Applied Sciences, Год журнала: 2024, Номер 14(22), С. 10546 - 10546

Опубликована: Ноя. 15, 2024

The paper considers the neural network application to detect microstructure defects in dissimilar welded joints using acoustic emission (AE) method. peculiarity of proposed approach is that defect detection carried out taking into account a priori information about properties AE source and waveguide parameters testing structure. Industrial process pipelines with were studied as object, diffusion interlayers formed fusion zones considered defects. simulation signals was hybrid method: signal waveform determined based on finite element model, while amplitudes hits physical experiment mechanical joints. Measurement data from industrial used noise realizations. As result, sample realistic signal-to-noise ratio. method allows for more accurate determination waveform, spectrum, amplitude signal. Greater certainty useful achieving reliable classification result. When backpropagation network, percentage correct than 90% obtained set which ratio less (−5 dB) cases.

Язык: Английский