No. LA-UR 87-2662 (1987). No. LA-U 88-418 (1988)
- Dynamic Replacement Sheet Rule 26
- Radial basis function network for example, M. Casdagli, Physica D, Vol. 35, pp. 335-356 (1989)
- simplex projection for example, Sugihara and RM May, Nature, Vol. 344) , pp.734-741 (1990)
- An abnormality detection method is a method for measuring a physical quantity that changes with time emitted by an object at predetermined time intervals, and detecting an abnormality of the object based on obtained time-series data.
- a plurality of first vectors of appropriate dimensions are generated from the time-series data, a second vector is generated by translating each first vector by an appropriate time, and the generated second vector and first vector are generated. Calculate the deviation from the torque, compare the calculated value with the set threshold, and judge whether or not there is an abnormality based on the comparison result.
- the method of calculating the deviation includes a step of selecting an appropriate vector from the first vector, and a step of selecting a predetermined number of vectors from the first vector in an ascending order of distance to the selected vector.
- a step of extracting the vector from the vector a step of generating a translation vector obtained by translating the extraction vector and the selection vector by an appropriate time, and a step of extracting the generated translation vector, the selection vector, and the extraction.
- the method includes a step of obtaining a difference vector from the vector and a step of calculating a variance of the obtained difference vector.
- N is the number of data.
- An appropriate embedding dimension D and sampling interval ⁇ t are set, and a vector X (t) expressed by the following equation (2) is generated from the time-series data. It has been proven that generating vectors X (t) for different embedding dimensions D is equivalent to obtaining time-series data for multiple types of physical quantities, respectively. Thus, abnormality detection can be performed with high accuracy.
- the vector X (t) is translated by an appropriate time interval ⁇ to generate a vector ⁇ (t) expressed by the following equation (3).
- Y (t) ⁇ X (t + ⁇ ⁇ t), x (t + TA t-At), ⁇ , ⁇ (t + ⁇ ⁇ t-(D- 1) ⁇ t) ⁇ ... (3 )
- the N vectors X (t) force and the appropriate vector X (t (0)) are selected, and the selected vector X (t (0)) and the other vectors (N-1) are selected.
- V (t (k)) X (t (k)) ⁇ (t (k))... (5)
- FIG. 1 is a block diagram showing a configuration of an abnormality detection system according to the present invention
- FIG. 2 is a flowchart showing an abnormality detection procedure using the computer of FIG. 1
- FIG. 3 is a flowchart of FIG. Fig. 4 is a flowchart showing the abnormality detection procedure by the computer shown in Fig. 4
- Fig. 4 is a flowchart showing the abnormality detection procedure by the computer shown in Fig. 1
- Fig. 5 is the flowchart shown in Fig. 1.
- 6 (A) and 6 (B) are graphs showing the time series data used in the comparative test
- FIG. 7 is a flowchart showing the abnormality detection procedure according to the method of the present invention.
- FIGS. 9 (A) and 9 (B) are graphs showing the time series data used in the comparative test
- FIG. 10 is shown in FIGS. 9 (A) and 9 (B) by the method of the present invention.
- Fig. 11 shows the results from the two time series data shown in Figs. 6 is a graph showing a result of creating a power spectrum.
- FIG. 1 is a block diagram showing the configuration of the abnormality detection system according to the present invention.
- reference numeral 1 denotes a sound sensor, an ultrasonic sensor, a vibration sensor, an optical sensor, a voltmeter, an ammeter, and the like.
- Measuring instrument The type of measuring instrument 1 is selected according to the physical quantity to be measured, and measuring instrument 1 measures the physical quantity of the object to be detected for abnormality at a predetermined cycle.
- the output of the measuring instrument 1 is supplied to an analog / digital (AZD) converter 2, where it is converted into a digital signal, which is then supplied to a memory 32 provided in a computer 3, where the time is output. Stored as sequence data.
- APD analog / digital
- the time series data stored in the memory 32 within a predetermined time is obtained by the central processing unit 31 by the vector deviation calculating unit 33, and the vector deviation calculating unit 33 outputs E (trans 'Calculate and give it to the failure diagnosis unit 34 that determines whether it is abnormal.
- the failure condition input unit 35 has a first threshold and a second threshold set, and the failure diagnosis unit 34 has a vector deviation.
- the E (trans) provided by the calculation unit 33 is compared with the first threshold or the second threshold set in the failure condition input unit 35, respectively, and E (trans) is the first ⁇ replacement sheet (Rule 26) If the value is equal to or more than the value, or if E (trans) is equal to or less than the second threshold value, it is determined that an abnormality has occurred, and the output unit 36 outputs an alarm to the external device 4 such as a system control device or an alarm device. .
- FIGS. 2 to 5 are flowcharts showing the abnormality detection procedure using the computer 3 shown in FIG.
- the vector deviation calculation part 33 of the computer 3 has an embedding dimension D, a sampling interval A t, a translation time TA t, a vector selection number M, and a repetition number K, Q.
- a first threshold value and a second threshold value are set in the input unit 35 (step S 1).
- the memory 32 of the computer 3 stores the digital signal given from the AZD converter 2 as time-series data (step S2), and stores the time-series data stored in the memory 32 within a predetermined time. Is supplied to the vector deviation calculation unit 33 by the central processing unit 31.
- the vector deviation calculation unit 33 outputs the given time series data ⁇ X (t) ⁇ ,! From (N is the number of data), a vector X (t) represented by the following equation (2) is generated (step S3).
- the vector deviation calculator 33 translates the vector X (t) by an appropriate time interval ⁇ , and 3) Generate a vector ⁇ (t) represented by the equation (step S4).
- the vector deviation calculating unit 33 generates an arbitrary uniform random number (step S5), and uses the generated uniform random number from the N vectors X (t) generated in step S3.
- M (M ⁇ N) vectors X (t (0)) are selected (step S6), and one vector X (t (0)) replacement sheet (rule 26) is appropriately selected from them. ) Is identified (step S7).
- h x ⁇ [x (t (0)-d ⁇ t) — x (t (k)-d ⁇ t)] 2
- the vector deviation calculator 33 calculates (K + 1) number of difference vectors V (t (k)) based on the following equation (5) (step S11), and calculates each difference vector V (t (k)). Based on t (k)) and the following equation (7), a first-order translation error E (trans)! is obtained (step S12).
- V (t (k)) X (t (k)) -Y (t (k))-(5)
- the vector deviation calculator 33 repeats the operations from step S7 to step S12 until it is determined in step S13 that the M primary translation errors E (trans), have been obtained. For the M vectors X (t (0)) selected in, obtain the first-order translation error E (trans) ,. Beq replacement paper (Rule 26) The torque deviation calculator 33 determines that M] translation errors E (trans) 1 have been obtained, calculates an intermediate value (median) thereof, and obtains a first-order average translation error E (trans). Is obtained (step S14).
- the vector deviation calculation unit 33 performs the operations from step S5 to step S14 until it is determined in step S15 that Q primary average translation errors E (trans),, are obtained. Repetition, Q first-order average translation errors
- the vector deviation calculation unit 33 gives it to the failure diagnosis unit 34.
- the failure diagnosis unit 34 compares E (trans) given from the vector deviation calculation unit 33 with the first threshold and Z or the second threshold set by the failure condition input unit 35 (step S17). If E (trans) is equal to or greater than the first threshold, or if E (trans) is equal to or less than the second threshold, it is determined that there is an abnormality (step S18), and the external device 4 To output an alarm (Step S19) o
- FIGS. 6 and 9 are graphs showing the time series data used in the comparative test, in which the vertical axis represents the signal intensity and the horizontal axis represents the time from the start of the measurement.
- Fig. 6 is a graph showing the vibration of the motor rotating part of the fan in a time series.
- Fig. 6 (A) shows the time series data when no abnormality occurs.
- Figure (B) shows time-series data when an error occurs.
- the signal strength on the vertical axis is converted into an electric signal by the vibration sensor, but the reflected wave of the vibration part may be detected by the optical sensor.
- the voltage characteristics of the motor measured by a voltmeter or motor switching ( ⁇ ⁇ Article 26) 1 O
- a similar oscillating component can be obtained from the current value of the evening current measured by an ammeter.
- Fig. 9 is a graph showing the sound signal as time-series data.
- Fig. 9 (A) shows the time-series data when no abnormality occurs
- Fig. 9 (B) shows the time when abnormality occurs. This is the time series data at the time of performing.
- the signal strength on the vertical axis is converted into an electric signal by a sound wave sensor.
- FIG. 7 is a graph showing the result of calculating E (trans) from the two time-series data shown in FIGS. 6 (A) and (B) by the method of the present invention.
- the symbols ⁇ indicate the abnormal cases. Note that the sampling interval ⁇ t is 1, the translation time ⁇ t T is 5, the vector selection number M is 300, the vector selection number K is 4, and the number of repetitions Q is The embedding dimension D is set to 5, 6, 6.7, 8, 9, and 10 in 20.
- Fig. 8 is a graph showing the result of creating a power spectrum from the two time-series data shown in Figs. 6 (A) and (B) by the conventional method. Indicates a case where there is an abnormality.
- the abnormal case and the non-abnormal case are clearly separated, and the abnormal condition is set by setting the threshold value to 0.05. Can be determined with high accuracy.
- the power spectrum in the case of an abnormality and the power spectrum in the case of a non-abnormal case often overlapped, and a judgment error occurred between the two.
- FIG. 10 is a graph showing the result of calculating E (trans) from the two time series data shown in FIGS. 9 (A) and (B) according to the present invention.
- the sampling interval ⁇ t is set to 1 and the translation time T ⁇ t is replaced (Rule 26).
- T to 5 vector selection ⁇ to 300, vector selection ⁇ to 4, repetition Q to 20 and embedding dimension D to 1 1, 12, 1 3, 1 4, 1 It is set to 5.
- Fig. 11 is a graph showing the result of creating a power spectrum from the two time-series data shown in Figs. 9 ( ⁇ ) and ( ⁇ ) by the conventional method. The broken line shows the case where there is an abnormality.
- the abnormal case and the non-abnormal case are clearly separated, and by setting 0.90 as the threshold value, the abnormal condition is determined. It can be determined with high accuracy whether or not there is.
- FIG. 11 in the conventional method, there are many places where the power spectrums of the abnormal case and the non-abnormal case overlap, and a judgment error occurs between the two.
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Description
Claims
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AU16716/97A AU1671697A (en) | 1996-02-16 | 1997-02-13 | Method of detecting abnormality and abnormality detection system |
EP97902662A EP0881467A4 (en) | 1996-02-16 | 1997-02-13 | METHOD AND SYSTEM FOR DETECTING ANOMALIES |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2973396 | 1996-02-16 | ||
JP8/29733 | 1996-02-16 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO1997030329A1 true WO1997030329A1 (fr) | 1997-08-21 |
Family
ID=12284315
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP1997/000367 WO1997030329A1 (fr) | 1996-02-16 | 1997-02-13 | Procede de detection d'anomalies et systeme de detection d'anomalies |
Country Status (6)
Country | Link |
---|---|
EP (1) | EP0881467A4 (ja) |
KR (1) | KR19990082532A (ja) |
CN (1) | CN1213429A (ja) |
AU (1) | AU1671697A (ja) |
TW (1) | TW327221B (ja) |
WO (1) | WO1997030329A1 (ja) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114184274A (zh) * | 2021-11-19 | 2022-03-15 | 哈尔滨工程大学 | 一种受高频噪声干扰振动信号的非线性特性诊断方法 |
Families Citing this family (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6526405B1 (en) * | 1999-12-17 | 2003-02-25 | Microsoft Corporation | Determining similarity between event types in sequences |
US7152068B2 (en) * | 2001-12-21 | 2006-12-19 | Honeywell International Inc. | Method and apparatus for retrieving time series data related to an activity |
JP4200332B2 (ja) * | 2006-08-29 | 2008-12-24 | パナソニック電工株式会社 | 異常監視装置、異常監視方法 |
KR101029394B1 (ko) | 2006-09-28 | 2011-04-15 | 미쓰비시덴키 가부시키가이샤 | 고장 검출 장치, 고장 검출 방법 및 고장 검출 프로그램 |
TWI474023B (zh) * | 2008-12-10 | 2015-02-21 | Ind Tech Res Inst | 馬達故障診斷方法及其診斷裝置 |
DE102009028325A1 (de) * | 2009-08-07 | 2011-02-10 | Robert Bosch Gmbh | Verfahren und Vorrichtung zur Detektion von Schwingungen |
KR101659989B1 (ko) * | 2014-12-10 | 2016-09-27 | 부산대학교 산학협력단 | 다차원 특징을 이용한 이상신호 분석장치 및 방법 |
CN105938011B (zh) * | 2016-07-15 | 2017-08-11 | 武汉理工大学 | 一种水力除焦过程钻杆工作状态实时判定方法 |
CN106872168A (zh) * | 2017-03-09 | 2017-06-20 | 武汉科技大学 | 一种用于滚动轴承的故障诊断方法 |
CN107703916B (zh) * | 2017-09-30 | 2019-06-07 | 英特尔产品(成都)有限公司 | 电子压力调节器诊断系统及方法 |
CN109861857A (zh) * | 2019-01-28 | 2019-06-07 | 网联清算有限公司 | 故障检测方法及装置 |
CN109919495A (zh) * | 2019-03-12 | 2019-06-21 | 中国计量大学 | 基于阈值计算的光伏电站监测数据异常与程度判断方法 |
CN111984827A (zh) * | 2019-05-24 | 2020-11-24 | 上海东方富联科技有限公司 | 门磁传感器数据异常检测方法及系统、存储介质及终端 |
KR102230463B1 (ko) * | 2019-12-12 | 2021-03-19 | 울산대학교 산학협력단 | 설비 부품의 결함 진단 시스템 및 이의 방법 |
CN112710446B (zh) * | 2020-12-21 | 2022-06-07 | 北京和中普方新能源科技有限公司 | 电动车电池系统振动试验的判断方法、系统及存储介质 |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH02226024A (ja) * | 1989-02-27 | 1990-09-07 | Toshiba Corp | 回転機械のラビング検出装置 |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4453407A (en) * | 1980-04-17 | 1984-06-12 | Hitachi, Ltd. | Vibration diagnosis method and apparatus for rotary machines |
JP2909134B2 (ja) * | 1990-04-02 | 1999-06-23 | 株式会社東芝 | 回転機械の回転部異常検知装置 |
-
1997
- 1997-02-13 WO PCT/JP1997/000367 patent/WO1997030329A1/ja not_active Application Discontinuation
- 1997-02-13 CN CN97192893A patent/CN1213429A/zh active Pending
- 1997-02-13 KR KR1019980706268A patent/KR19990082532A/ko active IP Right Grant
- 1997-02-13 AU AU16716/97A patent/AU1671697A/en not_active Abandoned
- 1997-02-13 EP EP97902662A patent/EP0881467A4/en not_active Withdrawn
- 1997-02-15 TW TW086101800A patent/TW327221B/zh active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH02226024A (ja) * | 1989-02-27 | 1990-09-07 | Toshiba Corp | 回転機械のラビング検出装置 |
Non-Patent Citations (1)
Title |
---|
See also references of EP0881467A4 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114184274A (zh) * | 2021-11-19 | 2022-03-15 | 哈尔滨工程大学 | 一种受高频噪声干扰振动信号的非线性特性诊断方法 |
Also Published As
Publication number | Publication date |
---|---|
KR19990082532A (ko) | 1999-11-25 |
AU1671697A (en) | 1997-09-02 |
CN1213429A (zh) | 1999-04-07 |
EP0881467A4 (en) | 1999-05-12 |
EP0881467A1 (en) | 1998-12-02 |
TW327221B (en) | 1998-02-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO1997030329A1 (fr) | Procede de detection d'anomalies et systeme de detection d'anomalies | |
Tanner Jr et al. | Definitions of d′ and η as psychophysical measures | |
JP3040471B2 (ja) | 力学系アナライザ | |
US11934184B2 (en) | Automated analysis of non-stationary machine performance | |
CN102016607B (zh) | 用于分析电力系统的波形信号的方法和装置 | |
KR101025163B1 (ko) | 진동 및 소음 전달경로 해석 시스템과 진동 및 소음 전달경로 해석 방법 | |
CN106199267B (zh) | 一种电力设备故障特征分析方法 | |
JP6316510B1 (ja) | 電動機の診断装置 | |
Balouji et al. | Partial discharge classification in power electronics applications using machine learning | |
JP2022094037A (ja) | 異常温度検知装置、異常温度検知方法、及び異常温度検知プログラム | |
Aggoune et al. | Fault detection in the distillation column process using Kullback Leibler divergence | |
Ferreira et al. | An application of machine learning approach to fault detection of a synchronous machine | |
US8140282B2 (en) | Determining a total length for conductive whiskers in computer systems | |
Roy et al. | Software reliability allocation of digital relay for transmission line protection using a combined system hierarchy and fault tree approach | |
Aggoune et al. | Modeling of a distillation column based on NARMAX and Hammerstein models | |
Wilson et al. | Uncertainty Quantification of Capacitor Switching Transient Location Using Machine Learning | |
Reger et al. | Pattern recognition in load profiles of electric drives in manufacturing plants | |
JP7230371B2 (ja) | 異常検出装置、異常検出方法、異常検出プログラム及び異常検出システム | |
CN104267259B (zh) | 电力谐波测量方法和系统 | |
RU2242164C2 (ru) | Способ выявления информативных параметров st-сегмента электрокардиосигнала и устройство для его осуществления | |
CN111679651A (zh) | 用于故障引起的变结构变参数系统的辨识方法及系统 | |
JP2002196783A (ja) | 時系列信号の識別方法及び装置 | |
EP0386820A1 (en) | Method and arrangement for probabilistic pitch measuring | |
JP2002196784A (ja) | 時系列信号の識別方法及び装置 | |
Abid et al. | Incipient Broken Bar Detection Using Information Entropy of Second Generation Wavelet Transform |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 97192893.2 Country of ref document: CN |
|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AL AM AT AU AZ BA BB BG BR BY CA CH CN CU CZ DE DK EE ES FI GB GE HU IL IS JP KE KG KR KZ LC LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK TJ TM TR TT UA UG US UZ VN YU |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): KE LS MW SD SZ UG AT BE CH DE DK ES FI FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN ML MR NE SN TD |
|
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 1997902662 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1019980706268 Country of ref document: KR |
|
WWP | Wipo information: published in national office |
Ref document number: 1997902662 Country of ref document: EP |
|
REG | Reference to national code |
Ref country code: DE Ref legal event code: 8642 |
|
WWP | Wipo information: published in national office |
Ref document number: 1019980706268 Country of ref document: KR |
|
WWG | Wipo information: grant in national office |
Ref document number: 1019980706268 Country of ref document: KR |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: 1997902662 Country of ref document: EP |