Using Neural Networks in the Maintenance and Operations REcommender (MORE)

May 6, 2004 - 12:00pm
Bldg. 90
Seminar Host/Point of Contact: 

We'll describe the design of Maintenance and Operations Recommender using Neural Networks. By integrating the information from Computerized Maintenance Management Systems (CMMS) and Energy Managementand Control System(EMCS), MORE can analyze and mine the data to make recommendations to the maintenance personnel and even automate the action if the precision of the prediction is good enough.Our previous research has shown that the relationship between the problem description and the actual action that's taken is non-linear. By introducing Neural Network to map this relationship, we improve the performance of the recommendation a lot. Also we experiment with methods to deal with the textual information in the system. One way is to use similarity index and the other is to use codified action.We demonstrate using codified action is a better choice to handle text.This work is part of the CITRIS Energy project, and the team is led by Dr. Clifford Federspiel.

Add event to Google Calendar

Announcement List

For notification about future seminars, contact Erin Bertiglia.

Site Access

Schedule subject to change without notice. If you are coming from off-site, please call first to verify. UC staff and guests are welcome. LBNL shuttle buses stop every few minutes at marked sidewalk locations along Bancroft and Hearst Avenues and Rockridge BART.