【作者】 周硕；

【导师】 崔宝侠；

【作者基本信息】 沈阳工业大学 ， 系统工程， 2007， 硕士

【摘要】 近年来我国石膏纤维板行业得到了飞速发展，已成为世界上石膏纤维板需求量的大国之一。但由于工业起步比较晚，我国石膏纤维板生产的技术水平与国际先进水平相比还有相当大的差距。石膏纤维板的厚度是石膏纤维板生产中重要的质量指标，板厚控制技术也自然成为现代压机中的关键。因此，板厚控制技术与板厚预测的智能化是石膏纤维板压机控制中的主要研究课题。压机是板厚控制系统中的主要设备，它的动态特性和稳态特性对于整个板厚控制系统起着至关重要的作用。石膏纤维板在压制过程中，板材的厚度受压机的运行速度、铺料高度、主压辊压力、一区压力、二区压力、三区压力等诸多因素的影响，而这种影响具有较大的非线性和时变性。因此，采用传统的数学建模工具来建立相对准确的模型是非常困难的，而且预测精度也不能满足板厚在线控制的要求。本文利用改进Elman神经网络建立了压机板厚预测模型，探索了一种非解析原理的板厚控制建模方法，解决了复杂系统建模带来的诸多困难，仿真结果表明：该神经网络模型基本上可以反映实际系统模型的输人输出对应关系。最后针对石膏纤维板压制控制系统具有非线性、时变性及大滞后等特点，设计了一种基于遗传算法的模糊控制器，利用遗传算法来优化模糊控制器的隶属函数，克服了传统获取隶属度函数方法的不足。将该控制器应用于压机控制回路中，仿真结果表明：板厚能够得到有效的控制，该控制器与传统的模糊控制器相比具有很小的超调量和调节时间，并且达到稳态时几乎没有振荡等控制性能。更多还原

【Abstract】 Recent years, the industry of Gypsum-Fibre board has developed at full speed in our country, and our country has become one of the countries whose output of Gypsum-Fibre board is most in the world. But the industry of gypsum material starts very late in our country, especially in Gypsum-Fibre board, so currently sizable disparity exists in our country’s Gypsum-Fibre board compared with international advanced level. The board thickness is main quality target in production of Gypsum-Fibre board. So the technology of thickness control has become pivotal point in modern pressing machine. As far as the technique is concerned, the key scientific problem in the world is thick forecast and the realization of thick intelligent control.The pressing machine system is a basic part, its dynamic characteristic and steady characteristic affect the capability of entire thick control system. There are many influencing factors in the process of pressing Gypsum-Fibre board, such as machine running rate、height of stuff、main roller pressure、one area pressure , two area pressure and three area pressure. But these relations have the characteristics of nonlinear and the time-variable. So it is difficult to express the relation of parameter and dynamic characteristic using traditional modeling way. Therefore, this way has the difficult to establish mathematics model of thickness, and the precision of prediction cannot satisfy the thick on-line control request. This paper established prediction model of pressing machine based on Elman dynamic recursion network algorithms to apply flatness online prediction for enhance the steady state performance and real time performance of the system in thick control process. The simulation results indicate that this network model may basically reflect the corresponding relations of actual model’s input and output.Finally aiming at the characteristics of nonlinear、time-variable and lag, this paper has designed a fuzzy controller base on genetic algorithms. CA is employed to optimize the membership functions of the Fuzzy Logic Controller .This way overcomes the insufficiency of traditional way. The simulation results indicate that the thickness of board can be efficiently controlled, the FLC optimized by GA has better control performance than conventional FLC and it has self-adaptive capability to a certain degree.更多还原