Abstract: Aiming at the problem of making well balance between computational efficiency and visual realism in the traditional coupling of blood flow and vessel wall, a data driven simulation method based on Periodic-corrected Network（PcNet） for coupling of blood flow and vessel wall is proposed. The feature vector of blood flow particle state based on Smooth Particle Hydrodynamics（SPH） is designed to model the mixed contributions of neighboring proxy particles on the blood vessel wall and neighboring blood particles. A semi-supervised neural network, also called improved periodic-corrected neural network is proposed. It predicts the acceleration of each particle in the next frame. The experimental results show that the simulation method achieves fast, stable and realistic coupling result of blood flow and vessel wall.

Key words: fluid-solid coupling, blood vessel, data-driven, periodic-corrected, Smooth Particle Hydrodynamics（SPH）

摘要： 针对传统的血流血管壁耦合难以兼顾计算效率和视觉真实感的问题，提出了一种基于周期性校正神经网络（Periodic-corrected Network，PcNet）的血流血管壁耦合数据驱动仿真方法。设计基于平滑粒子流体动力学（SPH）的血流粒子状态特征向量，对邻域血流粒子和血管壁代理粒子的混合贡献进行建模。提出一种半监督的神经网络——改进的周期性校正神经网络，预测每个粒子在下一帧的加速度。实验结果表明该仿真方法实现了快速、稳定、逼真的血流血管壁耦合。

关键词: 流固耦合, 血管壁, 数据驱动, 周期性校正, 平滑粒子流体动力学（SPH）