Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
Hvac_vibe

Member
Registered:
Posts: 20
Reply with quote  #1 
Anyone heard of these guys?  Seems to be a fairly new startup thats starting to creep into my particular field.
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #2 
Its just one of another hundred companies offering the same service, including mine!

My boss forwarded me their website once as an example of terrible advertising and marketing BS.

EDIT:

Talked to my boss about it just now, who had a few meetings with Augury to address if they had services worth packaging with our pumps in some way. He claims the services seem more useful and well-thought-out than competitors.
WWST

Sr. Member
Registered:
Posts: 86
Reply with quote  #3 
I am very familiar with them. They have a portable data collector that works through a smartphone.  They also have wireless vibration sensors. The difference is, they claim to run all data collected through their algorithms for "automated" analysis. They have started out by specializing in HVAC and associated equipment.

In my opinion, it will take years and many failures for the algos to learn specific machine's failure modes. At the moment, this is just a step below rules-based systems that companies like Emerson and AZIMA/DLI use. They will most likely have a completely different view. Machine learning is still in its infancy for predictive maintenance. Many of these types of companies have wayyyy more software/algo guys than they have vibration analysts, like 20 or 30 to 1. You do the math.
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #4 
Quote:
Originally Posted by WWST
In my opinion, it will take years and many failures for the algos to learn specific machine's failure modes.


That's not opinion, that's fact. The algos are useless without extensive training data and this training data does not exist. It is also likely that data will never be good enough to train competent algorithms because the secondary half of the data that needs to come from Mechanics (human entry) will probably never be reliable. Until then, many companies are trying what they CAN do to get good training data, and that is selling a useless algo that they then back up with human analysis until (they hope) sometime in the future when they will have a useful algo. Unfortunately, most clients don't understand that service companies or OEMs need years of data before we can help you! That's too long of an investment period for... everyone.
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.