Sunday, March 31, 2013

Big Data requires Pattern Awareness to Provide Situational Awareness


2012 saw the rise in what I call “Industrial Information Systems, Projects”. You may say “rubbish”, the whole historian, and information business has been around for years, and the answer is true. Today these projects are different dealing with a “lake of data”, delivering to more people, of different roles typically not even aware of what an historian is drawing data from many different sources including historians, xml files, transactional data sources such as MES and Batch systems, alarm, event systems, MS Excel and customer odd databases, as well real-time data. There is no one supplier, one source, or structure to this data. The challenge is when the context and knowledge of the data is retiring from the companies, but the size of internal community of roles and workers requiring access is increasing. How often I have been asked and discussed the issue of data validation and data awareness, vs architecture, and technologies at tossed into conversation hoping for a “silver bullet”, but I believe the solution comes with new capabilities like “Big Data” but also evolutions in existing industrial implementations, with a more holistic design!

I believe the growth will accelerate in “Industrial Information Systems, Projects” during 2013, and beyond, but this is not about delivering reports and information, it is about “empowering” the increased community in  business in making real-time decisions, based on real-time trustworthy, effective industrial information, no matter their location. I continue to get surprised by the notion that the solution is an Enterprise Historian on top of the existing system, acting as a data warehouse. I ‘Scratch my head” and usually “ask how to you know the data is valid, in context, and comparable. Too often it is a blank look they were lead to believe the data in or supplied from their SCADA, lower level historians, etc. is the only thing they need to access. Key to understanding is the ability to detect patterns, across data, but there needs to be enough context to allow the evolving big data tools to enable detection of patterns.
 
Last week this blog discussed exception based “self aware” models required in today’s proactive operational/ supervisory systems, especially as devices grow in intelligence capability. This model is also key to putting things in context enough to enable this analysis and patterns to be seen way further than process analysis. Big data concepts of pattern analysis, save that pattern, and now have it as auto detect on a similar pattern happening again, triggering an operational process that will continue a proven procedure to resolution, guiding the workers involved interacting in a consistent and pro active manner with the objective for early detection and fast resolution. This automatic pattern recognition, detection, and embedded procedure are one key aspect of the modern situational awareness concept. Building on last week’s blog concepts of the “self aware” model, if these smart devices and processes include a pattern recognition capability as part of the “self” intelligence, the shift is a response from the “as is” status to the ‘to be”. The diagram below shows a significant opportunity for improvement with the two “value of early corrective action” lines  effectively illustrating the value gain in early detection and pro active correct action.

The diagram illustrates how a condition over time “x axis” changes in cost/ value through time, and how the traditional alarm systems are in the “as is” state, and the whole objective to “SITUATIONAL AWARENESS” is to shift to the “to be “ state.

No comments:

Post a Comment