Defektoskopiâ

In ultrasonic inspection, digital aperture focusing (DAF) is increasingly being used to reconstruct reflector images. The reliability of inspection is determined by the quality of the DFA image — signal-to-noise ratio, ability to reconstruct the image of the entire reflector boundary, and resolution. Various methods are used to achieve super-resolution of echoes: maximum entropy method, methods of building autoregressive signal models, compressive sensing (CS) method, etc. To use these methods, it is important to know the impulse response of the ultrasound system, which can be measured or obtained using “blind” deconvolution methods used in image and signal processing. In this paper we consider the Minimum Entropy Deconvolution (MED) method for estimating the impulse response of an ultrasonic flaw detector and achieving the effect of image super-resolution, where knowledge of the transfer function of the system is critical. The effectiveness of the proposed method is confirmed by the results of model experiments

Deviation from the nominal wall thickness of a pipe—both during manufacturing and in operation—is a critical factor affecting the durability of equipment. This study proposes a model of acoustic wave propagation across a pipe cross-section with eccentricity, which forms the theoretical basis for a method of integral assessment of wall thickness non-uniformity in small-diameter pipes. The method is implemented using a specialized flexible piezoelectric transducer based on polyvinylidene fluoride (PVDF) on several samples of seamless pipes with varying thicknesses, and is validated by the results of local ultrasonic thickness measurements

This paper explores the application of linear regression models for estimating the hardness of heat-treated steels based on their static magnetic properties. The study uses pre-calculated datasets containing hysteresis loop parameters along with corresponding hardness values. Separate analyses were conducted for steels subjected to tempering and quenching to assess the influence of heat treatment on prediction accuracy. Several linear models were evaluated and compared using determination coefficients to identify the most effective correlations. The results confirm the feasibility of using regression-based approaches for rapid hardness estimation from magnetic parameters and highlight a way to choose the most informative features

The results of measurements of static magnetic characteristics using attachable transducers depend on the geometry of the test object and magnetic flux leakage, which requires additional study. The aim of the work was to analyze the influence of flat object dimensions and magnetic flux leakage on the results of measurements using attachable transducers. By the example of steel 3 specimens with different sizes, the change of static magnetic characteristics when varying the position of the transducer relative to the edge of the specimens was studied using the DIUS hardware and software system. The results revealed a paradoxical behavior: an increase in the maximum magnetic induction in the transducer was accompanied by a decrease in the field strength in the object, which is explained by the leakage of magnetic flux in the samples with a larger area. It was found that the edge effect and the geometry of the object significantly affect the measured parameters, including the coercive force by demagnetization current, which is used by most modern coercimeters. A possible way to solve this problem is to create regression models, including machine learning methods, to correct the influence of scattering and improve the accuracy of measurements in the conditions of varying the geometry of control objects