The Applied Research On Graphics&Image Processing And Fault Classification Of The In-service Pipeline

The pipeline is the most important transport carrier for oil and gas, which is one of the five major transportations. Because of the wide distribution and the complex working environment, the original quality defects of the base metal, the welding defects forming in the pipeline laying course and the fault defects caused by the internal medium corrosion and the external force can lead to the pipeline accidents. So there is great significance for the fault defects detection of the in-service pipeline.This paper studied the digital image of the X-ray detecting film of both the base metal’s and the welding line’s faults of the in-service pipeline. For the characteristics of the X-ray detecting films which are plenty of noise, low contrast and fuzzy edges contained in the digital images, it adopted self-adaptive median filtering to reduce the images’ noise and then enhance the images’ contrast. This paper took theoretical analysis of the principles of edge detection methods, and presented a new edge detecting method that is based on morphological gradient base on experiments, which attained good results.This paper generalized the types of the base metal’s and the welding line’s faults of the in-service pipeline and analyzed their causes after a lot of reading and under the help of fault detection experts, and then divided the base metal’s fault defects into4types which are crack, round-like holes, long-strip holes, irregular corrosion, and divided the welding lines’ fault defects into6types which are lack of fusion, air holes, incomplete penetration, crack, slag inclusion. And this paper uses aspect ratio, roundness, compactness, symmetry, steepness, gray contrast of defect and the background, position of the defect as the defect’s eigenvalues, the first5eigenvalues of which are used to describe the fault defects of base metal and all of the7eigenvalues are used to describe the fault defects of welding line, and gave their respective computations.The neural network of improved BP algorithm was applied to identify and classify the faults’types of the in-service pipeline, which are marked by the acquired eigenvalues in this paper. And experiments were used to get the best values of the number of the hidden nodes, momentum coefficient, the error level, the step and such network parameters.This paper applied Visual C++to program the system, and established the database with SQL Server2005. The establishment of both the system and the database can make the rechecking of in-service pipeline’s fault defects in the digital X-ray images and management of the parameter data as a matter of convenience greatly. And it also gets ready for the further study of the features and the frequency of the faults of the base metal and welding line of the in-service pipeline through the statistical analysis of the fault defects data in the database in the future.