Bearing Fault Diagnosis in IM Using STFT and J-48 Algorithm based on Vibration Signals in Dynamic Machines
K.M. Arun Kumar1, T.C. Manjunath2, G. Arun Kumar3

1K.M. Arun Kumar, Research Scholar, Assistant Professor, Department of ECE, VTU-RRC, Belagavi SJTJIT, Ranebennur (Karnataka) India.
2Dr. T.C. Manjunath, Professor & Head, Department of ECE, DSCE, Bangalore, (Karnataka) India.
3Dr. G. Arun Kumar, Associate Professor, Department of ECE, JSS Academy of Tech. Education, Noida (Uttar Pradesh), India.
Manuscript received on 05 February 2019 | Revised Manuscript received on 18 February 2019 | Manuscript Published on 04 March 2019 | PP: 68-79 | Volume-7 Issue-5S2 January 2019 | Retrieval Number: ES2009017519/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: The condition monitoring of bearing faults is carried out by analyzing the properties & the characteristics of the vibratory signal obtained from the machine. The detection of the fault from the signal which is extracted is still a challenging problem in the vibration control, which is one of the most important topics to be considered for the condition monitoring of machines and for research purpose. In this paper, diagnosis of bearing fault is done using STFT and J-48 algorithm. Short-Time-Fourier-Transform (STFT) can be used in order to identify the faults in the bearing from the vibration signal which is captured. STFT has the linear type phase characteristics and preserves the signal properties sharpness even when the sudden changes in the signal nature. Vibration signal is then divided into the different section so that relating to the ball bearing parts passage and thus exits from the bearing fault, allowing to estimating the faults occur in ball bearing element. The analysis of vibration signal is carried out in Lab VIEW. Machine learning is a method to enter the database for giving importance to the pleasant information. Machine learning consists of 3 stages, viz., FE, FS, FC. Then, the main important features were taken from the raw vibratory signal, selection of the features was obtained utilizing J-48 algorithm and to build the better classifier, the different parameter of J-48 algorithm are optimized. This algorithm is applied to the RT analysis & furthers the CMT is used as it is very much convenient since the time of computation required to analyze is very less, with an classification accuracy was found to be 94.5%
Keywords: Using STFT,FE, FS, FC. Then, CMT, RT, Analysis , Lab VIEW., Machine.
Scope of the Article: Design and Diagnosis