Evaluating the Probability of Detecting Acoustic Emission Signals Using the Amplitude Discrimination Method with a Low Signal-to-Noise Ratio

An approach to calculation of the probability of detecting acoustic emission (AE) signals in a wide range of signal-to-noise ratios (SNR) is proposed. This approach demonstrates good agreement with experimental data. A distinctive feature of the proposed approach is the frame-by-frame analysis of the distribution areas of amplitude spikes in the signal and noise that exceed the amplitude discrimination threshold making it possible to perform estimates with SNR < 6 dB. It is shown that the positive effect of applying the proposed approach is the transition from estimating the probability of missing a signal due to a defect to estimating the probability of detecting a signal of each type and a different degree of transformation (distortion); this allows one to objectively compare and select the options for optimal recording, processing, and filtering (noise reduction) procedures and for analysis of AE data, as well as to select settings for implementation of these procedures. The proposed approach is invariant to the method of obtaining AE recording frames and the form of acoustic signals, therefore, it is applicable to evaluating single- and multimodal pulsed and continuous AE recorded by frames in threshold and nonthreshold AE recording modes. The conditions affecting the discrepancy between the results of theoretical and experimental studies are established, and ways to improve the accuracy of applying the proposed approach are outlined. It is shown that the amplitude detection of signals is possible starting from the threshold level exceeding the noise level by 3 dB provided that the amplitude spikes in the noise are no higher than 7.5 dB of the root-mean-square deviation of noise distribution. Otherwise, high-quality and stable signal detection by the amplitude threshold cannot be ensured.