An image dataset for use in detecting unwanted bolt rotation

Area of Science:

• Industrial Engineering
• Materials Science
• Computer Vision

Background:

• Component degradation is critical for industrial safety and efficiency.
• Existing datasets lack data on mechanical fixings over time.
• Predictive maintenance requires data on component wear.

Purpose of the Study:

• To create a novel dataset for studying bolt degradation.
• To provide a resource for developing machine learning models for component monitoring.
• To address the gap in existing datasets for time-varying mechanical components.

Main Methods:

• Collected over 1,100 images of a bolted apparatus.
• Varied image acquisition angles and bolt rotation degrees.
• Conducted data collection under controlled laboratory conditions.

Main Results:

• A unique dataset of bolted apparatus images was created.
• The dataset captures variations in bolt condition and orientation.
• This resource is the first of its kind for time-varying bolt analysis.

Conclusions:

• The novel dataset supports research in industrial component monitoring.
• Enables development of AI for detecting degradation in mechanical fixings.
• Facilitates advancements in predictive maintenance strategies.