【作者】 刘延俊；

【导师】 张建华；

【作者基本信息】 山东大学 ， 机械制造及其自动化， 2008， 博士

【摘要】 气动系统以其结构简单、无污染、性价比高、维修方便及抗干扰能力强等优点,被广泛应用于化工、医药、纺织、微电子、生物工程等工业自动化领域中。气动比例技术的出现,使气动系统从逻辑控制领域扩展到比例/伺服控制领域。但是由于气动系统固有的非线性、刚度小、阻尼比小以及固有频率低等缺点,使得气动比例位置系统定位技术进展缓慢,其控制精度和工作性能难以达到理想的效果,从而限制了气动系统在工业领域中的推广及应用。本文主要以提高气动比例系统的控制精度为目标,通过分析其摩擦力及动态特性,对系统的摩擦非线性补偿及智能控制策略进行研究。本文综述了气动比例系统的特点及发展状况,分析了气动系统的缓冲定位技术,阐述了智能控制技术在该领域中的研究与应用。在深入分析气动比例系统的工作性能及特点的基础上,研究系统的摩擦机理,并且通过叠加高频低幅颤振信号补偿系统的摩擦力,结合气动比例系统的非线性特征,设计出模糊神经元网络控制器,使系统获得了良好的控制精度。首先,对气动比例系统进行了两种方法的数学建模研究。一种方法是机理建模,即通过分析气缸与比例方向阀的力平衡特性以及压力-流量特性,建立了系统的非线性数学模型,并对模型进行了系统辨识,得到了较为精确的数学模型,以便于为下一步的研究提供依据。另一种方法是基于图形化的物理建模,通过仿真软件AMESim,建立气动比例位置系统的数学模型,这种方法可以最大程度的考虑系统的细节问题,从而能够得到更加准确的数学模型。由于图形化建模是利用AMESim与Matlab/simulink的联合仿真平台,把两个优秀的专业仿真工具联合起来使用,既发挥AMESim突出的流体机械的仿真效能,又能借助MATLAB/Simulink强大的数值处理能力,取长补短,取得更加完美的互补效果,所以本文的所有研究都是在基于图形化物理建模的基础上进行的。其次,研究了补偿气动比例阀控缸系统摩擦力的理论。由系统的摩擦力带来的稳态误差和低速爬行问题,通常是通过提高运动部件的加工精度和改进系统的润滑条件来解决。在气动比例系统中,可以通过改进气缸的机械结构,或者采用高精度新型气缸等措施来减少非线性摩擦对系统运动性能的影响,但是由于这种方法会使成本显著增加,也不可能最终消除非线性摩擦,所以会影响系统的定位精度和轨迹跟踪精度。仿真及实验研究表明,通过叠加颤振信号补偿系统的摩擦力,可克服气动比例阀的中位死区,提高系统的灵敏度和动态响应特性。再次,针对系统的非线性特性,研究了气动比例系统的智能控制方法。文中分别对智能控制领域内的模糊控制、神经网络控制的方法进行了分析和仿真研究。模糊控制能仿效人的模糊逻辑思维方法,允许系统在工作过程中某些数值型量的不精确性存在。但是模糊规则的确定对操作人员的经验以及语言表达方式有一定的依赖性,不同人员对于问题认识的深度和综合能力直接影响到模糊控制系统的工作性能。神经元网络通过其结构的可变性,逐步适应外部环境的各种因素的作用,能够从不十分精确的输入/输出值描述中挖掘出研究对象之间的因果联系,从而达到解决问题的目的。为了减小控制系统对经验知识的依赖性,增强控制系统的学习能力以提高控制系统对运行过程中工况条件变化时的适应能力,针对比例阀控缸对象一类的非线性、时变不确定系统,考虑采用神经元网络技术或模糊神经元网络技术来解决。因此,提出了一种模糊神经元网络(FNN)的控制方法,即在基本模糊控制器的基础上,引入神经元网络技术,利用神经元网络的学习功能结合模糊逻辑推理,以进一步改善比例阀控缸系统性能、提高系统的适应能力。仿真结果表明,针对该系统设计的模糊神经元网络控制器,能够很好的克服外界负载扰动对比例阀控缸系统的影响,使系统的鲁棒性提高。最后,将模糊神经元网络的自学习功能和叠加颤振信号的补偿方法相结合,对比例阀的死区进行了补偿实验研究;在对高次曲线进行理论分析的基础上,提出了采用高次曲线作为理想曲线实现气动比例阀控缸位置控制系统轨迹跟踪的方法;通过研制两自由度气动比例系统的控制程序及控制界面,实现了气动比例系统在平面内的高精度轨迹跟踪研究;单自由度系统定位精度控制在±0.100mm以内,平面两自由度连续轨迹跟踪精度控制在±0.263mm以内,可以替代价格昂贵的伺服系统。更多还原

【Abstract】 Pneumatic systems have been widely used in industrial automation fields, such as chemical industry, medicine, textile, micro-electrics and bioengineering, for its advantages of simplicity in structure, anti-pollution, high performance to cost, easy maintenance and anti-jamming. With the development of the pneumatic proportional technology, pneumatic system control is extended from logical control to proportional/servo control area. However, pneumatic systems have some inherent disadvantages as nonlinearity, low stiffness, low damping ratio and low natural frequency, it is difficult to obtain satisfied control performance, and thus its uses in industry area are limited. In this paper, the study on improving the control precision for the pneumatic proportional system is presented. The system friction nonlinear compensation and intelligent control method are developed based on the study of its friction and dynamic characteristics.In this paper, the features and development of pneumatic proportional system are summarized, the buffer positioning technology in pneumatic system is analyzed, also the research and application of intelligent control technology in this domain is expounded. Based on the study on working performance and features of the system, the friction mechanism is developed, the friction is overcome by adding high frequency and low amplitude chatter signals to the system, the fuzzy neural network controller is presented according to its nonlinear, which achieves good precision for the system.Two mathematical modeling methods on pneumatic proportional system are presented. One method is called mechanism modeling, which contains the establishment of nonlinear mathematical model of the system through analyzing the balance of power and pressure - flow characteristics of the cylinder and the proportional direction valve, a more accurate mathematical model is attained according to the system identification in order to provide the basis for further research. The other method is based on the graphical modeling of physical model. Pneumatic proportional system model is established by using the modeling and simulation software AMESim, in this way, the details of the system can be considered at utmost, and more accurate system model can be gained. The graphical modeling is based on the joint simulation platform of AMESim and Matlab / Simulink. The outstanding performance of AMESim is its simulation on fluid mechanism, but MATLAB / Simulink has powerful numerical processing power. The two outstanding professional simulation tools are used together in this study and a perfect complementary effect will be gained. Therefore, all studies in this paper are based on the graphical modeling of physical model.Theoretical analysis on friction compensation based on valve-controlled cylinder system is presented. To overcome the system steady-state error and scrawl under low velocity caused by friction, we usually improve the machining accuracy and lubricating of the moving parts to reduce the system friction. However, in pneumatic proportional system, the nonlinear friction influence on the motion performance can be reduced by improving the mechanical structure of the cylinder, or using new high precision cylinders, but these methods will lead to high cost, the nonlinear friction can not eliminated thoroughly, and the system positioning precision and low-velocity tracking precision can not be improved finally. So that, it is necessary to use adding chatter signal method with friction compensation to overcome the dead area of the pneumatic proportional valve, which improves the sensitivity of the system and reflects the dynamic response characteristics.Based on the non-linearization of the system, intelligent control method on non-linearization of pneumatic system is presented. Fuzzy control and neural network control methods are analyzed and simulated separately. Fuzzy control can follow the fuzzy logic methods of human being and allows the existence of the non-precious numerical type. But the determinations of the fuzzy rules depend on the experience and the expression of operators, therefore, the depth of the understanding of the problem and comprehensive capacity of the operators will directly affect the system performance . Neural network can adapt to the external environment and other factors gradually by the aid of variability of its structures. Besides, it can gain the causal relation between the research objects from less accurate input / output description, then solve the problems.To reduce the dependence on prior knowledge of the control system, and improve the learning ability of the control system to enhance the adaptability to the change of working conditions, the nonlinear of valve-controlled cylinder, time-varying uncertainty, neural network technology or FNN technology can be used to solve these problems. Therefore, FNN control method is presented. By introducing neural network technology on the basis of fuzzy controller, using the learning function of neural network and fuzzy logic, the proportional valve-controlled cylinder system performance and adaptability are improved. The simulation results show that, the fuzzy-neural network controller designed for the system can overcome the outside load disturbance on the system well, and greatly enhance the robustness of the system.Finally, the self-learning function of the fuzzy-neural networks and compensation methods of adding chatter signal to the system are applied to the pneumatic system. The dead area compensation experimental of the proportional valve is presented. On the basis of theoretical analysis for high-order curve, the method which can achieve the tracking trajectory of pneumatic proportional valve-controlled cylinder position control system by using the high-order curve as the ideal curve is put forward. The tracking trajectory control of the pneumatic proportional system is realized according to setting up the control procedures and control interface of a two freedom pneumatic proportional system. the positioning precision of one freedom system is within±0.100mm,and the continuous tracking trajectory control precision of two freedom system is within±0.263mm,it is suitable for replacing some expensive servo system.更多还原