Active aerodynamic flow control system

Abstract

The invention proposes a system for controlling the flow behind an object using a jet stream, employing a neural network for enhanced aerodynamic drag reduction. The system collects flow state and actuator state data, then performs preprocessing and feature extraction to construct a state representation of the system. Based on this state, the neural network model infers the optimal control action for the jet nozzles, where the control signal is converted into physical parameters to control the jet system's impact on the flow structure behind the object. The controlled flow state is evaluated through aerodynamic indices to determine drag reduction efficiency and is used as feedback to improve the control strategy in subsequent cycles. As a result, the system is capable of self-learning and optimizing the flow control process, contributing to improved aerodynamic performance of the object in a flowing environment.

Images