making the distribution grid observable via deep learning-九游会平台

报告题目: making the distribution grid observable via deep learning

报 告 人: prof. lang tong

报告时间: 2018年12月17日，10:00 am– 12:00pm

报告地点: 清华大学西主楼2区203会议室

联 系 人: 孙宏斌 电话：62783086

lang tong is the irwin and joan jacobs professor of engineering of cornell university and the site director of power systems engineering research center (pserc). he received the b.e. degree from tsinghua university and the ph.d. degree in electrical engineering from the university of notre dame.? his current research focuses on data analytics, optimization, and economic problems in energy and power systems, smart grid, and electrified transportation systems.? a fellow of ieee, lang tong is the 2018 fulbright distinguished chair in alternative energy.

abstract:

unlike the transmission systems where redundant measurements are collected, current distribution systems have few installed meters.? the lack of real-time measurements makes the distribution grid unobservable for state estimation. the conventional weighted least squares (wls) method and its variants either fail numerically or produce misleading estimates.? in this talk, we present a machine learning approach to state estimation, bad-data detection, and bad-data cleansing.? the machine learning solution overcomes system unobservability and outperforms conventional wls-based pseudo-measurement techniques.