Fault disturbances classification analysis using adaptive neuro-fuzzy inferences system

Shahbudin, Shahrani; Kassim, Murizah; Mohamad, Rosfarizan; Suliman, Saiful Izwan; Yusof, Yushaizad

doi:10.11591/ijeecs.v16.i3.pp1196-1202

Cited by 1 publication

(1 citation statement)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another advance approach known as hybrid technique that combines more than one of the previous methods is also used to perform the fault analysis. For instance, a neurofuzzy technique called Adaptive based Fuzzy Inference System (ANFIS) is used to perform fault classification with excellent classification results [15][16][17]. Another hybrid technique combination used are wavelet and ANN approach (DWT-ANN) which was applied to the 765kV transmission line fault classification [18].…”

Section: Introductionmentioning

confidence: 99%

Fault classification on transmission line using LSTM network

Omar

Osman

Ibrahim

et al. 2020

IJEECS

View full text Add to dashboard Cite

Deep Learning has ignited great international attention in modern artificial intelligence techniques. The method has been widely applied in many power system applications and produced promising results. A few attempts have been made to classify fault on transmission lines using various deep learning methods. However, a type of deep learning called Long Short-Term Memory (LSTM) has not been reported in literature. Therefore, this paper presents fault classification on transmission line using LSTM network as a tool to classify different types of faults. In this study, a transmission line model with 400 kV and 100 km distance was modelled. Fault free and 10 types of fault signals are generated from the transmission line model. Fault signals are pre-processed by extracting post-fault current signals. Then, these signals are fed as input to the LSTM network and trained to classify 10 types of faults. The white Gaussian noise of level 20 dB and 30 dB signal to noise ratio (SNR) is also added to the fault current signals to evaluate the immunity of the proposed model. Simulation results show promising classification accuracy of 100%, 99.77% and 99.55% for ideal, 30 dB and 20 dB noise respectively. Results has been compared to four different methods which can be seen that the LSTM leading with the highest classification accuracy. In line with the purpose of the LSTM functions, it can be concluded that the method has a capability to classify fault signals with high accuracy.

show abstract