四旋翼ESO的RBF神经网络PID控制器研究  

ESO Based RBF Neural Network PID Controller for Quadrotor Aircrafts

作　　者：刘春玲[1] 王明 张瑾[1] LIU Chunling;WANG Ming;ZHANG Jin(College of Information Engineering,Dalian University,Dalian 116000,China)

机构地区：[1]大连大学信息工程学院,辽宁大连116000

出　　处：《电光与控制》2021年第9期84-88,93,共6页Electronics Optics & Control

基　　金：辽宁省科学技术基金(2019ZD0311)。

摘　　要：四旋翼飞行器因存在参数不确定性和环境干扰,会出现姿态不稳定的问题,而传统的PID控制对四旋翼的姿态稳定及机动性达不到控制需求。为此,提出了一种扩张状态观测器(ESO)的RBF神经网络PID控制器。首先,利用ESO的扩张特性和非线性函数对扰动进行估计和补偿,减少系统的误差;其次,将ESO对系统输出的估计值作为RBF神经网络的输入,使梯度信息更加精确,能够更好地优化增量PID的参数;最后,该神经网络的激励函数取高斯基函数,利用RBF神经网络的自适应性、自学习能力对模型控制参数进行调整。Matlab仿真实验表明,在未知干扰环境下,ESO的RBF神经网络PID控制器能够明显提高系统的抗干扰能力,且具有较小的超调量及较好的鲁棒性。The parameter uncertainty and environmental interferences may result in unstable attitude of quadrotor aircrafts,and the traditional PID control can’t meet the control requirements of the quadrotor’s attitude stability and maneuverability. Aiming at the problem,an Extended State Observer( ESO) neural network RBF PID controller is proposed. Firstly,the extension characteristics of ESO and nonlinear functions are used to estimate and compensate for disturbances to reduce system errors. Secondly,the ESO’s estimated value of the system output is used as the input of the RBF neural network,to make the gradient information more accurate and better optimize the parameters of the incremental PID. Finally,a Gaussian function is adopted as the excitation function of the neural network,and the model control parameters are adjusted by using the self-adaptability and self-learning ability of the RBF neural network. The Matlab simulation experiment shows that: in the unknown interference environment,the ESO’s RBF neural network PID controller can significantly improve the anti-interference ability of the system,and has a smaller overshoot and better robustness.

关 键 词：四旋翼控制 扩张状态观测器 径向基函数神经网络 比例积分微分控制 

分 类 号：V249.1[航空宇航科学与技术—飞行器设计]