With a healthy skepticism out there about cloud applications and plant
manufacturing playing together, a new system, called mVision, uses a combination of supervised
and unsupervised learning techniques to determine predictors of
equipment failure and then continuously monitors it.
Developed by team of
data-mining specialists, the platform includes pre-built adapters for maintenance, automation and
condition monitoring systems, which converts all data into the
Machinery Information Management Open Systems Alliance (MIMOSA)
open-standard model. With the open standards, the system can integrated
to MES and ERP packages for scheduling decisions based on capability
forecast.
Alex Bates, Chief Technology Officer, Mtelligence, says,"Machine
learning hasn't hit mainstream in manufacturing, in part due to the
effort required to build an accurate model and get the data needed to
train the system."
Other features of the platform are a library of intelligent processing
filters for sensor data, including statistical process control and
signal processing algorithms to improve the signal-to-noise ratio prior
to training the platform. That feature offers the platform to benchmark a
normal operating context and use it to compare "abnormal" operating
conditions.
The system also has the ability to correlate sensor data with equipment
assembly and transactional date coming from EAM/CMMS systems, and it
includes prebuilt adapters for SAP, IBM, Maximo, Infor EAM, Infor
Hansen, Ventyx EMPAC, JD Edwards and others. For operations, drivers are
provided for plant historians, including OSIsoft PI System, Wonderware
Historian, GE Proficy, Honeywell PHD and numerous other automation
packages. Mtelligencewww.mtelligence.netMIMOSAwww.mimosa.org
