Monitoring for invalid data from field instruments

• US 9,009,001 B2
• Filed: 05/26/2011
• Issued: 04/14/2015
• Est. Priority Date: 05/26/2011
• Status: Active Grant

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Vass, Jiri, Van Schie, Ard
• Primary Examiner(s): Lau, Tung S
• Application Number:
  US13/116,364
• Publication Number:
  US 20120303321A1
• Time in Patent Office: 1,419 Days
• Field of Search: 702/179
• US Class Current: 702/179
• CPC Class Codes: G05B 23/0221 Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods
• CPC Class Codes: G05B 9/02 electric

Abstract

Processing systems and methods used to identify faults in the operation of one or more field instruments are described and shown. In one embodiment, a sensing system includes one or more field instruments and a processing component configured to process data from the field instruments. This processing may include identifying invalid samples of data using an algorithm; correlating invalid samples of data to specific field instruments; and determining a likelihood of a fault occurring on the specific field instruments. The processing may occur in real-time using an online processing technique, or with an offline processing technique on data maintained in a data store.

Claims

1. A sensing system comprising:

a field instrument; and

a processing component operably coupled to the field instrument and configured to identify faults in operation of the field instrument, by;

processing data collected from the field instrument to produce a plurality of samples;

identifying a fault-indicating sample of the data using an algorithm, the algorithm configured to identify the fault-indicating sample of the data from the plurality of samples, wherein operations to perform the algorithm upon the plurality of samples include operations comprising;

performing a filtering operation on the data from the plurality of samples to generate filtered data samples for the field instrument, wherein the filtered data samples represent actual, non-simulated data measurements collected exclusively from the field instrument;

computing a mean value M and standard deviation S from the filtered data samples for the field instrument;

computing an upper control limit and lower control limit using the mean value M and standard deviation S; and

identifying the fault-indicating sample from the filtered data samples for the field instrument, the fault-indicating sample providing one or more values greater than the upper control limit or lower than the lower control limit and satisfying a set of one or more empirical rules, wherein the set of empirical rules includes a rule requiring a difference between two or more successive samples of the data measurements collected exclusively from the field instrument to exceed a threshold value; and

determining a numerical likelihood of a fault occurring in the field instrument from a combination of;

a previously determined numerical likelihood of a fault for the field instrument obtained from analysis of other data collected from the field instrument at a previous time, with a current numerical likelihood of a fault determined from processing the one or more values indicated by the fault-indicating sample of the data.

2. The system of claim 1, further comprising:

a plurality of additional field instruments;

wherein the processing component is further configured to identify faults in operation of the field instrument or the plurality of additional field instruments by;

processing the data collected from the field instrument and the plurality of additional field instruments using an online process in real time; and

correlating the fault-indicating sample of the data to one or more specific field instruments.

3. The system of claim 1, wherein the field instrument is one of:

a product level gauge, temperature probe, water level probe, pressure transmitter, or flowmeter.

4. The system of claim 1, wherein the data measured by the field instrument includes one or more of:

product level, product temperature, product pressure, product flow, vapor temperature, vapor pressure, product density, water level, or weather data.

5. The system of claim 1, wherein the field instrument comprises an array of field instrument elements, and wherein the processing component is further configured for identifying a detected change and the likelihood of fault for a specific element in the array of field instrument elements by factoring one or more measurements obtained from field instruments external to the array of field instrument elements.

6. The system of claim 1, wherein the filtering operation is performed by one or more filters, the filters including at least one of:

a low-pass frequency filter, a high-pass frequency filter, a band-pass frequency filter, a band-stop frequency filter, a median-based non-linear filter, a percentile-based non-linear filter, a weighted averaging filter, a wavelet filter, a frequency-domain thresholding filter, a wavelet-domain thresholding filter, a cepstrum-domain thresholding filter, an adaptive linear filter, and an adaptive non-linear filter.

7. The system of claim 1, the algorithm further comprising computing one or more statistical indicators from the data and the filtered data samples;

wherein the set of empirical rules is applied to the data, the filtered data samples and the one or more statistical indicators; and

wherein the set of empirical rules includes a rule requiring a difference between two or more successive samples of the filtered data samples to exceed a threshold value, a rule requiring one or more statistical indicators of the data to exceed a corresponding threshold value, and a rule requiring one or more statistical indicators of the filtered data samples to exceed a corresponding threshold value; and

wherein a threshold value is one of;

a fixed threshold, a fuzzy threshold, a time-varying threshold, and an adaptive threshold.

8. The system of claim 1, the processing component further configured to:

report incidences of the fault-indicating sample of the data to a user; and

notify a user after the likelihood of a fault exceeds a predefined threshold.

9. The system of claim 1, wherein determining a likelihood of a fault occurring in the field instrument includes factoring one or more of:

a previously determined likelihood of a fault for the field instrument, or

data collected from other field instruments located proximate to a location of the field instrument.

10. A processing system configured to process data collected from a plurality of field instruments, the processing system comprising:

a processor;

memory operably coupled to the processor;

instructions for execution within the processing system with use of the memory and the processor, the instructions configured to;

process data originating from a plurality of field instruments to produce a plurality of samples;

identify multiple fault-indicating samples of the data using an algorithm applied to individual data sets, the individual data sets produced from actual, non-simulated data measurements at each respective field instrument of the plurality of field instruments, the algorithm configured to identify the fault-indicating samples of the data within the plurality of samples of the data, by detecting data changes among the plurality of samples from the data that are exceeding a maximum limit within a period of time;

correlate the multiple fault-indicating samples of the data to one or more specific field instruments in the plurality of field instruments; and

determine a numerical likelihood of a fault occurring in each of the one or more specific field instruments based on a combination of;

values indicated by the multiple fault-indicating samples of the data,

a status of the specific field instruments, and

data collected from other instruments in the plurality of field instruments that are located proximate to the specific field instruments.

11. The processing system of claim 10, wherein the algorithm as applied to instrument data for a particular field instrument comprises:

performing a filtering operation on the instrument data to generate filtered data samples for the particular field instrument, wherein the filtered data samples represent actual, non-simulated data measurements collected exclusively from the particular field instrument;

computing a mean value M and standard deviation S from the filtered data samples for the field instrument;

computing an upper control limit and lower control limit using the mean value M and standard deviation S;

computing one or more statistical indicators from the instrument data and the filtered data samples; and

identifying the fault-indicating sample from the filtered data samples for the field instrument, the fault-indicating sample having values greater than the upper control limit or lower than the lower control limit and satisfying a set of empirical rules applied to the instrument data, the filtered data samples, and one or more statistical indicators;

wherein the filtering operation is performed by one or more filters, the filters including at least one of;

a low-pass frequency filter, a high-pass frequency filter, a band-pass frequency filter, a band-stop frequency filter, a median-based non-linear filter, a percentile-based non-linear filter, a weighted averaging filter, a wavelet filter, a frequency-domain thresholding filter, a wavelet-domain thresholding filter, a cepstrum-domain thresholding filter, an adaptive linear filter, or an adaptive non-linear filter; and

wherein the set of empirical rules includes a rule requiring a difference between two or more successive samples of the instrument data to be greater than a deadband parameter D, a rule requiring a difference between two or more successive samples of the filtered data samples to be greater than a deadband parameter, a rule requiring one or more statistical indicators of the instrument data to be greater than a corresponding maximum threshold, and a rule requiring one or more statistical indicators of the filtered data samples to be greater than a corresponding maximum threshold; and

wherein the threshold value is one of;

a fixed threshold, a fuzzy threshold, a time-varying threshold, or an adaptive threshold.

12. The processing system of claim 10, wherein determining a likelihood of a fault occurring in each of the one or more specific field instruments includes factoring one or more of:

a previously determined fault likelihood for the specific field instruments;

or

a location of the specific field instruments in relation to other instruments in plurality of field instruments.

13. The processing system of claim 10, wherein at least one of the specific field instruments comprises an array of field instrument elements, and wherein the processing component is further configured for identifying a detected change and the likelihood of fault for a specific element in the array of field instrument elements by factoring one or more measurements obtained from field instruments external to the array of field instrument elements.