An aircraft intelligent control method based on dynamic teaching deep reinforcement learning

By employing a dynamic teaching deep reinforcement learning method, a feedback control architecture and simulation environment for aircraft were constructed. Combined with expert controllers and neural network training, the control problem of implicit parameter uncertainty in high-speed aircraft was solved, and the dynamic performance and robustness of the controller were improved.

CN122085645BActive Publication Date: 2026-07-24BEIJING AUTOMATION CONTROL EQUIP INST
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
BEIJING AUTOMATION CONTROL EQUIP INST
Filing Date
2026-04-24
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing technologies are insufficient to effectively address the large uncertainties and complex models of implicit parameters in high-speed aircraft control, leading to a decline in controller quality or even instability.

Method used

A feedback control architecture for an aircraft is constructed using a dynamic teaching-based deep reinforcement learning approach. A deep neural network is used as the intelligent controller, combined with a simulation environment where implicit parameters are randomly biased and an expert controller. The system is trained using an Actor-Critic network structure and a hybrid experience pool to obtain an optimized control strategy.

Benefits of technology

It improves the dynamic performance and robustness of the aircraft under conditions of implicit parameter uncertainty, enhances the autonomy and intelligence of the controller, and strengthens its adaptability to complex environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122085645B_ABST
    Figure CN122085645B_ABST
Patent Text Reader

Abstract

The application provides a kind of intelligent control method of aircraft based on dynamic teaching deep reinforcement learning, comprising: constructing aircraft feedback control architecture: with deep neural network as aircraft intelligent controller, with instruction signal and real-time state as neural network input, with actuator control signal as neural network output;Constructing aircraft simulation environment containing implicit parameter random pull bias;According to implicit parameter, construct expert controller;Constructing intelligent controller neural network model: intelligent controller adopts Actor-Critic network structure when training;Constructing mixed experience pool, the data maintained by mixed experience pool includes: experience dataset, priority dataset and environment similarity dataset;Determined policy gradient is combined with dynamic expert teaching deep reinforcement learning algorithm to train deep neural network, and control strategy for maximizing the realization of flight index is obtained.The application can improve the control quality of aircraft control method facing high uncertainty of implicit dynamic parameters.
Need to check novelty before this filing date? Find Prior Art