人工神经网络预测管道冲蚀速率研究进展

Recent progress on ANN-based pipeline erosion predictions

本文系统综述了近年来采用人工智能方法对固体壁面受固体颗粒冲击摩擦导致质量损失现象的研究成果,回顾了现有的理论与实验研究结论,分析了人工智能方法的优势及与传统研究方法的互补性。从数据来源的角度分类,讨论了以实验数据为基础及以CFD模拟为基础的人工神经网络方法预测结果。同时,简要介绍了最新的应用支持向量机、随机森林等方法进行冲蚀研究的案例。近年来的研究结果表明,人工智能方法在管道冲蚀现象的研究中具有很高的应用潜力,将在未来管道完整性管理水平的提升及管道智能化建设的加速中发挥重要的作用。

We systematically review the recent studies on wear and friction of pipeline walls caused by solid particle impact using AI related techniques. We examine the existing theoretical and experimental approaches, and analyze the advantages of Artificial Intelligence methods and their complementarity with classic research methods. According to the classification of data sources, the prediction results of Artificial Neural Networks(ANNs) based on experimental data or CFD simulation are discussed separately. The latest cases of erosion research using Support Vector Machine, Random Forest and other methods are also briefly introduced. Recent research conclusions show that AI method has high potential for application in the prevention of pipeline erosion, and will play an increasingly important role in the improvement of pipeline integrity management and the acceleration of Intelligent Pipeline Construction in the future.