STATE MONITORING OF MACHINES AND TECHNICAL INSTALLATIONS

• US 20080052040A1
• Filed: 08/23/2007
• Published: 02/28/2008
• Est. Priority Date: 08/26/2006
• Status: Active Grant

• Current Assignee: Peter Renner
• Original Assignee: Peter Renner
• Inventors: Renner, Peter
• Granted Patent: US 7,774,165 B2
• Time in Patent Office: Days
• US Class Current: 702/182
• CPC Class Codes: G05B 19/406 characterised by monitoring or safety G05B19/19 takes precedence
• CPC Class Codes: G05B 21/02 electric
• CPC Class Codes: G05B 2219/33303 Expert system for diagnostic, monitoring use of tree and probability
• CPC Class Codes: G05B 2219/34477 Fault prediction, analyzing signal trends
• CPC Class Codes: G05B 23/0235 based on a comparison with predetermined threshold or range, e.g. "classical methods", carried out during normal operation; threshold adaptation or choice; when or how to compare with the threshold
• CPC Class Codes: G07C 3/08 Registering or indicating the production of the machine either with or without registering working or idle time

Abstract

In order further to improve the monitoring of technical processes, in particular still more precisely to differentiate the influences on monitoring which are caused by depreciation or wear from the external influences, there is described a method in which measurement data are detected by way of measurement channels and the measurement data of a measurement channel are processed in a plurality of time series of state values with mutually different time bases and wherein limit value ranges are ascertained in a learning phase and monitoring-related steps are triggered in a monitoring phase when limit values are exceeded. The method according to the invention is distinguished in that in the monitoring phase a good value extent in respect of the state values is calculated for a time series with an upper and a lower limit value, with the inclusion of a limit value range which is predetermined in the learning phase and a state base value of the time series, which is determined in the monitoring phase.

The invention further concerns an expert system for the implementation of such a method.

Claims

1. A method of monitoring technical processes in machines, installations or the like, wherein measurement data are detected by way of at least one measurement channel and the measurement data of a measurement channel are processed in a plurality of time series of state values with mutually different time bases and wherein limit value ranges are ascertained in a learning phase and monitoring-related steps are triggered in a monitoring phase when limit values are exceeded, characterised in that in the monitoring phase a good value extent in respect of the state values (ZW) is calculated for a time series with an upper and a lower limit value, with the inclusion of a limit value range which is predetermined in the learning phase and a state base value (ZB) of the time series, which is determined in the monitoring phase.

2. A method as set forth in claim 1 characterised in that the upper and the lower limit value of the time series is calculated by adding or subtracting half the limit value range (GS) to or from the state base value (ZB) of the time series respectively.

3. A method as set forth in claim 1 or claim 2 characterised in that a time series includes a predetermined number of values (test block) and the state base value (ZB) of the time series is calculated in the monitoring phase by data processing of at least two currently prevailing good values of the time series, in particular by averaging of the Max value (MAX) and the Min value (MIN) of the time series or by averaging of all good values of the time series.

4. A method as set forth in claim 1, claim 2 or claim 3 characterised in that the state values (ZW) of a time series are standardised on to the state base value (ZB).

5. A method as set forth in claim 4 characterised in that the standardised state values of a plurality of time series with the same time base are represented in a common standard value graph (80, 81, 82).

6. A method as set forth in one of claims 1 through 5 characterised in that a predetermined good value range (GWS) of a time series is calculated from the limit value range (GS) ascertained in the learning phase and an offset (OS).

7. A method as set forth in claim 6 characterised in that the limit value range (GS) of a time series is established by the range of fluctuation in the state values of the time series in the learning phase.

8. A method as set forth in one of claims 1 through 7 characterised in that in relation to a measurement channel in the learning phase a state value (ZW) of the time series with the lowest time base is continuously ascertained and incorporated into the time series and/or in the monitoring phase a good value of the time series with the lowest time base is continuously ascertained and incorporated into the time series.

9. A method as set forth in one of claims 1 through 8 characterised in that in relation to a measurement channel in the learning phase a state value (ZW) of a time series with a predetermined time base is ascertained by statistical evaluation, in particular by averaging, of state values of the time series with a lower time base and is incorporated into the first-mentioned time series and/or in the monitoring phase a good value of a time series with a predetermined time base is ascertained by statistical evaluation, in particular by averaging, of good values of the time series with a lower time base and is incorporated into the first-mentioned time series.

10. A method as set forth in one of claims 8 and 9 characterised in that when classifying the freshly ascertained value in the respective time series the oldest value is removed from the time series when a number of classified values which is predetermined for the time series has been reached.

11. A method as set forth in one of claims 9 and 10 characterised in that it is only after the presence of a predetermined minimum number of good values in the monitoring phase or state values in the learning phase within the time interval of a time series that a valid good value or a state value for the next higher time series is calculated.

12. A method as set forth in one of claims 8 through 11 characterised in that in the monitoring phase upon the occurrence of a state value in a time series with a predetermined time base which is outside the good value extent a monitoring-related step and in particular an observation indication for the time series in question is outputted.

13. A method as set forth in claim 12 characterised in that a parameter which is dependent on the configuration in respect of time of the time series, in particular a time derivative of the curve configuration, is detected, and a predetermined monitoring-related step and in particular an alarm indication for the time series in question is outputted in dependence on said parameter.

14. A method as set forth in one of claims 8 through 13 characterised in that in the monitoring phase a state value of a time series, which is outside the good value extent, is not used for ascertaining a state value of the time series with a higher time base.

15. A method as set forth in claim 12 characterised in that after the occurrence of a state value in the time series, which is outside the good value extent, the subsequently ascertained state values of a time series are rejected as invalid until a state value is in a value interval which is established solely by the limit value range and the state base value of the time series.

16. A method as set forth in one of claims 1 through 15 characterised in that for a time series there is established a minimum state range which replaces the limit value range (GS) ascertained in the learning phase when the latter is less than the minimum state range.

17. A method as set forth in one of claims 1 through 15 characterised in that state values of a hybrid channel are formed by forming the difference of measurement data from physically correlated measurement channels.

18. A method as set forth in one of claims 1 through 15 characterised in that state values of a hybrid channel are formed by forming the sum of measurement data from physically correlated measurement channels.

19. A method as set forth in one of claims 17 and 18 characterised in that the state values of a hybrid channel are processed in a plurality of time series of state values with mutually different time bases.

20. A method as set forth in one of claims 1 through 18 characterised in that at least two items of information are associated with each state value of a time series, in particular an item of information about the validity of the state value and an item of information as to whether the state value is within the good value extent of the respective time series.

23. A computer program product which is stored on a storage medium and which can be loaded into a memory of a computer and which includes software code portions with which a method as set forth in one of claims 1 through 18 is carried out when the product runs on the computer.

21. An expert system for monitoring technical processes comprising at least one measurement channel which has sensor, line and evaluation means for detecting measurement data, data processing means for processing the measurement data and for ascertaining a plurality of time series of state values with different time bases from the measurement data, storage means for storing the time series of state data, wherein in a learning phase limit value ranges of the time series can be calculated and stored, and a means for carrying out a monitoring-related step, in particular a display means for displaying that a state value has exceeded a limit value in a monitoring phase, characterised in that the data processing means (46-50) in the monitoring phase for a time series calculates a good value extent with an upper and a lower limit value with the incorporation of a limit value range (GS) which is predetermined in the learning phase and a state base value (ZB) of the time series, which is determined in the monitoring phase.

22. An expert system as set forth in claim 21 characterised in that state values are provided with a time mark.