〈Name of award and short explanation about the award〉
Excellent Student Award
It was given for the excellent poster at this symposium.
〈About awarded research〉
Numerical simulation for the large-scale systems commonly suffers from the exploding computational cost. Aimed at tackle this issue, the reduced-order model (ROM) has been drawn much attention in recent years due to its effectiveness to reduce the systems’ complexity. In the present work, a deep-learning-based reduced order modeling framework was developed for the large-scale simulation in a fluidized bed. In this framework, a Lanczos POD technique is employed to generate a set of reduced bases. Then, we exploit the time-series prediction capability on the POD coefficients using the long short-term memory (LSTM) recurrent neural network architecture. In this framework, no prior information about the underlying governing equations is required to generate the ROM. The framework proposed herein is shown to accurately predict the detailed flow features and provide a precise field of physical variables in a fluidized bed. Meanwhile, it can reduce the computational cost by several order of magnitudes. This framework would significantly contribute to the real-time simulation and optimization design in the large-scale simulations.
〈Your impression & future plan〉
It is my great honor to win this award. I would like to express my sincere appreciation to every member in my lab, especially Dr. Guangtao Duan and my supervisor Prof. Mikio Sakai. For future, I will make my best effort to continue my research about the numerical simulation.