Research on corrosion defect identification and risk assessment of well control equipment based on a machine learning algorithm

Research on corrosion defect identification and risk assessment of well control equipment based on a machine learning algorithm

K-means++

KNN

Machine learning

Pulse magnetic flux leakage testing

Risk assessment

In recent years, the risk assessment of well control equipment has faced some problems, such as shallow defect detection depth, large identification error of corrosion defect type, inaccurate equipment corrosion assessment, and so on. To solve the above problems, a corrosion defect classification and identification model based on an improved K Nearest Neighbor algorithm (KNN) is established for the well control pipeline in well control equipment. Firstly, the Pulsed Magnetic Flux Leakage (PMFL) sensor is used to detect the pipeline defects, and then the collected data are denoised. Then, the corrosion type identification model of well control pipeline based on K-means++ and KNN is established. Finally, the corrosion risk of well control pipeline is evaluated according to the type of corrosion output from the identification model. The experimental results show that the improved algorithm has high accuracy in identifying the corrosion type of well control pipeline, and the calculation speed is better than other algorithms described in this paper.

Zhang, He

Cao, Jiangna

Liang, Haibo

Cheng, Gang

Petroleum

2022

journalArticle

2405-6561

10.1016/j.petlm.2022.07.003