Nondestructive Test for Oil-well Tubing Defects Based on Neural Network and Multi-sensor Fusion

摘要

This paper provides the method of nondestructive test for oil-well tubing detects based on neural network and multi-sensor fusion. Oil-well tubing magnetism of leak signal with defects is gathered, and its characteristic quantities are analyzed in the experiment. Oil-well tubing defect detecting model, which is applied to detect the oil-well tubing defects, is built up by establishing a group of characteristic quantities reflecting the defects sizes and utilizes neural network to map the nonlinear connection of signal characteristic quantities and defects sizes. It adopts the different characteristic quantities and fusion arithmetic according to the different defects, and processes the flaws and orifices defect detecting experiments based on characteristic quantities and data fusion arithmetic of neural network.