• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    The invention relates to a method for automatically adjusting alarm parameters for a machine. The method comprises the steps of obtaining the historical measurement data of the machine which are normally regularly recorded, comparing on the one hand the historical measurement data at each alarm trigger and on the other hand the measurement data of the machine. current at each alarm trigger with recorded faults, identify a relationship between alarm triggers and detected faults, where a relationship is identified if an alarm has been recorded a set time before fault detection, set limits defined allowable for all such relationships on the basis of eligible historical measurement data, count these relationships from the current measurement data, compare the number of said relationships counted with the allowable defined limits. If the compared number of relationships is within the allowable defined limits, then the adjustment is complete. Otherwise the adjustment is not completed, and the adjustment is made automatically until the defined allowable limits are reached.
    公开号:FR3067154A1
    申请号:FR1854759
    申请日:2018-06-01
    公开日:2018-12-07
    发明作者:Per-Erik Larsson
    申请人:SKF AB；
    IPC主号:
    专利说明:

    (54) METHOD FOR TREND ANALYSIS AND AUTOMATIC ADJUSTMENT OF ALARM PARAMETERS.
    FR 3 067 154 - A1 (57) The invention relates to a method for automatically adjusting alarm parameters for a machine. The method comprises the steps of obtaining the historical measurement data of the machine which are normally recorded regularly, comparing on the one hand the historical measurement data on each alarm triggering and on the other hand machine measurement data current at each alarm trigger with recorded faults, identify a relationship between alarm triggers and detected faults, where a relationship is identified if an alarm has been recorded an established time before the detection of the fault, establish limits defined admissible for all said relations on the basis of admissible historical measurement data, counting said relations from current measurement data, comparing the number of said relations counted with the defined admissible limits. If the compared number of relationships is within the defined allowable limits, then the adjustment is complete. Otherwise the adjustment is not completed, and the adjustment is carried out automatically until the defined admissible limits are reached.
    Obi don
    I
    Establish dàfiriieE
    i
    METHOD FOR AUTOMATIC TREND ANALYSIS AND ADJUSTMENT OF ALARM PARAMETERS
    FIELD OF THE INVENTION
    A method of trend analysis and automatic adjustment is provided. In particular, this invention relates to a method of trend analysis and automatic adjustment of alarm parameters for a machine.
    BACKGROUND OF THE INVENTION
    In the field of state monitoring, a common problem is that the operating states of a machine are constantly changing and, therefore, the measurement value of a state indicator also changes. This makes it difficult to establish appropriate alarm thresholds and the risk of false alarms is relatively high. Significant efforts have been made these days to create reliable status indicators and to use alarm hysteresis to avoid false alarms.
    SUMMARY OF THE INVENTION
    The basic concept of the invention relates to a method of trend analysis and adjustment of alarm parameters for a machine.
    One aspect of the invention relates to a method for adjusting alarm parameters for a machine, and comprising the steps consisting in:
    - obtain the historical measurement data of the machine which is normally recorded regularly, this including the detected faults and the triggering of alarms, with M the number of times that the value of State indicator (IE) has been greater than a threshold during the last N measurements,
    - compare on the one hand the historical measurement data during each alarm triggering and on the other hand the current machine measurement data during each alarm triggering with the recorded faults,
    - identify a relationship between the triggering of alarms and the detected faults, where a relationship is identified if an alarm has been recorded a set time before the detection of the fault, and where:
    o if a relationship exists between the alarm and the fault, then the alarm is defined as a true positive alarm (VP), o if a fault has been detected but no alarm has been triggered, then the fault is defined as a false negative (FN), o if an alarm has been triggered without recording a fault, then the fault is defined as a false positive (FP),
    - establish defined admissible limits for true positives (VP), false negatives (FN) and false positives (FP) on the basis of admissible historical measurement data,
    - count the number of true positives (VP), false negatives (FN) and false positives (FP) from the current measurement data,
    - compare the number of true positives (VP), false negatives (FN) and false positives (FP) counted with the defined allowable limits, where:
    o if the compared number of true positives (VP), false negatives (FN) and false positives (FP) counted is within the defined admissible limits, then the adjustment is finished, o if the compared number of true positives (VP), false negatives (FN) and false positives (FP) counted is outside the defined defined limits, so the adjustment is not complete, and where:
    - the adjustment of the calculation of the State Indicators (IE), the threshold, M and N is carried out automatically until the defined admissible limits are reached, where:
    o if the number of false positives (FP) exceeds the admissible defined limits, then at least one of the threshold, of M and N is increased, o if the number of false negatives (FN) exceeds the defined admissible limits, then at least one of the threshold, M and N is decreased, and the step is repeated until the alarm parameters are adjusted.
    Advantageously, the adjustment process is carried out following a trend analysis process of the alarm parameters for a machine, and comprising the steps consisting in:
    - obtain a series of measurements relating to machine states, where the State indicator value (IE) relates to a machine state with respect to the fault,
    - check the value of the State indicator (IE) in relation to an established threshold,
    - calculate M as the number of times that the State indicator value (IE) has been greater than the threshold during the last N measurements,
    - present the number M of times that the State indicator value (IE) has been greater than the threshold during the last N measurements in a graph with the date / time on the x-axis and the number of times on the y-axis, and
    - trigger the alarm if the State indicator value (IE) has been greater than the threshold established a number of times M greater than the alarm level during the last N measurements.
    According to another aspect of the present invention, it is provided in the method that during the step of obtaining a series of measurements relating to states, each measurement contains a time series of data points which creates a form of 'time wave.
    According to another aspect of the invention, the method includes performing a time waveform (TFR) signal analysis to create a spectrum.
    According to another aspect of the present invention, the method determines a fault associated with the measurements relating to states using spectral peaks located on the time waveform.
    According to another aspect of the present invention, the method comprises the identification of the spectral peaks relating to a particular defect to be analyzed.
    According to another aspect of the present invention, the method includes calculating the state indicator value (IE) from the identified spectral peaks.
    According to another aspect of the present invention, the method provides for multiple values of state indicators (IE), each designed to detect a fault and calculated from each measurement.
    According to another aspect of the present invention, the series of measurements relating to states of the mechanism corresponds to a vibration and / or a temperature.
    According to another aspect of the present invention, each measurement contains a time series from 1024 to 16384 data points.
    Those skilled in the art will better understand and appreciate these and other advantages of the invention upon examination of the written description and claims and the accompanying drawings.
    SHORT DESCRIPTION OF THE MULTIPLE VIEWS OF THE DRAWINGS
    The invention will now be described, by way of illustration, with reference to the accompanying drawings, in which:
    - Figure 1 is a flow diagram of a trend analysis and automatic adjustment method according to the present invention;
    - Figure 2 is a graph of an example of initial trends from a defective bearing comprising a chipped outer ring having defects difficult to detect according to the prior art; and
    FIG. 3 is a graph of an example of new trends originating from a defect, the new trends being stable, according to the present invention.
    DETAILED DESCRIPTION OF THE INVENTION
    The following detailed description refers to the accompanying drawings. Where possible, the same reference numbers have been used in the drawings and the following description refers to the same or similar parts. If multiple embodiments and illustrative elements of this disclosure are described here, modifications, adaptations and other embodiments are possible without departing from the spirit or scope of this disclosure. Accordingly, the following detailed description does not limit this disclosure.
    FIG. 1 is a flow diagram illustrating a method 100 of trend analysis 200 and adjustment 300 of alarm parameters for a machine.
    In the present case, the invention examines a machine for which a certain number of measurements relating to states are carried out. Such a measure, e.g. a vibration measurement can be taken for explanatory purposes, but the measurement can be any number of measurements related to states, including temperature, force, dynamic pressure, etc.
    The trend analysis method 100 200 comprises a step 210 consisting in obtaining a series of measurements relating to states of the machine. Each measure can contain a time series from 1024 to 16384 data points, which represents a time waveform. The method includes performing signal analysis on this waveform, called TER, which creates a spectrum.
    From this, spectral peaks relating to a particular fault are selected and used to calculate a State Indicator (IE) value which reflects the state of the machine with respect to this fault. It should be noted that several IE values, each designed to detect a fault, can be calculated from each measurement. So there is an IE value for each measurement.
    In a next step 220, the state indicator value (IE) is checked against a set threshold.
    The system then calculates the number of times the value of IE has been greater than the threshold during the last N measurements in step 230.
    This number of times is then presented in a graph with the date / time on the x-axis and the number of times on the y-axis in step 240.
    In step 250, if the State Indicator (IE) value has been greater than the threshold a number of times greater than the alarm level during the last N measurements, an alarm is triggered.
    The adjustment method 100 300 comprises a step 310 consisting in obtaining historical measurement data from the machine which is normally recorded regularly by a technician. The historical measurement data includes detected faults and alarm triggers according to the preceding steps 210 to 250. For example, a wind farm and faults can be chosen at the main bearing of each wind turbine.
    Then, in step 320, one can compare on the one hand the historical measurement data at each alarm triggering and on the other hand current machine measurement data at each alarm triggering with the detected faults and saved.
    Step 330 consists in identifying a relationship between the triggering of alarms and the detected faults. A relationship is identified if an alarm has been recorded an established time before the fault is detected. Thus, if we see a correlation between the alarm and the fault, that is to say that an alarm was recorded one month before the detection of the fault, then the alarm is defined as a true positive alarm (VP ). OK! If a fault has been detected but no alarm has been triggered, then the fault is defined as a false negative (EN). INCORRECT! Finally, if an alarm has been triggered without recording a fault, then the fault is defined as a false positive (EP). INCORRECT!
    Step 340 is to establish allowable defined limits for true positives (VP), false negatives (FN), and false positives (FP) based on admissible historical measurement data.
    Step 350 consists of counting the number of true positives (VP), false negatives (FN) and false positives (FP) from the current measurement data and step 360 to compare the number of true positives (VP) ), false negatives (FN) and false positives (FP) counted with what has been defined as allowable limits. For example, for a wind farm of 100 wind turbines, it would be admissible to trigger a maximum of 10 FP and 2 FN in a year. However, to feel safe, at least 10 VPs should have occurred during this same period.
    Therefore, if the compared number of true positives (VP), false negatives (FN) and false positives (FP) counted is within the defined allowable limits, then the adjustment is completed in accordance with step 370.
    However, if the compared number of true positives (VP), false negatives (FN) and false positives (FP) counted is outside the defined allowable limits, then the adjustment is not complete.
    Consequently, the adjustment of the calculation of the State Indicators (IE), the threshold, M and N is carried out automatically until the defined admissible limits are reached. This is done by returning to process step 340 of establishing / adjusting the defined allowable limits until the adjustment is complete.
    Figure 2 illustrates trends from a defective bearing with a chipped outer ring.
    It is difficult to identify a trend for this defect, but it is clearly visible in the new trend illustrated in Figure 3. After replacing the bearing (2014-09-15), the initial trend in Figure 2 indicates unstable behavior , while the new trend in Figure 3 is very stable.
    权利要求:
    Claims (9)
    [1" id="c-fr-0001]
    1. Method for automatically adjusting alarm parameters for a machine, the method comprising the steps consisting in:
    - obtain the historical measurement data of the machine which is normally recorded regularly, this including the detected faults and the triggering of alarms, with M the number of times that the value of State indicator (IE) has been greater than a threshold during the last N measurements,
    - compare on the one hand the historical measurement data at each alarm triggering and on the other hand the current machine measurement data at each alarm triggering with the recorded faults,
    - identify a relationship between the triggering of alarms and the detected faults, where a relationship is identified if an alarm has been recorded a time established before the detection of the fault, and where:
    o if a relationship exists between the alarm and the fault, then the alarm is defined as a true positive alarm (VP), o if a fault has been detected but no alarm has been triggered, then the fault is defined as a false negative (FN), o if an alarm has been triggered without recording a fault, then the fault is defined as a false positive (FP),
    - establish defined admissible limits for true positives (VP), false negatives (FN) and false positives (FP) on the basis of admissible historical measurement data,
    - count the number of true positives (VP), false negatives (FN) and false positives (FP) from the current measurement data,
    - compare the number of true positives (VP), false negatives (FN) and false positives (FP) counted with the defined allowable limits, where:
    o if the compared number of true positives (VP), false negatives (FN) and false positives (FP) counted is within the defined admissible limits, then the adjustment is finished, o if the compared number of true positives (VP), false negatives (FN) and false positives (FP) counted is outside the defined admissible limits, then the adjustment is not completed, and where:
    - the adjustment of the calculation of the State Indicators (IE), of the threshold, of M and of N is carried out automatically until the admissible defined limits are reached, where:
    o if the number of false positives (FP) exceeds the admissible defined limits, then at least one of the threshold, of M and N is increased, o if the number of false negatives (FN) exceeds the defined admissible limits, then at least one of the threshold, M and N is decreased, and the step is repeated until the alarm parameters are adjusted.
    [2" id="c-fr-0002]
    2. Method comprising a method of trend analysis of alarm parameters for a machine, followed by the steps of the method of automatic adjustment of alarm parameters for a machine according to claim 1, said method of trend analysis alarm parameters for a machine comprising the steps of:
    - obtain a series of measurements relating to machine states, where the State indicator value (IE) relates to a machine state with respect to the fault,
    - check the value of the State indicator (IE) in relation to an established threshold,
    - calculate M as the number of times that the State indicator value (IE) has been greater than the threshold during the last N measurements,
    - present the number M of times that the State indicator value (IE) has been greater than the threshold during the last N measurements in a graph with the date / time on the x-axis and the number of times on the y-axis, and
    - trigger the alarm if the State indicator value (IE) has been greater than the threshold established a number of times M greater than the alarm level during the last N measurements.
    [3" id="c-fr-0003]
    The method of claim 2, wherein in the step of obtaining a series of measurements relating to states, each measurement contains a time series of data points which creates a time waveform.
    [4" id="c-fr-0004]
    4. A method according to any one of claims 2 or 3, further comprising performing a signal analysis on the time waveform (TFR) to create a spectrum.
    [5" id="c-fr-0005]
    5. Method according to any one of claims 2 to 4, further comprising determining a defect associated with the measurements relating to states using spectral peaks located on the time waveform.
    [6" id="c-fr-0006]
    6. Method according to any one of claims 2 to 5, further comprising the fact of identifying the spectral peaks relating to a particular defect to be analyzed.
    [7" id="c-fr-0007]
    7. A method according to any of claims 2 to 6, further comprising calculating the State Indicator (IE) value from the identified spectral peaks.
    [8" id="c-fr-0008]
    8. Method according to any one of claims 2 to 7, further comprising
    10 multiple State Indicator (IE) values, each designed to detect a fault and calculated from each measurement.
    [9" id="c-fr-0009]
    9. Method according to any one of claims 2 to 8, in which the series of measurements relating to states of the mechanism corresponds to a vibration and / or a temperature.
    类似技术:
    公开号 | 公开日 | 专利标题
    FR3067154A1|2018-12-07|METHOD FOR TREND ANALYSIS AND AUTOMATIC ADJUSTMENT OF ALARM PARAMETERS
    CA2851124C|2020-11-03|Method for the preventive detection of failure in an apparatus, computer program, system and module for the preventive detection of failure in an apparatus
    US6796709B2|2004-09-28|Turbine blade | health monitoring and prognosis using infrared camera
    US20070067114A1|2007-03-22|System and method for monitoring degradation
    CA2528900C|2016-02-02|Process for monitoring the performance of industrial equipment
    CA2744977C|2018-04-24|Detection of anomalies in an aircraft engine
    EP2693176B1|2019-11-06|Method for detecting defects of a bearing by vibrational analysis
    JP2004150439A|2004-05-27|Method for determining failure phenomenon
    JP2017106908A|2017-06-15|System for automated in-process inspection of welds
    FR2939924A1|2010-06-18|IDENTIFICATION OF FAILURES IN AN AIRCRAFT ENGINE
    FR3019295A1|2015-10-02|METHOD FOR ESTIMATING THE NORMAL OR NON-MEASURED VALUE OF A PHYSICAL PARAMETER OF AN AIRCRAFT ENGINE
    FR2983529A1|2013-06-07|METHOD FOR MONITORING A DEVICE FOR CONTROLLING A FUEL TANK OF A TURBOJET ENGINE
    FR2983528A1|2013-06-07|Method for monitoring measuring equipment of turbojet of aircraft, involves measuring position of fuel metering device of turbojet, and standardizing degradation indicator depending on position of fuel metering device
    EP3215903B1|2019-12-18|Tool for validating a system for monitoring an aircraft engine
    EP2872956B1|2016-10-12|Method for detecting deterioration in a turbomachine by monitoring the performance of said turbomachine
    EP2269015B1|2017-05-31|Optronic infrared system with predictive maintenance following a sudden drift
    CA2918215A1|2015-01-29|Method of estimation on a curve of a relevant point for the detection of an anomaly of a motor and data processing system for the implementation thereof
    FR2957170A1|2011-09-09|Equipment monitoring system designing tool for engine of aircraft, involves processing unit evaluating quantification of monitoring system based on quality value of result corresponding to output quality value associated with output module
    JP2009176024A|2009-08-06|Production process abnormality-detecting method, production process abnormality-detecting system, program for making computer execute the production process abnormality-detecting method, and computer-readable recording medium to which the program is recorded
    CN109469896B|2020-06-09|Industrial boiler fault diagnosis method and system based on time series analysis
    FR2966928A1|2012-05-04|METHOD AND DEVICE FOR MONITORING A REDUNDANT CHAIN OF MEASUREMENT
    WO2020052147A1|2020-03-19|Monitoring device fault detection method and apparatus
    JP6879888B2|2021-06-02|Information processing equipment, information processing methods, and programs
    EP3789687A1|2021-03-10|Method for determining a state of a heating system
    US20210191372A1|2021-06-24|Analysis of additive manufacturing processes
    同族专利:
    公开号 | 公开日
    DE102017221168A1|2018-06-07|
    FR3059812B1|2021-01-08|
    FR3059812A1|2018-06-08|
    US20180158314A1|2018-06-07|
    CN108153211A|2018-06-12|
    US10115298B2|2018-10-30|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US6681633B2|2000-11-07|2004-01-27|Halliburton Energy Services, Inc.|Spectral power ratio method and system for detecting drill bit failure and signaling surface operator|
    US9606520B2|2009-06-22|2017-03-28|Johnson Controls Technology Company|Automated fault detection and diagnostics in a building management system|
    US8988238B2|2012-08-21|2015-03-24|General Electric Company|Change detection system using frequency analysis and method|
    CN104880247B|2015-05-04|2016-01-20|华中科技大学|A kind of group alarm method for rotating machinery on-line monitoring system|
    US10317875B2|2015-09-30|2019-06-11|Bj Services, Llc|Pump integrity detection, monitoring and alarm generation|
    CN105894703B|2016-05-11|2019-03-08|优乐（武汉）健康科技有限公司|A kind of intelligent monitoring method based on WiFi, apparatus and system|
    CN106066209B|2016-05-31|2018-09-28|郑州海威光电科技有限公司|Temperature alarming judgment method based on infrared measurement of temperature monitoring system|US10234855B2|2017-04-17|2019-03-19|Honeywell International Inc.|Apparatus and method for rationalizing and resolving alarms in industrial process control and automation systems|
    CN108986418A|2018-09-18|2018-12-11|广东电网有限责任公司|intelligent alarm method, device, equipment and storage medium|
    CN111127849B|2019-12-19|2021-04-09|浙江天禄环境科技有限公司|Accident early warning method comprehensively considering meteorological and chemical plant peripheral area|
    CN112185050B|2020-09-25|2022-03-04|珠海格力电器股份有限公司|Security level confirmation method and device and fire fighting system|
    法律状态:
    2018-11-30| PLFP| Fee payment|Year of fee payment: 2 |
    2020-05-13| PLFP| Fee payment|Year of fee payment: 3 |
    2020-11-26| PLFP| Fee payment|Year of fee payment: 4 |
    2021-08-06| PLSC| Publication of the preliminary search report|Effective date: 20210806 |
    2021-11-26| PLFP| Fee payment|Year of fee payment: 5 |
    优先权:
    申请号 | 申请日 | 专利标题
    US15367652|2016-12-02|
    US15/367,652|US10115298B2|2016-12-02|2016-12-02|Method of trend analysis and automatic tuning of alarm parameters|
    FR1761203A|FR3059812B1|2016-12-02|2017-11-27|METHOD OF TREND ANALYSIS AND AUTOMATIC ADJUSTMENT OF ALARM PARAMETERS|
    [返回顶部]