FIELD OF THE INVENTION
The present invention relates to a system and a method adapted to monitor and diagnose the
health of a machine. More particularly, the present invention relates to a system and a method
adapted to monitor and diagnose the bearing defect of machine which is running at a low speed.
BACKGROUND AND PRIOR ART OF THE INVENTION
In Industries lot of equipments of huge size are used for regular production. A few of these
equipment move / rotate at a slow speed as per the requirement of the process. The bearings used
by such equipment are of special type. If bearing fails, the out put of equipment is stopped. It
may lead the engineers to stop the production of entire unit as out put of one is the input to the
other equipment. Thus it is desired to detect the incipient problem of the bearing before real
breakdown occurs and plan the repair job at a suitable and convenient time to optimize the
bearing life and make the production hour loss to zero. Detection of fault at the early stage
requires monitoring of equipment through vibration signals and examining it physically or by
instrument.
In recent years, a variety of condition monitoring techniques has been developed to detect the
incipient machine fault. One of the principal tools for diagnosing this kind of faults is vibration
analysis. The general principle is that any machine is subjected to some level of vibration during
its operation. When fault develops, significant deviation in the vibration pattern occurs.
Therefore by employing suitable data processing technique, it is feasible to detect the changes in
the vibration signature caused by faulty component. A time series of the vibration signal can
indicate sufficient difference between healthy and faulty conditions, however, it does not provide
any information about the position of fault at their early stages. On the other hand, a frequency
spectrum analysis of the vibration signal reveals frequency components which can be assigned to
a specific fault type. Traditionally, Fourier transform is used for in-depth analysis of the
vibration signal. However, the accuracy and reliability of this technique depend upon the length
of the data with the requirement that the signal should be stationary. In practice, the vibration
data collected from the machine structure may be non-stationary in nature. Therefore, FFT of the

vibration signal does not provide any information about the presence of drift and abrupt changes
appearing in the vibration signal due to fault/supply variation.
"Simulation of vibrations produced by localized faults in rolling elements of bearings in
gearboxes" by N.Sawalhil and R .B.Randall, 5th Australasian Congress on Applied Mechanics,
AC AM 2007 10-12 December 2007, Brisbane, Australia discloses a simulation model for a
localized fault in a rolling element in one of the bearings of a single stage gearbox. The
simulated fault produced signals, which are reasonably comparable to the measured ones. The
simulated and measured signals reacted similarly to signal processing techniques normally
applied to rolling element bearings (power spectrum comparison and envelope analysis). The
five degree-of-freedom model gearbox test rig is a lumped mass parameter model, which has the
capacity to model different fault types (localized inner/outer race faults and extended faults). The
gear/bearing model takes into consideration the slippage in the bearings, the Hertzian contact and
the nonlinearity of the bearing stiffness (time variant).
"A review of vibration and acoustic measurement methods for the detection of defects in
rolling element bearings" by N. Tandon a,*, A. Choudhury, Tribology International 32 (1999)
469-480 discloses a review of vibration analysis methods for the detection of defects in rolling
element bearings is described in this paper. The vibration detection techniques take into account
the Hertizan's theory and the time domain plot. Further accelerometers are used for the detection
of the vibration signals. But does not talk about the sensitivity of the accelerometer further the
paper gives a general view about the vibration analysis and does not talk about the assembly for
such detection.
US 6,289,735 disclose a diagnostic system for a dynamoelectric machine, comprising: at least
one vibration sensor (accelerometer) mounted on the machine to sense vibration; and a processor
adapted to receive a vibration signal from the at least one vibration sensor, the processor
scanning a vibration signature corresponding to the vibration signal over a series of frequencies
to identify a resonant frequency for a transmission path from a vibration source to the at least one
vibration sensor and to evaluate the vibration signature (using FFT) in the vicinity of the
resonant frequency to determine an operating state of the machine.

US 5,566,092 discloses a fault diagnostic system, comprising a data acquisition module that
collects sensory signals, from the sensors, such signals are then transformed into a time domain
using autoregressive technique.
There are systems available for diagnosing bearing problem. The data acquisition is done at
lower sampling rate (50 K cycles per second max) and the sensor / accelerometer used is of
lower sensitivity (100 -200 mv/g) in those systems. These types of systems are quite useful for
the equipment running at higher speed. But there are not useful for the system running at slow
speed because of the poor sampling rate and lower frequency cut off of the sensor.
The system available in the market consists of data logger accelerometer and analysis software in
built with the data logger itself (in most cases). Standard test / analysis results are indicated by
this type of system. Special tests as one required for equipment running in slow speed are not
permitted. That is why some times recommendation based on it goes wrong. Sometimes testing
over the period i.e. trending may be helpful. But in doing so the bearing may get damaged
severely. Available system is useful in higher speed i.e. running at > 1000 RPM.
The accelerometer supplied with the available system has to be used. No freedom over selection
of accelerometer.
Data logger of the available system has the sampling frequency 20,000 to 30,000 sample per
second. The accelerometer provided is of low sensitivity. These are not suitable for the system
running with slow speed.
Thus there is a need to provide a system and a method to detect bearing defect of machine which
is running at a low speed.
The present inventors have developed system where the above mentioned problems are well
taken care of. The data acquisition card of higher sampling rate (<50 K cycles per second) and
the higher sensitivity (<500mv/g) sensors are used so that no faulty signature escaped.
4

OBJECTS OF THE INVENTION
One object of the present invention is to overcome the disadvantages / drawbacks of the prior art.
A basic object of the present invention is to provide a system which is useful to monitor the
equipment running at slow speed (< 500 rpm).
Another object of the present invention is to provide a system adapted detect the location of
bearing fault and enhancing the bearing life.
Yet another object of the present invention is to provide a system adapted for the detection of
incipient fault helps in eliminating the production hour loss.
Yet another object of the present invention is to provide a method adapted to monitor and
diagnose the bearing defect of machine running at slow speed.
These and other advantages of the present invention will become readily apparent from the
following detailed description taken in conjunction with the accompanying drawings.
SUMMARY OF THE INVENTION
According to one of the aspect of the present invention the system is provided to monitor and
diagnose the bearing defect/fault of a machine/equipment which is running at a low speed, said
system comprising:
a model means adapted to process (simulation) a set of predetermined data(input parameter -
bearing parameter) and output time domain (time- acceleration/amplitude) plot of said
(simulation output) set of data ;

atleast one accelerometer/sensor operatively connected with said machine means adapted to
capture the fault signal of the bearing of said machine;
atleast one data acquisition card means operatively connected with said accelerometer/sensor
adapted to collect data from said sensor means and send the data to said processor means via
suitable cable connections and
an analysis means comprising a processor means adapted to diagnose the time domain data to
monitor and diagnose the bearing defect of said machine.
According to another the aspect of the present invention there is provided a method to monitor
and diagnose the bearing defect of a machine/equipment which is running at a low speed, said
method comprising steps of:
inputting a first set of data to a model means adapted to process (simulation) said inputs;
outputting a second set of time domain (output of simulation) plot from said model means;
analyzing said time domain (data) plots to monitor and diagnose the bearing defect of said
machine and
final outputting a third set of data to monitor and diagnose the bearing defect of said machine
from said time domain (data) plots.
DETAILED DESCRIPTION OF THE INVENTION
The present invention relates to a system and a method adapted to monitor and diagnose the
bearing defect of machine which is running at a low speed.

The system comprises a model that has been developed based on the basic principles of vibration
theory. It generates data for various faults at different conditions and then data is analyzed. Its
basic purpose is to get the fault characteristics of the bearing arising out of defects at outer / inner
race, rolling elements etc. Peak value at the characteristic frequency not only indicates the
presence fault but also indicates the time for replacing depending upon the peak value. The
accelerometer, Ni card (data acquisition card) is used to collect data in laboratory putting the
faulty bearing in different working condition in Machinery Fault Simulator. This data is then
analyzed to detect the fault. Both laboratory data and the experimental data are required to
develop the diagnostic software and subsequently for validation of the entire system.
The model is operated with field condition i.e. bearing parameter, load condition, operating
speed (<500 RPM) etc. The generated data is analyzed to get fault characteristic of that bearing
at the condition.
In experiment the accelerometer is placed in proper place so that noise is minimum from the
drive. Then using the data logger i.e. the NI card data is collected and stored in PC for analysis
immediately or at suitable convenient time. The peak value at characteristic frequency is located
and compare with that found through model and then recommendation is made.
The system can detect the problem arising out of problem like misalignment, unbalance and
damaged bearing for the equipment running at slow speed. Specially it is very useful in industry
for detecting the bearing problem of Converter, Mill stands and other heavy equipment which are
running at slow speed.
Normally vibration at slow speed is of low energy and absorbed by the heavy structure of the
machine. So incipient fault caused at bearing is very difficult to identify through vibration
signature. The model plays here an important role. It is operated with bearing geometry, load,
and speed and defect size. The analysis of the data generated through model gives an idea of
Characteristic frequency and the peak value (amplitude) which helps in finding the incipient fault
and criteria to recommend for undertaking the repair work or replacing the bearing. The data
acquisition card and accelerometer are required to collect the information form field. The

processor i.e. Computer or laptop is used for analysis of information from field as well as from
model. Basically those are the tools for the system.
The vibration generated due to the defect of equipment running at slow speed is of low energy
and part of it is absorbed by the various structural path of the equipment. Capturing this type of
signal and the analysis for detection of incipient fault is very difficult and there is no such system
available in the market. The system described here is capable of this type of signature and it can
detect the incipient fault for the equipment running at slow speed. Specially it works best for the
equipment running a slow speed as low as fraction of RPM also.
Two degrees of freedom are for shaft (inner race) and pedestal (outer race) respectively. Another
degree of freedom is for sprung system i.e. housing. These are the total five degrees of freedom
are considered for the model. The model takes care of lubrication effect, slippage, clearance and
positioning of rolling elements. The model calculates the contact stress and the overall deflection
with various defects in various element of the bearing.
The parameters required for the model are fed and then model generate the data. The data is two
column data i.e. time and amplitude. The mode can also present the data as time domain plot.
These data then denoised and analysis further to detect the fault in it. This helps in getting the
faulty characteristic of the bearing in laboratory as well in actual working field.
Accelerometer is place in proper place to capture the vibration signature with data acquisition
connected in it. This is then transfer to process in PC or laptop/processor to analyze the data and
diagnose the incipient fault with the bearing using diagnostic software developed for it.
According to one of the embodiments of the present invention there is provided system to
monitor and diagnose the bearing defect/fault of a machine/equipment which is running at a low
speed. The system comprises a model means to process (simulation) a set of predetermined field
data(input parameter -bearing parameter) and output time domain (time- acceleration/amplitude)
plot of said (simulation output) set of data ; at least one accelerometer/sensor operatively
connected with machine means to capture the fault signal of the bearing of said machine; at least

one data acquisition card means operatively connected with accelerometer/sensor to collect data
from sensor means and send the data to the processor (PC) means via suitable cable connections;
an analysis means comprising a processor means adapted to diagnose the time domain plot to
monitor and diagnose the bearing defect of said machine.
The processor means of said analysis means comprising a transformer means adapted to break
the signal(data) in five different levels; a envelope means to produce an enveloped signal from
signal produced from transformer means and a spectrum means convert the enveloped signal to
envelop spectrum adapted to detect fault.
According to another embodiment of the present invention there is provided a method to monitor
and diagnose the bearing defect of a machine/equipment which is running at a low speed. The
method comprises steps of inputting a first set of data to a model means to process said inputs;
outputting a second set of time domain plot (data) from the model means; analyzing the time
domain plots to monitor and diagnose the bearing defect of said machine and final outputting a
third set of data to monitor and diagnose the bearing defect of said machine from said time
domain (data)plots.
The step analyzing the time domain plots to monitor and diagnose the bearing comprising steps
of transforming time domain plots adapted to break the signal(data) in five different levels by a
transformer means; enveloping to produce an enveloped signal from signal produced from the
transformer means and converting enveloped signal to envelop spectrum so as to detect fault.
Features:
- Finding the stiffness of the bearing
- Incorporation of defect and its design in the model
- Consideration of some of intricate parameters for industrial use

- Special type of accelerometer (other than sensitivity) used to capture the low energy
signal generated in slow speed running equipment due to presence of defects in
bearing
- Data acquisition with high sampling rate and special type of card used
Special parameter considered in developing the diagnostic software for industrial use
- New technique (no filter) used to denoise the signal
- Combination of techniques used to develop the diagnostic software
BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWING
Other features as well as the advantages of the invention will be clear from the following
description.
In the appended drawing:
Figure 1 illustrates schematic representation of a rolling element bearing with pedestal and rotor.
Figure 2 illustrates the definition of a fault on the outer race.
Figure 3 illustrates definition of the spall and its model.
Figure 4 illustrates geometric representation of ball spall.
Figure 5 illustrates scheme for model data Generation.
Figure 6 illustrates field data Generation scheme.
Figure 7 illustrates field Time Domain Plot (Model Data)
Figure 8 illustrates field Wavelet Transformation (Model data).
Figure 9 illustrates field Wavelet Coefficient Plot (Model data).

Figure 10 illustrates field Envelope Capturing (Model Data).
Figure 11 illustrates field Envelope Spectrum of d3, d4 and d5 levels (Model Data).
Figure 12 illustrates field Time Domain Plot (Field Data).
Figure 13 illustrates field wavelet Transformation (Field Data).
Figure 14 illustrates field Wavelet Coefficient Plot (Field data).
Figure 15 illustrates field Envelope Capturing (Field Data).
Figure 16 illustrates field Envelope Spectrum of d3, d4 and d5 levels (Field Data).
DETAILED DESCRIPTION OF THE ACCOMPANYING DRAWING
In the following detailed description, reference is made to the accompanying drawings that form
a part hereof, and illustrate the best mode presently contemplated for carrying out the invention.
Further functioning of the system and method has been discussed below to describe the way it
operates. However, such description should not be considered as any limitation of scope of the
present system. The structure thus conceived is susceptible of numerous modifications and
variations, all the details may furthermore be replaced with elements having technical
equivalence. In practice the materials and dimensions may be any according to the requirements,
which will still be comprised within its true spirit.
The purpose of the invention is to detect the incipient bearing fault for the equipment running at
slow speed (<500 RPM). The first step in the system, to detect the incipient fault at slow speed is
the development of a model. This is five degrees of freedom model based on Hertzian contact
theory which relates the raceway displacement to the bearing load.

The equations of motions according to the above model (Fig.1) are:

Where the forces fx,fy are calculated according to Hertzian theory (non- linear stiffness) as
detailed in equation (9), g is the acceleration due to gravity and e is the eccentricity of the rotor.
The case frequency is calculated as

The model is developed from the above-mentioned formulations. The output of the model is
basically time domain (time - acceleration) plot. The data of the plot is used as the input of the
programme /software developed for diagnosis of the kind of faults with the bearing.

The model has been developed to detect various faults like defects in inner / outer /ball or rolling
elements. The model is operated for one kind of faults at time. The variables used in developing
the model are listed below

The values for these variables are the input parameter.
Once these basic parameters are fed to the model it produces a set of data i.e. a two column data
of time and amplitude (Acceleration). The model produces this data as plot of time vs. amplitude
(acceleration) which is nothing but the time domain Plot. This is the output of the model. This
data is further processed / analyzed to detect the fault.

The model generated data fed to the developed software embedded /installed in processor as an
input to it for processing and diagnosing the fault. In the first step these data is processed through
wavelet transformation. This breaks the signal (data) in five different levels (field details). Then
required levels are (here it is d3-d5) enveloped. Finally this enveloped signal (data) is converted
to enveloped spectrum. The presence of higher amplitude at fault characteristic frequency in this
envelope spectrum indicates the fault. Figure 7 shows the time domain plot of model data (the
output of model). Figure 8 is the wavelet transformation of the model output or data. Figure 9
shows the five illuminated lines. This indicates the impacts due to defect in bearing. Figure 10 is
the envelope capturing. Figure 11 shows the envelope spectrum. Here three peaks (high
amplitudes) are quite prominent at frequencies 5Hz.10Hz and 15Hz. The characteristic frequency
for this case is 5 Hz and 10 Hz, 15 Hz are its harmonics. The higher amplitudes at characteristic
frequency and at its harmonics are observed which indicate the bearing is damaged.
The use of wavelet transformation and breaking the signal in five different levels is new
technique. The required levels are then further processed for detecting the presence of fault.
Another new aspect is to use wavelet coefficients to detect the occurrence of impacts. This
indicates some damage in the bearing.
The existing system uses the enveloping only to detect bearing fault running at higher speed
(>100RPM). Some time it also suggest the filtering for denoise the undesired signals presents in
the collected information (data).
The Machinery Fault Simulator (MFS) is used for experimental data collection / generation. The
sensor / accelerometer are generally placed at suitable position on bearing housing connected to a
data acquisition card. The data is generated at higher sampling rate (50 kcs - 100 kcs) and is
saved to a PC connected with the card. This is then analyzed using the developed software
embedded in processor of the analysis means as is done for the data generated through the
model to detect the problem
Figure 12 shows the time domain plot of model data (the output of model). Figure 13 is the
wavelet transformation of the model output or data. Figure 14 shows the five illuminated lines.
This indicates the impacts due to defect in bearing. Figure 15 is the envelope capturing. Figure
16 shows the envelope spectrum. Here three peaks (high amplitudes) are quite prominent at

frequencies 5Hz.10Hz and 15Hz. The characteristic frequency for this case is 5 Hz and 10 Hz, 15
Hz are its harmonics. The higher amplitudes at characteristic frequency and at its harmonics are
observed which indicate the bearing is damaged
Figure 7 shows the time domain plot of model data (the output of model). Figure 8 is the wavelet
transformation of the model output or data. Figure 9 shows the five illuminated lines. This
indicates the impacts due to defect in bearing. Figure 10 is the envelope capturing. Figure 11
shows the envelope spectrum. Here three peaks (high amplitudes) are quite prominent at
frequencies 5Hz.10Hz and 15Hz. The characteristic frequency for this case is 5 Hz and 10 Hz, 15
Hz are its harmonics. The higher amplitudes at characteristic frequency and at its harmonics are
observed which indicate the bearing is damaged
In the next phase of the work field data were generated using the Machinery Fault Simulator
(MFS). The simulator was operated at different speed (within slow range) and the bearings with
different combination fault mounted with it. The data were collected using data acquisition card
with higher sampling rate and accelerometer of higher sensitivity so that fault signal doesn't
escape. These data were analyzed using the same developed program /software that were used to
analyze the model data.
The data collected from MFS is used as the input data for the software embedded in the model.
The model analyzes it to detect the fault as follows
- The data is broken in five different levels using the wavelet transformation
- Enveloping is carried out for d3 - d5 levels
- Envelop data is transformed to envelope spectrum.
The plots are shown in Figures 13,15 and 16.
In industry it is always not possible to reach the equipment. In addition to that some time it
becomes difficult or not possible to fix the sensor / accelerometer on equipment. In such cases
remote sensors like laser sensor can be used. It helps in collecting the data from a distance apart
depending on the type or specification of the sensor. This data is used in the system to detect the
incipient problem of the bearing used slow moving equipment

The model developed based on basic equations as explained above. The model was then run with
basic input like bearing and shaft parameters, running speed, value of 'g' etc. for a period of
predetermined time (10 sec or 20 sec) the model generated two column data time and amplitude.
It was represented in the form of plot also called time domain plot. The data were analyzed using
the developed software. The condition of the bearing obtained after this analysis. The analysis
also provided other information like time domain and FFT plot, the characteristic / fault
frequencies and shows the peaks at those frequencies
The field data was collected from Machinery Fault Simulator (MFS). The MFS was operated at
different speed (within slow range) and putting good / defective bearing at rear end of it. The
accelerometer of higher sensitivity and data acquisition card with higher sampling rate are used
so that fault signal doesn't escape.
The data were directly saved in the laptop as shown in the above Fig. 6 the data were then
analyzed using the developed software. This indicates condition of the bearing. The analysis
provided other information like time domain and FFT plot, the characteristic / fault frequencies
and shows the peaks at those frequencies
The model data and its analysis indicated the condition and type of fault with bearing. This is
merely theoretical but indicates that the similar result should be obtained from the actual / field
data. In this innovation similar results were obtained. The test with this developed system in field
(MFS) detected various bearing problem in the field
The system comprising of laptop, data acquisition card, accelerometer with connecting cables
System has an in built model with it. Given the parameters like bearing and shaft geometry,
running speed etc it can provide information using the developed program / software on bearing
fault characteristics. This helps in detecting the incipient fault of bearing in field / industry.
The system is useful for the equipment running at slow speed. It is useful for equipment like
Converter, Mill stand and other heavy equipment in industry. The system accepts wide range of
sensors available in the market.

WE CLAIM
1. A system to monitor and diagnose the bearing defect/fault of a machine/equipment
which is running at a low speed, said system comprising :
a model means adapted to process (simulation) a set of predetermined field data(input
parameter -bearing parameter) and output time domain (time- acceleration/amplitude)
plot of said (simulation output) set of data ;
at least one accelerometer/sensor operatively connected with said machine means adapted
to capture the fault signal of the bearing of said machine;
at least one data acquisition card means operatively connected with said
accelerometer/sensor adapted to collect data from said sensor means and send the data to
said processor (PC) means via suitable cable connections;
an analysis means comprising a processor means adapted to diagnose the time domain
plot to monitor and diagnose the bearing defect of said machine.
2. System as claimed in claim 1 wherein said accelerometer/sensor is having a sensitivity of
of 500mv/gto 1000mv/g).
3. System as claimed in claim 1 wherein said data acquisition card of higher sampling rate
of <50 K cycles per second to 100 K cycles per second.
4. System as claimed in claim 1 wherein said machine is running at slow/low speed of <
500 RPM.
5. System as claimed in claim 1 wherein said predetermined inputs of the model comprising
bearing parameters, shaft parameters, running speed, value of g and the like.

6. System as claimed in claim 1 wherein said processor means of said analysis means
comprising a transformer means adapted to break the signal(data) in five different levels;
envelope means to produce an enveloped signal from signal produced from transformer
means and
a spectrum means convert the enveloped signal to envelop spectrum adapted to detect
fault.
7. System as claimed in claim 1 wherein said processor means of said analysis means
analyses said output of model means to give dnoised TDP, dnoised FFT, bearing
status/fault, peaks at characteristics frequency/fault frequency and the like.
8. System as claimed in claim 1 wherein said machine place good/defective bearing means
substantially at its rear end.
9. A method to monitor and diagnose the bearing defect of a machine/equipment which is
running at a low speed, said method comprising steps of:
inputting a first set of data to a model means adapted to process said inputs;
outputting a second set of time domain plot from said model means;
analyzing said time domain plots to monitor and diagnose the bearing defect of said
machine.
final outputting a third set of data to monitor and diagnose the bearing defect of said
machine from said time domain (data)plots
10. Method as claimed in claim 9 wherein said machine is running at slow/low speed < 500
RPM.

11. Method as claimed in claim 9 wherein said first set of data comprising bearing
parameters, shaft parameters, running speed, value of g and the like.
12. Method as claimed in claim 9 wherein said step analyzing said time domain plots to
monitor and diagnose the bearing comprising steps of
transforming time domain plots by a transformer means adapted to break the signal(data)
in five different levels;
enveloping to produce an enveloped signal from signal produced from the transformer
means and
converting enveloped signal to envelop spectrum adapted to detect fault.
13. Method as claimed in claim 9 wherein said third set of data comprising dnoised TDP,
dnoised FFT, bearing status/fault, peaks at characteristics frequency/fault frequency and
the like.

The present invention relates to a system and a method adapted to monitor and diagnose the
bearing defect of machine which is running at a low speed. The system is provided to monitor
and diagnose the bearing defect/fault of a machine/equipment which is running at a low speed.
The system comprising a model means to process (simulation) a set of predetermined data(input
parameter -bearing parameter) and output time domain (time- acceleration/amplitude) plot of
said (simulation output) set of data, atleast one accelerometer/sensor operatively connected with
the machine means adapted to capture the fault signal of the bearing of the machine, atleast one
data acquisition card means operatively connected with the accelerometer/sensor to collect data
from sensor means and send the data to processor means via suitable cable connections and an
analysis means comprising a processor means to diagnose the time domain data to monitor and
diagnose the bearing defect of the machine.