Distributed Industrial Performance Monitoring and Analysis

Abstract

A technique for providing early fault detection in a process plant using process control data generated by control equipment is provided. The technique determines leading indicators of conditions (eg, failures, abnormalities, or performance degradation) within the process plant. Leading indicators can be determined using principal component analysis. A process signal indicative of 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 that is utilized to monitor the process plant. When the presence of leading indicators is detected in the time series data, warnings or other predictions of the situation can be generated. Therefore, process failures can be identified using fluctuations and anomalies as leading predictors.

Description

[0001] Cross References to Related Applications [0002] This application claims priority to U.S. Provisional Application No. 32 / 239,620, filed October 9, 2015, entitled DISTRIBUTED INDUSTRIALPERFORMANCE MONITORING AND ANALYTICS, which is hereby expressly incorporated by reference in its entirety. This application is related to (i) U.S. Application No. 13 / 784,041, filed March 4, 2013, entitled "BIG DATA IN PROCESS CONTROL SYSTEMS," and (ii) filed February 6, 2014 U.S. Application No. 14 / 174,413, entitled "COLLECTING AND DELIVERING DATA TO A BIG DATAMACHINE IN A PROCESS CONTROL SYSTEM", (iii) filed on August 11, 2014, entitled "SECURING DEVICES TO PROCESS CONTROL SYSTEMS U.S. Application No. 14 / 456,763, (iv) filed March 17, 2014, and titled "DATA MODELING STUDIO," U.S. Application No. 14 / 216,823, (v) filed January 31, 2014 U.S. Application No. 14 / 169,965, entitled "MANAGING BIG DATA IN PROCESSCONTROL SYSTEMS," (vi) filed March 14, 2014, and entitled "DETERMINING ASSOCIATIONS AND...

AI Technical Summary

Problems solved by technology

[0010] Still further, the size of data sets for industrial or process control plants has steadily increased to the point where current data processing and analysis applications are insufficient

In general, known analysis techniques only attempt to extract values ​​from the data, but do not address the specific size of the data set from which the value will be extracted, and notably, very large data sets (such as those produced by factory all process data generated) to operate on

Furthermore, known analytics techniques often cannot handle streaming or streaming data

Images