ICSP2012 Proceedings A Novel Method for Wire Fault Location Using Relectometry and Iterative Deconvolution Qinghai Shi, Olfa Kanoun Chair of Measurement and Sensor Technology, Chemnitz University of Technology, Chemnitz, Germany qinghai.shi@etit.tu-chemnitz.de, kanoun@ieee.org Abstract-The novelt y of this p a p er is to p ro p ose an eicient method for the detection and location of wiring faults with transfer function anal y sis using time domain relectometr y (TDR), iterative deconvolution and o p timization techniques. The p a p er shows how iterative deconvolution and o ptimization techniques are used to locate hard faults, small discontinuities and enhance the signal-to-noise ratio. The develo p ed method can also be a pp lied to characterize the wire faults in the branches of the network. The ada p tive ilters are a pp lied to reduce the deconvolution noise. The p erformance of three frequenc y domain deconvolution techniques, the short window function, the o p timum com p ensation and the automated-regularization technique are a pp lied to solve the ill- p osed inverse p roblem for the transfer function between incident and relected waves. The p ro p osed a pp roach was a pp lied to locate faults on the shielded coaxial cables and twisted p air cables. Keywords-wire faults location; iterative deconvolution; opimization criteria; time domain relectometry; trasfer function between incident and relected waves I. INTRODUCTION The growing use of the wiring in cars, trains, aircrat, communication systems and power distribution systems etc., has caused an increased need for electrical characterization of wiring networks and location of wire faults. There are several emerging approaches for wire faults location and characterization. The most widely used technique for wire fault location is relectometry. Generally, a high-frequency signal is send down the cable. The relected signal includes information about changes of cable impedance and can be therefore used to detect open and short circuits. Furthermore different techniques are available for detecting rays, joints and other small anomalies. The nature of the incident signal is used to distinguish each type of relectometry: Time Domain Relectometry (TDR) [1] uses a pulse or half sine signal, Frequency Domain Relectometry (FDR) [2] uses a set of stepped sine waves, Ultra Wide-Band (UWB) based TDR [3] or Time-Frequency Domain Relectometry (TFDR) [4] called uses a linearly modulated chip signal with a Gaussian envelope, and Sequence Time Domain Relectometry (STDR) uses a pseudo-noise (PN) code [5]. However, these techniques can locate wiring hard faults (open and short) that produce large relection but they are not always able to locate the small anomalies such as rays or chafes, whose small relections are conused by the noise of 978-1-4673-2197-6/12/$31.00 ©2012 IEEE 2139 measurement. Baseline measurement is available for small anomalies location but it is practically unreliable because of the continuously changing cable characteristic and types in wiring systems. A data post processing can be applied to extract wire faults information and locate both the hard faults and small discontinuities. Analytical [6] and estimated [7] transfer unction between incident and relected signals are used to reduce the measurement noise effects and work well in the presence of random noise only and with hard faults. Cepstrum [8] and Pecstrum [9] algorithms have been also applied to detect and locate cable faults. These techniques were applied to estimate cable with hard faults that causes a large relection and thus a high signal-to-noise ratio in the TDR measurement. Correlation algorithms have been applied to detect and locate the small discontinuities. An intrinsic limitation to the applicability of this technique is the attenuation and dispersion of relected signal that can limit the maximum distance of wire fault detection and can affect the accuracy of fault location. In this paper, a novel approach is presented, which is based on using TDR, transfer unction analysis, iterative deconvolution and optimization techniques. The paper shows how iterative deconvolution and optimization techniques are applied to detect and locate hard faults, small discontinuities and to enhance the maximum distance of wire fault detection and signal-to-noise ratio. The developed method can also be available for characterization of wire faults in the branches of the network. The adaptive ilters are applied to reduce the deconvolution noise. Three deconvolution techniques: the short window unction, the optimum compensation and the automated-regularization technique are applied to solve the ill posed inverse problem for the transfer unction between incident and relected waves. The proposed approach was applied to locate faults on the shielded coaxial cables and twisted pair cables. Section II describes the deinition of deconvolution and three frequency domain deconvolution techniques and their optimization criteria. Section III and IV show the experimental measurement system and results with TDR method. Finally, Section V gives the conclusion. II. DECONVOLUTION For a linear, time invariant and causal electromagnetic system with a zero initial state, the relation between input