Bearing Faults Diagnosis Based on B-spline Neurofuzzy Networks

As the common part of all kinds of rotating machinery, roller bearings are also vulnerable. Theories and methodologies for roller bearing fault diagnosis are more and more important for modern industry. This paper puts forward data processing methods and intelligent pattern recognition techniques for predicting bearing faults. A bearing monitoring platform is built and vibration signal of bearings of different working states are collected. Time-delayed correlation and demodulation technique is applied to extract effective signal features and B-spline neurofuzzy networks are then employed to recognize bearing faults. Experimental results prove the accuracy and reliability of the monitoring process.