en-de
en-es
en-fr
en-sl
en
en-zh
0.25
0.5
0.75
1.25
1.5
1.75
2
Predicting Electricity Distribution Feeder Failures Using Boosting and Online Learning
Published on Feb 25, 20075331 Views
Related categories
Chapter list
Predicting Electricity Distribution Feeder Failures using Machine Learning00:00
Overview of the Talk02:24
The Electrical System03:12
Electricity Distribution: Feeders05:04
Problem05:52
Our Solution: Machine Learning07:59
New York City09:20
Some facts about feeders and failures pt 109:41
Some facts about feeders and failures pt 210:28
Some facts about feeders and failures pt 311:08
Feeder data12:03
Feeder Ranking Application13:23
Application Structure13:54
Goal: rank feeders according to likelihood to failure14:08
Overview of the Talk14:32
(pseudo) ROC pt 115:11
(pseudo) ROC pt 216:12
(pseudo) ROC pt 316:57
Some observations about the (p)ROC17:23
MartiRank pt 119:16
MartiRank pt 220:23
MartiRank pt 322:30
Using MartiRank for real-time ranking of feeders23:26
Performance Metric24:56
Performance Metric Example26:20
How to measure performance over time26:43
MartiRank Comparison: training every 2 weeks27:17
Using MartiRank for real-time ranking of feeders29:09
Overview of the Talk30:50
Learning from expert advice pt 132:05
Learning from expert advice pt 232:38
Weighted Majority Algorithm [Littlestone & Warmuth ‘88]33:41
In our case, can’t use WM directly35:00
Dealing with ranking vs. binary classification35:22
Dealing with a moving set of experts35:48
Other parameters37:36
Performance38:45
Failures’ rank distribution39:28
Daily average rank of failures40:01
Other things that I have not talked about but took a significant amount of time40:37
Current Status42:12
Related work-in-progress42:58
Other related projects within collaboration with Con Edison46:50
Acknowledgments47:50