Acoustic Anomaly Detection on UAM Propeller Defect with Acoustic dataset for Crack of drone Propeller (ADCP)

2025-03-02

AI Summary

This work addresses the safety challenge of non-contact, real-time detection of early-stage cracks in unmanned aerial vehicle (UAV) propellers for urban air mobility (UAM). We propose an acoustic-based non-destructive anomaly detection method. To this end, we introduce ADCP, a dedicated acoustic dataset comprising multi-angle microphone recordings under variable-throttle excitation. A novel time-frequency joint preprocessing paradigm—integrating Fast Fourier Transform (FFT) and Short-Time Fourier Transform (STFT)—is designed to simultaneously capture global spectral characteristics and local transient responses, thereby significantly enhancing discriminability of subtle crack-induced acoustic signatures. The resulting model robustly classifies propeller states into normal, torn, and fractured categories. Experimental results demonstrate high sensitivity of acoustic signals to incipient cracks, validating the feasibility and effectiveness of the approach. This study establishes a deployable intelligent predictive maintenance framework for UAM systems.

Technology Category

• Application Domains: Mobility, Driving & Flight
• Planning, Routing, and Scheduling: Activity and Plan Recognition
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

The imminent commercialization of UAM requires stable, AI-based maintenance systems to ensure safety for both passengers and pedestrians. This paper presents a methodology for non-destructively detecting cracks in UAM propellers using drone propeller sound datasets. Normal operating sounds were recorded, and abnormal sounds (categorized as ripped and broken) were differentiated by varying the microphone-propeller angle and throttle power. Our novel approach integrates FFT and STFT preprocessing techniques to capture both global frequency patterns and local time-frequency variations, thereby enhancing anomaly detection performance. The constructed Acoustic Dataset for Crack of Drone Propeller (ADCP) demonstrates the potential for detecting propeller cracks and lays the groundwork for future UAM maintenance applications.

Problem

• Detects cracks in UAM propellers using sound datasets.
• Differentiates normal and abnormal sounds for maintenance.
• Enhances anomaly detection with FFT and STFT techniques.

Innovation

• Non-destructive crack detection using sound datasets
• Integration of FFT and STFT preprocessing techniques
• Enhanced anomaly detection with microphone-propeller angle variations