Deep Learning-Based Aircraft Engine Anomaly Detection and Health Classification System

This paper presents an advanced deep learning framework for automated anomaly detection and health clas- sification in aircraft engines. The system introduces a novel integrated approach combining multiple AI techniques: Masked Multi-scale Reconstruction (MMR) for blade defect detection, VGG16-based classification for anomaly categorization, and Ran- dom Forest Classification for engine health assessment using sensor data. Our framework represents a significant advancement in automated engine inspection, combining computer vision tech- niques with comprehensive sensor data analysis to create a robust health monitoring system. The solution enables real-time detec- tion of developing faults and implements predictive maintenance strategies that substantially improve upon traditional inspection methods. Through extensive testing across multiple aircraft types and operating conditions, our system demonstrates marked im- provements in early fault detection, maintenance efficiency, and operational reliability. This innovation contributes significantly to aviation safety while substantially reducing maintenance costs and operational downtime. The framework’s modular architec- ture and scalable design make it suitable for deployment across various engine types and maintenance environments, presenting a promising solution for the aviation industry’s growing demand for automated inspection systems.