Estimation of Tool Wear and Surface Roughness Development Using Deep Learning and Sensors Fusion.

Area of Science:

• Manufacturing Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Accurate estimation of tool wear and surface roughness is crucial for optimizing machining processes.
• Traditional methods often rely on direct measurement, which can be time-consuming and disruptive.
• Sensor fusion and deep learning offer promising avenues for non-invasive, real-time monitoring.

Purpose of the Study:

• To develop and validate a deep learning-based approach for estimating tool wear and surface roughness.
• To investigate the effectiveness of fusing vibration and sound sensor signals.
• To implement an efficient sensor selection strategy to minimize computational load while maintaining accuracy.

Main Methods:

• Utilized a one-dimensional convolutional neural network (1D-CNN) for the estimation model.
• Employed sensor fusion techniques combining X/Y vibration and sound signals.
• Implemented a uniform experimental design (UED) for data collection and an accelerated experiment for tool wear.
• Developed an influential sensor selection analysis to optimize sensor input.

Main Results:

• The proposed sensor fusion and 1D-CNN model achieved accurate estimations of tool wear and surface roughness.
• Sensor influence analysis effectively identified key sensors, reducing complexity without sacrificing performance.
• The integrated system demonstrated high accuracy and computational efficiency.

Conclusions:

• The developed deep learning approach effectively estimates tool wear and surface roughness using fused sensor data.
• The method is suitable for on-line monitoring applications, providing timely alerts for tool conditions.
• This approach offers a robust and efficient solution for enhancing manufacturing quality control.