Identification of Load Categories in Rotor System Based on Vibration Analysis

Area of Science:

• Mechanical Engineering
• Vibration Analysis
• Machine Learning

Background:

• Rotating machinery operates under variable loads, necessitating effective monitoring and identification.
• Accurate load type identification is crucial for operational efficiency and preventing equipment failure.

Purpose of the Study:

• To develop and validate a novel method for identifying different load types in a rotor system using vibration signals.
• To assess the accuracy and robustness of the proposed load identification technique.

Main Methods:

• Utilized ensemble empirical mode decomposition (EEMD) for signal reconstruction.
• Extracted energy features from vibration signals.
• Employed a back propagation (BP) neural network for load classification.

Main Results:

• Achieved a general identification rate of 94.54% for five distinct load types.
• Demonstrated high accuracy and good robustness in load identification.
• Experimentally validated the feasibility of the proposed method on a dedicated test bench.

Conclusions:

• The proposed EEMD, energy feature extraction, and BP neural network method is a feasible and effective approach for rotor system load identification.
• The validated results support the application of this method for load identification and fault diagnosis in engineering rotor equipment.