Weak Signal Processing in Dual-Beam Interferometric Velocity Measurements

Accurate extraction of real velocity signals from Doppler velocity measurements is challenging, particularly in environments with low signal-to-noise ratios (SNRs). To address this issue, we propose a novel algorithm based on spectral peak density searching, designed for reliable Doppler signal processing under noisy conditions. First, we apply a Gaussian filter to reduce the interference of Gaussian white noise while preserving key characteristics of the Doppler signal. Next, a peak-searching operation identifies all spectral peaks in the denoised signals, and potential Doppler shift signals are selected based on a specific bandwidth threshold. The densest spectral peak band is then determined, and nonlinear least-squares smoothing is applied to precisely identify the velocity shift signal. Compared to conventional Doppler signal processing algorithms, this method significantly improves processing accuracy, reducing execution time by approximately 56%, decreasing the average relative error from 3.41% to 1.92%, and increasing the detection probability from 58% to 88%. We validated the performance of the algorithm through experiments simulating weak signals with substantial noise under conditions of low emission intensity and high-speed water flow. The results demonstrate the effectiveness and reliability of the proposed algorithm. The novelty lies in its ability to quickly transmit the processed flow velocity value back to the user interface instead of a massive amount of raw flow velocity data, emphasizing its potential for practical applications in challenging environments.