Phase unwrapping noise suppression method based on BEAST-SPA in distributed acoustic sensing system

In the scenario of measuring a signal under strong interference using distributed acoustic sensing (DAS) system based on phase-sensitive optical time domain reflectometry (φ-OTDR), the unwrapped DAS exhibits change point distortion due to noise interference. In order to improve the signal-to-noise ratio (SNR), a phase unwrapping noise suppression method based on Bayesian Estimator of Abrupt change, Seasonality, and Trend (BEAST) and Smoothness prior approach (SPA) is proposed and studied. The BEAST-SPA method begins by employing the BEAST algorithm to decompose the time series data into four components: abrupt changes, seasonal signals, trend signals and remainder signals. The points of abrupt changes correspond to the distortion points where change point distortion occurred in the unwrapped signal. Subsequently, the time series data is partitioned into several segments using distortion points. Finally, the SPA algorithm is applied to fit and remove the non-linear trend components in each segment of the time series data. The BEAST-SPA method not only excels in effectively removing non-linear trend components without the need to assume the type of trend, but it is also applicable for distortion recovery in long-term scale signals. In order to validate the feasibility of this method, we conducted multiple experiments using Piezoelectric Transducer (PZT) as active signal source and actions such as desk tapping and stamping as strong interference. Experiments were designed for single-frequency vibration at 300 Hz. The recovered signals closely matched the PZT reference signals, indicating the effectiveness of the BEAST-SPA method in eliminating change point distortions caused by strong interference. In conclusion, the BEAST-SPA method effectively suppresses phase noise, significantly enhances the SNR.