Riduzione del rumore in segnali ultrasonori tramite algoritmi di cross-correlazione per il monitoraggio di componenti in servizio

In this work, a method to enhance the signal-to-noise ratio (SNR) of ultrasonic signals is proposed, based on the use of cross-correlation between the signal itself and a proper signal, made of one or more square pulses, used to trigger ultrasound pulses. The improvement in SNR was evaluated using two types of trigger signals: with single and multiple square pulses. The procedure was compared to other denoising methods: ensemble averaging, moving averaging and wavelet-based filtering. It is shown that, using a trigger signal with a single pulse δ points long, an improvement in SNR proportional to √ can be reached. Using a trigger signal with m pulses, each δ points wide, yields better results, since an improvement in SNR proportional to √· can be obtained; this depends also on the noise frequency band in relation to the signal band. The procedure allows the treatment of very noisy signals, in order to calculate the ultrasound time of flight (Tof).