Reducing ambient sensor noise in wind tunnel tests using spectral subtraction method

Abstract

Ambient electrical and mechanical noises can have an impact on sensor measurements during wind tunnel tests, potentially causing inaccuracies, particularly in older sensors. This study introduces a spectral subtraction method to mitigate ambient sensor noise in wind tunnel tests. The proposed method was implemented for three types of measurements in wind tunnel tests: pressure, acceleration, and strain. When applied to the pressure measurement of the CAARC building model, the standard deviations of pressure, which initially exhibited variation in the data collected by pressure scanners with varying levels of ambient noise, became uniform across all scanners after the implementation of the spectral subtraction method. The denoised pressure showed good agreement with previously published results in both mean and standard deviation. In the case of acceleration and strain measurements, a significant improvement in signal-to-noise ratio at low wind speeds was observed after denoising. Overall, the application of the spectral subtraction method led to substantial reductions in ambient noise without affecting the mean values of the measurements. The responses in both the background component and resonant component have been well preserved even after the implementation of the spectral subtraction method.