Monitor and Diagnose Framework for Manufacturing Process
(MDFMP) is stream computing optimization analytical software
designed to increase yield and leverage overall production performance
in the manufacturing industry. With its parameter optimization
technology, MDFMP first identifies the key parameters of the
production process, collects and analyzes heterogeneous data from
various machines, then applies mathematical models to suggest best
resource combinations for the product lines. Throughout the course of
data collection, the system provides more accurate analysis each time.
The manufacturing process becomes more flexible which contributes to
better production performance and shorten the time of manufacturing.
MDFMP is able to process production data including machine status,
sensors and visual images to build analytics models. The scalable
storage framework allows manufacturers to collect a large number of
non-synchronization, heterogeneous and fast-paced characterized data
◗ For more info: https://web.iii.org.tw
OAK RIDGE NATIONAL LABORATORY
Scalable Algorithm for Deep Learning
Deep learning has the potential to revolutionize scientific discovery.
However, designing an optimal network topology with optimal
hyperparameter values for scientific data is a challenge. Multinode
Evolutionary Neural Networks for Deep Learning (MENNDL) is
a scalable evolutionary algorithm, using the Oak Ridge Leadership
Computing Facility’s Titan supercomputer, to design the optimal
hyperparameters and topology of deep convolutional neural networks.
This scalable evolutionary algorithm can automatically design a deep
learning network capable of very high accuracy in classification or
prediction tasks. Using this technology, researchers can now design an
optimal deep learning network within a matter of hours as opposed
to months and have successfully executed this algorithm on 100
percent of Titan’s 18,688 nodes for periods up to 24 consecutive hours.
This technology is the first to provide researchers with the ability
to optimize the interactions between parameters and their values as
applied to different data sets. It also significantly reduces the expertise
and knowledge required for
parameter tuning. Solving the
problem provides the ability
for scientists to use deep
learning for scientific discovery, especially as an in situ data analysis tool
for scientific simulations or data collection.
◗ For more info: www.ornl.gov
Understanding Pipeline Risk
PRIME-One is a predictive model that analyzes pipelines on an
individual joint (pipe section) and weld level to provide more granular
risk assessments. The technology performs detailed risk assessments
based on over 100 parameters and relationships known to contribute
to pipeline failures. This analysis includes the determination of high
consequence areas based upon population density and environmental
sensitivity, the creation of verifiable
and traceable data sets, and a failure
analysis based on operational
and geographic factors such
as landslides, floodplains, soil
characteristics, and earthquakes.
With the granularity of the analysis
at the individual joint and weld
level, the tool provides this analysis
for each component of the pipeline
at its specific location, as opposed
to a system-by-system approach.
In performing analysis at this level,
inspection, repair, and replacement
efforts can be prioritized based on
areas of highest risk of failure and/
or potential consequence at the
individual component level.
◗ For more info: www.bechtel.com
PACIFIC NORTHWEST NATIONAL LABORATORY
Better Understanding Cyber Attacks
Stream Works is a unique technology that discovers emerging patterns
of sophisticated cyber-attacks by analyzing multi-source streaming
data in near-real time. It alerts analysts to top-pattern incidents, and
provides an easy-to-understand description of the potential threat
and a rationale for why the threat was chosen. It shifts cyber defense
postures from reactive (the style of current postures) to proactive (the
style of postures needed). Most security breaches are eventually detected
by existing cybersecurity tools deployed within enterprise networks.
However, analysts in most operations centers are often overwhelmed
by the volume of alerts typical systems constantly generate, the need to
manually triage such alerts and the need to understand why a system
generated such alerts. Stream Works addresses these challenges by
providing a cyber-defender with new capabilities and tools.
◗ For more info: www.pnnl.gov