基于物理信息神经网络的油气管道内腐蚀预测方法

周涛涛, 刘迎正, 郑文培, 姜恒良, 刘海鹏, 夏刚

Pipeline Corrosion Prediction Method Based on Physics-informed Neural Networks

ZHOU Taotao, LIU Yingzheng, ZHENG Wenpei, JIANG Hengliang, LIU Haipeng, XIA Gang

图4 各个模型中预测点对温度和pCO2偏导数的分布图

Fig.4 Distributions of partial derivatives of prediction points by SVM (a), XGBoost (b), ANN (c) and PINN (d) models with respect to temperature and pCO2