Introduction to Vibration Screening
In the world of Predictive Maintenance, vibration screening is the first line of defense against machine failure. By monitoring the "health" of an asset through its vibration signature, we can detect anomalies before they turn into costly breakdowns.
One of the most effective methods for automated screening is Frequency Thresholding. Unlike simple time-domain alarms, frequency-based alerts allow us to pinpoint specific faults like imbalance, misalignment, or bearing wear by looking at the Fast Fourier Transform (FFT) spectrum.
Why Use Frequency Thresholding?
- Precision: Distinguish between normal operating noise and specific fault frequencies.
- Automation: Filter out massive amounts of data and only alert when a specific frequency peak exceeds a limit.
- Efficiency: Focus maintenance efforts on machines showing genuine signs of distress.
Implementing Frequency Thresholding with Python
To implement this, we typically convert raw time-series data into the frequency domain using FFT (Fast Fourier Transform). Below is a simplified Python implementation using numpy and scipy.
import numpy as np
import matplotlib.pyplot as plt
# 1. Simulate Vibration Data
fs = 1000 # Sampling frequency (Hz)
t = np.arange(0, 1.0, 1/fs)
# Normal signal + specific fault frequency at 120Hz
signal = 0.5 * np.sin(2 * np.pi * 50 * t) + 1.2 * np.sin(2 * np.pi * 120 * t)
noise = np.random.normal(0, 0.2, len(t))
vibration_data = signal + noise
# 2. Perform FFT
fft_values = np.abs(np.fft.rfft(vibration_data))
frequencies = np.fft.rfftfreq(len(vibration_data), 1/fs)
# 3. Define Threshold
threshold = 0.8
# 4. Screening Logic
anomalies = frequencies[fft_values > threshold]
# Plotting
plt.figure(figsize=(10, 4))
plt.plot(frequencies, fft_values, label='Vibration Spectrum')
plt.axhline(y=threshold, color='r', linestyle='--', label='Warning Threshold')
plt.title('Vibration Screening via Frequency Thresholding')
plt.xlabel('Frequency (Hz)')
plt.ylabel('Amplitude')
plt.legend()
plt.show()
if len(anomalies) > 0:
print(f"Alert! Anomalies detected at: {anomalies} Hz")
Key Steps in the Process
- Data Acquisition: High-frequency sampling is crucial to capture high-order harmonics.
- Signal Transformation: Using FFT to move from time-domain to frequency-domain.
- Establishing Baselines: You must know what a "healthy" machine looks like to set an accurate threshold level.
- Alarm Triggering: If the amplitude at a specific frequency (e.g., 1x or 2x RPM) crosses the line, it's time for an inspection.
Conclusion
Implementing Frequency Thresholding for vibration screening is a powerful way to enhance your maintenance strategy. By automating the detection process, you ensure that small issues are caught before they lead to catastrophic equipment failure.
