A Study on Tool Breakage Detection Technology Based on Current Sensing and Non-Contact Signal Analysis

Area of Science:

• Manufacturing Engineering
• Mechanical Engineering
• Signal Processing

Background:

• Tool breakage in CNC machining significantly impacts productivity and incurs high maintenance expenses.
• Existing detection methods often require complex hardware modifications or are unreliable.

Purpose of the Study:

• To propose and validate a non-contact tool breakage detection method for CNC machines.
• To utilize spindle current signals and frequency domain analysis for real-time monitoring.

Main Methods:

• Employed an SCT013 current sensor for non-invasive capture of spindle motor current signals.
• Applied Fast Fourier Transform (FFT) for spectral feature extraction, focusing on high-frequency anomalies.
• Evaluated Artificial Neural Network (ANN), Deep Neural Network (DNN), and Convolutional Neural Network (CNN) for automated detection.

Main Results:

• Consistent spectral anomalies were observed in spindle current signals during tool breakage events across 20 experiments.
• The proposed system reliably detected tool breakage by identifying frequency domain anomalies within 1-3 seconds post-event.
• Deep learning models showed varying inference times (15-58 s) but the core detection mechanism identified breakage characteristics early.

Conclusions:

• The non-contact tool breakage detection method using spindle current signals is effective and reliable.
• Early detection of tool breakage enables timely tool condition evaluation and proactive maintenance strategies.
• This approach offers a cost-effective solution for enhancing CNC machining efficiency and reducing operational costs.