To identify and predict which external factors affect machine efficiency as a key performance indicator and how to make recommendations to optimize such factors to maximize performance
2. To identify and predict which external factors affect machine efficiency as
a key performance indicator and how to make recommendations to
optimize such factors to maximize performance
3. The Industrial Internet-of-Things (IIoT) demands dynamic approaches in
operationalization strategies that can handle the mass scale of data
driven by the explosion of smart-enabled, network connected sensors,
actuators, motors, devices, machinery and other ‘things’. Forward-thinking
manufacturers are rapidly adopting the disruptive curve in order to
connect their machines to the Industrial IoT to gain predictive analytics,
actionable insights and cognitive recommendations for their strategic
advantage.
While the world is enamored with the promise of leveraging data from the
IIoT, the automotive giants are facing dramatic challenges in actually
operationalizing such an approach into their day-to-day business flows as
a closed-loop system. As a result, establishing a continuous and
repeatable process to resolve the data deluge and to further increase
machine efficiency also becomes a serious challenge – a classic case of
“human scale vs machine scale”.
The client, a Fortune 100 automotive manufacturer, wanted to analyze
their Engine Manufacturing Center (EMC) by identifying which external
factors could impact its production and predict how they could optimize
the efficiency of their machines while reducing maintenance costs. The
Engine Manufacturing Center (EMC) has 4 sections namely Block Line,
Head Line, Crank Line and Assembly Line, which together produce
equipment to be shipped to vehicles for assembly. The efficiency of a
machine - termed OEE - is a measure of the percentage of Productive
Time Multiplied by the percentage of good parts.
OEE = (% of Productive Time) * (% of Good Parts)
The overall objective was to identify and draw inferences about how
various external factors influence machine efficiency. By combining
machine data with external data from influencing factors in a way that
requires minimal human intervention, DataRPM empowered the client to
achieve its strategic initiative in “Exploiting The Unknown” (ETU).
4. DataRPM’s CPdM Platform identified segments with Low OEE and High
Power consumption and recommended prescriptions for achieving higher
machine efficiency
The core of the analysis was to monitor the efficiency of each machine
over a period of time and to identify the external factors that influence the
maximum and minimum efficiencies. These factors were then used to
predict machine breakdowns / failures and then plan production
schedules in a much more efficient manner to achieve an optimal desired
outcome.
DataRPM’s CPdM (Cognitive Predictive Maintenance) Platform combined
and analyzed the data, automatically, from production log with external
factors such as local weather , traffic (Productivity of Labour), electrical
power (as consumed by each machine) and temperature data (the
temperature details and layout of the EMC center) based on machine and
time (hourly, 3-hourly and daily) to gain a holistic view of the probable
factors affecting machine efficiency and production quality.
However, the machine data itself had multiple challenges, such as arriving
at an apt definition for OEE, calculating OEE and aggregating data from
multiple machines to calculate effective OEE. Additionally there were 30
different AHU file structures in temperature data, inconsistency in time
granularity of data sources encoded in multiple data formats and electrical
data with PSD delimiters.
Every machine has a point at which its OEE is maximum, energy
consumption is minimum and it produces the highest number of good
parts. DataRPM had to identify this point for each machine so that
efficiency could be maximized at minimum costs.
DataRPM’s CPdM Platform processed thousands of machine records
aggregated from multiple sources and formats and automatically built a
prediction model that was scalable and extensible. Additionally the model
could also be scheduled and automated to allow for new data with great
flexibility in adding new sources and types of data-files.
Within 5 days, DataRPM’s Meta-Learning engine was able to provide
actionable insights that established relations between the external factors
and the machine data and thus was able to make highly accurate
predictions. This amazingly fast time-to-market was only possible
because the CPdM platform automated a lot of complex, time-consuming
and otherwise manual processes to run many models in parallel, which is
beyond the human scale of capability.
The client thus had insights that were operationalized in its day-to- day
decision making systems for increasing machine efficiency.
5. Delivered Hourly, 3-Hourly and Daily rollups of Production Log
with weather, traffic, electrical, temperature and other
parameters
Based on external factors, the client was able to predict weeks in advance
what exact factors were driving higher consumption and would lead to
failure states of machines.
The company could thus modify its production plan to improve efficiency,
increase the production of good quality parts and ensure that they weren’t
running behind schedule.
Machine generated insights based on all the data itself for
users who didn’t know where to start their analysis, thus
resulting in a significant 3% improvement in operational
performance
DataRPM’s CPdM Platform powered by the Cognitive Data
Science engine automatically identified multiple factors that
gave the client a deeper insight into how various external
factors were impacting machine efficiency
Overall, by predicting factors that would impact machine efficiency, the
global automotive client was empowered to safely alter their production plan
to achieve a 3% improvement in machine efficiency and a solid 5%
reduction in operational costs, and hence greater profitability and market
competitiveness.
6. DataRPM’s award-winning Cognitive Predictive Maintenance Platform
automates Data Science for the Industrial IoT by using Meta-Learning, which
is forging a new frontier as the next evolution in Machine Learning.
DataRPM’s patent-pending technologies help asset-based industries predict
& prevent asset failures, minimize maintenance costs, optimize inventory &
resource utilization, predict & prevent quality issues in production, forecast &
minimize warranty claims, and reduce risks in the most rapid manner with
high accuracy and while delivering multiple significant annual cost savings
to companies’ bottom lines. Company is headquartered in Redwood City,
California with offices in US, UK, and India. Customers include industry
leaders like Jaguar Land Rover, GE, Cisco, Orange, & fortune 1000 global
industrial companies.
To Know More
Email : marketing@datarpm.com