Here we are in the world of Big Data and all of its possibilities. Just look at all the data we have available to us: production, maintenance, distribution, personnel, finances – real-time, historical and predictive. There is more data being collected more quickly and from more sources than ever before. We are swimming in it.
So, now what? Now that we’ve gathered all of this data, what does it mean to us? Personally, having reams of integers, floats, strings and timestamps in my hands doesn’t make me feel any smarter. As the old adage goes: Data is not information. Data without context offers no insight. Data without structure reveals no opportunities. How do we get from data to information? How do we get from information to knowledge? And how do we get from knowledge to action?
Finding the Anomalies
The US Department of Defense employs a process known as Activity-Based Intelligence (ABI) to find useful details in large sets of data. For example, in 2013, when two bombs exploded near the finish line of the Boston Marathon, investigators immediately had at their disposal hundreds of hours of surveillance footage, cell phone photos, and time-stamped video from dozens of angles. To manually review all of this media would require thousands of man-hours – time that is obviously not available in a situation like this.
To make use of this constellation of data, investigators were forced to find a way of automating the investigation. They decided to establish a specific set of details they wanted to locate in all of these photos and videos. Namely, they were looking for any individuals at the scene of the bombing who were not running away or looked unafraid. The behavior recognition technology existed, so it was a simple matter to enter a set of variables into a program and to let the software review the footage in an effort to find the activity that matched these variables. Soon, two suspects were revealed.
While it would have been nearly impossible for human analysts to review all of this footage in a timely fashion, investigators discovered that Big Data could in fact be very useful if combined with a mechanism to compare and contrast the thousands of data points being reviewed.
A similar technique is now being employed in cancer research. A so-called “Big Mechanism” has been created to review the vast and complex medical records of cancer patients that have been established over the years to find overlapping patterns or consistencies that can lead to a new understanding of root causes or precipitating circumstances. By automating the research, we are now able to analyze data sets of much greater size and complexity than would be possible using only human analysts.
Can Similar Techniques be Employed in Industrial Automation?
Today’s industrial enterprises find themselves in a situation similar to those described above. Huge amounts of data are being recorded and opportunities for improvement are known to exist, but how do we know what to look for and how do we find it? The same sort of ABI employed by the DoD may well have a place in the commercial world.
If we can review our historical process data to define the circumstances surrounding certain conditions (unplanned downtime, spikes in energy consumption, etc.), we may be able to recognize repeated patterns or anomalous activity related to these specific circumstances, thereby enabling us to take action to correct the situation before it happens again. By finding the data that stands out from the rest, detailing the characteristics of that data, and looking for those characteristics elsewhere, we may be able to pinpoint causal relationships that were previously obscure or misleading.
On the flipside, the same techniques can be employed to define the circumstances surrounding periods of extended productivity or energy efficiency. The same techniques used to discern the cause of deficiencies can be used to optimize asset performance and improve the quality and efficiency of our processes.
By creating analytic mechanisms aligned with the principles of ABI, we are able to create a safer, more efficient, more productive work environment. Of course, some of this runs counter to the way most of us are programmed to think. We tend to put more stock in consistent, reliable information, while discounting the anomalies. ABI encourages us to find the anomalies and focus on them.
The key to navigating the world of Big Data may not lie in the massive set of data, but in the tiny subset of data that teaches us about the abnormalities or anomalies we find. Look for the data points that stand out from the rest and ask yourself why. Consider the circumstances surrounding the collection of that data; can we map certain plant floor conditions to specific results?
Thus far, the Big Data movement has been a combination of hype and optimism, with very little practical value in daily operations. Some companies are finding ways to take advantage of the opportunities, while others have fallen behind.
Can you find the opportunities?
