Study on High-Performance Gear Fatigue Life Prediction Method Based on Deep Learning Theories

Abstract

This paper studies fatigue application scenarios for high-performance gears and other mechanical components. It addresses the limitations of internal encapsulation detection and challenges of long-cycle tests. The paper proposes an intelligent prediction method for fatigue features, utilizing visual detection and accelerated degradation life. It integrates conventional test benches and environmental reliability accelerated test conditions, conducts in-depth research on fatigue life estimation algorithms, and explores the feasibility of employing deep learning algorithms and failure prediction models for fatigue life prediction. The paper also establishes an algorithmic system architecture that integrates and processes information from multiple systems and sensors, including gear fatigue performance driving and fatigue monitoring. This approach enables the rapid identification of early micro-motion fatigue characteristics, online autonomous detection, and intelligent failure estimation by integrating information from various systems and sensors. It can accurately predict fatigue degradation and provide a basis for adopting a rational anti-fatigue optimization design.