System and method for training region selection for deep learning models

A system for training region selection for deep learning model comprising:

• a processor;
• and
• a memory coupled to the processor, wherein the memory comprises a processing subsystem configured to be executed by the processor, wherein the processing subsystem is hosted on a server and configured to execute on a network to control bidirectional communications among a plurality of modules comprising:
• an identification module configured to:
• identify a shutdown of at least one equipment by analyzing operational data associated with the at least one equipment, wherein the operational data comprises sensor readings, performance metrics and status indicators collected during real-time operation of the at least one equipment;
• identify a buffer region around the detected shutdown, wherein the buffer region comprises a predefined period immediately preceding and immediately following the detected shutdown;
• and
• obtain a first data from the operational data by excluding a data corresponding to the buffer region;
• a segmentation module operatively coupled to the identification module, and configured to:
• determine an optimal number of clusters for the first data using an elbow method, and partition optimal number of clusters into a predefined operating regions using K-means clustering technique;
• a feature analysis module operatively coupled to the segmentation module, and configured to:
• select, for each operating region, a predetermined percentage of data points closest to a centroid of the each operating region;
• compute corresponding median values of tags associated with the selected data points;
• identify top-contributing tags for the each operating region by sorting the tags based on corresponding absolute values of the corresponding computed median values;
• and
• identify low and high values of the top-contributing tags within the each operating region based on the corresponding absolute values;
• a domain comparison module operatively coupled to the feature analysis module, and configured to compare the top-contributing tags of the each operating region with corresponding domain knowledge, and exclude one or more of the operating regions whose top-contributing tags are associated with mapped failures based on the corresponding domain knowledge from the predefined operating regions to obtain remaining operating regions;
• a data cleaning module operatively coupled to the domain comparison module, and configured to determine a mean and standard deviation within the remaining operating regions upon eliminating erroneous data by recalculating statistical measures, and eliminate outlier data points from the remaining operating regions by applying statistical outlier detection methods to obtain cleansed operating regions;
• and
• a filtering module operatively coupled to the data cleaning module, and configured to apply a filter over the cleansed operating regions to refine the cleansed operating regions further and consolidate operating regions among the cleansed operating regions that are within a predetermined distance to reduce a number of training regions.