Vous êtes sur la page 1sur 13


James Hofmeister and Sonia Vohnout

Ridgetop Group, Inc.
3580 West Ina Road
Tucson, AZ 85741
Telephone: (520) 742-3300

Abstract: Cabling degradation represents a large segment of complex systemmaintenance budgets. In addition to submarines, unmanned underwater vehicles (UUVs)
use submersible cables. Surface ships, aircraft, ground vehicles and industrial machines
that encounter vibration or adverse environmental conditions can also benefit from
Ridgetops cable failure detection and PHM toolset. This paper presents a non-invasive,
ruggedized toolset consisting of a personal hand-held device that uses an innovative,
frequency sweep-based analysis approach to detect damage to multi-conductor
submersible high data rate (SubHDR) cables. The toolset also includes an innovative
adaptive time-to-failure (ATTF) prognostic algorithm to generate accurate remaining
useful life (RUL) estimates for detected damaged cables.
An electronic hand-held device to detect degradation, such as a cable kink and/or broken
strands in a multi-strand conductor in SubHDR cables, greatly reduces maintenance costs
and improves mission reliability. SubHDR cables, such as those used in the sensor masts
on the Virginia class of submarines, are integral to the sensor masts, and physical
inspection requires removal and disassembly of the large, heavy masts themselves. The
addition of a prognostic algorithm to provide fast, accurate time-to-failure (TTF) or
remaining useful life (RUL) estimates supports deferred maintenance paradigms: as
opposed to the costly and time-consuming immediate removal and replacement of a mast
that, although the SubHDR cable is damaged, its damage might not be significant enough
to cause cable failure before completion of the next mission. The new electronic detection
method can readily be adapted for commercial use, including multi-strand cables that are
not submersible or integrated in antennas.
Key words: Diagnostics; prognostics; health management; submersible high data
rate cable; HDR; SubHDR; remaining useful life; time-to-failure
Introduction: SubHDR cables, such as those used in the sensor masts on the Virginia
class of submarines (as shown in Figure 1), are physically integrated in the sensor masts.
Current electronic checking methods are limited to only continuity and resistance
measurements, and unless the wire is completely broken, continuity can still be verified.
The current methods do not detect whether any strands in a wire are broken or whether a

coaxial conductor is fractured. Broken strands in a wire could lead to cable failure in a
matter of days or weeks, depending on the severity of the wire fatigue.

Figure 1: Virginia class submarine with sensor mast attached to the sail
Periodic inspection of outboard multi-conductor cables with a means to identify potential
failures allows the U.S. Navy to predict the service life and replace cables before a
complete open-circuit failure occurs during at-sea operations. Physical inspection is not
an acceptable detection method because doing so requires the removal and disassembly
of the large, heavy sensor masts themselves. A hoisted cartridge system of sensor mast is
shown in Figure 2).

Figure 2: Hoisting a universal modular mast (UMM) cartridge

What the Navy requires is a personal hand-held, ruggedized device capable of
pinpointing potential failures in outboard cables that are subjected to harsh environments
day in and day out, including submersion, extreme temperatures, high hydrostatic
pressure, bending, or any other external force that decreases useful life. Furthermore, the
inspection method(s) must be non-invasive. The Navy wants the solution set to include an
accompanying prognostic algorithm to accurately predict the remaining service life of
damaged, but not yet failed, outboard cables; remaining service life is also called
remaining useful life (RUL).
A personal hand-held device will provide the methods and means to support periodic
inspection of outboard multi-conductor cables with the capability to identify and locate

potential failures without having to remove and disassemble the sensor mast. The
proposed prognostic algorithm will provide the Navy with an RUL capability (accurate
estimation of the remaining service life). Cables can be replaced before a complete opencircuit failure (severed cable) occurs during at-sea operations.
The innovation described in this paper will meet the Navys need: a non-invasive method
to detect non-open-circuit damage and an RUL calculation engine coupled with an
advanced prognostic algorithm to meet a Navy desire for a service life prediction
capability, and provide a damage location capability. These methods, housed in a handheld tool, will provide a significant improvement in maintenance methods, and reduce the
Navys support costs.
Background: Submersible high-data rate (SubHDR) cables, such as those used in the
sensor masts on the Virginia class of submarines (as shown in Figure 1 and Figure 3), are
physically integrated in the sensor masts. A lingering malfunction is that they are subject
to kinking and eventual open-conductor failures. The cable is drawn through a roller
mechanism as the mast is raised and lowered, and a kink occurs when a cable twist forms
due to torsion and tension action, and then tightens under subsequent tension increase.

Figure 3: Mast to cable-roller relationship

One antenna-mast cable consists of three twisted, stranded conductors within a grounded,
braided mesh: 19 copper strands comprise each conductor. Figure 4 shows a section of
the three conductors with the grounded mesh removed; the circled area shows an example
of the type of kink that occurs [1]. The damage progresses until the wire strands within
the conductor begin to fracture (see Figure 5), eventually resulting in a complete open.
The cables are failing much sooner than the expected life of five years.

Figure 4: Twisted triplet 24-AWG segment showing a kink (grounded shield is removed)

Figure 5: Microscopic camera view of a fracture failure

Current electronic checking methods are limited to continuity and resistance
measurements, which may still be verifiable unless the wire is completely broken.
Current assessment methods do not detect whether any strands in a wire are broken or a
coaxial conductor is fractured. Periodic inspection of outboard multi-conductor cables
with a means to identify potential failures allows the Navy to predict the service life and
replace cables before a complete open-circuit failure occurs during at-sea operations.
Physical inspection is not an acceptable detection method because doing so requires the
removal and disassembly of the large, heavy sensor masts.
This paper will discuss Ridgetops two innovative approaches, which are discussed in the
following sections.
Sample Mode Response Technique (SMRT) Sweeping Probe and Sensor
Processing Unit (SPU): SMRT is a method of applying a stimulus signal to a device or
unit for a short period of time (the sample), capturing the response to the stimulus, and
using digital signal processing techniques to characterize the response, then extracting
eigenvalues that comprise a fault-to-failure progression (FFP) signature (see Figure 6).
The SMRT sweeping probe and SPU is a replacement solution to a time domain
reflectometry (TDR) method, which determines whether the cable is damaged and
provides information about some of the damage locations.


Cable: Digital Signals
Amplified RF reflections

SubHDR Cable Type


Operator Input
Electronic Input

SPU Display

E value
SoH %

Sensor Processing Unit

------------------- Digitize Received RF
Save Digitized data
Perform FFT (DFT)
Save FFT output
Perform IFFT
Save IFFT output
Perform DSP Analysis
Find Reflections
Calculate E = ER/EI
Calculate SoH
------------------- Cable Specifications
Last Adapted Model(s)
Saved Files

[time E] to ARULE
Bulk Data Download to PC

Figure 6: SMRT Sweeping Probe and SPU

This innovative approach1:
1. Provides the same functionality as TDR.
2. Overcomes a deficiency in the TDR approach. The TDR approach requires a great
deal of transmission energy to identify all of the fault locations that might be
present in, for example, a 50-foot cable.
3. Is a proven technology:
o it uses a well-known frequency sweep in the GHz range
o it uses well-known signal amplification and analog-to-digital data conversion
o it uses well-known (digital) fast Fourier transforms (FFT) and the well-known
inverse fast Fourier transforms (IFFT).
4. Uses the easy-to-characterize E-ratio as an FFP signature. This method is easy
compared to using either the TDR reflection amplitudes or time locations.
5. Overcomes the inability of TDR to locate a large number of damage locations. In
the tests we performed, using commercial TDR equipment, we were able to
identify only three of the eight damage locations in an 11.5-foot cable sample
provided by the Navy.
Adaptive Remaining Useful Life Estimator (ARULE) Program and the
Advanced Time-to-Failure (ATTF) Algorithm2: ARULE [3] is a program
consisting of an application-specific front-end module that interfaces to an ATTF kernel
module (see Figure 7) to adapt a fault-to-failure progression (FFP) signature model to the
data, and uses the adapted model to produce estimates of the RUL of the device,

Patent applied for.

Patent pending final approval.

component, or assembly to which the FFP refers. The ARULE program and the ATTF
algorithm support the creation of a model using E-ratio data, and they verify that the
model is correct.
From SPU

SoH: 75%
RUL: 29


Retrieve Data

ATTF Kernel


ARULE Program


Figure 7: ARULE and the ATTF kernel

This method:
1. Provides a real-time, condition-based remaining service life capability to the
2. Is a proven technology. ARULE, with an earlier version of a prognostic
algorithm, has been delivered to the U.S. Army for use with power supplies. The
enhanced prognostic algorithm (ATTF) has been fully tested and benchmarked.
Experiment and Results: The design and architecture of the solution determines the
state of health (SOH), the number and locations of damage, and estimated RUL of
SubHDR antenna-mast cables. This solution set is a SMRT Sweeping Probe connected to
sensor processing unit (SPU), which sends FFP signature data to an ARULE program
with an ATTF RUL estimating algorithm. The SMRT Sweeping Probe, under the control
of the SPU, generates a sweeping radio frequency (RF) of up to 2 GHz into the antennamast cable (see Figure 8). Damage in the form of kinks, fraying, crushing, and broken
conductor strands introduces a change in impedance at the damage locations, and
impedance change causes RF waves to be reflected back to the source.
The RF reflections are captured, amplified, and sent by the probe to the SPU. The SPU
then digitizes and saves the amplified response from the probe and uses digital signal
processing functions, such as FFT (fast Fourier transform) and IFFT (inverse FFT), and
other DSP processing to find the locations and amplitudes of reflected energy. The SPU
then calculates an energy ratio, E, of the maximal reflected energy to the transmitted

Figure 8: SMRT Sweeping Probe block diagram

Experiments were conducted on cables provided to Ridgetop by the Navy. A damaged
cable had eight visible kinks. Figure 9 shows the plots of inverse fast Fourier transforms
for an undamaged cable (with unmatched input and output impedances) on the left, and
for a damaged cable on the right.

Figure 9: Plots of the IFF transform undamaged cable (left) and damaged cable (right)
The results and analysis of the experiment on an undamaged cable exhibit three peaks,
each caused by the characteristic inductive impedance of the cable, the load, and the
reflected signal. An undamaged cable is expected to produce three such peaks. The
undamaged cable was a twisted three-wire conductor with 18 AWG stranded wires.
The results and analysis of the experiment on the damaged cable show that there is the
incident peak of the input (blue circle) and nine reflection peaks (red circles): one for the
end of the cable and eight for each kink in the cable.
The SPU calculated maximal value of E and the time of the sweeping sample are sent to
ARULE to calculate an estimated RUL. For quick diagnostic purposes, the SPU will send
digitized IFFT data to a small display along with human-readable values of the E-ratio, a
state-of-health assessment, and a pass, damaged, or failed condition. The SPU saves data
for both processing purposes and for downloading to support post-processing analyses.
Prognostic Results: As a first step in preparing ARULE for making RUL estimations for
SubHDR cables, we plotted the candidate FFP signatures. We then selected one of the
three candidate signatures and created a prototype model. We then created a test data set
for ARULE modeling. We used the prototype model in ARULE to produce RUL
estimates and verified the modeling by processing the data collected for cut cable strands
at the 1.2 foot and 6.4 foot locations.

Figure 10 shows the plot of each of the three candidate FFP signatures, and the plots
reveal all three signatures have a very similar characteristic shape, which is good because
it simplifies our choice of which to use as the FFP signature.

Figure 10: Plot of the three candidate FFP signatures

There were three candidate measurements for use as FFP signatures: Maximum reflected
energy (Rmax), the ratio of reflected energy to incident energy (I-ratio), and the ratio of
transmitted energy to reflected energy (E-ratio). We selected E-ratio because it is a
normalized ratio unlike Rmax, which uses absolute dB measurements or the I-ratio,
which is not normalized with respect to amplitude. The E-ratio is a normalized,
dimensionless value with 1.0 as the maximum value. A test data set was then created (see
Figure 11) for ARULE modeling, using averaging and extrapolation, so we could analyze
the accuracy of the model and because the two data sets from our investigative
experiments were somewhat sparse. As seen, the test data set (squares) retains the
characteristic shape of the FFP signature and the amplitude and time [event] information.

Figure 11: Plot of experiment (circles and diamonds) and test (squares) data

After creating the model definition in our ARULE front-end program, we ran the test.
Figure 12 shows the input data set, the floor threshold (bottom horizontal line), the
ceiling threshold (top horizontal line), the input model definition (blue line), and the
adapted model (pink line). As seen, and as expected for a fitted model, there is only a
slight adjustment between the initial model and the adapted model.

Figure 12: Test data set showing floor, ceiling, and models
Figure 13 shows the plots of the RUL estimates and the differential nonlinearity (DNL)
for those estimates. The results are a positive DNL of 1.9%, a negative DNL of -5.3%,
and an overall DNL of 7.2%. The small red circle on the horizontal line of the RUL plot
is the time-at-failure point.

Figure 13: RUL and DNL plots for the SubHDR test data

The last step was to verify the original data can be accurately processed by ARULE using
the model we created. Figure 14 shows the data collected from cutting strands of the
cable conductor at the 1.2-foot location.

Figure 14: E-ratio data for the 1.2 cut location

This is a very sparse data set, yet the RUL estimates are very accurate with a total DNL
of only 2.0% as seen in Figure 15.

Figure 15: RUL estimates and DNL for data from the 1.2 cut location
Figure 16 shows the E-ratio data collected from cutting strands of the cable conductor at
the 6.4-foot location. This is also a very sparse data set.


Figure 16: E-ratio data for the 6.4 cut location

As seen in Figure 17, the RUL estimates are also accurate: the total DNL is only 3.5%.

Figure 17: RUL estimates and DNL for data from the 6.4 cut location
Conclusion and future developments: The authors identified and verified the feasibility
of developing two non-invasive significant and innovative approaches to providing the
Navy with hand-held device(s) capable of pinpointing potential failures in outboard
cables that are subjected to harsh environments including submersion, extreme
temperatures, high hydrostatic pressure, bending, or any other day in and day out external
force that decreases useful life. In addition we presented a prognostic algorithm to
accurately predict the remaining service life of damaged, but not yet failed, outboard
cables. The solutions and the prognostic algorithm options are the following: (1) SMRT
Sweeping Probe connected to sensor processing unit, and (2) ARULE program using an
ATTF algorithm.
As we move forward toward the goal of providing a hand-held device for naval
personnel, some of the near-term future work includes the design and simulation of a
prototype probe device and processing unit that differs from the vector network analyzer

(VNA) equipment used in our s-parameter experiments described in this paper. The
significant differences are the following:
1. Reduced complexity
2. Reduced disk support
3. Reduced graphical display support
4. Reduced weight, power, and cost. The previously described reduced functionality
and support translates into reduced weight, reduced power consumption, and
reduced cost3 to design, build and test.
5. Increase in ease of use. By reducing the complexity and display options, and by
using built-in tables and calculations, our SMRT Sweeping Probe and SPU will
offer naval maintenance personnel equipment that does not require RF and
network expertise in order to operate the unit and understand the results.
6. Cables and connectors will be designed for below-the-hull SubHDR connectivity
instead of very expensive BNC coaxial connections.
[1] Bryan Hoffman, NAVSEA technical point of contact for this SBIR program,
provided us with pictures and an 11.5-foot sample of a kinked, 3-conductor cable.
[2] B.P. Lathi, Modern Digital and Analog Communication Systems; HRW Series in
Electrical and Computer Engineering. 1983
[3] Time Domain Reflectometry Theory; Application Note 1304-2, Agilent Technologies
[4] James Hofmeister and Sonia Vohnout; Adaptive Remaining Useful Life Estimator;
ISHM 2009
[5] Kojovic, L.A.; Williams, C.W., Jr.; Sub-cycle detection of incipient cable splice
faults to prevent cable damage; Power Engineering Society Summer Meeting, 2000.
IEEE Volume 2, 16-20 July 2000
[6] Cooper, E.S.; Dissado, L.A.; Fothergill, J.C.; Application of thermoelectric aging
models to polymeric insulation in cable geometry; Dielectrics and Electrical
Insulation, IEEE Transactions on [see also Electrical Insulation, IEEE Transactions
on] Volume 12, Issue 1, Feb. 2005
We greatly appreciate the assistance and support given to us by our Navy technical points
of contact during initial Small Business Innovation Research (SBIR) contract # N6553810-C-0039 to design and develop this technology.
James Hofmeister is a senior principal research engineer and engineering
manager. He has been a software developer, designer and architect for IBM and
represented IBM as a member of the board of directors of the Southern Arizona Center

The full-function Agilent VNA equipment we used in our s-parameter experiments has a base price of
approximately $125,000 with options ranging to $5,000 and more. And that equipment has no prognostic or
state-of-health capability.


for Software Excellence. At Ridgetop Group, he is a principal investigator and lead

design engineer, specializing in analog and digital circuit designs for electronic
prognostics. He is a co-author on six U.S. patents (three IBM and three Ridgetop), two
other pending Ridgetop patents, and four filed invention disclosures. He retired from
IBM in 1998 after a 30-year career and joined Ridgetop Group in 2003. He earned a BS
in electrical engineering from the University of Hawaii, Manoa Campus, and an MS in
electrical and computer engineering from the University of Arizona.
Sonia Vohnout earned her MS in Systems Engineering from the University of
Arizona in Tucson. With a diverse background and experience, Sonia is well-suited to
manage Ridgetops commercialization efforts from its many government-funded projects.
Sonia joined Ridgetop after successfully building an electronic subassembly business in
Mexico, working as a Systems Engineer at IBM, and handling overseas installations of
software with Modular Mining Systems (now part of Komatsu). During her career, she
has held executive management and senior technical positions. In addition, Sonia has cofounded several companies. Sonia is a board member of the Society for Machinery
Failure Prevention Technology (MFPT) (www.mfpt.org), an interdisciplinary technical
organization strongly oriented toward practical applications. Sonia recently founded the
Prognostic and Health Management (PHM) Professionals LinkedIn Group
(www.linkedin.com), a fast-growing group whose objectives are to discuss PHM-related
topics, network with others in the PHM community, and increase awareness of PHM.