Open AccessWordbook‐based light‐duty time series learning machine for short‐term voltage stability assessment
Author(s) -
Zhu Lipeng,
Lu Chao,
Liu Yongjun,
Wu Wei,
Hong Chao
Publication year - 2017
Publication title -
iet generation, transmission and distribution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 110
eISSN - 1751-8695
pISSN - 1751-8687
DOI - 10.1049/iet-gtd.2016.2074
Subject(s) - computer science , software portability , artificial intelligence , term (time) , stability (learning theory) , adaptability , machine learning , benchmark (surveying) , series (stratigraphy) , feature (linguistics) , voltage , engineering , electrical engineering , paleontology , philosophy , physics , quantum mechanics , biology , ecology , linguistics , geodesy , programming language , geography
Focusing on high‐efficiency sequential feature learning for short‐term voltage stability (SVS) assessment, this study develops a light‐duty time series (TS) learning machine based on symbolic TS datasets, namely wordbooks. Numerous cumbersome TS acquired from post‐contingency synchrophasor measurements are first tactfully transformed into short symbolic words, constituting a compact and light wordbook. By centralising the intra‐class words in the wordbook, a series of keywords are quickly extracted without iteration to perform wordbook learning. Owing to the portability of the wordbook, online SVS assessment models can be derived extremely fast from it. With such merits, wordbook learning is designed to be periodically conducted by studying new cases from up‐to‐minute measurements, resulting in enhanced adaptability to inconstant unknown situations. Fractional affixes that frequently occur in the wordbook are further extracted for pattern discovery of voltage instability. Test results on the realistic Hong Kong power grid illustrate the effectiveness and advantages of the proposed learning machine.