2
FIELD OF INVENTION
The present invention generally relates to a method of fault identification in
rotating mechanical components of a mechanical system. More particularly, the
invention relates to an apparatus for monitoring and detecting defects in rotating
mechanical components of a mechanical system, in particular remote monitoring
of the rotating mechanical components in their non - intrusive disposition for
defect detection.
BACKGROUND OF THE INVENTION
Fault identification in a mechanical component is generally carried out using
various condition monitoring techniques such as vibration signature analysis,
acoustics signature analysis, infrared thermal imaging, and wear and debris
analysis. These techniques require in-situ data acquisition but further processing
of data subsequently in the laboratory. Hence real-time fault detection is difficult
to be carried out by the prior art devices. Moreover, mounting of transducer;
such as accelerometer and microphone in case of vibration and noise analysis
respectively, poses problem due to the remotely located or interiorly installed
machinery or components. In a machine, if several components are made of the
same material, then identifying the faulty component is difficult by wear and
debris analysis.
When an induction motor drives a machinery or components, the motor current
signature analysis (MCSA) can be generally helpful in monitoring defects. The
characteristic frequencies of such mechanical components are reflected in the

3
motor current signature as sidebands across the supply line frequency of the
induction motor. The advantage of such a technique is that a real-time fault
diagnosis can easily be carried out by monitoring the current signal from a
remote location. Motor current signature analysis has already been used in
health monitoring of induction motor and its bearings, motor operated valves,
and multi-stage gearbox. However, there are disadvantages in this technique, for
example, some of the frequency components of the line current waveform gets
altered through the current transformer and the associated current to voltage
converter. Nonlinearity in circuits due to current transformer saturation during
overload or fault condition substantially modifies the frequency contents.
The published documents for example, by Wear & Debris and Contaminant
analysis [1], Prabhakaran & Jagga [2] have disclosed use of contaminant
analysis to characterize wear in a turbine-generator. Vibration including various
process data such as temperature and pressure have also been used for turbine-
generator monitoring [3]. The publications by Martin and Duffeau [4] teaches an
analysis of the modes of vibration in order to investigate the end-winding failure.
The publication by H. Ma and others [5] teaches an insulation deterioration
monitoring which has been one of the most important aspect in generator fault
analysis. Insulation problem will lead to the partial discharge (PD) in the
generator, which is originated from cavity, end windings and broken strand in
stator winding. Partial discharge is measured by the broad-band current
transducers that are clipped to the neutral wire of the stator. Wu and Park [6]
have also suggested a system for PD measurement for better signal to noise
ratio as the noise dominates the partial discharge.

4
Yonggang et al. [7] have disclosed a process of monitoring the rotor inter turn
short-circuit by determining the relative level of exciting current. The disclosure
further teaches monitoring higher harmonics of the stator circular current and
voltages for detecting this fault.
The prior art disclose a number of techniques for detection of faults in
mechanical components such as gears, bearings, turbines, pumps etc. For
example, the publication by P. J. Dempsey [8] suggests a method of monitoring
defect in a gearbox, the techniques used are wear and debris analysis. Noise
signature analysis [9], acoustics emission analysis [10], vibration signature
analysis and motor current signature analysis [11-13] published respectively by
N. Byder and others, T. Toutountzakis and others, and C. Kar and A. R. Mohanty,
are also being used in detection of faults in mechanical components.
Current signature analysis is another frequently used technique for monitoring
defects in an induction motor and its bearings as published by R. Yacamini and
others and L. Eren and others [14-15].
US patents numbers US 5629870, US 57425222, US 4965513, US 2384,
US 6727725 and 6709240 although teach various aspects on motor current
signature analysis (MCSA), the areas of these disclosures are however, induction
motor [16-17], bearings of the induction motor [18], motor operated valves [19],
Fuel pumps [20], and centrifugal pumps [21]. There are also a number of
improved processing techniques for MCSA disclosed in US 6199023, US 5461329
and US 4978909.

5
For monitoring AC generator faults such as open and short circuit faults, current
of the three phases were analyzed as disclosed in EP 352951 Bl [25]. The
fundamental frequency and its harmonics of the voltage output of an alternative
tacho-generator and starter generator in order to monitor aircraft components
has been taught by US 5483833 [26].
However, Williams [25] disclosed the method of monitoring the three phases of
current supplied by an AC generator for fault detection. These faults are open
and short circuit faults in the AC generator. Thus, this reference fails to teach the
method of detecting fault in other mechanical systems.
Similarly, Dickens et al. [26] have studied fundamental frequency and its
harmonics of the voltage supplied by AC generator for studying aircraft faults
such as imbalance in turbo-generator. But attention has not been focused on the
characteristics frequencies of the other mechanical components such as blowers,
pumps, bearings and automotive transmission gearbox. The ripple current or
voltage over a DC component of any DC tacho-generator or DC generator has
not been considered.
Yonggang et al. [7] monitored the excitation current for monitoring fault in that
generator only. This fault is rotor inter turn short-circuit of the generator.
There are a number of limitations of MCSA for example, in case of unbalanced
and non-sinusoidal supply voltage, MCSA does not provide faithful information.
Furthermore, some frequency components are contaminated due to the current
transformer and nonlinearity of current to voltage converter circuits.
The reference of the background of the prior art are briefly described and
enclosed with the specification.

6
OBJECTS OF THE INVENTION
It is therefore an object of the invention to propose an apparatus for monitoring
and detecting defects in rotating mechanical components of a mechanical
system, in particular remote monitoring of the rotating mechanical components
in their non-intrusive disposition for defect detection which can monitor on-line
from the control room the defects in the mechanical components.
Another object of the invention to propose an apparatus for monitoring and
detecting defects in rotating mechanical components of a mechanical system, in
particular remote monitoring of the rotating mechanical components in their
non-intrusive disposition for defect detection which can conveniently adapt
transducers and noise monitoring respectively in respect of the mechanical
components.
A further object of the invention to propose an apparatus for monitoring and
detecting defects in rotating mechanical components of a mechanical system, in
particular remote monitoring of the rotating mechanical components in their
non-intrusive disposition for defect detection which efficiently detect the faults in
mechanical components even in case of unbalanced and non-sinusoidal supply
voltage.
A still further object of the invention to propose an apparatus for monitoring and
detecting defects in rotating mechanical components of a mechanical system, in
particular remote monitoring of the rotating mechanical components in their

7
non-intrusive disposition for defect detection which can overcome the possibility
of contamination of frequency components due to the current transformer and
non-linearity of current to voltage converter circuits.
An yet further object of the invention to propose an apparatus for monitoring
and detecting defects in rotating mechanical components of a mechanical
system, in particular remote monitoring of the rotating mechanical components
in their non-intrusive disposition for defect detection which is adaptable to any
mechanical system including I. C. Engine, Pump an Rotors.
An yet another object of the invention to propose an apparatus for monitoring
and detecting defects in rotating mechanical components of a mechanical
system, in particular remote monitoring of the rotating mechanical components
in their non-intrusive disposition for defect detection which overcome the
disadvantages of prior art systems.
A still another object of the invention to propose an apparatus for monitoring
and detecting defects in rotating mechanical components of a mechanical
system, in particular remote monitoring of the rotating mechanical components
in their non-intrusive disposition for defect detection which is easy to handle and
cost effective.
SUMMARY OF THE INVENTION
An electric motor drives various mechanical systems such as pumps,
compressors, gearbox etc. This invention deals with monitoring defects in these
rotating mechanical components by the measurement of a developed ripple in

8
the voltage or current caused by using a small DC/AC generator which is
connected to the rotating shaft. Spectrum of the ripple voltage developed by the
DC generator and/or alternator contains information of the characteristic
frequencies of the mechanical components. These characteristic frequencies are
rotating shaft frequencies, gear mesh frequencies in a multi-stage gearbox,
inner-race, outer race and ball spin frequencies in bearings, vane pass frequency
in blowers or compressors, and pump impeller frequency in pumps. The ripple
superimposed in the DC supply of the DC generator contains all the characteristic
frequencies along with the 50 Hz line frequency. In case of a balanced
alternator, these frequencies appear as sidebands across the generated
frequency which is same as the shaft speed. At higher loads, the distinction of
defective condition is easier than that in low load condition. Besides the
spectrum analysis, other signal processing techniques used in these invention are
wavelet transform and a corrected multiresolution fourier transform. A real-time
embedded digital signal processing (DSP) platform has been developed to
provide alarm whenever a defect is sensed by the sensor.
According to the invention, the apparatus comprises atleast one electromagnetic
device, a signal conditioning module, and a real-time embedded DSP platform.
The atleast one electromagnetic device is one of a DC tacho generator, a DC
generator and an alternator. The atleast one electromagnetic device is connected
to the output shaft of any mechanical system and the supplied ripple current and
ripple voltage are acquired in the DSP platform. By applying corrected
multiresolution fourier transform (MFT) with constant window, the characteristic
frequencies of the ripple current or voltage are highlighted. A corrected
multiresolution fourier transform with moving window tracks the energy
concentration around a frequency bandwidth. Whenever there is a defect or

9
changes in load, the amplitude levels of the characteristic frequencies present in
the mechanical system will change, which is tracked by the apparatus and which
is representative of the presence of fault in a component of the mechanical
system which is indicated in the form of a display board, or an alarm indicator or
through network interface.
A defect-free mechanical component gives a baseline voltage or current ripple
whereas the same component with defects indicate a change in the amplitude of
some specific frequencies of the ripple current or voltage. Those frequencies
represent the general characteristics frequencies of the mechanical components,
for example, inner race, outer race or ball spin frequencies for a bearing, various
gear mesh frequencies for a gearbox, vane pass frequency for a blower or
compressors, and blade pass frequency of the impeller of a pump. Even for a
separately excited generator, the supply line frequency of the exciting current
plays an important role in the fault diagnosis of a mechanical component.
BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWING
Figure 1 - Shows the developed electromagnetic system having real-time
embedded DSP platform.
Figure 2a - Depicts a line diagram of a test set-up for testing a multi-stage
gearbox.
Figure 2b - A schematic details of the test set-up of figure 2(a) indicating the
instrumentations employed.
Figure 3 - Line diagram of the multi-stage gearbox used in the test set - up.

10
Figure 4 - A block diagram of another test set-up for testing a bearing with
instrumentations employed.
Figure 5 - A flow diagram depicting decomposition of ripple current / voltage
signals using Discrete Wavelet Transform.
Figure 6a - Shows the frequencies in the Approximate 4 level of the
decomposed ripple voltage supplied by DC tacho-generator in respect
of defect-free gears (dO) having undergone a corrected
multiresolution fourier transform with a constant window.
Figure 6b - Shows the frequencies in the Approximate 4 level of the
decomposed ripple voltage supplied by DC tacho-generator in respect
of two-teeth missing in one gear (dl) having undergone a corrected
multiresolution fourier transform with constant window.
Figures 7a & 7b - Show the frequencies similar to figures 6a and 6b, but at
details 2 level of decomposed ripple voltage.
Figure 8 - Shows a bar chart indicating variation in amplitude of two
characteristics frequencies (f3 and fm) of the ripple voltage in respect
of defective gears, the ripple voltage being supplied by the D.C.
tachogenerator.

11
Figures 9a & 9b - Show the frequencies of the decomposed ripple voltage
supplied by DC generator connected to the gearbox after having
undergone a corrected MFT with constant window of approximate 4
level in respect of defect-free gear (dO) and two-teeth missing in
one gear (dl).
Figures 10a & 10b - Show the frequencies of the decomposed ripple voltage
supplied by DC generator connected to the gearbox after having
undergone a corrected MFT with constant window of details 2 level in
respect of defect-free gear and two-teeth missing in one gear.
Figure 11 - Shows similar bar chart of figure 8, the ripple voltage being supplied
by the DC generator with characteristic frequencies f3 and fe.
Figures 12a & 12b - Show frequencies under similar conditions, of figures 6a and
6b the ripple voltage being supplied by an alternator.
Figures 13a & 13b - Show frequencies under similar conditions, of figures 7a and
7b the ripple voltage being supplied by an alternator.
Figure 14 - Shows similar bar chart of figure 8, the ripple voltage being supplied
by alternator.
Figures 15a & 15b - show the frequencies of ripple voltage supplied by an
alternator in respect of defect-free gear (dO) and two-teeth missing
in one gear (dl), having undergone a corrected MFT with moving
window of details 2 level.

12
Figures 16a & 16b - show the frequencies of the decomposed ripple current
supplied by DC generator connected to the gearbox having two
missing teeth in the second gear, having undergone MFT with
constant window at approximate 4 levels and details 2 levels
respectively.
Figures 17a & 17b - Show the frequencies of the ripple voltage supplied by DC
tacho-generator connected to a bearing test set-up for defect-free
bearing and one scratch in the outer race of the bearing, having
undergone FFT-based Hilbert transform.
DETAILED DESCRIPTION OF THE INVENTION
A real-time embedded digital signal processing (DSP) platform (item 1); as
illustrated in FIGURE 1; has been developed to provide alarm whenever a
defect is sensed by the sensor. Referring to FIGURE 1, the electromagnetic
device may be a DC tacho generator, an alternator or a DC generator. The
signals of signal conditioning module (item 2) represents the ripple voltage or
current supplied by one of these electromagnetic devices, which are acquired in
a DSP board (4) through an A/D converter (3). The embedded DSP board (4)
has three choices for example, either to display (6) the peak amplitude of any
characteristic frequency or to glow an alarm (5) whenever there is a large
change in this peak level or to send information through network (7).

13
Although the invention has been described in reference to testing of a gearbox
test rig; illustrated in FIGURE 2; having a multi-stage gearbox shown in
FIGURE 3, the present invention is applicable for other mechanical components
including the bearing as described in respect of a bearing test rig illustrated in
FIGURE 4. The test rig shown in FIGURE 2a comprises of an induction motor
(8) that drives a multi-stage gearbox (9), which is connected to a DC Generator
(10) for loading purpose (11). An alternator and a DC tacho generator (12) are
also connected to the gearbox. FIGURE 2b illustrated the sensors and
instruments used and signal conditioning steps used in the testing. Vibration
signature from an accelerometer (13), speed signal from a photoelectric probe
(14), and current and voltage signals from the electromagnetic devices (12) are
acquired using a current probe (15) and a change amplifier (16). All the data is
stored in DAT recorder (17) which are acquired through an antialising filter (18).
The details of the gearbox (9) is shown in the line diagram of FIGURE 3 or 21.
The gearbox is such that there are three rotating shaft frequency and three gear
mesh frequencies because of the presence of a counter shaft and synchro-
meshed conditions.
As illustrated in FIGURE 4, an induction motor (22) rotates a bearings (23), to
which again a DC generator or alternator (12) is connected. The photoelectric
probe (14) is connected to a frequency counter to know the speed. A second
accelerometer (24) is also used to acquire vibration data. The current or voltage
from the electromagnetic device (12) is also acquired using the antialising filter
(18) and the P.C. (20).

14
The DC component is eliminated using the AC coupling of the filter (18).
Moreover, the bandwidth of the signal is kept from 1 Hz to 2 kHz. 8192 number
of data of this ripple current and voltage are acquired from both the test rig with
a sampling frequency of 4.096 Hz. The testing is done after introducing one
artificial defect each in the gearbox and in the bearing. The artificial defect
introduced in the gearbox is two teeth missing in 2nd main gear, whereas the
same in bearing is one scratch in inner race. These defects are referred as dl
defect whereas the defect-free case is termed as dO in the respective test rig.
FIGURE 5 depicts a discrete wavelet transform (DWT) tree where the ripple
current or voltage signal are decomposed into 4 number of levels, where all the
rotating shaft frequencies and gear mesh frequencies are confined to
Approximate 4 level and Details 2 level respectively. Hence, the corrected
multiresolution fourier transform (MFT) will be applied to the signals at these
levels. In case of the bearing test rigs, all the frequencies will lie in the
Approximate 2 level. For applying corrected MFT, the scaling is done as per
FIGURE 5. Then the resulting signal in that level is convolved with a hanning
window for the whole time record to result in constant window MFT. However, if
a hanning window is moved with some overlapping, then it is known as moving
window MFT.
The result of the MFT with constant window applied to Approximate 4 level of
the ripple voltage supplied by the DC tacho generator connected to the gearbox
is illustrated in FIGURE 6. The operating load is 5.625 kW. All the rotating shaft
frequencies such as input shaft (fi), lay shaft (f2) and output shaft (f3) are noted.
There are also prominent harmonics of the output shaft (f3). Similarly in FIGURE

15
7, all the gear mesh frequencies could be found. A number of sidebands of
rotating shaft frequencies also appear across the gear mesh frequencies. The
amplitude levels of these frequencies for dO and dl shown in FIGURE 6 a-b and
FIGURE 7 a-b. The summary of this variation for two frequencies such as f3
and fm2 are illustrated in FIGURE 8. There is 56% and 89% decline in these
amplitudes for defective cases.
FIGURE 9, FIGURE 10 and FIGURE 11 illustrates the MFT with constant
window for ripple voltage of DC generator. The DC generator is found to possess
a supply line frequency of 50 Hz (fe), the reason of which can be attributed to
fact that it is a separately excited generator drawing current from the main
supply of 50 Hz. The summary reveals that there is a maximum rise in amplitude
levels of f3 and fe in the respective order of 123% and 1328%.
FIGURE 12, FIGURE 13 and FIGURE 14 illustrates the MFT with constant
window for an alternator. The alternator is found to possess sidebands across
the output shaft speed (f3), the reason of which has been explained analytically
in Ref. [11-13] for an induction motor. Another important revelation is that the
higher frequency region i.e. Details 2 level is interspersed with smearing of
energy and noise even after applying constant window MFT, the reason of which
can be attributed to the slip while connecting to the gearbox. Therefore a
moving window MFT is applied as illustrated in FIGURE 15. The summary in
FIGURE 14 reveals that there is a decline in amplitude levels of f3 and 2f3 in the
respective order of 12% and 50% respectively. FIGURE 15 shows that the
energy is distributed across all the frequencies in the 500-1000 bandwidth
(Details 2 level) for defect-free gears (FIGURE 15a), whereas an abrupt rise in
energy is focused around 2nd gear mesh frequency (FIGURE 15b). The energy
difference is more that 4 times in favor of defective case.

16
FIGURE 16 a-b show the result of the constant window MFT for the
approximate 4 and Detail 2 level of the ripple current of the DC generator in dl
defect case respectively.
FIGURE 17 a-b illustrate the FFT based Hilbert transform of ripple voltage from
the DC tacho generator connected to the bearing in the bearing test rig
illustrated in FIGURE 4. The ball spin frequency and resonant frequency could
be detected clearly. The other bearing frequencies such as inner race and outer
race; which have negligible difference with the harmonics of the rotating shaft;
have also been traced for both defect-free and defective cases.
This technology can be extended to monitoring defects in other mechanical
components such as IC engines, pumps, blowers etc. where the tacho-generator
can easily be fixed at the output shaft because of its small size and the voltage
output can be investigated for monitoring any defect in these components. A DC
generator can also be used for the same purpose as its output will again be
indicative of the information of defects in the mechanical components.

-17-
REFERENCES
1. H. S. Ahn, E. S. Yoon, D. G. Sohn, 0. K. Kwon, K. S. Shin and C. H. Nam
(1996). Practical contaminant analysis of lubricating oil in a steam
turbine-generator. Tribology International, 29(2), 161-168.
2. A. Prabhakaran and C. R. Jagga (1999). Condition monitoring of a steam
turbine generator through contaminant analysis of used lubricating oil.
Tribology International, 33,145-152.
3. W. Xue and Y. Shuzi (1996). A parallel distributed knowledge-based
system for turbine-generator fault diagnosis. Artificial Intelligence in
Engineering, 10, 335-341.
4. T. Fortin and F. Duffeau (1997). Large generator vibration monitoring.
Proceedings of IEE International Conference on Electrical Machines and
Drives, 155-159.
5. H. Ma, J. Ma, G. Wu, Z. Huang, X. Wu, H. Xie, D. Sun and R. Zhang
(1995). The study of broad-band current transducer system for on-line
monitoring of partial discharges in turbine-generator. International
Symposium on Electrical Insulating Materials, 217-220.
6. G. Wu and D. H. Park (1999). On-line monitoring instrument of fault
discharge in large generators. IEE High Voltage Engineering Symposium,
Publication no. 487, 5.340P5-5.343P5.
7. L Yonggang, Z. Hua and L. Heming (2003). The new method on rotor
winding inter-turn short-circuit fault measure of turbine-generator.
Proceedings of IEEE International on Electrical Machines and Drives
Conference, 3,1483-1487.
8. P. J. Dempsey (2003). Integrating oil debris and vibration measurement
for intelligent machine health monitoring. Glenn Research Center,
Cleveland, Ohio, NASA/TM-2003-2 11307,1-122.

-18-
9. N. Byder and A. Ball (2003). Detection of gear failures via vibration and
acoustics signals using wavelet transform. Mechanical System Signal
Processing, 17(4), 787-804.
10. T. Toutountzakis, C. K. Tan and D. Mba (2005). Application of acoustics
emission to seeded gear fault detection. NOT & E International, 38(1),
27-36.
11. C. Kar and A.R. Mohanty (2005). Monitoring gear vibrations through
motor current signature analysis and wavelet transform, Mechanical
Systems Signal Processing, In Press.
12. C. Kar and A.R. Mohanty (2005). Multi-stage gearbox condition
monitoring using motor current signature analysis and Kolmogorov-
SmirnovTest, Journal of Sound and Vibration, In Press.
13. A. R. Mohanty, and C. Kar (2005). "Fault detection in a multi-stage
gearbox by demodulation of motor current waveform", IEEE
Transactions on Industrial Electronics, In Press.
14. R. Yacamini, K. S. Smith and L Ran (1998). Monitoring torsional
vibrations of electro-mechanical systems using stator currents. ASME
Journal of Vibration and Acoustics Vol.120, 72-79.
15. L. Eren and M. J. Devany (2004). Bearing damage detection via wavelet
packet decomposition of the starting current, IEEE Transaction on
Instrumentation and Measurement, Vol. 53(2), 431-436.
16. S. F. Farag, T.G. Habetler and J. H. Schlag (1997). Method and
apparatus for predicting electric induction motor failure during operation.
US Patent 5,629,870.
17. B. Yazici and G. B. Kliman (1998). Adaptive on line, statistical method
and apparatus for detection of broken rotor bars in motor by passive
motor current monitoring and digital torque estimation. US Patent
5,742,522.

18. H. D. Haynes and D. M. Eissenberg (1990). Motor current signature
analysis method for diagnosing motor operated devices. US Patent
4,965,513, 1990.
19. H. D. Haynes, D. F. Cox and D.E. Welch (2004). Electrical fuel pump
condition monitor system using electrical signature analysis, US patent
2004/023002384A1.
20. L. Eren and M. J. Devany (2004). Motor bearing damage detection via
wavelet analysis of the starting current transient. US Patent 6,727,725.
21. S. C. Schmaltz and R. P. Schuchmann (2004). Method and apparatus of
detecting low flow/cavitation in a centrifugal pump. US Patent
6,709,240.
22. G. B. Kliman (2001). System for removing spurious signatures in motor
current signature analysis. US Patent 6,199,023.
23. D. 3. Linehan, S. L. Bunch and C. T. Lyster (1995). Method and
apparatus for generating motor current spectra to enhance motor
system fault detection. US Patent 5,461,329.
24. D. E. Hendrix, S. F. Smith (1990).Demodulation circuit for AC motor
current spectral analysis. US Patent 4,978,909.
25. D.C. Williams (1994). A.C. Generator fault detector. EU Patent
O,352,951B1.
26. L. M. Dickens, H. D. Haynes and C. W. Ayers (1996). Method and
apparatus for monitoring aircraft components Patent 5,483,833.

-20-
We Claim
1. An apparatus for monitoring and detecting defects in rotating mechanical
components of a mechanical system, in particular remote monitoring of
the rotating mechanical components in their non-intrusive disposition for
defect detection, the mechanical system having a gearbox (9) including at
least one bearing (23) being driven by an induction motor (8, 22), the
apparatus comprising: -
- At least one electromagnetic device (10, 12) for developing a
spectrum of ripple in the voltage or current being connected to a
rotating shaft of the mechanical system, the mechanical system
having a gearbox (9) including at least one bearing (23) being
driven by an induction motor (8, 22), the developed spectrum of
the ripple voltage representing the characteristic frequencies of the
mechanical components;
- a signal conditioning module (2) generating signals representing
the ripple voltage or current being supplied by the at least one
electromagnetic device (10, 12), the signal conditioning module
(2) having a current probe (15) and a charge-amplifier (16) for
acquiring vibration signature from a first accelerometer (13), speed
signals from a photoelectric probe (14), and current and voltage
signals from the electromagnetic device (12);

-21-
- a real-time embedded digital signal processing platform (DSP) (1)
for acquiring the ripple current and voltage being generated via the
monitorable mechanical components of the mechanical system, the
DSP (1) acquiring data from the signal conditioning module (2)
through an antialising filter (18) which is stored in a P.C. (20)
assigned to the DSP, the acquired data being processed by a
processor (4) of the DSP by application of a corrected
multiresolution fourier transform (MFT) with constant window
which generates a first characteristic frequencies of the ripple
current or voltage, the first characteristic frequencies being further
processed by application of a corrected multiresolution fourier
transform with moving window which generates a second
characteristic frequencies of the ripple current or voltage indicating
the defects in the mechanical components, the defects being
notified via one of a display board (6), an alarm indicator (5), and
a network interface (7).
2. The apparatus as claimed in claim 1, wherein the signal conditioning
module (2) optionally comprise a second accelerometer (24) to acquire
vibration data.
3. The apparatus as claimed in claim 1, wherein the signals from the signal
conditioning module (2) are acquired by the DSP (1) via an A / D
converter (3).

-22-
4. The apparatus as claimed in claim 1, wherein the ripple current or voltage
signal are decomposed by a discrete wavelet transform (DWT) technique
in registration with the rotating shaft frequencies and gear mesh
frequencies preferably confined to approximate 4 level and details 2 level.
5. An apparatus for monitoring and detecting defects in rotating mechanical
components of a mechanical system, in particular remote monitoring of
the rotating mechanical components in their non-intrusive disposition for
defect detection as substantially herein described and illustrated with the
accompanying drawings.