An efficient technique based on DORT method to locate

multiple soft faults in wiring networks

Moussa Kafal, Andrea Cozza, Lionel Pichon

To cite this version:

Moussa Kafal, Andrea Cozza, Lionel Pichon. An efficient technique based on DORT method
to locate multiple soft faults in wiring networks. IEEE Instrumentation and Measurement
Magazine, Institute of Electrical and Electronics Engineers, 2016, 19 (4), pp.10-14. .

HAL Id: hal-01282600

https://hal-centralesupelec.archives-ouvertes.fr/hal-01282600
Submitted on 7 Apr 2016

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 An Efficient Technique Based on DORT Method to Locate Multiple Soft Faults in Wiring
Networks

Moussa Kafal, Andrea Cozza, Lionel Pichon

Pôle PIEM, GeePs, UMR8507, 11 rue Joliot-Curie, 91190 Gif sur Yvette

Decomposition of the time reversal operator (DORT) recently adopted to wiring fault detection
and localization presented effectual results when dealing with a single soft fault along with
complex network configurations. On the other hand, it failed in handling the task of locating
multiple faults within even simple ones. In this article, we propose an enhanced version of the
standard DORT technique (EDORT) which is based on a complementary procedure enabling the
accurate and selective localization of multiple soft faults in various wiring systems.

Introduction

The emergence of electrical employments geared the indispensable use of electrical cables in
nearly all modern systems. They became the backbone structures that are widely utilized to
distribute power and communication signals throughout nuclear power plants, industrial
machinery, human facilities and transportation systems which made their reliable and safe
operation a pivotal issue. The increasing complexity of modern systems was accompanied with a
huge increase in the lengths of cables. From few hundreds of meters in vehicles few decades ago
to about 4 km in modern ones, wires lengths increased to several hundred km in civil and military
airplanes. Moreover, estimates show that about 5000 km are present in power plants which
skyrockets to 40000 km of cables in railway infrastructures of large countries [1]. All these
examples, show that the massive use of wiring networks in nowadays systems became crucial,
especially that some of these cables are integrated in control and safety operations of many of these
systems. Consequently, the searching phase for the detection and repair of a wiring fault counting
on the human capabilities intervention is both expensive, time consuming and non-efficient.
Besides, this was accompanied with an inability of detection in many of the cases especially that
most of the wires are embedded in the body structure of the systems where as an example, studies
reported that 70 % of checked vehicles were returned fault free although they were defected. Thus,
proper and efficient wire diagnosis tool is strongly recommended to prevent the drastic
consequences that may occur [1].

Whatever the application is, wires are subject to unwanted modifications and breakdowns that can
be classified into two main categories: hard faults (open or short circuits) or soft faults, typically
featured by minor alterations as insulation defects, cracks, or frays which can lead to drastic
consequences if not early rectified. It is worthy to note that our interest in soft faults shoved off
from the studies that were conducted on cables and showed that 30% to 50% of detected wiring
faults are considered to be soft which might be far dangerous than hard ones due to their detection
difficulty [1]. Several techniques have been lately innovated for the task of cable testing, but the
widely used ones were the reflectometry based methods whose aim is to detect the presence of an
impedance discontinuity by submitting a testing signal and monitoring the reflected one [2]. These
high frequency methods, include two main families: Time Domain Reflectometry (TDR) and
Frequency Domain Reflectometry (FRD); they differ in the form of the injected wave, or in the
post processing techniques used to extract information from the reflected wave. For example, in
the TDR family, standard TDR injects a step or a pulse into the wire network to be tested, and
relies on the analysis of the reflected wave in the temporal domain in order to extract the fault’s
signature [3]. On the other hand, in FDR techniques stepped frequency sine waves are injected into
the network under test (NUT), and determines the location of the fault depending on the parameter
that is to be measured (frequency, phase, or stationary wave amplitude measurements) [3]. In fact,
these reflectometry methods showed promising results when dealing with hard faults but failed
when addressing soft ones depicted by weak reflectivities [3]. Moreover, they started flopping
when complex networks evolved due to the appearance of junction related echoes that increased
the complexity of analyzing the reflected signal.

Differential DORT, an acoustically originating technique [4] was adopted to guided wave
propagation on transmission lines [5] and succeeded in detecting and locating a single soft fault
within an NUT. Besides, it has showed a great feasibility when addressing complex wiring
configurations composed from several branches and junctions. On the other hand, it started failing
when handling the mission of detecting multiple faults in different NUTs. Within this context, the
aim of the article is to recall the main constraints faced by the standard DORT as presented in [6]
and extend the applicability of this technique to the case of multiple soft faults in transmission
lines. The second section will elaborate the general concepts of DORT method along with
presenting the main restrictions it faced when addressing multiple faults. This will be followed by
presenting the new enhanced version of the standard DORT explaining the different steps it is
composed of and interpreting the procedure of locating the multiple faults. After that, the third
section will validate the new approach numerically and experimentally on different complexity
NUTs with considering the presence of two different soft faults embedded in the network. Finally,
the last section will draw a conclusion and a summary of this paper.

DORT METHOD

General Concept

Soft fault localization in complex NUT’s is to a great extent similar to the problem of target
location used in radar communication where the target’s positioning becomes tough when its
signature is relatively weak compared to other scatterers present in the medium [7]. It was proven
that by using the time reversal (TR) properties, the reflected echo from a target can be maximized
while using the same amount of energy of the excitation signal thus leading to focal spots on the
target’s location after a series of several iterations. DORT method, derived from the iterative TR
isolates and classifies different scattering spots without the need for iterations, which is achieved
by the use of the time reversal operator (TRO) [8] [9]. This TRO contains valuable information of
the scattering centers in the media encapsulated in its eigenspace structure where propagating the
corresponding signals conducts a focusing on the desired scatterer thus providing separate images
of the scatterers in a media. Taking advantage of the properties of DORT method adopted in a
standard version in guided wave propagation, it became possible to synthesize signals intended to
focus on the position of an eventual fault [5]. It is worthy to note that, all subsequent quantities are
function of frequency thus the frequency variable will be dropped in all of what follows.

In order to approach the benefits of the standard DORT technique (SDORT) in locating soft faults,
we shall introduce first S, the baselined scattering matrix of the NUT given by:

𝑺 = 𝑺𝒇 − 𝑺𝒉 ( 1)
where 𝑺𝒇 is the scattering matrix measured by observing the testing ports of the NUT containing
the soft fault whereas 𝑺𝒉 being the scattering matrix measured in the non-faulty (healthy) version
of the same NUT. This operation allows removing the spurious echoes generated by impedance
discontinuities like junctions, leaving only those echoes initially generated by the interaction
between the testing signals and the faults. The resulting scattering matrix 𝑺 will thus only contain
data related to the signals scattered by the fault, which will be used to synthesize the testing signals
aiming to focus on the fault’s position [5]. This is accomplished by creating the TRO 𝑲 = 𝑺𝑺† ,
given that † refers to the Hermitian transpose, followed by an eigendecomposition process which
produces eigenvalues with their corresponding eigenvectors. As a matter of fact, the number of
non-zero eigenvalues hints at the number of faults present in the NUT. Whereas, the eigenvectors
form the fundamentals of the testing signals which once propagated in the network will form focal
spots indicating the position of the fault. Besides, theoretically DORT method as presented in open
space target location techniques shows that it is possible to focus on any of the scatterers in the
medium simply by propagating the eigenvectors corresponding to its eigenvalue as long as the
scatterers are ideally resolved [8].

This property no longer holds in guided wave propagation along transmission lines, while waves
attenuate as a factor of 1⁄𝑟 in free space [10], guided waves experience negligible attenuation. On
the contrary, a new mechanism of energy exchange appears where waves cannot focus solely on a
fault without interacting with the others. As a matter of this fact, ideal resolve can no longer be
satisfied preventing the selective focusing on each fault separately by using its eigenvalue
components. Besides, as we are dealing with soft faults, SDORT faced an impasse whenever a
considerable high severity difference is present between the faults in a system. In other words, a
relatively weak soft fault in the NUT with respect to the other stronger one in the configuration
will have a masked signature thus reducing the chances of localizing it. In addition to that, it has
stood still when the complexity of the network increases that is manifested by the appearance of
junction related echoes which are able to mask the eventual position of a weak soft fault.
Eventually, these restrictions are well detailed in [6] in a clear manner.

Enhanced DORT Method

While SDORT does not perform well in locating multiple faults in an NUT, it gives accurate and
effective results when dealing with a single fault even in complex network configurations. This
fact motivated us to maximize the profit we can gain from the DORT concept in this field which
resulted in an enhanced version (EDORT) of this method. The EDORT adopts the main steps
followed by the standard method to locate a single fault in the NUT, and proceeds to a
complementary procedure enabling the localization of other faults present in a configuration and
thus bypassing the restrictions encountered by the SDORT.

This new method commences by measuring the global baselined scattering matrix of the multiple
faulty NUT 𝑺𝟏𝑮 , which contains information of the signals scattered by all the faults in the system.
Applying the SDORT steps using this matrix will lead to several focusings on some of the faults’
positions while others will not appear depending on the configuration’s complexity and on the
fault’s severities. As a matter of what has been presented in [6], increasing the system’s complexity
may lead to masking the signature of the very weak soft faults. Besides, as the intensity of the soft
fault becomes relatively low with respect to the others, its detection becomes very hard and
sometimes impossible [6]. On the other hand, locating the position of the strongest soft fault in the
NUT is always possible in the first step which is usually recognized by a dominant focal spot.
Thus, after localizing the first strongest fault we will move to a complementary similar process
allowing locating other faults present in the wiring network. As a matter of what has been presented
in [6], the global scattering matrix 𝑺𝟏𝑮 containing all the faults’ signatures, can be divided into
multiple scattering matrices each containing one of the fault’s response that is

𝑁
𝑺𝟏𝑮 = ∑𝑖−1
𝑓
𝛼𝑖 𝑺𝒊𝒅 ( 2)

where 𝑁𝑓 being the number of faults in the NUT, 𝛼𝑖 the weighting factor which will be later
presented, while 𝑺𝒊𝒅 designating the scattering matrix intending to focus on each soft fault
separately [6]. Thus, we will become capable of providing selective focusing on each fault
separately by monitoring each 𝛼𝑖 𝑺𝒊𝒅 corresponding to one of the faults. So, after localizing the
strongest soft fault present in the NUT, a cloned network similar in structure to the original one
will be created, but with a dummy soft fault on the position localized. The corresponding scattering
matrix 𝑺𝟏𝒅 will embrace the response of the strongest fault solely. After that, a weighted version
𝛼1 𝑺𝟏𝒅 will be subtracted from 𝑺𝟏𝑮 leading to new scattering matrix containing the remaining faults’
signatures. The SDORT steps will be then applied in order to locate the new strongest fault in the
NUT, after which the procedure described before will be followed to localize all faults in the
configuration.

For the sake of clarity, the steps required in the EDORT process are summarized in what follows:

1) measure the global scattering matrix 𝑺𝟏𝑮 of the (eventually) multiple faulty NUT;
2) apply the SDORT method on 𝑺𝒊𝑮 , in order to estimate the position of the strongest soft fault;
3) include a dummy soft fault on the guessed position and compute its scattering matrix 𝑺𝒊𝒅 ;
4) compute the only significant eigenvalue of 𝑺𝒊𝒅 and its eigenvector 𝜇𝑖 and 𝜔𝑖 respectively;
𝜔𝑖𝑇 𝑺𝑮 𝜔𝑖
5) compute the weighting factor 𝛼𝑖 = 2 and 𝑺𝒊 = 𝛼𝑖 𝑺𝒊𝒅 ;
𝜇𝑖 (𝜔𝑖𝑇 𝜔𝑖 )

6) remove the 𝑖𝑡ℎ fault contribution by computing 𝑺𝒊+𝟏 𝒊

𝑮 = 𝑺𝑮 − 𝑺𝒊 ;

7) increase 𝑖 and repeat from step 2) until the update 𝑺𝒊𝑮 has no significant eigenvalue.

RESULTS

In order to better illustrate the benefits introduced by EDORT over its standard version, a single
junction network configuration will be analyzed and its results will be presented in this section.

Analyzed Configuration & Standards

In order to support the EDORT, some rules and standards have been adopted in our study in order
to organize the process of obtaining results and their display. If we consider the general network
structure, illustrated in Fig. 1, we first begin by assigning a number to each of the extremities where
each designates a testing port. We will consider that the one numbered by (1), be the origin, and
distances will be measured according to this origin, and consequently all graphs will plotted
accordingly. Concerning the path designation, each path will be assigned number that is the part
of the network between the origin testing port and any of the other ports. As an example, in Fig.
1, the third path is the part of the network linking the origin to the testing port numbered (3).
Figure 1 Layout of the NUT considered for the experimental validation where two different severity soft faults were created by
partial crushing of 30-cm semi-rigid coaxial cables as presented in Fig. 2 (fault’s severity depends on the radius of the circle and
their positions are measured with respect to the reference port (1)).

All results obtained are treated by frequency domain simulation on Matlab, using the transmission
line theory (TLT), where the transmission lines are considered to be uniform lossless lines with
matched-ends in the simulation model created. This simulation model will be used for post-
processing the experimental data collected. Moreover, in order to display the results for
determining the actual position of the fault, a time-space (ZT diagram) tool is used which allows
the observing the voltage propagation in the network both in space and time.

Experimental results

Experimental tests were conducted, considering the NUT single junction structure depicted in Fig.
1 and implemented using standard 50 Ω coaxial cables as transmission lines as shown in Fig. 2.
Besides, 30 cm semi-rigid coaxial cables were used for the sake of creating the soft faults which
is manifested by a crushing of a small portion of it whose severity depends on the length crushed
as reproduced in Fig. 3. The stronger soft fault is produced by crushing the line for a distance of 2
cm whereas the weaker is designated by a 1 cm crushing. The ends of the cables were the testing
ports for the purpose of simplicity, where three ends of the NUT were connected to a Rohde &
Schwarz ZVB8 vector network analyzer (VNA) capable of covering a frequency range from 300
KHz to 8 GHz. Measurements of the scattering matrices was performed over a total bandwidth of
2 GHz and a sampling frequency of 20 MHz After calibrating the VNA using the calibration kit
provided by the manufacturer studying the network consisted of two steps:
 1. Measurement of 𝑺ℎ (𝜔) of the reference system (without a fault) for a frequency range from
300 KHz to 2 GHz, where we considered the network with unaltered 30-cm semi-rigid
sections.
2. Measurement of 𝑺𝑓 (𝜔)of the faulty system, where the unaltered cable samples were
replaced by the faulty ones.

Having measured the two transfer matrices of the considered NUT, the detailed EDORT steps are
applied followed by post processing using the Matlab simulation model. The ZT diagram of Fig.
4 shows a focal spot pointing the location of the first strongest soft fault in the NUT after the first
iteration of the EDORT. Consequently, the second iteration was enough to clearly localize the
weaker soft fault as designated in the ZT diagram of Fig. 5. The experimental results presented,
allowed the localization of the two soft faults in the single junction network of Fig. 1 which imply
the feasibility and efficiency of the proposed method in locating multiple faults within wiring
networks which shall be extended to include more complex NUT’s.

Figure 2 Implementation of the NUT described in Fig. 1 and connected to a VNA for experimental tests.
Figure 3 The two soft faults considered in the test whose severity depends on the length of crushing of the 30-cm semi-rigid
cables.

Figure 4 The voltage propagation along the second path of the network of Fig. 1 after the first EDORT iteration.
Figure 5 The ZT diagram corresponding to the NUT of Fig. 2 after the second iteration of the EDORT showing clearly the position
of the second soft fault in the configuration.

CONCLUSIONS

EDORT, a new enhanced method based on the standard DORT technique, was extended to
embrace the detection and localization of multiple soft faults within wiring systems. It conquered
the restrictions faced by the standard technique and showed a great feasibility in the precise
localization of multiple faults. Besides, it was able to preserve the main advantage gained by
DORT and manifested by its efficiency when dealing with complex network configurations
composed of several branches and junctions which was validated numerically and by experimental
results. Further work will be needed to assess the sensitivity of the proposed technique with
increased network complexity, as well as its robustness with respect to noisy signals propagating
along live-tested NUT’s.

REFERENCES

[1] F. Auzanneau, “Wire troubleshooting and diagnosis: Review and perspectives,” Progress In
Electromagnetics Research B, vol. 49, pp. 253– 279, 2013.
[2] C. Furse and R. Haupt, “Down to the wire [aircraft wiring],” Spectrum, IEEE, vol. 38, no. 2,
pp. 34–39, 2001.
[3] C. Furse, Y. C. Chung, C. Lo, and P. Pendayala, “A critical comparison of reflectometry
methods for location of wiring faults,” Smart Structures and Systems, vol. 2, no. 1, pp. 25–46,
2006.
[4] M. Fink, C. Prada, F. Wu, and D. Cassereau, “Self focusing in inhomogeneous media with
time reversal acoustic mirrors,” in Ultrasonics Symposium, 1989. Proceedings. IEEE 1989. IEEE,
1989, pp. 681–686.
[5] L. Abboud, A. Cozza, and L. Pichon, “A noniterative method for locating soft faults in
complex wire networks,” Vehicular Technology, IEEE Transactions on, vol. 62, no. 3, pp. 1010–
1019, 2013.
[6] M. Kafal, A. Cozza, and L. Pichon, “Locating multiple soft faults in wire networks using an
alternative dort implementation,” Instrumentation and Measurement, IEEE Transactions on, vol.
65, no. 2, pp. 399–406, 2016.
[7] M. Fink et al., “Time-reversed acoustics,” Scientific American, vol. 281, no. 5, pp. 91–97,
1999.
[8] C. Prada and M. Fink, “Eigenmodes of the time reversal operator: A solution to selective
focusing in multiple-target media,” Wave motion, vol. 20, no. 2, pp. 151–163, 1994.
[9] C. Prada, S. Manneville, D. Spoliansky, and M. Fink, “Decomposition of the time reversal
operator: Detection and selective focusing on two scatterers,” The Journal of the Acoustical
Society of America, vol. 99, no. 4, pp. 2067–2076, 1996.
[10] H. W. Ott, Electromagnetic compatibility engineering. John Wiley & Sons, 2011.
1.