International Journal of Science, Engineering and TechnologyAuthors: Rishi Kinger, Atharva Singh
Abstract: Low-cost MEMS accelerometers and microcontrollers make continuous vibration monitoring affordable for small wind turbines, but published low-cost systems seldom document their signal chain well enough for a reader to judge what the displayed indicators mean. This paper describes a monitor built from an ESP32 microcontroller and an MPU6050 three-axis accelerometer mounted on a motor-driven small-turbine rotor. The firmware samples at 100 Hz and removes gravity with a first-order recursive high-pass filter (−3 dB at 0.86 Hz); a MATLAB dashboard computes the RMS level, a Welch power spectral density (PSD) and the X–Y cross-correlation over a 10-s sliding window. Each processing stage is characterized analytically, including the 95% confidence interval of the PSD estimate (−3.0 dB to +4.6 dB, 11 equivalent degrees of freedom). In a recorded window, the RMS levels were 163, 224 and 170 mg on the X, Y and Z axes, and the integrated Z-axis PSD (166 mg RMS) agreed with the time-domain value within 3%. The Z-axis spectrum peaked at 6.7 Hz, 21 dB above the 12–50 Hz floor. Segment-wise analysis showed, however, that this peak and the oscillation of the X–Y correlation came mainly from a 1.6-s transient, while the less disturbed parts of the record oscillated at 2.8–3.8 Hz. The single-window indicators therefore mixed transient and steady vibration. The paper also identifies three configuration risks common to such systems — a sensor low-pass filter set above the Nyquist frequency, possible clipping on the gravity-bearing axis and the absence of sample timestamps — and specifies the experiments needed before the monitor can support fault diagnosis.
DOI: https://doi.org/10.5281/zenodo.23756593
