Fault Evolution Characteristics Analysis of Spur Gear Based on Dimensional Technique and Support Vector Machine

Abstract Spur gears are critical components in industrial machinery, where undetected faults can lead to catastrophic failures and costly downtime. While existing fault diagnosis methods often rely on standalone signal processing or machine learning techniques. To address this challenge, we propose an innovative hybrid approach that combines dimensional analysis (DA) with support vector machines (SVM) to achieve superior fault detection and classification. Vibration signals are acquired from a controlled experimental setup and processed using the fast Fourier transform (FFT) to extract fault-specific frequency features. A DA model is developed to simulate gear fault dynamics, deriving dimensionless parameters correlating with fault severity. Results show that the model achieves high accuracy with fault detection, with an average error of 3.15%, outperforming conventional empirical methods. The extracted features are then classified using a multiclass SVM with an optimized kernel, achieving 97.84% accuracy in distinguishing healthy gears from pitting, addendum cracks, and broken teeth. The results confirm the reliability of the proposed method, highlighting its potential for accurate fault detection and quantification in industrial spur gear systems.