DISTRIBUTED INDUSTRIAL PERFORMANCE MONITORING AND ANALYTICS

• US 20170102696A1
• Filed: 09/23/2016
• Published: 04/13/2017
• Est. Priority Date: 03/04/2013
• Status: Active Grant

• Current Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Original Assignee: Fisher-Rosemount Systems Incorporated (Emerson Electric Company)
• Inventors: Bell, Noel Howard, Nixon, Mark John, Enver, Alper Turhan, Kidd, Joshua Brian, Muston, Paul R.
• Granted Patent: US 10,649,449 B2
• Time in Patent Office: Days
• CPC Class Codes: G05B 2219/31356 Automatic fault detection and isolation
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
• CPC Class Codes: G05B 23/0275 Fault isolation and identification, e.g. classify fault; estimate cause or root of failure
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
• CPC Class Codes: Y02P 90/80 Management or planning

Abstract

A technique is provided for providing early fault detection using process control data generated by control devices in a process plant. The technique determines a leading indicator of a condition within the process plant, such as a fault, abnormality, or decrease in performance. The leading indicator may be determined using principal component analysis. A process signal indicating a process variable corresponding to the leading indicator is then obtained and analyzed. A rolling fast Fourier transform (FFT) may be performed on the process signal to generate time-series data with which to monitor the process plant. When the presence of the leading indicator is detected in the time-series data, an alert or other prediction of the condition may be generated. Thus, process faults may be identified using fluctuations and abnormalities as leading predictors.

Claims

1. A method for providing early fault detection in a process plant, the method comprising:

obtaining a process signal comprising time domain data generated as a result of one or more process control devices operating to control a process in the process plant;

determining, based on the obtained process signal, a leading indicator of a condition of the process plant;

defining a process variable corresponding to the leading indicator;

performing a rolling fast Fourier transform (FFT) on a subsequent process signal to determine time-series data corresponding to the defined process variable;

monitoring the time-series data corresponding to the defined process variable for a presence of the leading indicator; and

generating an indication corresponding to a prediction of the condition based on a detected presence of the leading indicator in the monitored time-series data.

2. The method of claim 1, wherein obtaining the process signal comprising the time domain data comprises obtaining time domain data generated by a data source included in the process plant.

3. The method of claim 2, wherein obtaining the process signal comprising the time domain data comprises obtaining on-line time domain data generated by the one or more process control devices.

4. The method of claim 1, wherein determining the leading indicator of the condition comprises performing one or more statistical analytic techniques on the process signal, the one or more statistical analytic techniques include at least one of principal component analysis, partial least squares regression, clustering, or cross-correlation.

5. The method of claim 1, wherein defining the process variable corresponding to the leading indicator comprises defining a process variable corresponding to a change in the process signal that occurs prior to an occurrence of the condition and that contributes to the occurrence of the condition.

6. The method of claim 1, wherein defining the process variable comprises causing a tag to be assigned to the new process variable and storing the tag in a process control database of the process plant.

7. The method of claim 1, further comprising obtaining the subsequent signal by obtaining on-line process data generated by a data source within the process plant, the subsequent signal comprising additional time domain data generated as a result of the one or more process control devices operating to control the process in the process plant.

8. The method of claim 1, wherein performing the rolling FFT on the subsequent process signal comprises performing the rolling FFT on time domain data included in the subsequent process signal, thereby generating frequency domain data.

9. The method of claim 8, further comprising determining values of the defined process variable from the generated frequency domain data, and transforming the determined values into time domain data that includes the time-series data corresponding to the defined process variable.

10. The method of claim 8, further comprising:

performing one or more analytics techniques on the generated frequency domain data;

defining another process variable corresponding to the leading indicator based on a result of the one or more analytics techniques performed on the generated frequency domain data; and

monitoring time-series data corresponding to the another defined process variable for the presence of the leading indicator.

11. The method of claim 1, further comprising detecting the presence of the leading indicator in the monitored time-series data corresponding to the defined process variable based on a threshold corresponding to the values of the monitored time-series data, wherein the threshold corresponding to the values of the monitored time-series data comprises a threshold corresponding to at least one of a particular value or a duration of the particular value.

12. The method of claim 1, further comprising generating an indication of a time frame during which the condition is predicted to occur based on the detected presence of the leading indicator.

13. The method of claim 1, further comprising:

generating a process signal based on the detected presence of the leading indicator; and

providing the process signal to the process plant to change a behavior of the process plant to prevent or mitigate the effects of an occurrence of the predicted condition by changing at least one of a parameter, a value, a configuration, a state, or an executing application within the process plant.

14. The method of claim 1, further comprising generating a data analytics signal based on the detected presence of the leading indicator, and providing the data analytics signal to at least one of a distributed data engine or an analytics function of a data analytics system operating in conjunction with the process plant.

15. The method of claim 1, wherein at least a portion of the method is performed by a data analytics system operating in conjunction with the process plant.

16. The method of claim 15, wherein the at least the portion of the method is performed by one or more distributed data engines (DDEs) of the data analytics system.

17. The method of claim 15, wherein the instructions cause the processor to monitor the time-series data corresponding to the defined process variable for the presence of the leading indicator by a routine executing within the process control system of the process plant.

18. The method of claim 1, wherein monitoring the time-series data corresponding to the defined process variable comprises monitoring streaming process data from the process plant.

19. The method of claim 1, wherein obtaining the process signal comprising the time domain data generated as the result of the one or more process control devices operating to control the process comprises obtaining the process signal comprising time domain data generated by one or more data sources within the process plant, the one or more data sources excluding any data source within the process plant that generates first-order process data.

20. A system for providing early fault detection in a process plant, the system comprising:

a plurality of process control devices operating in a process plant;

a processor;

a program memory, communicatively coupled to the processor, and comprising a tangible, non-transitory storage medium storing instructions that, when executed by the processor, cause the processor to;

obtain a process signal comprising time domain data generated as a result of the plurality of process control devices operating in the process plant;

determine a leading indicator of a condition of the process plant based on the obtained process signal;

define a process variable corresponding to the leading indicator;

perform a rolling fast Fourier transform (FFT) on a subsequent process signal to determine time-series data corresponding to the defined process variable;

monitor the time-series data corresponding to the defined process variable for a presence of the leading indicator; and

generate an indication corresponding to a prediction of the condition based on a detected presence of the leading indicator in the monitored time-series data.

21. The system of claim 20, wherein the instructions that cause the processor to obtain the process signal comprising the time domain data include instructions that cause the processor to obtain time domain data generated by the one or more process control devices.

22. The system of claim 20, wherein the instructions that cause the processor to perform the rolling FFT on the subsequent process signal include instructions that cause the processor to perform the rolling FFT on time domain data included in the subsequent process signal, thereby generating frequency domain data, and the instructions further cause the processor to:

perform one or more analytics techniques on the generated frequency domain data;

define another process variable corresponding to the leading indicator based on a result of the one or more analytics techniques performed on the generated frequency domain data; and

monitor time-series data corresponding to the another defined process variable for the presence of the leading indicator.

23. The system of claim 20, wherein the instructions further cause the processor to generate an indication of a time frame during which the condition is predicted to occur based on the detected presence of the leading indicator.

24. The system of claim 20, wherein the instructions further cause the processor to generate a process signal based on the detected presence of the leading indicator and provide the process signal to the process plant to change the behavior of the process plant by changing at least one of a parameter, a value, a configuration, a state, or an executing application within the process plant.

25. The system of claim 20, wherein the processor is disposed within a data analytics system operating in conjunction with the process plant, the data analytics system including one or more distributed data engines (DDEs), and wherein the instructions are part of one or more data analytics functions executing in the data analytics system.

26. The system of claim 25, wherein the rolling FFT is performed by at least one of a data block or a data module of the data analytics system.

27. The system of claim 25, wherein the data analytics system include a data block definition library storing a data block definition of the rolling FFT.

28. A tangible, non-transitory computer-readable medium storing instructions for providing early fault detection in a process plant that, when executed by a processor of a computer system, cause the computer system to:

obtain a process signal comprising time domain data generated as a result of one or more process control devices operating to control a process in the process plant;

determine a leading indicator of a condition of the process plant based on the obtained process signal;

define a process variable corresponding to the leading indicator;

perform a rolling fast Fourier transform (FFT) on a subsequent process signal to determine time-series data corresponding to the defined process variable;

monitor the time-series data corresponding to the defined process variable for a presence of the leading indicator; and

generate an indication corresponding to a prediction of the condition based on a detected presence of the leading indicator in the monitored time-series data.

29. The tangible, non-transitory computer-readable medium of claim 28, wherein the instructions that cause the computer system to obtain the process signal include instructions that cause the computer system to the time domain data comprises obtaining on-line time domain data generated by the one or more process control devices.

30. The tangible, non-transitory computer-readable medium of claim 28, wherein the instructions that cause the computer system to perform the rolling FFT on the subsequent process signal include instructions that cause the computer system to perform the rolling FFT on time domain data included in the subsequent process signal, thereby generating frequency domain data, and further storing instructions that cause the computer system to:

perform one or more analytics techniques on the generated frequency domain data;

define another process variable corresponding to the leading indicator based on a result of the one or more analytics techniques performed on the generated frequency domain data; and

monitor time-series data corresponding to the another defined process variable for the presence of the leading indicator.

31. The tangible, non-transitory computer-readable medium of claim 28, further storing instructions that cause the computer system to generate an indication of a time frame during which the condition is predicted to occur based on the detected presence of the leading indicator.

32. The tangible, non-transitory computer-readable medium of claim 28, further storing instructions that cause the computer system to generate a process signal based on the detected presence of the leading indicator and provide the process signal to the process plant to change the behavior of the process plant by changing at least one of a parameter, a value, a configuration, a state, or an executing application within the process plant.