Monitoring for invalid data from field instruments

摘要

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.

主权项

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; andidentifying 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.