Energy and entropy-based feature extraction for locating fault on transmission lines by using neural network and wavelet packet decomposition

https://doi.org/10.1016/j.eswa.2007.05.011Get rights and content

Abstract

The aim of this paper is to estimate the fault location on transmission lines quickly and accurately. The faulty current and voltage signals obtained from a simulation are decomposed by wavelet packet transform (WPT). The extracted features are applied to artificial neural network (ANN) for estimating fault location. As data sets increase in size, their analysis become more complicated and time consuming. The energy and entropy criterion are applied to wavelet packet coefficients to decrease the size of feature vectors. The test results of ANN demonstrate that the applying of energy criterion to current signals after WPT is a very powerful and reliable method for reducing data sets in size and hence estimating fault locations on transmission lines quickly and accurately.

Introduction

A fault occurs when two or more conductors come in contact with each other or ground Hawary (1995). In three phase systems, faults are classified as:

  • Single line-to-ground faults.

  • Line-to-line faults.

  • Double line-to-ground faults.

  • Three phase symmetrical faults.

Ground faults have been considered as one of the main problems in power systems and account for more than 80% of all faults (Tungkanawanich, Kawasaki, Matsuura, & Kuno, 2000). These faults give rise to serious damage on power system equipments. A ground fault which occurs on transmission lines not only effects the equipments but also the power quality. So, it is necessary to determine the fault location on the line and clear the fault as soon as possible in order not to cause such damages. Flashover, lightning strikes, birds, wind, snow and ice-load lead to short circuits. Deformation of insulator materials also leads to short circuit faults. It is essential to detect the fault quickly and separate the faulty section of the transmission line. Locating ground faults quickly is very important for safety, economy and power quality.

There are several methods such as using the variation of line impedance, measuring faulty current and voltage signals and a lot of study has been continued with advances in computer technology. When fault location is estimated by using current and voltage wave information, methods based on traveling waves, faulty line impedance calculations, neural network and wavelet transform (WT) are used widely (Hannienen & Lethonen, 2002). In traveling wave method, fault location is determined by using time difference between incident and reflecting waves (A, 2004, Jian et al., 1998, Liang et al., 2000). This method has been restricted because of the difficulty in analysis. Traveling waves require a very high sampling rate as well. Calculating characteristic reactance is another method which is used for estimating fault distance (Fikri & El-Sayed, 1988). This method calculates reactance between relay point and fault represents an accurate estimation of fault distance when the fault resistance is very small at line ends. One of the other techniques employed is Fourier transform which obtain line impedance in the frequency domain (Tawfik & Morcos, 1998). Recently the most used technique in estimation of fault distance is artificial neural network (Lai et al., 2000, Yu and Song, 1998). Artificial neural network is trained with the samples of faulty current and voltage signals. In spite of Fourier transform, recently WT has been used to obtain the best information of current and voltage signals. The main advantage of WT is that the band of analysis can be finely adjusted and the results obtained from WT are shown on both the time and frequency domain.

In spite of the technological development and efforts in the literatures for fault location studies, providing a more reliable and accurate fault location algorithm is still considered a challenge. In this paper, a new extension of WT which is named wavelet packet transform (WPT) has been used for feature extraction. WPT is capable of dividing the whole time-frequency plane while the classical WT can determine analysis only for low band frequency (Hong, Chiang, & Elangovan, 1999). The features obtained from WPT will be used for training and testing of ANN. On the other hand, training of ANN is very difficult as data sets increase in size. So, it is essential to reduce the size of data sets. For this purpose, wavelet energy and entropy criterion have been used. After the wavelet packet coefficients are obtained for each faulty current and voltage signals, the energy and entropy values have been calculated for each signal separately. In this way, the size of data sets decreases in a remarkable ratio. In this paper, it is intended to find out the best solution method which yields the best results for estimating transmission lines fault location by using the above mentioned techniques.

Section snippets

Wavelet packet decomposition

The wavelet packet method is an expansion of classical wavelet decomposition that presents more possibilities for signal processing (Misiti, Misiti, Oppenheim, & Poggi, 2004). In wavelet transform, signals split into a detail and an approximation. The approximation obtained from first-level is split into new detail and approximation and this process is repeated. Because of the fact that WT decomposes only the approximations of the signal, it may cause problems while applying WT to in certain

Artificial neural network

Artificial neural network is widely used in the engineering areas such as telecommunication, medicine, control and power systems (Hong et al., 1999). ANN is made up of many computational processing elements called neurons or nodes (Mori et al., 2002, Saha et al., 1999). These nodes operate in parallel and are connected together in topologies that are loosely modeled after biological neural systems (Joorabian, 1998). The training of ANN is carried out to associate correct output responses to

The implementation of the fault location system

A 380 kV and 360 km long power system which is simulated by using alternative transient program (ATP) is shown in Fig. 3. The simulation parameters are chosen with 1:1000 scale factors. Considering the scale factor, line resistance, line inductance, mutual capacitance, earth resistance and earth capacitance were selected as 13 ohm, 290 mH, 1 μF, 5 ohm and 2 μF, respectively. The simulation time is 110 ms with 10 μs time step. It was assumed that faults occurred at 20 different locations randomly. For

Conclusions

This paper presents a data reduction technique for estimation of fault locations on transmission lines rapidly and accurately. In the literature, there was a lot of study in this area. In recent years, the methods based on wavelet transform started to be used in signal processing applications widely. In this work, the proposed method uses wavelet packet decomposition which provides more features about the signal than classical wavelet. After wavelet packet decomposition with three-level, the

References (23)

  • R.E. Learned et al.

    A wavelet packet approach to transient signal classification

    Applied and Computational Harmonic Analysis

    (1995)
  • J. Liang et al.

    Application of wavelet transform in travelling wave protection

    International Journal of Electrical Power and Energy Systems

    (2000)
  • I.K. Yu et al.

    Wavelet analysis and neural networks based adaptive single-pole autoreclosure scheme for EHV transmission systems

    International Journal of Electrical Power and Energy Systems

    (1998)
  • R.R. Coifman et al.

    Entropy-based algorithms for best basis selection

    IEEE Transactions on Information Theory

    (1992)
  • Elhaffar, A., & Lehtonen, M. (2004). Travelling waves based earth fault location in 400kV transmission network using...
  • M. Fikri et al.

    New algorithm for distance protection of high voltage transmission lines

    IEE Proceedings C

    (1988)
  • Hannienen, S., & Lethonen, M. (2002). Earth fault distance computation with fundamental frequency signals based on...
  • M.E. Hawary

    Electrical power systems

    (1995)
  • Hong, C., Chiang, L.P., & Elangovan, S. (1999). Wavelet packet analysis and artificial intelligence based adaptive...
  • Q. Jian et al.

    Traveling wave fault location of transmission line using wavelet transform

    Power System Technology

    (1998)
  • Joorabian, M. (1998). Artificial intelligent based fault location technique for EHV series-compensated lines. In...
  • Cited by (189)

    View all citing articles on Scopus
    View full text