A Novel Multi-Source Fault Diagnosis Strategy Based on Knowledge and Data Dual-Drive for a Planetary Gearbox.

Area of Science:

• Mechanical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Traditional fault diagnosis methods struggle with incomplete fault information and a trade-off between accuracy and computational cost.
• Effective fault diagnosis is critical for wind turbine reliability and operational efficiency.

Purpose of the Study:

• To propose a novel multi-source fault diagnosis strategy addressing limitations of traditional methods.
• To enhance fault information expression and achieve high accuracy with efficient computation.

Main Methods:

• A multi-source information correlation matrix (MICM) was designed to integrate time, frequency, and channel features.
• Kernel Principal Component Analysis (KPCA) was employed for dimensionality reduction of the MICM.
• A hybrid classifier combining Softmax logical regression (SLR) for pre-classification and K nearest neighbor (KNN) for accurate classification was developed.

Main Results:

• The proposed MICM-SLR-KNN algorithm effectively enhanced fault information expression.
• The algorithm demonstrated superior classification accuracy compared to other methods.
• Experimental validation using a planetary gearbox dataset confirmed the algorithm's effectiveness and efficiency.

Conclusions:

• The knowledge and data dual-drive algorithm offers a superior approach to multi-source fault diagnosis.
• The MICM-SLR-KNN strategy provides an effective solution for wind turbine fault diagnosis, balancing accuracy and computational load.