Loading Events

« All Events

  • This event has passed.

RAEL Lunch Talk: Efficient Clustering Methods for Wind Turbine Performance Across the U.S

March 21, 2018 @ 12:00 pm - 12:30 pm

Cur­rently there are more than 50,000 oper­at­ing wind tur­bines across the entire U.S. Their per­for­mances varies based on the design, weather con­di­tions, and geo­graphic loca­tions. Study­ing the capac­ity for each wind tur­bine indi­vid­u­ally is cum­ber­some and unable to scale with the rapid growth in wind energy deploy­ment. We pro­pose a solu­tion that group the wind tur­bines with close geo­graph­i­cal loca­tions into wind farms of sim­i­lar per­for­mances. This enables us to study the high-​​level zone behav­ior by treat­ing all wind tur­bines within a zone to be iden­ti­cal. It not only reduces the total num­ber of data (around 7TB) we need to analy­sis but also takes the build­ing of future wind tur­bines into con­sid­er­a­tion. Dur­ing this pre­sen­ta­tion, we will intro­duce the cur­rent clus­ter­ing tech­niques and dis­cuss the method used to clus­ter the wind tur­bines, includ­ing what para­me­ters are taken into con­sid­er­a­tion, what makes a “good” clus­ter, and how to batch clus­ter this 7TB data across the U.S. Finally, the talk will present some pre­lim­i­nary results of clus­tered zones with var­i­ous met­rics to eval­u­ate these clus­ters. This research has wide­spread applic­a­bil­ity and is par­tic­u­larly rel­e­vant in group­ing other renew­able energy systems.

 

Zhaohong(Roy) Jin is a fourth-​​year under­grad­u­ate stu­dent at UC Berke­ley dou­ble major­ing in Com­puter Sci­ence and Eco­nom­ics. He Joined the Pro­fes­sor Daniel Kammen’s Renew­able & Appro­pri­ate Energy Lab­o­ra­tory in Spring 2017 and worked with Pro­fes­sor Deb­o­rah Sunter on the SWITCH USA project. His focus is on the real-​​world appli­ca­tion of machine learn­ing tech­nolo­gies. Renew­able energy is an ideal field for him to study because of the huge amount of data gen­er­ated every­day. Roy’s goal to help intro­duce the cut­ting edge machine learn­ing researches to the clean energy field by quickly pro­to­typ­ing appli­ca­tions and build reli­able com­pu­ta­tional pipelines. He is also inter­ested in var­i­ous deep learn­ing tech­niques in com­puter vision and nat­ural lan­guage pre­pro­cess­ing. Roy has pre­vi­ously worked as a soft­ware engi­neer in Moody’s Ana­lyt­ics and Ama­zon. He will be attend­ing grad­u­ate school after his Bach­e­lor degree.

Roy

Details

Date:
March 21, 2018
Time:
12:00 pm - 12:30 pm

Venue

ERG Reading Room
310 Barrows Hall, Berkeley, CA 94720 United States
+ Google Map
Phone:
510-642-1640
Website:
http://rael.berkeley.edu
Main Menu
RAEL Info

Energy & Resources Group
310 Barrows Hall
University of California
Berkeley, CA 94720-3050
Phone: (510) 642-1640
Fax: (510) 642-1085
Email: ergdeskb@berkeley.edu


Projects

  • Open the Main Menu
  • People at RAEL

  • Open the Main Menu