Cycle-Informed Triaxial Sensor for Smart and Sustainable Manufacturing

Area of Science:

• Manufacturing Engineering
• Signal Processing
• Machine Health Monitoring

Background:

• Industry 4.0 and 5.0 drive demand for intelligent, sustainable manufacturing.
• Robust predictive maintenance is hindered by operational variability and limited data access.
• Existing methods struggle with unstable process data segmentation, leading to misinterpretation.

Purpose of the Study:

• To introduce a lightweight, embedded-compatible framework for health status signature extraction.
• To enable predictive maintenance using only single-axis accelerometer data.
• To address challenges in segmenting unstable process data for accurate diagnostics.

Main Methods:

• Empirical Mode Decomposition (EMD) for health status signature extraction.
• Cycle-synchronized segmentation using accelerometer-derived velocity and cross-correlation.
• Frequency domain analysis of extracted intrinsic mode functions (IMFs).

Main Results:

• The proposed framework effectively detects subtle machine degradations.
• The method demonstrates a low computational footprint, suitable for embedded systems.
• Successful evaluation on a real-world manufacturing benchmark.

Conclusions:

• The developed framework is effective for health status signature extraction in dynamic machining conditions.
• The cycle-synchronized segmentation strategy overcomes limitations of controller-less systems.
• The approach is suitable for deployment in embedded predictive maintenance systems, especially on legacy machinery.