Defects Detection and Recognition in Aviation Riveted Joints by Using Ultrasonic Echo Signals of Non-Destructive Testing

The statement of the problem of defects detection and recognition in aviation riveted joints is given. The solution is proposed to be implemented using a deep neural network with a recurrent LSTM layer. An example of solving the problem of detecting and recognizing a defect in rivets is considered. The accuracy of the result obtained in the simulation was 100% for two -class and 96.25% for four-class recognition by echo signals of an ultrasonic flaw detector. It is shown that it is possible to use the fractal properties of the non-destructive testing echo signal when detecting defects. Copyright (C) 2021 The Authors.