It has been an exciting two weeks. We went directly from an on-site electrical infrastructure survey in Kentucky back to Illinois then off to Raleigh, NC for the SMRP 30th Annual Conference followed by an indirect route back to Illinois to stop in Sandusky, OH for the Midwest Vibration Institute meeting on the Friday. Yes, we drove. Unfortunately, air travel is not quite back to the level of reliability where I can trust that we will be where we need to be when we need to be. But that is an article we’re working on for the next newsletter.
Following some unusual challenges at the 29
th annual conference, and the size of the initial attendee list, I was looking forward to a relaxing few days presenting, kicking back in the booth, getting caught up on emails, catching up with friends and associates, and attending a few presentations. Unfortunately, the conference went exceedingly well with close to 1200 attendees (including vendors), an extremely active show floor – that went on 40-60 minutes after the end of the show hours each day, including holding up teardown, and continuous meetings with potential clients. Heck, we even did a little troubleshooting at another vendor’s booth (see
Machine Learning IoT in a Box: Lessons from the SMRP Conference - MotorDocAI) and walked out of there with a number of projects and orders. So, I had to put relaxing off until the weekend (or maybe the next one).
Sandwiched between our friends from R&G Labs and ReliabilityX, with RDI at our backs, when any of us had a chance to breathe we caught up on how things within the industry have been going. Then, of course, the coffee and food being available for vendors before and after the attendees were released en masse into the hall and the open Innovation Lab space. How dare a conference set up something so benevolent for us vendors – we actually felt included. With this kind of treatment we’ll definitely have to attend next year – in protest, of course.
I did get to open up the Innovation Labs on the first day with a discussion of how incoming power effects electrical and electrically driven equipment in-plant, which included a few new and very fresh case studies. The concerns about the quality of power from local distribution, utilities, and internal ground/neutrals are rising to a level we did not see in the past. One of the unintended consequences of adding more and more electronic devices to our systems without reviewing those areas that are invisibly impacted. To my humbled surprise the session was extremely well attended – standing room only well into the eating area. Thank you to those who hung out and listened to me ramble on for an hour about another favorite topic. I will be recording a version of the session over the next few weeks for those who were unable to attend – well, actually, it will be exclusive for our MDMH readers (hint).
There was, of course, the opportunity to have a lesson-learned occur on the last day. In this case, it was a reminder why expert systems and basic tools and knowledge are critically important over just automated ML/AI systems. Don’t get me wrong, these have their place – I’ve used and taught variations of ML/AI/digital twin, etc. since the 1990s (before anyone could design stuff on a cheap laptop or PC) for analytics and data science. Over the past few years we even discovered new methodologies (see IEEE paper:
Machine Learning for Electric Machine Prognostics IEEE Paper - MotorDocAI ) and published the concepts. However, it became apparent during a troubleshooting opportunity at another booth that relying upon AI/ML can have dire consequences. As it turns out, you cannot replace personnel with AI/ML yet – and most likely never. ML can only show you what it has been taught, and associated human bias, and falls apart as soon as something that it has not been trained for occurs – it is explicit. Human beings remain implicit in that we adjust to varying circumstances.
Don’t get me wrong, I use ML/AI and data science, have since the end of the 1980s when it was just a few decades old. Those of us in the reliability and maintenance industry have – by definition. I mean, yesterday I was building a model related to adjusting our 3 and 5 year forecasting in Python, reviewing some probability analysis in R, and working on a new model concept in MatLab for Time to Failure Estimation (what better way to spend a Saturday?). Over the past week we’ve also utilized data collected at a site to do some data drilling and pattern analysis, etc. which was more of a data science task. The use of some of the new tools to increase my (and our) effectiveness is impressive – if you are in reliability engineering I highly recommend getting some exposure to the concepts surrounding data science and machine learning. If you don’t have time for in-person classes there is a lot of information in online coursework and YouTube. I recommend the University courses on systems such as Coursera as they spend time on the concepts and methods instead of selling short cuts (always good to understand the tech when you have to implement it, or you won’t know the strengths and weaknesses).
Of course, we also had initial pre-release copies of “Practical Electrical and Current Signature Analysis of Electrical Machinery and Systems,” which provides details on ESA/MCSA analysis including insight into the automation built into the EMPATH system. You will be able to utilize your ESA device, should it be capable of anything other than linear analysis (i.e.: it only shows FFTs in current or voltage (linear) versus dB, which is not load dependent), with what is in the book. It is expected to be on Amazon and other online sales before the end of the year and will be available in eBook form within the next few weeks online. However, if you are in one of our training courses you will receive a copy.
After tearing down the booth we got some much needed rest before the 10 hour trip to Sandusky, OH for the 90 minute class at the Midwest Vibration Institute on the Friday. In that presentation we covered the use of ESA for the detection of mechanical systems in the electric motor and driven equipment, or the generator and powertrain (of course). This class was also well-attended and ended two days of great classes on vibration, motion amplification, IoT, alignment, etc. with lots of great interaction.
Now, we are back in the field and getting caught up. There are a lot of new things that will hit the wind industry this year and the rest of industry next year. We’ve been working with our partners and a number of the projects we have been working on are about to be released. We paid attention during 2020-2022 and had already started accounting for changes – watch for announcements.
See you on the road!
Howard W Penrose, Ph.D., CMRP
President, MotorDoc LLC
Howard has been a member of the National Writers Union since 2008.
MotorDoc LLC is a registered Veteran Owned Small Business
