Application of the regression neural network for the analysis of the results of ultrasonic testing

Abstract

Conducting a study on this topic becomes relevant due to the great importance of the safety of critical infrastructure facilities and the presence of operational defects in equipment elements and pipelines, which poses serious threats, including the possibility of equipment destruction and negative environmental impact. The purpose of this work is to study the possibility of using the diffraction-time technique of ultrasonic non-destructive testing together with a deep convolutional neural network to accurately determine the numerical value of the height of an operational crack. The methods used include the analytical method, classification method, functional method, statistical method, synthesis method, and others. The study found that an automated approach to measuring crack height, based on diffraction signals and the use of neural networks, significantly improved the quality and accuracy of non-destructive testing. Ultrasonic testing is one of the most common inspection methods for detecting service cracks and is considered to be the most effective. It allows for reliable detection of defects and determination of their size without destroying the product. The results of the study emphasize the high potential and efficiency of the method in analysing the data obtained and provide confirmation of its applicability for determining the condition of objects during ultrasonic inspection. The paper emphasizes that these technologies are particularly important and effective. It is noted that their widespread use in various industries, such as medicine, aviation, and machine learning, demonstrates their power in solving complex problems. The practical significance of the work lies in the development of advanced approaches that provide new insights and methods to improve the efficiency of analysing the results, which can be applied in industry to improve the quality of control and reliability of technical facilities

Keywords

• non-destructive testing
• operational defects
• diagnostic efficiency
• data processing
• automation of procedures
• industrial monitoring