Fault detection of taper roller bearings using tunable Q-factor wavelet transform and fault classification using...

Area of Science:

• Mechanical Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Taper roller bearings are critical components in heavy industrial machinery.
• Early fault detection and prognosis are essential for operational reliability and maintenance.
• Advancements in AI and big data offer potential for improved fault diagnosis methods.

Purpose of the Study:

• To develop and evaluate an advanced fault diagnosis and classification strategy for taper roller bearings.
• To leverage signal processing and machine learning for enhanced bearing health monitoring.
• To compare the proposed method against existing techniques.

Main Methods:

• Utilized Tunable Q-factor Wavelet Transform (TQWT) for effective signal processing of bearing data.
• Employed a Long-Short-Term Memory (LSTM) neural network for accurate fault classification.
• Experimentally validated the performance of the TQWT-LSTM combination.

Main Results:

• The TQWT and LSTM combination demonstrated highly efficient and reliable diagnosis of bearing faults.
• Achieved one hundred percent accuracy in classifying different types of bearing faults.
• Outperformed other combinations, including Variational Mode Decomposition (VMD) with Convolutional Neural Networks (CNN).

Conclusions:

• The proposed TQWT-LSTM method is a superior approach for bearing fault diagnosis and classification.
• This AI-driven strategy offers significant improvements in reliability and accuracy for industrial machinery maintenance.
• The findings highlight the potential of advanced signal processing and deep learning in predictive maintenance.