IDENTIFICATION OF THE STRESS-STRAIN STATE AND DAMAGE TO THE STRUCTURE OF METALS BY THE PARAMETERS OF THE ELECTROMAGNETIC-ACOUSTIC SIGNAL

Abstract

Ensuring industrial safety at hazardous oil and gas facilities is inextricably linked to the need for timely detection of stress concentration zones in process pipelines and pressure vessels. Statistical analysis confirms that over 30 % of accidents are caused by the hidden accumulation of microstructural damage preceding macroscopic cracks. Traditional ultrasonic methods are limited by surface preparation requirements and the need for stable fluid acoustic contact. This study examines the improvement of the electromagnetic-acoustic (EMA) method for contactless identification of stress-strain state (SSS) stages and damage levels in structural steels. The objective is to develop an intelligent neural network algorithm for metal state classification based on EMA response spectral characteristics. A second-order transfer function obtained via dynamic pulse identification (method of moments) is proposed as an integral parameter. To increase informativity, frequency models in the form of amplitude-phase frequency response characteristic (APFRC) hodographs were utilized. Experimental research involved static tension testing of St3sp, 09G2S, and 12Kh18N10T steel samples using a universal Walter + Bai machine and an EM4000 gauge. For automated data interpretation, an artificial neural network (ANN) was designed in C# using the ML.NET machine learning library. The ANN is based on a polynomial logistic regression algorithm with a Softmax activation function. The amplitudes and phases of the first 15 signal harmonics, obtained through spectral decomposition, were selected as input features for the neural network. The developed technique allows for contactless recording of the transition to the plastic region and pre-defect zones directly through protective coatings. The findings confirm the effectiveness of integrating frequency analysis and machine learning technologies to prevent accidental failure and enhance the technosphere safety of oil and gas production.