摘要
提出了一种层叠小脑模型关节控制器(CMAC)神经网络同传统PD控制相结合的控制策略。该策略利用改进CMAC学习控制系统的不确定信息,并作为前馈补偿来确保跟踪误差的快速收敛,采用PD控制系统实现反馈控制,保证系统的稳定性,且抑制扰动。并采用跟踪-微分器对信号进行了滤波。仿真及实验结果表明:该控制策略提高了系统的控制精度和响应速度,并且具备较强的抗干扰能力和鲁棒性。
This paper proposed a new control scheme which is compose of cascaded Cerebellar Model Articulation Controller （CMAC） and PD controller. Firstly, the scheme learn the uncertainty information of controlled system by improved CMAC, which is adopted as forward compensator to reduce the traced error immediately. Secondly, it adopts PD as feedback controller to ensure the controlled object stability and increase the capacity of anti - interfere. The scheme also adopts tracking differentiator to filter the signal. The result of simulation and experiment shows that this control scheme promotes the control precision and real time speed, and possesses the better ability of anti - interfere and robustness.
出处
《电子机械工程》
2007年第2期62-64,共3页
Electro-Mechanical Engineering
基金
中国高等教育学会"十一五"教育科学研究规划资助项目(06AIP0090046)
