In-service pipeline corrosion is an important reason for failure of oil pipeline transport.Timely detection of corrosion of pipes in service is essential.Ultrasonic guided waves are used for long-distance pipeline inspection

and the artificial neural networks are used for intelligent recognition of pipeline defects.Pipes defect detection experiments are performed by the ultrasonic guided wave detection equipments.Based on the original detection signal

signal processing is performed

from the result of which sample-features are extracted.It builds and trains a BP neural networks for defect identification.The experiment results demonstrate that the BP neural networks identifies the defect type and achieves good effect.