Artificial neural network for the classification of steel hollow pipe

Abstract

Within industry, piping is a very important system that used to convey fluid (liquid and gases) from one location to another. Steel pipe is one of the commonly type of pipe that has been used since before. Crack on pipe is one of the things that always happen on pipe due to transfer fluid. Non-destructive (NDT) testing is responsible to detect the damage on pipe to avoid from bursting. From this, the paper presents a NDT method to detect damage in pipe by using Artificial Intelligence Neural Network (ANN) to compare Frequency Response Function (FRF) derived from impact testing on intact and damage pipe. Carbon Steel pipe with different hollow through the pipe in free-free condition is considered as a specimen. A simple feedforward with multilayer backpropagation neural network models is developed for the recognition of intact and damage steel pipe. FRF data presented on variation of amplitude load vs. frequency wave depends on disposition features can be very useful in crack detection in pipelines knowing the frequencies. This indicates that the representation of intact and damage pipe by the frequency using Artificial Neural Network (ANN) is reasonably accurate. Experimental results demonstrate that the recognition rate of the proposed neural network models is about 91.48%