【作者】 陈星；

【导师】 王益群；

【作者基本信息】 燕山大学 ， 机械电子工程， 2002， 硕士

【摘要】 现代热轧卷取机的发展主要着眼于提高产品的产量和质量。电液伺服系统在现代的热轧卷取机中的应用主要是实现带头的自动步进回避，这样就有效地消除冲击，同时也使卷取过程中各种事故大为减少，大大提高了轧制过程的自动化水平，产品产量和质量。 本文以热轧卷取机步进控制系统为研究对象，针对实际现场卷取机卷取工艺，在实验室建造了热轧卷取机系统物理实验模型，在国内首次研制了针对卷取机步进控制的电液伺服控制系统，建立了电液伺服系统的数学模型。 本论文详细研究了神经网络自适应控制结构。人工神经网络是一门发展十分迅速的交叉学科，它具有信息的分布存储，并行处理以及自学习能力等优点，遗传算法是一种借鉴生物界自然选择和自然遗传机制的随机化搜索算法，其主要特点是群体搜索策略和群体中个体之间的信息交换，搜索不依赖于梯度信息，人工神经网络和遗传算法在智能控制领域中有着广泛的应用。本文利用人工神经网络和遗传算法自身的特点结合自适应控制来实现对液压伺服系统的实时控制。 本文建立了液压伺服步进控制系统数学模型，利用遗传算法优化神经网络自适应控制模型，以Windows98为操作平台，利用Visual C++编程语言开发了可视化控制软件。该软件人机交互性能好，操作简单，实用性强。 实验证明，该控制系统能实现带头的自动步进回避，提高材料利用率，为研制高水平热轧卷取机奠定基础。更多还原

【Abstract】 The development of modern coiler system is mainly aiming at improving the yield and quality of product. Hydraulic servo system used in coiler is mainly to realize the step-by-step control. By using the servo system, vibration and striking are largely decreased, the accidents are reduced, and the level of automation is largely improved, as well as the quality of production.This paper is focused on step-by-step control system for coiler of hot rolling mill, and based on the real work process of coiler at worksite, a mechanical model for coiler control system is built at laboratory used for research work. This is the first time in China to develop an electro-hydraulic servo system for this step-by-step control system.In the thesis, neural network adaptive control method in the coiler control system is studied in detail. Artificial neural network (ANN), which is a rapidly developing crossing subjects, has many advantages such as parallel storing, parallel processing of information, teaching itself etc. Genetic algorithm (GA), a probability search algorithm, simulates the mechanism of natural choose and natural genetic and its main features are group research strategy and individual information exchange which do not depend on gradient information. ANN and GA have a broad application in intellectual control. This thesis realizes the real-time control of hydraulic servo system by using neural network adaptive control method, which is used the features and virtues of ANN and GA.This thesis establishes a GA-NNI model for identification of hydraulic system. Utilizing Windows98 as its O.S., the control software is developed by using Visual C++ language. The results show that the software has strong application value and its interface is interactive and easy to operate.Experiments show that in the system, step-by-step control is realized, the utilization rate of materials is gained. This model lays the basis for developing the better coiler.更多还原