Active flow control system based on mask deep neural network and deep reinforcement learning
The invention discloses an active flow control system based on a mask deep neural network and deep reinforcement learning. The main implementation process of the active flow control system comprises the following steps: step 1, establishing a database; 2, constructing a training set and a test set a...
Saved in:
Main Authors | , , , |
---|---|
Format | Patent |
Language | Chinese English |
Published |
26.04.2024
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | The invention discloses an active flow control system based on a mask deep neural network and deep reinforcement learning. The main implementation process of the active flow control system comprises the following steps: step 1, establishing a database; 2, constructing a training set and a test set according to the time sequence; 3, building a mask deep neural network and training the mask deep neural network; and 4, designing an active flow control strategy by combining a mask deep neural network and a deep reinforcement learning algorithm. By establishing a flow field reduced-order model based on a mask deep neural network, unsteady flow field prediction of a control body under any motion interference is provided, traditional computational fluid dynamics solution is replaced, interaction with a deep reinforcement learning algorithm is carried out as an environment, rapid solution of an active flow control strategy is realized, and the flow field prediction efficiency is improved. The problems that a traditio |
---|---|
Bibliography: | Application Number: CN202410196667 |