Periodic-Filtering Method for Low-SNR Vibration Radar Signal

Radar is a non-contact, high-precision vibration measurement device and an important tool for bridge monitoring. Vibration information needs to be extracted from the radar phase, but phase sensitive noise. Under low signal-to-noise ratio (SNR) data acquisition conditions, such as transmission power or long observation distance, differential jump errors occur clutter estimation becomes difficult, which leads inaccurate inversion of deformation. Traditional low-pass filtering methods can filter out noise improve SNR, they require oversampling. The sampling rate several times higher than Doppler bandwidth, frequency. This puts high requirements on systems causes large volumes. Therefore, this paper proposes novel signal method called periodic method. uses periodicity feature signals without derives time-domain frequency-domain expressions presents deformation process based them. involves extracting frequency in domain, suppressing through filtering, estimating using circle fitting complex plane, inverting final with phase. verified simulation experiments, calibration experiments. results show that new SNR by five times, resolve jumps, accurately estimate clutter, thus getting submillimeter-level at SNR.