基于Hammerstein模型压电陶瓷执行器迟滞非线性建模及控制方法
发布时间：2018-08-13 14:48
【摘要】：微纳米执行器是微纳米定位技术的核心元件,寻找性能优良的执行器材料对提高微纳米定位精度具有十分重要的意义。压电陶瓷是一种能使机械能量和电能量之间相互转换的智能执行器材料,它具有价格低、质量轻、频率响应快、位移精度高、驱动力大等优越点,近些年成为了一个研究热点,并广泛应用于航天航空、机器人、微控制工程等领域。然而压电陶瓷材料存在迟滞现象,该特性导致压电陶瓷执行器响应速度变慢、可控性变差,严重阻碍压电陶瓷执行器的应用与发展。因此,如何消除迟滞现象成为实现压电陶瓷执行器精密位移控制必须要攻克的难题。消除迟滞非线性常用的思路可归纳为建模和控制两种,针对压电陶瓷执行器中率相关迟滞非线性的现象,本文首先对其建立率相关迟滞非线性模型,由于传统KP模型属于率不相关的迟滞非线性模型,本文因此在经典KP模型的基础上加以改进,构造出Hammerstein模型,该模型可以描述压电陶瓷执行器中频率相关迟滞非线性(率相关迟滞非线性)。此模型由经典KP模型与传递函数串级组成,KP模型用来表征迟滞非线性,传递函数部分用来描述动态率相关性特性。其次,本文分步对其进行参数辨识,并采用和声搜索优化算法和蝙蝠和声混合优化算法对KP模型的密度函数进行辨识,选取误差较小的一种辨识算法。最后,为了减小甚至消除率相关迟滞非线性对位移控制精度的影响,本文提出了三种控制方案:第一种控制方案是迟滞逆模型补偿的前馈控制方案,实验表明,本文所提出的前馈控制器有效地减少了率相关迟滞非线性对位移控制精度的不利影响。开环控制虽然提高了跟踪精度,但是系统的抗干扰能力较弱,于是本文提出第二种控制方案,采用BP算法整定PID参数与前馈相结合的复合控制方案,进一步提高了控制精度,而且增强了系统的抗干扰能力。由于复合控制仍然是基于模型的控制方法,那么建模不确定性依然会削弱控制效果。第三种方案是一种无需逆模型的控制方案,主要思想是基于神经网络逼近非线性函数的超强能力,用两个神经网络分别逼近控制系统函数关系中的两个未知量,以达成系统输出无差跟踪系统输入的理想目的。实验数据表明,同复合控制相比,神经网络控制的最大误差和均方根误差均有减小。实验结果证明了神经网逼近非线性函数控制系统的跟踪精度优于压电陶瓷执行器复合控制系统的精度。
[Abstract]:Micro / nano actuators are the core components of micro / nano positioning technology. It is very important to find good actuator materials to improve the precision of micro / nano positioning. Piezoelectric ceramic is a kind of intelligent actuator material which can convert mechanical and electrical energy. It has many advantages such as low price, light mass, fast frequency response, high displacement precision, high driving force and so on. In recent years, piezoelectric ceramics has become a research hotspot. And widely used in aerospace, robotics, micro-control engineering and other fields. However, the hysteresis phenomenon of piezoelectric ceramic materials leads to the slow response speed and poor controllability of piezoelectric actuators, which seriously hinders the application and development of piezoelectric actuators. Therefore, how to eliminate hysteresis becomes a difficult problem to realize the precision displacement control of piezoelectric actuator. The common ideas for eliminating hysteresis nonlinearity can be summarized as modeling and control. Aiming at the phenomenon of rate-dependent hysteresis nonlinearity in piezoelectric ceramic actuators, a rate-dependent hysteresis nonlinear model is established in this paper. Because the traditional KP model belongs to the rate independent hysteretic nonlinear model, this paper improves on the classical KP model and constructs the Hammerstein model. The model can be used to describe the frequency dependent hysteresis nonlinearity in piezoelectric actuator. The model is composed of the classical KP model and the transfer function cascade to characterize the hysteresis nonlinearity, and the transfer function part is used to describe the dynamic rate correlation. Secondly, the parameter identification of KP model is carried out step by step, and the density function of KP model is identified by harmony search optimization algorithm and bat harmony sound hybrid optimization algorithm, and a less error identification algorithm is selected. Finally, in order to reduce or even eliminate the influence of rate-dependent hysteresis nonlinearity on displacement control accuracy, three control schemes are proposed in this paper: the first control scheme is feedforward control scheme with hysteresis inverse model compensation. The feedforward controller presented in this paper can effectively reduce the adverse effect of the rate dependent hysteresis nonlinearity on the displacement control accuracy. Although the open-loop control improves the tracking accuracy, the anti-jamming ability of the system is weak, so the second control scheme is proposed in this paper. The BP algorithm is used to adjust the PID parameters and the feedforward compound control scheme, which further improves the control accuracy. Moreover, it enhances the anti-interference ability of the system. Because compound control is still a model-based control method, modeling uncertainty will still weaken the control effect. The third scheme is a control scheme without inverse model. The main idea is to use two neural networks to approximate the two unknown variables in the function relationship of the control system, based on the super ability of the neural network to approximate the nonlinear function. In order to achieve the system output differential tracking system input ideal purpose. The experimental data show that the maximum error and root mean square error of neural network control are smaller than that of compound control. The experimental results show that the tracking accuracy of the neural network approximating nonlinear function control system is better than that of the piezoelectric ceramic actuator composite control system.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN384;TP273
本文编号：2181302
[Abstract]:Micro / nano actuators are the core components of micro / nano positioning technology. It is very important to find good actuator materials to improve the precision of micro / nano positioning. Piezoelectric ceramic is a kind of intelligent actuator material which can convert mechanical and electrical energy. It has many advantages such as low price, light mass, fast frequency response, high displacement precision, high driving force and so on. In recent years, piezoelectric ceramics has become a research hotspot. And widely used in aerospace, robotics, micro-control engineering and other fields. However, the hysteresis phenomenon of piezoelectric ceramic materials leads to the slow response speed and poor controllability of piezoelectric actuators, which seriously hinders the application and development of piezoelectric actuators. Therefore, how to eliminate hysteresis becomes a difficult problem to realize the precision displacement control of piezoelectric actuator. The common ideas for eliminating hysteresis nonlinearity can be summarized as modeling and control. Aiming at the phenomenon of rate-dependent hysteresis nonlinearity in piezoelectric ceramic actuators, a rate-dependent hysteresis nonlinear model is established in this paper. Because the traditional KP model belongs to the rate independent hysteretic nonlinear model, this paper improves on the classical KP model and constructs the Hammerstein model. The model can be used to describe the frequency dependent hysteresis nonlinearity in piezoelectric actuator. The model is composed of the classical KP model and the transfer function cascade to characterize the hysteresis nonlinearity, and the transfer function part is used to describe the dynamic rate correlation. Secondly, the parameter identification of KP model is carried out step by step, and the density function of KP model is identified by harmony search optimization algorithm and bat harmony sound hybrid optimization algorithm, and a less error identification algorithm is selected. Finally, in order to reduce or even eliminate the influence of rate-dependent hysteresis nonlinearity on displacement control accuracy, three control schemes are proposed in this paper: the first control scheme is feedforward control scheme with hysteresis inverse model compensation. The feedforward controller presented in this paper can effectively reduce the adverse effect of the rate dependent hysteresis nonlinearity on the displacement control accuracy. Although the open-loop control improves the tracking accuracy, the anti-jamming ability of the system is weak, so the second control scheme is proposed in this paper. The BP algorithm is used to adjust the PID parameters and the feedforward compound control scheme, which further improves the control accuracy. Moreover, it enhances the anti-interference ability of the system. Because compound control is still a model-based control method, modeling uncertainty will still weaken the control effect. The third scheme is a control scheme without inverse model. The main idea is to use two neural networks to approximate the two unknown variables in the function relationship of the control system, based on the super ability of the neural network to approximate the nonlinear function. In order to achieve the system output differential tracking system input ideal purpose. The experimental data show that the maximum error and root mean square error of neural network control are smaller than that of compound control. The experimental results show that the tracking accuracy of the neural network approximating nonlinear function control system is better than that of the piezoelectric ceramic actuator composite control system.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN384;TP273
【参考文献】
中国期刊全文数据库 前10条
1 孙丽;甘博;崔迪;;形状记忆合金作为钢框架混凝土墙结构抗剪连接件性能研究[J];沈阳建筑大学学报(自然科学版);2016年03期
2 王扬威;于凯;王振龙;;形状记忆合金丝驱动仿生波动鳍推进器设计与实验研究[J];中国机械工程;2015年08期
3 郭咏新;张臻;毛剑琴;周克敏;;超磁致伸缩作动器的率相关Hammerstein模型与H_∞鲁棒跟踪控制[J];自动化学报;2014年02期
4 董晓蓉;左孝青;钟子龙;谢香云;;医用多孔Ni-Ti形状记忆合金的制备和性能研究进展[J];材料导报;2014年03期
5 陈云;吕西林;蒋欢军;;新型耗能增强型形状记忆合金阻尼器减震性能研究[J];湖南大学学报(自然科学版);2013年02期
6 李朋志;葛川;苏志德;闫丰;隋永新;杨怀江;;基于动态模糊系统模型的压电陶瓷驱动器控制[J];光学精密工程;2013年02期
7 赖志林;刘向东;耿洁;;压电陶瓷执行器的类Hammerstein模型及其参数辨识[J];光学精密工程;2012年09期
8 陈辉;谭永红;周杏鹏;张亚红;董瑞丽;;压电陶瓷执行器的动态模型辨识与控制[J];光学精密工程;2012年01期
9 赖志林;刘向东;耿洁;李黎;;压电陶瓷执行器迟滞的滑模逆补偿控制[J];光学精密工程;2011年06期
10 裴先茹;高海荣;;压电材料的研究和应用现状[J];安徽化工;2010年03期
中国硕士学位论文全文数据库 前1条
1 李康;压电陶瓷驱动器的迟滞非线性建模与控制[D];山东大学;2014年
，本文编号：2181302
本文链接：https://www.wllwen.com/shoufeilunwen/xixikjs/2181302.html
