The Gearbox Fault Diagnosis Based on Wavelet Transform
0 likes211 views
This study investigates the fault diagnosis of pump gearboxes using wavelet transform and vibration analysis, comparing two gearboxes identified as 1# and 2#. Results indicate that the 2# gearbox exhibits a significantly higher amplitude and prominent frequency components indicative of minor faults. The findings suggest that monitoring the modulation signal frequencies is a more sensitive approach for diagnosing gearbox failures.
![IOSR Journal of Mechanical and Civil Engineering (IOSR-JMCE)
e-ISSN: 2278-1684,p-ISSN: 2320-334X, Volume 12, Issue 6 Ver.III (Nov. - Dec. 2015), PP 06-08
www.iosrjournals.org
DOI: 10.9790/1684-12630608 www.iosrjournals.org 6 | Page
The Gearbox Fault Diagnosis Based on Wavelet Transform
Jinyu Wang1 Shibo Hou1
Wenchao Zhang2
1Department of Electrical & Information Engineering, Northeast Petroleum University, Heilongjiang Daqing,
163318, China
2The Fourth Oil Extraction Plant of Daqing Oilfield Company Limited,163000,China
Abstract: The main gearbox is an important driving force for a pump. It needs to carry out regular monitoring.
In this paper, it measures the vibration acceleration signal of a pumping unit’s main gearbox, names them 1#
and 2#. After comparing the time course curve and spectral of the two main gearbox, it has found the 2#
gearbox’s maximum amplitude is about 2.5 times than the 1#, and the frequency which corresponds to the
spectral value is close to the meshing frequency. The narrowband signals are filtered by Hilbert transform to
obtain the envelope of the narrow band signal. The envelope signals are significant different between the two
gear boxes, 1# gearbox has more spectrum of frequency, and 2# gearbox is more outstanding single frequency
component. The results show that the presence of the 2#gearbox has some minor faults. So does the oil analysis
show. The 2# gearbox is more wear than 1# gearbox, which explains the rationality of the vibration analysis.
Keywords: Wavelet Transform; Gearbox; Vibration; Fault-diagnosis
I. Introduction
A pump’s power plant is an equipment of four machines, which has biaxial propelled diesel-diesel
combined propulsion. Two one each side of the diesel engine output single axial power through the main
gearbox. The main gearbox is the main component of the power[1]
. To vibration monitor it can detect faults,
eliminate the hidden dangers and introduce the use of equipment, management and maintenance. However, the
gearbox vibration signal is complicated. It has so much frequency components. So analysis and process the
vibration signal is the key to fault diagnosis[2]
.
1. Data Collection and Analysis
After collecting the gearbox vibration signal, set the sample frequency to 12.8 kHz, collect continuous duration
of 5.12s, collect 65536 points. The measuring points are arranged in 1# gearbox right free end of the pinion and
2# gearbox right free pinion end, each measurement point has three directions of perpendicular, horizontal radial
and horizontal axis[3]
. The vibration acceleration time history is shown in Figure 1. For clarity, it only draws the
front 0.2s data. As can be seen, the more frequency components of acceleration, the more maximum
acceleration reaches to 10 m/s2
. The different of the two gearbox vibration acceleration time history curve is
less.
(a) 1# Gearbox result (b) 2#Gearbox result
Fig.1. 1# and 2# Gearbox vertical acceleration time histories
The vibration acceleration spectrum is shown in Figures 2. It can be seen from the spectrums, the
acceleration spectrum consists of a wideband spectrum and a narrowband spectrum around the meshing
frequency. It has found the 2# gearbox’s maximum amplitude is about 2.5 times than the 1#, and the frequency
which corresponds to the spectral value is close to the meshing frequency.](https://image.slidesharecdn.com/b012630608-151204072313-lva1-app6891/85/The-Gearbox-Fault-Diagnosis-Based-on-Wavelet-Transform-1-320.jpg)
![The Gearbox Fault Diagnosis Based on Wavelet Transform
DOI: 10.9790/1684-12630608 www.iosrjournals.org 7 | Page
a) 1# Gearbox result (b)2#Gearbox result
Fig.2. 1# and 2# Gearbox vertical acceleration time histories
II. Analysis of Vibration Acceleration Signals Around the Gearbox Meshing Frequency
2.1 The Modulation Phenomenon at the Mesh Frequency
During the processing of the gear transmission, the load, the stiff, the speed fluctuations and errors will
change the gear vibration, resulting in the modulation phenomenon of meshing frequency[4]. The changing
modulation signal can reflect in the change state of the gear, so that it can reflect the gearbox malfunction.
Therefore, the meshing frequency modulation analysis is an important part of gearbox vibration condition
monitoring and fault diagnosis.
2.2 Band-pass filtering
To move out the interferences from other frequency components, it is necessary for first band-pass
filter, to remain the component frequencies in the vicinity of the mesh frequency. The meshing frequency is
575Hz, the lower cut-off frequency of the filter is 512Hz, the upper cut-off frequency is taken to 640Hz and the
filter order is 1000. After using MATLAB filters, the results are shown in Figures 3. The filter causes the front
0.04s delay distortion, so the portion of the signal may not be considered. As it can be seen, an amplitude
modulation frequency phenomenon shows significant in the engagement. However, 1# and 2# gearbox meshing
frequency modulated signal is significantly different. 2# gearbox modulation signal peak is larger than 1#; 2#
gearbox modulation signal has a significant prominent frequency component.
(a) 1# Gearbox result (b)2#Gearbox result
Fig.3. 1# and 2# Gearbox vertical acceleration time histories
2.3 Converting the modulated signal based on the Hilebert analysis
Using the Hilebert transform, turn the signal into:
1 ( )
ˆ( )
x
x t d
t
(1)
Order the plural: ，then is called the analytic signal of. The amplitude of is corresponding to the
envelope of, The phase of is corresponding to the instantaneous phase of . So, The Hilbert transformation is a
basic tool for the analysis of the modulated signal. In MATLAB, the Hilbert transformation is. It is needed to be
noted that the output is the analytic signal of the array .The order of seeking envelope is.](https://image.slidesharecdn.com/b012630608-151204072313-lva1-app6891/85/The-Gearbox-Fault-Diagnosis-Based-on-Wavelet-Transform-2-320.jpg)
![The Gearbox Fault Diagnosis Based on Wavelet Transform
DOI: 10.9790/1684-12630608 www.iosrjournals.org 8 | Page
The most substantial value of The engagement envelope of the right free end pinion of the 2# gearbox
vertical acceleration signal is 4m/s2, and 1# gearbox is 1.8m/s2. The envelope signal spectrum removing the
direct current is shown in Figure 4. The envelope signal spectrum frequency of 1# gearbox right pinion output is
8.8Hz, corresponding to the amplitude of 0.11m/s2. And the envelope signal spectrum frequency of 2 # gearbox
small right gear output is 25.4Hz, corresponding to the amplitude of 0.87m/s2,so the frequency component is
very prominent. Comprehensive analysis of the results: the 2# gearbox meshing vibration frequency is greater
than 1# gearbox, and there is a single frequency component projection of the modulation signal. So there is a
minor faulty of 2# gearbox, the operation state is slightly worse than 1# gearbox. And the oil monitoring results
also reflect that 2# gearbox wear slightly larger than 1# gearbox, as shown in Table 1. The source of the
prominent single frequency component requires further analysis. 2# gearbox vibration intensity of 1.2mm/s, 1#
gearbox is 1.3mm/s. Two gearbox vibration intensity different is small, indicating that the fault is not serious,
but also shows: Compared to the vibration intensity, the modulated signal frequencies are more sensitive to
gearbox failure, is a better monitoring and diagnostic parameters
Table1 gearbox oil monitoring results
device eF rC bP uC aN
1# gearbox 11.3 0.2 9.5 11.7 86.9
2# gearbox 21.7 0.6 18.2 20.5 176
III. Conclusion
(1) The different between the two gearbox vibration acceleration signal is not obvious in time course curve, but
the maximum amplitude differ of the spectrum is significant, the frequency which is corresponding to the
maximum amplitude is approximately to the gear mesh frequency, that indicating a fault with the mesh
frequency are closely related.
(2) In the vicinity of the meshing frequency band-pass filter, and then take the Hilbert transform envelope, two
gearbox vibration signal difference is more obvious. The presence of a single frequency component was
prominent in the 2# gearbox envelope signal, that results of failure, fluid analyzed for the presence of the
gearbox also verified this conclusion.
(a) 1# Gearbox result (b) 2#Gearbox result
Fig.4. 1# and 2# Gearbox vertical vibration acceleration envelope signal spectrum near mesh frequency
(3) The vibration intensity different of the two gearbox is less, indicating a slight fault, but also shows the
modulation signal at the frequency of engagement gearbox failure is more sensitive, is a better monitoring and
diagnostic parameters.
(4) The 25.4Hz is neither the 2# gearbox modulation frequency nor the diesel transfer frequency. The source of
the frequency components still need more test data for further analysis to determine the specific source of the
fault.
References
[1]. Ding Kang, Li Weihua, Zhu Xiaoyong. Gears and Gearbox Fault Diagnosis Practical Technology. [M]. Beijing: Machinery Industry
Press, 2005.
[2]. Fei-Sike Technology R&D Center. MATLAB7 Auxiliary Signal Processing Technology and Applications. [M]. Beijing: Electronic
Industry Press. 2005.
[3]. Liu Shulin, Wang Jingdong, Li Fengming, etc. Shock and Vibration Handbook. [M]. Beijing: China Petrochemical Press. 2008.](https://image.slidesharecdn.com/b012630608-151204072313-lva1-app6891/85/The-Gearbox-Fault-Diagnosis-Based-on-Wavelet-Transform-3-320.jpg)
The Gearbox Fault Diagnosis Based on Wavelet Transform
- 1. IOSR Journal of Mechanical and Civil Engineering (IOSR-JMCE) e-ISSN: 2278-1684,p-ISSN: 2320-334X, Volume 12, Issue 6 Ver.III (Nov. - Dec. 2015), PP 06-08 www.iosrjournals.org DOI: 10.9790/1684-12630608 www.iosrjournals.org 6 | Page The Gearbox Fault Diagnosis Based on Wavelet Transform Jinyu Wang1 Shibo Hou1 Wenchao Zhang2 1Department of Electrical & Information Engineering, Northeast Petroleum University, Heilongjiang Daqing, 163318, China 2The Fourth Oil Extraction Plant of Daqing Oilfield Company Limited,163000,China Abstract: The main gearbox is an important driving force for a pump. It needs to carry out regular monitoring. In this paper, it measures the vibration acceleration signal of a pumping unit’s main gearbox, names them 1# and 2#. After comparing the time course curve and spectral of the two main gearbox, it has found the 2# gearbox’s maximum amplitude is about 2.5 times than the 1#, and the frequency which corresponds to the spectral value is close to the meshing frequency. The narrowband signals are filtered by Hilbert transform to obtain the envelope of the narrow band signal. The envelope signals are significant different between the two gear boxes, 1# gearbox has more spectrum of frequency, and 2# gearbox is more outstanding single frequency component. The results show that the presence of the 2#gearbox has some minor faults. So does the oil analysis show. The 2# gearbox is more wear than 1# gearbox, which explains the rationality of the vibration analysis. Keywords: Wavelet Transform; Gearbox; Vibration; Fault-diagnosis I. Introduction A pump’s power plant is an equipment of four machines, which has biaxial propelled diesel-diesel combined propulsion. Two one each side of the diesel engine output single axial power through the main gearbox. The main gearbox is the main component of the power[1] . To vibration monitor it can detect faults, eliminate the hidden dangers and introduce the use of equipment, management and maintenance. However, the gearbox vibration signal is complicated. It has so much frequency components. So analysis and process the vibration signal is the key to fault diagnosis[2] . 1. Data Collection and Analysis After collecting the gearbox vibration signal, set the sample frequency to 12.8 kHz, collect continuous duration of 5.12s, collect 65536 points. The measuring points are arranged in 1# gearbox right free end of the pinion and 2# gearbox right free pinion end, each measurement point has three directions of perpendicular, horizontal radial and horizontal axis[3] . The vibration acceleration time history is shown in Figure 1. For clarity, it only draws the front 0.2s data. As can be seen, the more frequency components of acceleration, the more maximum acceleration reaches to 10 m/s2 . The different of the two gearbox vibration acceleration time history curve is less. (a) 1# Gearbox result (b) 2#Gearbox result Fig.1. 1# and 2# Gearbox vertical acceleration time histories The vibration acceleration spectrum is shown in Figures 2. It can be seen from the spectrums, the acceleration spectrum consists of a wideband spectrum and a narrowband spectrum around the meshing frequency. It has found the 2# gearbox’s maximum amplitude is about 2.5 times than the 1#, and the frequency which corresponds to the spectral value is close to the meshing frequency.
- 2. The Gearbox Fault Diagnosis Based on Wavelet Transform DOI: 10.9790/1684-12630608 www.iosrjournals.org 7 | Page a) 1# Gearbox result (b)2#Gearbox result Fig.2. 1# and 2# Gearbox vertical acceleration time histories II. Analysis of Vibration Acceleration Signals Around the Gearbox Meshing Frequency 2.1 The Modulation Phenomenon at the Mesh Frequency During the processing of the gear transmission, the load, the stiff, the speed fluctuations and errors will change the gear vibration, resulting in the modulation phenomenon of meshing frequency[4]. The changing modulation signal can reflect in the change state of the gear, so that it can reflect the gearbox malfunction. Therefore, the meshing frequency modulation analysis is an important part of gearbox vibration condition monitoring and fault diagnosis. 2.2 Band-pass filtering To move out the interferences from other frequency components, it is necessary for first band-pass filter, to remain the component frequencies in the vicinity of the mesh frequency. The meshing frequency is 575Hz, the lower cut-off frequency of the filter is 512Hz, the upper cut-off frequency is taken to 640Hz and the filter order is 1000. After using MATLAB filters, the results are shown in Figures 3. The filter causes the front 0.04s delay distortion, so the portion of the signal may not be considered. As it can be seen, an amplitude modulation frequency phenomenon shows significant in the engagement. However, 1# and 2# gearbox meshing frequency modulated signal is significantly different. 2# gearbox modulation signal peak is larger than 1#; 2# gearbox modulation signal has a significant prominent frequency component. (a) 1# Gearbox result (b)2#Gearbox result Fig.3. 1# and 2# Gearbox vertical acceleration time histories 2.3 Converting the modulated signal based on the Hilebert analysis Using the Hilebert transform, turn the signal into: 1 ( ) ˆ( ) x x t d t (1) Order the plural: ，then is called the analytic signal of. The amplitude of is corresponding to the envelope of, The phase of is corresponding to the instantaneous phase of . So, The Hilbert transformation is a basic tool for the analysis of the modulated signal. In MATLAB, the Hilbert transformation is. It is needed to be noted that the output is the analytic signal of the array .The order of seeking envelope is.
- 3. The Gearbox Fault Diagnosis Based on Wavelet Transform DOI: 10.9790/1684-12630608 www.iosrjournals.org 8 | Page The most substantial value of The engagement envelope of the right free end pinion of the 2# gearbox vertical acceleration signal is 4m/s2, and 1# gearbox is 1.8m/s2. The envelope signal spectrum removing the direct current is shown in Figure 4. The envelope signal spectrum frequency of 1# gearbox right pinion output is 8.8Hz, corresponding to the amplitude of 0.11m/s2. And the envelope signal spectrum frequency of 2 # gearbox small right gear output is 25.4Hz, corresponding to the amplitude of 0.87m/s2,so the frequency component is very prominent. Comprehensive analysis of the results: the 2# gearbox meshing vibration frequency is greater than 1# gearbox, and there is a single frequency component projection of the modulation signal. So there is a minor faulty of 2# gearbox, the operation state is slightly worse than 1# gearbox. And the oil monitoring results also reflect that 2# gearbox wear slightly larger than 1# gearbox, as shown in Table 1. The source of the prominent single frequency component requires further analysis. 2# gearbox vibration intensity of 1.2mm/s, 1# gearbox is 1.3mm/s. Two gearbox vibration intensity different is small, indicating that the fault is not serious, but also shows: Compared to the vibration intensity, the modulated signal frequencies are more sensitive to gearbox failure, is a better monitoring and diagnostic parameters Table1 gearbox oil monitoring results device eF rC bP uC aN 1# gearbox 11.3 0.2 9.5 11.7 86.9 2# gearbox 21.7 0.6 18.2 20.5 176 III. Conclusion (1) The different between the two gearbox vibration acceleration signal is not obvious in time course curve, but the maximum amplitude differ of the spectrum is significant, the frequency which is corresponding to the maximum amplitude is approximately to the gear mesh frequency, that indicating a fault with the mesh frequency are closely related. (2) In the vicinity of the meshing frequency band-pass filter, and then take the Hilbert transform envelope, two gearbox vibration signal difference is more obvious. The presence of a single frequency component was prominent in the 2# gearbox envelope signal, that results of failure, fluid analyzed for the presence of the gearbox also verified this conclusion. (a) 1# Gearbox result (b) 2#Gearbox result Fig.4. 1# and 2# Gearbox vertical vibration acceleration envelope signal spectrum near mesh frequency (3) The vibration intensity different of the two gearbox is less, indicating a slight fault, but also shows the modulation signal at the frequency of engagement gearbox failure is more sensitive, is a better monitoring and diagnostic parameters. (4) The 25.4Hz is neither the 2# gearbox modulation frequency nor the diesel transfer frequency. The source of the frequency components still need more test data for further analysis to determine the specific source of the fault. References [1]. Ding Kang, Li Weihua, Zhu Xiaoyong. Gears and Gearbox Fault Diagnosis Practical Technology. [M]. Beijing: Machinery Industry Press, 2005. [2]. Fei-Sike Technology R&D Center. MATLAB7 Auxiliary Signal Processing Technology and Applications. [M]. Beijing: Electronic Industry Press. 2005. [3]. Liu Shulin, Wang Jingdong, Li Fengming, etc. Shock and Vibration Handbook. [M]. Beijing: China Petrochemical Press. 2008.