一种自抗扰控制器参数的学习算法

A Learning Algorithm for Parameters of Automatic Disturbances Rejection Controller

针对自抗扰控制器(Automatic disturbance rejection controller,ADRC)参数多且耦合性强,参数难于被确定的问题,提出了一种ADRC参数的自动调整算法.该算法以构造的控制性能函数为学习目标,根据参数对性能指标的影响,通过惩罚函数在线不断更新参数在有界区间内的概率密度分布,使得控制参数最优值的概率密度值最大.通过开环不稳定系统算例和对工业机电驱动器单元(Industrial mechatronic drives unit,IMDU)的控制实验,仿真和实验结果证明了该算法的有效性.

Considering the special characteristics of the auto- matic disturbance rejection controller （ADRC）, with emphasis on the parameters and strong coupling among them, an algo- rithm is presented in this paper for tuning the parameters of the ADRC automatically. Aiming at the minimization of the con- trol performance function, the algorithm learns an optimal set of controller parameter values of the ADRC by cost function, updating each parameter of the controller within a bounded in- terval probability density distribution constantly and making the probability density of the optimal control parameter maximum. The algorithm is applied to an open-loop unstable system and the industrial mechatronic drives unit （IMDU）, the results of simulation and experiment show its validity.