Assessment for burst failure of subsea production pipeline systems based on machine learning
OCEAN ENGINEERING(2024)
摘要
Subsea production system occupies an important position in the development of deep-sea resources. As a key part of the subsea production system, the safety and reliability of the pipeline system of the subsea production system is crucially important. In recent years, with the development of artificial intelligence technology, the use of machine learning algorithms for large-scale complex pipeline system to provide a possibility of safety assessment. This paper collects the burst pressure of the pipeline in the experiments and uses various machine learning methods to evaluate and predict. The XGBoost model with the best performance after preliminary selection, and the accuracy of this method is improved through hyperparameter optimization with Bayesian principle. The improved XGBoost model are explained based on the Forward Feature Selection method and SHAP analysis, compared with commonly pipeline industry standards. The results show that the improved XGBoost model can significantly improve the prediction of the burst pressure and can effectively assess the structural integrity and reliability of pipelines. Moreover, the influencing factors of the burst pressure of the pipeline are analysed, and it is concluded that t (thickness of the pipeline), L (length of corrosion) are the two most important factors influencing the pipeline burst pressure.
更多查看译文
关键词
Subsea production,Pipeline system,Machine learning,Burst pressure,XGBoost
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要