【作者】 黄镇海；

【导师】 方一鸣；

【作者基本信息】 燕山大学 ， 控制理论与控制工程， 2002， 硕士

【摘要】 本课题是在已完成的国家“九五”科技攻关项目“结晶器非正弦振动的研究”的基础上，针对该项目中的液压伺服驱动的连铸结晶器振动模拟装置所做的控制方面的研究。由于液压伺服驱动位置系统中存在参数摄动、外干扰和伺服阀的死区非线性等棘手的问题，而模糊控制不需要精确的数学模型，而神经网络可以有效地解决非线性问题，因此本文采用模糊控制和神经网络等控制方法对液压伺服位置系统进行控制，具有较大的实际意义和理论价值。 本文首先利用1250频率特性测试仪所测的实际液压伺服位置系统被控对象的高阶模型，然后结合实测数据得到被控对象的简化机理数学模型，并以此模型作为控制器设计的依据。在对模糊控制和神经网络理论的特点和应用等进行概述的基础上，分别设计了自调整模糊控制器、模型参考模糊自适应控制器、基于模糊逐级误差逼近算法的模糊神经网络控制器和神经网络变结构控制器。其中，神经网络变结构控制器是在利用神经网络对伺服阀摩擦死区非线性进行了辨识，并利用神经网络拟合的死区非线性的逆模型进行了补偿，然后再用变结构控制对液压伺服位置系统进行了控制器的设计，解决了液压伺服系统中伺服阀的死区非线性问题。通过对所设计控制器的仿真研究，结果表明，本文所设计的液压伺服位置控制系统具有较好的动静态性能、较好的跟踪精度，证明了所设计控制器的可行性。更多还原

【Abstract】 This paper is a theoretical research that is based on the having been finished project of the state ’’the ninth five years" tackling key scientific and technological problems and facing the continuous casting analogy device in the laboratory of our school. It is difficult for the hydraulic servo position system to solve the problem that is caused by the parameters variety, outside disturbance and the friction nonlinear dead zone of the valve. It is well known that it is not necessary to get an accurate model of the system for fuzzy control. We also know that neural network can solve the problems caused by non-linear effectively. So the research that two methods are applied to control the hydraulic servo position system has practical meaning and theory value.According to analysis the frequency graph tested by 1250 frequency detector, the high order mathematical model can be deduced. Then we get a simplified mechanism model by combining the practical model and use the simplified model to design the controllers. On the basis of the summary of the characteristics and application of fuzzy control and neural network theory, a self-tuning fuzzy control, a model reference fuzzy adaptive controller, a fuzzy neural network based on fuzzy hierarchy error approach and a neural network variable structure controller are proposed. In the fourth controller, a neural network is used to identify the friction nonlinear dead zone of the valve, and then another neural network is used to fit the inverse model of the nonlinear for compensating the nonlinear and a variable structure controller is designed to control the hydraulic servo position system. And the problem of the friction nonlinear dead zone of the valve is solved. The result of simulation shows that hydraulic servo position control system has good performances in the dynamic and static state control accuracy, which shows the feasibility of these control methods.更多还原