Diagnostics and recognition of types of mechanical defects in complex metal structures by acoustic emission method

The paper shows that understanding the wear and prevailing mechanisms of structural damage is crucial for devices and systems that are critical to safety, which include complex metal structures. During long-term operation, they are faced with the task of constantly monitoring the locations of microcracks and trends in their occurrence. The aim of the paper is to develop a procedure for optimal selection of the design of the diagnostic system for complex metal structures and a procedure for recognizing the type of mechanical deformation in such structures, based on the difference in the shapes of acoustic emission signals, which would allow monitoring the condition of structures with differentiation of the damage type. It is proposed to increase the efficiency of monitoring such structures by performing preliminary works aimed at determining the characteristics of technical means that affect the parameters of the entire monitoring system. A method for determining the optimal number of measurement channels is proposed, which would allow optimizing the design of the diagnostic system and ensuring high responsivity of this system and reliability of defect detection. A procedure for collecting data based on acoustic emission signals and their processing with their subsequent digitization and recognition using neural network technologies using the Kohonen card is also proposed. A distinctive feature of this method is the possibility to classify results into three categories: fatigue crack; plastic deformation; friction (mechanical noise). The practical significance lies in the possibility to create optimal monitoring systems for various complex structures, which would allow obtaining differential signs of mechanical damage.