Difference between revisions of "GasDay Alarm Prediction"

From REU@MU
Jump to: navigation, search
 
(One intermediate revision by the same user not shown)
Line 1: Line 1:
  
 
'''Researcher: [[user:jaired|Jaired Collins]]'''   
 
'''Researcher: [[user:jaired|Jaired Collins]]'''   
<br>'''Mentors: '''[http://povinelli.eece.mu.edu/ Dr. Richard Povinelli], Dr. Corliss
+
<br>'''Mentors: '''[http://www.marquette.edu/electrical-computer-engineering/povinelli-richard.php Dr. Richard Povinelli], [http://www.marquette.edu/mscs/facstaff-corliss.shtml Dr. Corliss]
 +
 
 
== Abstract ==
 
== Abstract ==
 
As an infrastructure, pipelines are vulnerable to damaging conditions that warrant action, which results in a loss of profit and extra labor. Sensors are placed along a pipe which can set off an alarm if a measured value exceeds safe levels. At this point, it is reactive instead of preventative maintenance. Using data from the sensors, we anticipate the ability to forecast when an alarm will go off within an hour. With the ability to accurately predict alarms, action can be taken to avoid a loss of profit. Without notification of anomalies, preventable maintenance, cleaning, or catastrophic failure may take place.
 
As an infrastructure, pipelines are vulnerable to damaging conditions that warrant action, which results in a loss of profit and extra labor. Sensors are placed along a pipe which can set off an alarm if a measured value exceeds safe levels. At this point, it is reactive instead of preventative maintenance. Using data from the sensors, we anticipate the ability to forecast when an alarm will go off within an hour. With the ability to accurately predict alarms, action can be taken to avoid a loss of profit. Without notification of anomalies, preventable maintenance, cleaning, or catastrophic failure may take place.

Latest revision as of 13:28, 1 June 2018

Researcher: Jaired Collins
Mentors: Dr. Richard Povinelli, Dr. Corliss

Abstract

As an infrastructure, pipelines are vulnerable to damaging conditions that warrant action, which results in a loss of profit and extra labor. Sensors are placed along a pipe which can set off an alarm if a measured value exceeds safe levels. At this point, it is reactive instead of preventative maintenance. Using data from the sensors, we anticipate the ability to forecast when an alarm will go off within an hour. With the ability to accurately predict alarms, action can be taken to avoid a loss of profit. Without notification of anomalies, preventable maintenance, cleaning, or catastrophic failure may take place.

Milestones and Goals

Week 1 Literature search
Week 2 Literature search and data cleaning
Week 3 Literature search and code state-of-the-art algorithms
Week 4 Run first set of experiments
Week 5 First pass of research paper
Week 6 Generate new ideas for algorithms
Week 7 Code and test new ideas
Week 8 Experiment with new algorithms
Week 9 Finish final draft of research paper
Week 10 Prepare poster and final presentation