基于深度学习的全三维气固两相流时空耦合智能预测
Deep Learning-based Intelligent Spatio-Temporal Prediction of Full 3D Gas-solid Two-phase Flow
谢心喻 1王晓放 1郝祎琛 1赵普 1谢蓉 1刘海涛1
作者信息
- 1. 大连理工大学能源与动力学院,大连 116024
- 折叠
摘要
气固流化床在化工、冶金及制药等领域得到了广泛的研究与应用.对流化床内气固两相流的动力学行为进行深入研究有利于流化床设备的设计和性能优化.本文利用深度学习技术构建了数据驱动的全三维深度时空序列模型,对流化床内气固两相流三维空间和时间维度的复杂动力学行为进行学习,并实现了对未知来流速度条件下流化床内气相和颗粒相速度场的合理预测.测试结果表明,该全三维智能模型的预测结果与CFD计算结果高度一致,具有较好的泛化能力;此外,该模型比传统的数值仿真速度快数百倍,可以用于流场的快速预测,以缓解数值仿真耗时问题.
Abstract
Gas-solid fluidized beds have been widely studied and applied in chemical,metallurgical and pharmaceutical fields.An in-depth study of the kinetic behavior of gas-solid two-phase flow in fluidized beds is beneficial to the design and performance optimization of fluidized bed equipments.In this study,a data-driven full 3D deep time-series model is constructed using the deep learning technology to learn the complex kinetic behavior of 3D gas-solid two-phase temporal flow fields in the fluidized bed.With this model,it is able to achieve a reasonable prediction of the velocity fields of gas and particle phases in the fluidized bed under unknown incoming flow velocity conditions.The test results show that the prediction results of this full 3D intelligent model are highly consistent with the CFD calculation results,and have good generalization ability.In addition,the model is hundreds of times faster than the traditional numerical simulation and can be used for fast prediction of the flow field to alleviate the time-consuming issue of numerical simulation.
关键词
深度学习/气固两相流/三维时空预测/流化床Key words
deep learning/gas-solid two-phase flow/3D spatio-temporal prediction/fluidized bed引用本文复制引用
基金项目
国家重点研发计划资助(2020YFA0714403)
国家自然科学基金青年项目资助(52005074)
中央高校基本科研业务费资助(DUT19RC3070)
出版年
2024
NETL
NSTL
万方数据