We’ve all heard about “machine learning.” At its simplest, “Machine learning is an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.” (1) Basically, we construct algorithms, instead of programmed instructions.
In today’s industrial world we are constantly asked to do more with less resources. We need to be more efficient in our work life to be able to accomplish all the tasks that are presented to us. If you are a Process Engineer, you may need to quickly analyze some process data to help diagnose a current quality problem. If you are a Production Manager, you may need to get a clear update on production progress no matter where you are or the time of day. If you are in maintenance, you may need some historical operations information to help identify an equipment issue. With all these scenarios you need personalized data readily available from your processes…at the perfect time. And then often, you need time to digest what the data means. You do not have time to wait to have someone from IT create a report for you. In fact, don’t you really want your data to “read you mind” and let you know when there is something you need to be aware of? What’s the latest “news?”
Today, tools with machine learning are serving up “the News” for you. I think you will find this to be very helpful.
In the past, when I was working in the aluminum industry we would capture real-time rolling mill parameters so the process and maintenance engineers could then manually review the data looking for issues that would affect coil stock quality. This process of manually pouring over key process parameters such as, coil temperature, roll force, motor amps and vibration, was a time consuming and tedious task. Today, it would be simple and easy for me to see that data in real time using Cloud Historian tools and even use a News Feed app (Machine learning algorithms that are analyzing the patterns in my data – See Figure 1) to tell me when it detects anomalies that may need my attention…even without me asking or programming alarm limits. Looking back, if we would have had a News Feed tool that could automatically detect when a critical variable’s data pattern was behaving “strangely”, we would have saved a tremendous amount of time. I would even wager that with this tool we would have detected a deteriorating set of the Hot Mill’s work rolls before off quality coils were produced and thus scrapped.
I’ve prepared a video to share what this cool capability looks like.
Figure 1 – Example News Feed Alerts from Wonderware Online Insight