Impact of Earthquake Monitoring on CNC Machine Tool Leveling Accuracy

By Wu Meng-Hsien, Graduate Student, Department of Mechanical Engineering, National Chin-Yi University of Technology
Prof. Chen Shao-Hsien, Department of Mechanical Engineering, National Chin-Yi University of Technology

Research Objective

Taiwan and Japan are both earthquake-prone regions, making the monitoring of machine tool leveling accuracy critically important. This study developed a long-term leveling measurement and monitoring system (see Figure 1). Through continuous, high-precision data collection and analysis, the system evaluates whether seismic activity causes machine leveling deviations, thereby affecting machining and measurement accuracy. The research also analyzes the potential impact of earthquakes on the stability of precision machinery, providing a reference for future earthquake compensation technologies and precision equipment design.
The study confirms that the impact of earthquakes on CNC machine leveling accuracy is minimal, with deviations controlled within 0.02 mm/m, eliminating concerns about reliability degradation after seismic events.

Research Methodology

Equipment Principle

The system uses an electrolytic tilt sensor (Figure 2). The sensor consists of electrodes and an electrolyte inside a metal or glass housing. Its principle (Figure 3): three electrodes—one sensing electrode and two excitation electrodes—measure voltage differences caused by tilt. When level, the electrolyte covers both excitation electrodes equally, resulting in equal resistance (R1 = R2). When tilted, coverage changes, creating a voltage difference (Figure 4).
An Analog-to-Digital Converter (ADC) converts voltage to digital signals. Resolution is calculated as:

where $𝑅$ = measurement range, $𝑛$ = ADC bit resolution. For a 16-bit ADC and ±3° range, resolution = 0.0001°.
Important: excitation electrodes require alternating signals with 50% duty cycle (Figure 5); DC input causes electrolysis and sensor damage.

System Development

The system includes:

• Real-time leveling measurement
• Visualization charts
• Internal/external temperature monitoring

Hardware: aluminum base plate, acrylic housing (Figure 7).
Software: Windows Forms (C language) for UI, featuring real-time curves, leveling status, temperature, runtime, and data logging (Figure 8).
Firmware: PCB designed via JLCEDA, integrating sensors, ADC, microcontroller, and LDO regulator.

Accuracy Verification

Model Building

Placed on a sine bar, the sensor output was compared to gauge readings. Python’s itertools generated segmentation combinations; AIC determined optimal breakpoints. Multi-segment linear regression models were built for X-axis (6 segments) and Y-axis (10 segments) (Figures 9 & 10).

Stability Test

Seven tests per axis at 1 Hz sampling (360 points). Results (Tables 3 & 4) achieved Grade A per process capability standards.

Repeatability Test

21 repeated measurements showed X-axis and Y-axis precision and accuracy both met Grade A (Table 5).

Resolution Test

Compared with Wyler electronic level; resolution reached 0.002 mm/m (Figure 11).

Signal Filtering

A second-order low-pass filter was applied. Cutoff frequency = 0.01 Hz (based on FFT analysis). Filtering reduced noise standard deviation from 32.83 to 3.44, an 89.52% improvement (Figures 12–15).

Experimental Results

1. Long-Term Operation Impact

Tests on granite and actual machine environments showed no divergence; leveling remained stable at 0.04 mm/m over 14 days. Strong correlation with temperature was observed (Tables 6 & 7; Figures 16–20).

2. Earthquake Impact

Monitoring from Jan 15–31, 2025 captured 7 felt earthquakes in Taichung.

• Major quake (M6.4, Jan 21): temporary deviation of 0.04 mm/m, recovered within 4 hours (Figure 22; Table 8).
• Smaller quakes (M5.6–4.8): minimal impact; deviations corrected within 1 minute (Figures 23–24; Table 9).
  Conclusion: earthquakes cause temporary, not permanent leveling changes.

Conclusion & Future Outlook

The developed system enables remote monitoring and cloud-based data logging with high precision (Cp, Ca, Cpk all Grade A). Noise reduction improved by 89.52%.
Earthquake impact on CNC machine leveling is negligible (≤0.02 mm/m), alleviating reliability concerns.
Future work:

• Develop nonlinear compensation models for temperature and micro-vibrations
• Multi-point measurements to reduce random errors
• Earthquake alert integration and analysis of seismic energy vs. permanent leveling impact

References:
Fredericks Company technical notes; academic theses on precision measurement and compensation; Taiwan earthquake data portal; quality management literature.