Laser-Wire-DED-ThermalAudio-Dataset

Description

The Laser-Wire-DED-ThermalAudio-Dataset provides multi-modal sensor data for real-time monitoring and anomaly detection in Laser Wire-Directed Energy Deposition (LW-DED) processes. This dataset is designed to support research in acoustic-based and thermal-based defect detection, particularly for monitoring aluminum alloy (Al7075) deposition.

This dataset corresponds to the MultiSensor-Monitoring-LW-DED repository on GitHub:
🔗 GitHub Repository

The dataset consists of synchronized thermal imaging and acoustic recordings collected during LW-DED experiments, providing a comprehensive resource for investigating defect formation mechanisms. The dataset enables deep learning-based anomaly detection by utilizing melt pool audio signals and thermal features.

A recorded experimental video is available at:
🎥 YouTube Video

Key Features

• Synchronized Multi-Sensor Data: Captures thermal and acoustic signals from LW-DED processes.
• Melt Pool Audio Processing: Includes preprocessed and raw acoustic signals for defect analysis.
• Anomaly Detection Labels: Contains labeled defects such as dripping anomalies.
• High-Fidelity Thermal Imaging: Captured via Xiris thermal camera and synchronized with acoustic data.

Dataset Structure

The dataset is structured as follows:

Laser-Wire-DED-ThermalAudio-Dataset/
│── Annotation/             # Contains labeled defect annotations
│── Visualization_demo/     # Scripts and visualizations of results
│── Raw_Video/              # Unprocessed video data
│── segmented_videos/       # Processed video segments
│── segmented_audio/        # Processed audio segments
│── Dataset/                # Core dataset containing thermal and acoustic signals

Usage

This dataset can be used for:

• Machine learning-based defect classification (audio-thermal fusion)
• LW-DED process monitoring and optimization
• Deep learning model training for acoustic and thermal data fusion
• Anomaly detection in metal additive manufacturing

Citation

If you use this dataset in your research, please cite:

@dataset{chen2024lw_ded,
  author    = {Chen, Lequn},
  title     = {Laser-Wire-DED-ThermalAudio-Dataset},
  year      = {2024},
  publisher = {Zenodo},
  url       = {https://zenodo.org/record/[Dataset-ID]}
}