介电型电活性聚合物驱动转动关节控制技术研究  

Research on Actuation Control of Dielectric Electroactive Polymer Rotary Joint

作　　者：朱银龙[1] 张鑫焱[1] 王化明[2] 周宏平[1] 

机构地区：[1]南京林业大学机械电子工程学院,南京210037 [2]南京航空航天大学机电学院,南京210016

出　　处：《农业机械学报》2015年第7期333-338,332,共7页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金资助项目(51305209);江苏省自然科学基金资助项目(BK20130979;BK2011735);中国博士后科学基金资助项目(2013M541678);南京航空航天大学基本科研业务费资助项目(NS2014047);江苏省高等学校大学生创新创业训练计划资助项目(201410298012Z)

摘　　要：介电型电活性聚合物(Dielectric electroactive polymer,DEAP)具有大应变、高效率、高能量密度等优点,在仿生机器人领域具有广泛的应用前景。基于生物关节驱动原理及圆柱形驱动器特点设计了单自由度转动关节。根据弹性大变形理论建立驱动器DEAP膜的机电耦合模型,通过微分方程组的迭代求解可获得关节输入电压与输出角度之间的非线性关系,但计算工作量大。将小脑模型关节控制器CMAC(Cerebellar model articulation controller)神经网络与PD控制器相结合实现输出角度与输入电压之间的非线性映射,避免了复杂计算过程。阶跃和正弦参考信号的跟踪控制试验验证了CMAC神经网络控制的可行性。Dielectric electroactive polymers have wide application prospects in research field of biomimetic robots due to the advantages such as large strain,high efficiency,high energy density,etc.1-DOF rotary joint was designed based on the actuation principle of bionic joint and the characteristic of cylinder actuator. The electromechanical coupling model of DEAP（ Dielectric electroactive polymer） of actuator was constructed based on the large elastic deformation theory,then the nonlinear relationship between input voltage and output angle of joint could be obtained by solving the equations iteratively,but this process of computing was time-consuming. To avoid the complicated calculation,CMAC（ Cerebellar model articulation controller） neural network and PD controller were combined to realize the nonlinear mapping between the output angle of joint and input voltage. Tracking control experiments were performed in tracking step and sinusoidal reference signals,which showed the feasibility of CMAC neural network control method,while reasons for long response time and phase difference were discussed in details.

关 键 词：介电型电活性聚合物 转动关节 跟踪控制 小脑模型关节控制器神经网络 PD控制器 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置] TB34[自动化与计算机技术—控制科学与工程]