CAJ | 학술논문

提出了在采用最近邻聚类算法在线构造RBF(NN1)正向辨识器,并在线辨识被控对象的Jacobian阵的基础上,引入RBF(NN2)对PID控制器参数进行在线调整的算法.该算法可以实现PID控制参数的自动在线整定和优化.通过和BP网络调整控制器参数的方法的对比和仿真实验证明,该方法控制精度高、响应速度快,并且具备较强的自适应性和鲁棒性.
제출료재채용최근린취류산법재선구조RBF(NN1)정향변식기,병재선변식피공대상적Jacobian진적기출상,인입RBF(NN2)대PID공제기삼수진행재선조정적산법.해산법가이실현PID공제삼수적자동재선정정화우화.통과화BP망락조정공제기삼수적방법적대비화방진실험증명,해방법공제정도고、향응속도쾌,병차구비교강적자괄응성화로봉성.
A new method is proposed, in which parameters of the PID controller are optimized online by RBF (NN2)based on con-strutting forward RBF (NN1) identifier online to identify the Jacobian matrix of the controlled plant by means of adopting the nearest neighbors clustering algorithm. The deep analyzing and comparison to BP net are made. The simulation results demonstrate that the method has perfect dynamic performanances, quick response speed, and upper precision, and possesses great adaptability and robustness.