Real‐time approach for oscillatory stability assessment in large‐scale power systems based on MRMR classifier. Issue 19 (4th October 2019)

• Record Type: Journal Article
• Authors: Li, Xin; Zheng, Zhiyi; Ma, Zhicheng; Guo, Panfeng; Shao, Kaixuan; Quan, Siping
• Is Part Of: IET generation, transmission & distribution. Volume 13:Issue 19(2019)
• Page Start: 4431
• Page End: 4442
• Publication Date: 2019-10-04
• Subjects: feature extraction -- filtering theory -- information theory -- pattern classification -- data mining -- power system stability; time approach -- large‐scale power systems -- MRMR classifier -- integrated approach -- real‐time oscillatory stability assessment -- OSA -- mutual information theory -- advanced maximum‐relevance minimum‐redundancy ensemble scheme -- internal relations -- operation variables -- oscillatory stability margin -- OSM -- multiple MRMR procedures -- multiple different feature subsets -- relevant description -- complementary description -- functional expression -- 21‐bus system -- 1648‐bus system -- data mining methods -- feature sets -- computation time -- accurate assessment results -- real‐time adaptive approach; Electric power production -- Periodicals; Electric power transmission -- Periodicals; Electric power distribution -- Periodicals; 621.3105
• DOI: 10.1049/iet-gtd.2019.0180
• Languages: English
• ISSNs: 1751-8687

Abstract

Abstract : An integrated approach for real‐time oscillatory stability assessment (OSA) based on mutual information theory is proposed in this study. An advanced maximum‐relevance minimum‐redundancy (MRMR) ensemble scheme is designed to explore the internal relations between operation variables and the oscillatory stability margin (OSM). Multiple MRMR procedures are generated in parallel to select multiple different feature subsets, in which each feature presents a relevant and complementary description of OSM. The functional expression of the relationships is obtained by curve fitting. The 21‐bus system and 1648‐bus system are implemented to test the performance of the proposed approach. A compared investigation is made with some other data mining methods. The impacts of the number of feature sets, size of feature sets, size of training set, invalid data and computation time are studied. Experimental results reveal that the proposed approach provides faster and more accurate assessment results and is a real‐time adaptive approach for OSA.