Method and apparatus for online identification of safe operation and advance detection of unsafe operation of a system or process

• US 6,826,513 B1
• Filed: 08/13/2003
• Issued: 11/30/2004
• Est. Priority Date: 11/29/2002
• Status: Active Grant

• Current Assignee: Benhov GmbH, LLC (Intellectual Ventures LLC)
• Original Assignee: Council of Scientific & Industrial Research
• Inventors: Kumar, Viruthiamparambath Ravi, Ghosh, Anandamohan, Kulkarni, Bhaskar Dattatraya
• Primary Examiner(s): Hoff, Marc S.
• Assistant Examiner(s): Tsai, Carol S W
• Application Number:
  US10/639,475
• Time in Patent Office: 475 Days
• Field of Search: 702/32-35, 702/56, 702/57, 702/66-79, 702/81, 702/104, 702/105, 702/122, 702/124, 702/126, 702/179, 702/180, 702/182-185, 702/189, 702/190, 702/191, 702/194-198, 702/FOR-103, 702/FOR-107, 702/FOR.164, 702/FOR.166, 702/FOR.168, 702/FOR-176, 714/11, 714/12, 714/15, 714/798, 714/799, 714/819, 340/3.42, 340/3.43, 340/3.44, 700/9-12, 700/214, 700/79, 700/83
• US Class Current: 702/185
• CPC Class Codes: H04L 43/00 Arrangements for monitoring or testing data switching networks
• CPC Class Codes: H04L 43/045 for graphical visualisation of monitoring data

Abstract

A method and apparatus for online identification of safe operation and advance detection of unsafe operation of systems or processes from signals containing noise wherein the method and apparatus uses wavelet transformations to accurately infer from criteria about the changes in the monitored process variables and its acceleration as a basis for advance detection of unsafe operation, the said process of online identification of safe operation and advance detection of unsafe operation has applications in a wide variety of situations where digitized data could be made available.

Claims

1. A method for an online identification of safe operation and advance detection of unsafe operation of a system or a process in presence of noise in sensor measurements or fluctuations in variables measured, said method consisting the steps of:

(a) obtaining an online signal from one or more sensors monitoring/measuring process states or operation of the system at specified time intervals, wherein said online signal being in the form of a time series data relating to variation in one or more process variables;

(b) digitizing said online signal of step (a) to obtain a digitized data set;

(c) differentiating the digitized data set of step (b) to obtain a first derivative data set;

(d) taking wavelet transformation of the first derivative data set of step (c) to obtain scalogram in terms of wavelet coefficients;

(e) computing power distribution for individual wavelet scale of step (d) from said wavelet coefficients at all wavelet scales;

(f) computing from the power distribution of step (e) a critical wavelet scale at which finer scales are separable with respect to coarser scales, wherein said finer scales have wavelet coefficients attributable to noise and said coarser scales have wavelet coefficients attributable to true first derivative data (from step (d));

(g) setting the value of the wavelet scales as zero, when the wavelet coefficients corresponding to the finer scales are less than or equal to the critical wavelet scale of step (f);

or retaining the value of the wavelet scales, when the wavelet coefficients corresponding to the coarser scales are greater than the critical wavelet scale;

(h) computing inverse wavelet transform of the wavelet coefficients of step (g) at different scales to get de-noised first derivative data;

(i) repeating the steps (d) through (h) on de-noised first derivative data to obtain de-noised second derivative data;

(j) testing criteria for unsafe operation based on simultaneous positivity in time of the de-noised first derivative data and de-noised second derivative data, subject to the condition that the power in the coarser scales for the de-noised second derivative data is finite;

(k) detecting unsafe process operation and initiating necessary corrective measures when the criteria in step (j) are satisfied, and

(l) identifying safe process operation and repeating steps (a) to (j) when the criteria in step (j) are not satisfied.

2. A method as claimed in claim 1, wherein the system or process for obtaining time-series data is independent of the parameter values of operation.

3. A method as claimed in claim 1, wherein the system or process includes chemical reactors including batch, continuous stirred tank, fixed-bed, biochemical, polymerization, fluidized-bed, catalytic/noncatalytic, combustion and other physical multi-phase systems, flow systems with simple/complex hydrodynamic systems including turbulence and seismographic instruments.

4. The method as claimed in claim 1, wherein sensors in step (a) are located on apparatus/equipments taking part in the process.

5. The method as claimed in claim 1, wherein the online signal is obtained from multiple sensors monitoring/measuring the variables at different spatial positions simultaneously from said apparatus/equipment.

6. The method as claimed in claim 1, wherein the online signal is obtained from multiple sensors monitoring/measuring the variables at unequal sampling time from said apparatus/equipment.

7. A method as claimed in claim 1, wherein the online signal explicitly denotes the variables.

8. A method as claimed in claim 1, wherein the online signal implicitly denotes the variables.

9. A method as claimed in claim 1, wherein the online signal denotes re-scaled values of monitored process variables.

10. A method as claimed in claim 7, wherein the online signal implicitly denoting the variable includes data on spectroscopic absorption, potential difference, conductance, temperature, concentration, flow rates and viscosity.

11. A method as claimed in claim 1, wherein the wavelet transformation methods used include orthogonal wavelet transformation, bi-orthogonal wavelet transformation and continuous wavelet transformation.

12. A method as claimed in claim 11, wherein the wavelet types are selected from Daubechies, Haar, spline, Morlet, Battle-Lemarie, Meyer, Mexican-hat, Coiflet and Symlet.

13. A method as claimed in claim 1, wherein in step (b), the analog signal is converted digitized signal feeding the analog signal into a programmed device with a storage medium.

14. A method as claimed in claim 1, wherein in step (k), the necessary corrective measure includes switching ON an alarm device, providing feed back to corrective system.

15. A method as claimed in claim 1, wherein digitized data are converted to analog signals using a digital to analog converter for purposes of sounding alarms and activating corrective systems.

16. An apparatus for online identification of safe operation and advance detection of unsafe operation of a system or a process in presence of noise in sensor measurements or fluctuations in variables measured, said apparatus consisting:

(a) one or more sensors monitoring/measuring process states or operation of the system at specified time intervals for providing an online signal, wherein said online signal being in the form of a time series data relating to variation in one or more process variables;

(b) a digitizing means connected to the sensors for digitizing said online signal of step (a);

(c) a differentiating means coupled to said digitizing means for differentiating the digitized data set of and obtain a first derivative data set;

(d) a first computing means programmed to receive the first derivative data set and compute wavelet transform of the first derivative data set to obtain scalogram in terms of wavelet coefficients;

(e) a second computing means programmed to compute power distribution for individual wavelet scale of step (d) from said wavelet coefficients at all wavelet scales;

(f) a third computing means configured to receive the power distribution from the second computing means and programmed to compute a critical wavelet scale at which finer scales are separable with respect to coarser scales, wherein said finer scales have wavelet coefficients attributable to noise and said coarser scales have wavelet coefficients attributable to true first derivative data;

(g) a means for setting the value of the wavelet scales as zero, when the wavelet coefficients corresponding to the finer scales are less than or equal to the critical wavelet scale of step (f);

or retaining the value of the wavelet scales, when the wavelet coefficients corresponding to the coarser scales are greater than the critical wavelet scale;

(h) a means for computing inverse wavelet transform of the wavelet coefficients of step-(g) at different scales to get de-noised first derivative data;

(i) a testing means for testing criteria for unsafe operation based on simultaneous positivity in time of the de-noised first derivative data and de-noised second derivative data, subject to the condition that the power in the coarser scales for the de-noised second derivative data is finite;

(j) an identifying means for identifying the operation as unsafe operation/safe operation, said identifying means being connected to the testing means at its input end and being connected to an alarm/error correcting system at its out put end.

17. An apparatus as claimed in claim 16, wherein the system or process from which time-series data is obtained is independent of the parameter values of operation.

18. A, apparatus as claimed in claim 16, wherein the system or process from which time-series data is obtained includes chemical reactors including batch, continuous stirred tank, fixed-bed, biochemical, polymerization, fluidized-bed, catalytic/noncatalytic, combustion and other physical multi-phase systems, flow systems with simple/complex hydrodynamic systems including turbulence and seismographic instruments.

19. An apparatus as claimed in claim 16, wherein sensors in step (a) are located on system/equipments taking part in the process.

20. An apparatus as claimed in claim 16, wherein the online signal is obtained from multiple sensors monitoring/measuring the variables at different spatial positions simultaneously from said system/equipments taking part in the process.

21. An apparatus as claimed in claim 16, wherein the online signal is obtained from multiple sensors monitoring/measuring the variables at unequal sampling time from said system/equipments taking part in the process.

22. An apparatus as claimed in claim 16, wherein the online signal explicitly denotes the variables.

23. An apparatus as claimed in claim 16, wherein the online signal implicitly denotes the variables.

24. An apparatus as claimed in claim 16, wherein the online signal denotes re-scaled values of monitored variables.

25. An apparatus as claimed in claim 23, wherein the online signal implicitly denoting the variable includes data on spectroscopic absorption, potential difference, conductance, temperature, concentration, flow rates and viscosity.

26. An apparatus as claimed in claim 16, wherein the means for computing wavelet transformation include orthogonal wavelet transformation means, bi-orthogonal wavelet transformation means and continuous wavelet transformation means.

27. An apparatus as claimed in claim 26, wherein the wavelet types are selected from Daubechies, Haar, spline, Morlet, Battle-Lemarie, Meyer, Mexican-hat, Coiflet and Symlet.

28. An apparatus as claimed in claim 16, wherein digitized data are converted to analog signals using a digital to analog converter for purposes of sounding alarms and activating corrective systems.