Electric Power Systems Research 78 (2008) 1686–1692 Contents lists available at ScienceDirect Electric Power Systems Research journal homepage: www.elsevier.com/locate/epsr Prediction of leakage current of non-ceramic insulators in early aging period Ayman H. El-Hag a,∗ , Ali Naderian Jahromi b , Majid Sanaye-Pasand c a Electrical Engineering Department, American University of Sharjah, Sharjah, United Arab Emirates b Kinectrics Inc., Transmission & Distribution Technologies, Toronto, Canada c Electrical and Computer Engineering Department, University of Tehran, Iran article info Article history: Received 10 September 2007 Received in revised form 11 February 2008 Accepted 24 February 2008 Available online 8 April 2008 Keywords: Outdoor insulators Aging Leakage current Neural network abstract The paper presents a neural network based prediction technique for the leakage current (LC) of non- ceramic insulators during salt-fog test. Nearly 50 distribution class silicone rubber (SIR) insulators with three different voltage classes have been tested in a salt-fog chamber, where the LC has been continuously recorded for at least 100h. A boundary for early aging period is defined by the rate of change of the LC instead of a fixed threshold value. Consequently, the Gaussian radial basis network has been adopted to predict the level of LC at the early stage of aging of the SIR insulators and is compared with a classical network. The initial values of LC and its rate of change at 10min intervals for the first 5h are selected as the input to the network, and the final value of LC of the early aging period is considered as the output of the network. It is found that Gaussian radial basis function network with a random optimizing training method is an appropriate network to predict the LC with a 3.5–5.3% accuracy, if the training data and the testing data are selected from the same type of SIR insulators. © 2008 Elsevier B.V. All rights reserved. 1. Introduction One of the main causes of aging of silicone rubber (SIR) insu- lators is the development of leakage current (LC) on the insulator surface leading to dry-band arcing. Therefore, LC is usually moni- tored to evaluate the insulator’s surface condition under both field and accelerated aging test conditions. Several studies have been conducted to understand the relation between the LC and degra- dation of SIR insulators [1–11]. It has been found that the level of LC low frequency harmonics (mainly the fundamental and third harmonic components) is highly correlated to the degree of insu- lator surface damage [2,3]. When the fundamental component of LC exceeds 1mA during salt-fog test, erosion is evident on the sur- face of the SIR [2]. Another study has been carried out by using the rotating wheel dip test as the accelerating aging technique to monitor the early aging period of SIR insulators [6]. It has been reported that if the peak value of the LC attains 1 mA, the insula- tors lose their hydrophobicity and the damage on the surface begins when the LC approaches 4 mA [6]. Kumagai and Yoshimura [11] sep- arated the leakage current during salt-fog test into three different components: sinusoidal, transition, and local arc. They have shown that the cumulative charges of these components are sensitive to the hydrophobicity and the contamination level of the insulating surfaces. ∗ Corresponding author. E-mail address: aelhag@aus.edu (A.H. El-Hag). Moreover, it has been observed that the surface degradation of non-ceramic insulators occurs when the LC exceeds certain values in the field as well [7,8]. This is particularly relevant for insulators that are exposed to high humidity, salt, and other pollutants [6–8]. This value depends on the nominal voltage, pollution level, humid- ity, and the period of the dry intervals without moisture so that the insulator may or may not recover its hydrophobicity. So it is evident from the previous studies that there is a correla- tion between the level of LC and surface conditions of non-ceramic insulators. As a result, if the level of LC is predictable it will be possible to forecast the surface degradation of SIR insulators. Arti- ficial neural network (ANN) has been employed in several studies related to outdoor insulators. Classification of LC waveforms mea- sured in clean-fog test has been investigated [3]. An ANN has been used to categorize the LC into four classes, based on the magnitude of the fundamental and harmonic components of the LC. Another network has been trained to classify the waveform as sinusoidal, nonlinear, or containing discharge. A feed-forward back-propagation ANN with two layers has been employed for the classification [3]. Although the classification has been successful, no attempts have been made to predict the level of LC. The classification of surface conditions for polymeric materi- als, assessed in inclined plane test, has been studied by using ANN [14]. The process of identifying of the surface condition of non- ceramic insulators was automated in this study. An ANN based classifier was used to categorize the LC measured in the inclined plane test. A multilayer feed-forward ANN with a back-propagation learning algorithm has been employed. Although the authors have 0378-7796/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.epsr.2008.02.010