A fault-tolerant control method for quadrotor formation based on adaptive neural network

Abstract

The invention discloses a four-rotor formation fault-tolerant control method based on an adaptive neural network. Aiming at a formation control system containing a plurality of four-rotor aircraft, the external interference that may occur in the system is considered, and the error information of the formation is used to design A quadrotor aircraft formation flight controller, using the neural network to compensate the loss caused by the failure, and using the Lyapunov stability theory to prove the stability, robustness and fault tolerance of the formation controller; based on the adaptive neural network control method , designed the attitude fault-tolerant controller of the inner loop of the quadrotor, which effectively improved the effect of attitude fault-tolerant control, and used the Lyapunov stability theory to prove that the system can still track the desired attitude in the case of actuator failure.

Description

technical field [0001] The invention relates to a four-rotor formation fault-tolerant control method based on an adaptive neural network, and belongs to the technical field of multi-agent systems. Background technique [0002] The quadrotor is a rotary-wing unmanned aircraft that can take off and land vertically. The basic structure of the quadrotor consists of a cross-shaped frame and four rotors mounted on the top of the frame. This special structure makes the control method of the quadrotor aircraft simpler and more convenient than other types of aircraft, and the attitude and position of the aircraft can be controlled by changing the rotation speed of the four rotors. The mutual coupling between the various control channels of the quadrotor aircraft is relatively weak and more flexible, so it has been widely used in power line patrol, SLAM, logistics transportation, aerial photography and other fields. [0003] With the rapid development of control theory and computer t...

AI Technical Summary

Problems solved by technology

Ordinary neural networks can obtain the weight matrix of the neural network through online or offline learning, but it is often difficult to analyze the stability of the system, while the adaptive neural network regards the weight matrix of the neural network as an adaptive variable, through The error information of the system updates the weight of the neural network. The adaptive neural network can not only realize the online self-learning of the neural network, but also ensure the stability of the system tracking error and the boundedness of the weight parameters of the neural network through the Lyapunov criterion. sex

Images