An active fault-tolerant control method for unmanned aerial vehicles based on reinforcement learning

Abstract

The invention discloses an active fault-tolerant control method for unmanned aerial vehicles based on reinforcement learning, which specifically includes two stages, the offline training stage in the early stage: by collecting the historical attitude generated when the unmanned aerial vehicle is running and the data output by the controller, the reinforcement learning The evaluation network of the fault-tolerant controller is trained and updated, and the evaluation network in the present invention is optimized by using the genetic algorithm to optimize the extreme learning machine, which improves the training speed and training accuracy; system operation and online training phase: during the operation of the drone, The reinforcement learning evaluation network is used for real-time online update, and the self-learning and self-improvement of the reinforcement learning fault-tolerant controller is realized through online update in the process of UAV active fault-tolerant control, and the real-time online update of the extreme learning machine is realized through the dynamic expansion update algorithm. The invention adopts an incremental strategy to optimize the reinforcement learning method, realizes the asymptotic approach to the optimal fault-tolerant control strategy, and can better realize the fault-tolerant control of the unmanned aerial vehicle.

Description

technical field [0001] The invention relates to an active fault-tolerant control method for unmanned aerial vehicle based on reinforcement learning, in particular to an active fault-tolerant control method for unmanned aerial vehicle based on extreme learning machine and incremental strategy reinforcement learning, belonging to the technical field of active fault-tolerant control for unmanned aerial vehicle . Background technique [0002] With the continuous development of aerospace technology, the scale of the flight control system is becoming larger and larger, and the complexity of the system is also increasing. While the flight control system is constantly improving, the stability of the system is also facing great challenges. Any type of failure can lead to system performance degradation or even paralysis, resulting in instability of the control system, resulting in huge losses. Therefore, how to reduce or even eliminate the danger caused by system failure is a proble...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is to provide an active fault-tolerant control method for unmanned aerial vehicles based on reinforcement learning, which solves the problems existing in the prior art, such as the accuracy of mathematical modeling will greatly affect the fault-tolerant effect and the traditional deterministic strategy reinforcement learning For problems such as poor fault-tolerant control effect, it has strong real-time and adaptability

Images