Open AccessTransmission line fault classification using modified fuzzy Q learning
Author(s) -
Malik Hasmat,
Sharma Rajneesh
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.2017.0331
Subject(s) - identifier , classifier (uml) , computer science , artificial intelligence , c4.5 algorithm , fuzzy logic , machine learning , transmission line , pattern recognition (psychology) , support vector machine , data mining , telecommunications , programming language , naive bayes classifier
The authors propose an adaptive, self‐learning fault classifier based on modified fuzzy Q learning (MFQL) for transmission lines. Proposed MFQL fault classifier is able to achieve very high classification accuracy with relatively small number of samples. The authors’ is a first attempt at designing a fault identifier using reinforcement learning for fault segregation in transmission lines. The authors’ identifier does not assume prior knowledge of transmission line model or target fault information. Raw voltage and current data (supply and load side) is processed using empirical mode decomposition to generate 13 intrinsic mode functions (IMFs’). Classifier employs the J48 algorithm to further prune these 13 IMF's to eight most relevant input variables, which serve as inputs to the MFQL fault classifier. The authors compare performance of the proposed MFQL classifier to other contemporary AI‐based classifiers, e.g. neural networks and support vector machines. Simulation results and performance comparison against other AI‐based classifiers elucidates that the proposed MFQL‐based identifier achieves a significantly higher performance level and could serve as an important tool for transmission line fault diagnosis.