Discrete Wavelet Transform, Support Vector Machine and Adaptive Neuro-Fuzzy Inference in Bearing Fault Diagnosis

Discrete Wavelet Transform, Support Vector Machine and Adaptive Neuro-Fuzzy Inference in Bearing Fault Diagnosis

This paper presents an useful intelligent system for bearing fault diagnosis based on vibration analysis using discrete wavelet transform (DWT) and two classifiers, namely, support vector machine (SVM) and adaptive neuro-fuzzy inference (ANFIS). Discrete wavelet transform with Daubechies-3 wavelet was used to identify abrupt changes in the vibration signals. Thirty features were extracted from the wavelet coefficients of vibration signals using different feature parameters. A data mining technique was implemented to obtain the superior features and to reduce the dimension of features and the training time of two classifiers. The accuracy rate of the SVM and ANFIS was calculated by applying the testing dataset. The experimental results showed that the performance of the SVM and ANFIS was perfect in fault diagnosis of rolling element bearing. This method allowed identification at a 100.0% level of efficiency. Therefore, it can be seen that the proposed procedure can be successfully applied for condition monitoring and fault diagnosis of bearings in industry