For the effective detection of acoustic emission (AE) impulses against a noisy background, the correct assessment of AE parameters, and an increase in defect location accuracy during data processing are needed. For these goals, it is necessary to consider the waveform of the AE impulse. The results of numerous studies have shown that the waveforms of AE impulses mainly depend on the properties of the waveguide, the path along which the signal propagates from the source to the sensor. In this paper, the analytical method for modeling of AE signals is considered. This model allows one to obtain model signals that have the same spectrum and waveform as real signals. Based on the obtained results, the attenuation parameters of the AE waves for various characteristics of the waveguide are obtained and the probability of defect detection at various distances between the AE source and sensor utilized for evaluation.

The paper studies the possibility of carbide and decarburized ferrite interlayers detection formed during the production and operation of dissimilar welded joints of austenitic to pearlitic steels. Diffusion interlayers may be interpreted as welded joint microstructure defects, since their formation can cause premature failure of the product under high-temperature operation conditions. In addition, with a certain chemical and structural-phase composition of the welded metals, brittle interlayers formation and development of cracks in the weld zone may be occurred. To identify diffusion interlayers, the acoustic emission (AE) method is used in this research. Specimens cut from welded joints were tested by tension until rupture with simultaneous recording of AE signals. Based on the research results, particular AE data signatures corresponding to the specimens with diffusion interlayers were identified. The main feature indicating the presence of diffusion interlayers is an increase in the amplitudes of AE impulses and AE activity at a stress value of 300 MPa, corresponding to the ultimate strength of the ferrite phase.

The article addresses the topical issues of achieving more efficient operation of the module for identifying the signals obtained during acoustic-emission monitoring of complex technical objects that involves case-based reasoning. Monitoring systems process large amounts of information that require significant time resources for manually processing them by a human (an expert or operator). Therefore, for rendering help to experts (operators), it is proposed to use machine learning methods and data mining tools. Unfortunately, in these tasks it is not always possible to obtain the necessary data set (a training sample) for effectively using the machine learning tools. In view of the above, it is proposed to apply the advanced approach known as case-based reasoning which is actively used in artificial intelligence systems, and which is able to operate with small volumes of the training sample and accumulate experience in solving such problems. However, this approach has a significant drawback: with an increase in the number of precedents, so does the number of comparison operations for processing a new situation (task). This can be solved by using the case similarity matrix. Instead of periodically reducing the base of accumulated cases, which is a time-consuming procedure, it is proposed to limit the number of accumulated use-cases and maintain the most significant (characteristic) ones. In addition, a case presentation of the signal spectrum is proposed to reduce the time spent on one operation of comparing two use-cases. Not only does the proposed method of representing the signal spectrum reduce the amount of processed information, but it also simplifies the metric for comparing use-cases. These modifications can significantly increase the efficiency (throughput) of the case-based module for identifying acoustic emission signals developed in the MS Visual Studio environment in the C# language. The results of testing the module on real acoustic emission monitoring data have confirmed the validity of applying the proposed modifications for the developed case-based module. Работа посвящена актуальным вопросам повышения эффективности работы модуля для идентификации сигналов, получаемых во время акустико-эмиссионного мониторинга сложных технических объектов, использующего рассуждения на основе прецедентов. Системы мониторинга обрабатывают большие объёмы информации, требующие затрат значительных временных ресурсов при ручной обработке человеком (экспертом, оператором), поэтому для помощи экспертам (операторам) предлагается использовать методы машинного обучения и средства интеллектуального анализа данных. К сожалению, в указанных задачах не всегда удаётся получить необходимый набор данных (обучающую выборку) для эффективного применения средств машинного обучения. По этой причине предложено использовать перспективный и активно применяемый в системах искусственного интеллекта (интеллектуальных системах) подход на основе прецедентов, который способен работать при малых объёмах обучающей выборки и накапливать опыт решения подобных задач. Однако он имеет существенный недостаток — с ростом числа прецедентов растет и число операций сравнения для обработки новой ситуации (задачи). Указанную проблему возможно решить за счёт использования матрицы сходства прецедентов. Вместо периодического сокращения базы накопленных прецедентов, что является затратной по временным ресурсам процедурой, следует ограничить количество накопленных прецедентов и сохранить наиболее значимые (характерные) из них. Помимо этого, рекомендовано прецедентное представление спектра сигнала для снижения временных затрат на одну операцию сравнения двух прецедентов. Описанный метод представления спектра сигнала не только позволяет сократить объемы обрабатываемой информации, но и позволяет упростить метрику для сравнения прецедентов. Данные модификации позволяют повысить эффективность (быстродействие) разработанного в среде MS Visual Studio на языке C# прецедентного модуля для идентификации сигналов акустической эмиссии. Результаты апробации на реальных данных акустико-эмиссионного мониторинга подтвердили обоснованность применения предложенных модификаций для...

The possibility of detecting specific structural defects in dissimilar welded joints—carbide and decarburized ferrite interlayers—by the acoustic emission (AE) method is investigated. These interlayers are formed in dissimilar welded joints of austenitic and pearlitic steels during welding and subsequent operation.

The paper investigates the possibility of using the acoustic emission (AE) method to detect carbide and decarburized interlayers formed in dissimilar welded joints of austenitic and pearlitic steels during welding and subsequent operation.