Industrial Bottleneck Detection and Management Method and System
AssigneeThroughput, Inc.
InventorsAli Hussan RAZA, Bhaskar Subbarao BALLAPRAGADA, Seth Rom PAGE, Anzar KAMDAR
Life of this patent
Estimating the term…
Abstract
The present invention includes: (a) a method for improving data to be processed for bottleneck detection, by cleaning corrupt or outlier data; (b) a method for improved analysis of bottleneck data using a plurality of rules for categorization; and (c) a method for improved display and/or allowing improved user feedback for bottleneck data using multivariate analysis and display. These methods can be used alone, or preferably be combined in whole or in part together to improve performance of an industrial process. A system is also provided.
Claims14
A method for improving data to be processed by a bottleneck detection system comprising the steps of:
- receiving in a processor a plurality of data items related to an industrial process, each data item being time stamped so that each data item includes time stamp and industrial process data regarding an industrial process occurring at the time;
- identifying in the processor corrupt or outlier industrial process data or corrupt time stamp data of the data items;
- providing the plurality of data items excluding data items having the identified corrupt or outlier industrial process data or corrupt time stamp data to permit performing in the processor at least one statistical calculation;
- discarding data items with corrupt industrial process data or time stamp data;
- and
- providing the plurality of data items including items with the outlier industrial process data but without the discarded data items to be able to be used to identify abnormal patterns or behaviors.
The method as recited in claim 3 wherein the identifying step includes first identifying a type of data distribution being received and comparing a cumulative density function to a cutoff to identify outliers.
The method as recited in claim 4 wherein the outlier industrial process data is identified when the cumulative density function is beyond a cutoff of mean plus 3 standard deviations.
A method for improved analysis of bottleneck data comprising:
- receiving in a processor a plurality of data items related to an industrial process, each data item being time stamped so that each data item includes time stamp and industrial process data regarding an industrial process occurring at the time, the plurality of data items preferably being the data items used for the statistical calculation discussed above;
- analyzing the plurality of data items in a processor via a plurality of rules, the analyzing identifying at least type of pattern shift to create pattern shift data;
- and
- performing a multivariate analysis on both the plurality of data items and the pattern shift data to aid in identifying causes of a bottleneck.
A method for improved display and/or allowing improved user feedback for bottleneck data, comprising:
- receiving in a processor a plurality of data items related to an industrial process, each data item being time stamped so that each data item includes time stamp and industrial process data regarding an industrial process occurring at the time;
- performing a multivariate analysis on the plurality of data items in order to identify a bottleneck and contributing factors in the industrial process;
- and
- displaying the multivariate analysis including so that data showing the bottleneck and data showing little or no bottleneck are displayed together.
Dependencies are inferred from the claim text.