Previse Data Sheet

Industries work hard to create machine reliability,
but breakdowns continue to happen that are sometimes catastrophic. More often than not, the fault
lies with the way we do maintenance and not with
the machine. Mtell™ Previse brings a fresh approach
to condition monitoring with “evidence-based” hard
data that replaces customary policy and subjective
opinions regarding maintenance best practices.
The truth is plain and simple: maintenance departments
want to keep machines running and prevent breakdowns.
Maintenance best practices have evolved over time to accommodate this, progressing from run-to-failure, to calendar/
condition-based, to reliability- centered maintenance (RCM).
But are these processes “best-practices” if improvements are
hard won, take a lot of time, and require great expertise and
expense? Industry leaders suggest something is not right:
63% of all maintenance is
unnecessary and causes
more problems than it fixes.
– Emerson
85% of all equipment failures
happen on a time-random
basis regardless of inspection
and service. – Boeing
The Mtell Previse Prescriptive Maintenance solution uses Autonomous Agents
that interpret every failure signal, every minute of the day … forever – and
notify you when equipment failure will occur and how to avoid it.
Contemporary condition monitoring applications use
techniques to “trap” anomalies or changes in operational
behavior of a machine that might indicate a problem.
Such methods are complex, limited to certain equipment,
prone to error, and ALWAYS require further expert investigation and validation; producing high levels of false positives.
Mtell™ Previse uses machine learning agents to learn
operational behavioral patterns using actual data from
sensors on and around a machine or manufacturing process.
Mtell Previse recognizes diverse patterns in the sensor
signals that indicate degradation, failure, and root cause.
Mtell Previse with
Autonomous Agents
Other Condition Monitoring
Agents do the work so you do not have to
Experts skills needed to build solutions
Agents sample, cleanse, analyze, diagnose,
predict, initiate prescriptive action and advise
These are generally manually developed processes,
that require expert intervention at all stages
End users can create Agents in minutes
Requires models/rules and experts
Learning and adaptive as conditions change –
minimizes false positives
Always need human intervention – no adaptive or
learning built in and tends to false positives
Patterns are accurate, evidence-based, derived
from real, hard, collected data from the machines
Patterns are estimates (derived from engineering
models, equations, and rules)
Agents recognize minuscule multi-dimensional and
temporal patterns across many sensor signals that
humans and other technologies cannot see
Built on models incorporating rules engines
to attempt to correct model errors – they are
difficult to build and prone to inaccuracies
Agents learn normal behavior and very specific
failure patters tied to the root cause
Usually, only performs anomaly detection. Cannot
detect precise patterns that lead to specific failures
Works 24/7 without breaks and retains
knowledge…forever
Requires extensive “care-and-feeding” by experts
Mtell Previse Features
™
Architecture Integration
Mtell Previse extracts sensor data from
Predictive Scheduling
Agents detect failure signatures very early
the plant historian, and failure work order
and immediately send prescriptive work
history from the enterprise asset manage-
orders to the EAM system with time to plan
ment (EAM) system. Previse requires only
and organize. Messages contain the full
a few equipment-connected sensors and
scope of work including: cost, tools, labor,
to activate machine learning.
time, and safety concerns.
Root Cause
Self-Learning and Training
Failure Agents autonomously detect the
Agents get smarter over time. They send
exact patterns that lead to specific failures
alerts to forewarn warn of irregular behavior
issues earlier and with greater clarity than
and follow up by refining baseline normal
anomaly detection only solutions. Agents
or by adding new, more accurate failure
detect the root cause and show which
Agents. All behavioral knowledge of the
sensors contribute most to the event.
plant operation is retained…forever.
Early Warning &
Accurate Time-to-Failure
Transfer Learning
Anomaly Agents ask, “Is this normal?” to
recognize irregular behavior. But, when it
is a failure, they then ask, “What pattern
led to this failure?” Thus, a 7-day anomaly
warning can be transformed into a Failure
Learned behaviors (normal, degradation,
and failure) captured on one are readily
transferred to similar assets with the same
sensor configuration. After a very short
retraining period, they all share the same
safety and breakdown protection.
Agent providing 30-day plus warnings.
Operations Systems
Asset Management Systems
*All trademarks are the property of their respective owners.
The Mtell™ Previse Difference
Unexpected Mtell Previse
A Customer Experience
Manufacturing Processes
At a major US oil and gas company, Mtell installed, configured,
Leading-edge customers drive even greater
and implemented Mtell Previse on five major rotating assets in
potential from Mtell’s Previse’s machine learning
less than three days. Mtell Machine Learning Agents actively
technology. Manufacturing processes are
monitored the assets and found the cause of a major problem
monitored for deterioration, or process degradation
that had plagued the client for years. The customer respond-
that may lead to product quantity and quality
ed with surprise, “We’ve never seen such a rapid plug-n-play
discrepancies. Mtell Previse can facilitate action
enterprise product deployed before!” Mtell Previse made
to correct operating deviations that would
the discovery within days, whereas the competition budgeted
otherwise spoil or waste multi-million dollar
3-4 months just for implementation.
product batches.
Self-Learning
Interoperates
with Maintenance
Less Cost
Less Risk
Far Earlier
Warnings
Big Data
Mtell Previse is creating a major positive change in maintenance culture.
Crews become less dependent on ineffective tools and methodologies.
Production lines run smoothly with less downtime, higher quality,
and increased net product output. Overall there is less risk to people,
the environment, and to the bottom line.
www.mtell.com . 1550 Hotel Circle North . Suite 120 . San Diego, CA 92108 . +1 (619) 295.0022
©2015 Mtelligence Corporation (dba. Mtell).
All Rights Reserved.
MTL-132