Failure Prediction of Underground Pipeline Based on Artificial Neural Network

Abstract

Artificial neural network has become a useful tool for many engineering problems. For the prediction and analysis of underground pipeline failure, an ANN model is established as basis of the data of buried pipelines in non-uniform settlement soil. Therefore, the failure of underground pipeline in non-uniform settlement soil is treated as a nonlinear function with several variables. Six influence factors, such as buried depth, wall thickness, pipe diameter, precipitation level, soil modulus of elasticity, and soil density, are considered in this ANN model. The ANN model is a back propagation (BP) network, and model structure is designed based on MATLAB, in which Neuron number in hidden layer and calculating function are selected. Finally, the accuracy of this ANN model and predictive results are investigated, and some suggestions are offered for the protection of underground pipeline in non-uniform settlement soil.