中文摘要：

在冷轧板带的控制中,针对液压APC系统多变量、高阶次和时变性等特点,提出一种引入记忆因子（Memory Factor,MF）的RBF神经网络.为提高网络精度,采用改进的混洗蛙跳算法（Improved Shuffled Frog Leaping Algorithm,ISFLA）全优化MF-RBF神经网络,并把优化前后的神经网络进行CFC（Continuous Functional Diagram）封装,得到基于ISFLA全优化的MF-RBF神经网络智能控制器.分析轧制厚控APC系统的原理及组成,采用机理方法建立液压APC系统数学模型,并搭建模拟电路作负载,将上述控制器在西门子FM458平台进行实验研究.通过将各控制器的实验结果作比较,发现ISFLA优化后的MF-RBFNN控制器的响应速度快、精度高、适应性好,具有较高的实际应用价值.

英文摘要：

In the controlling of cold-rolled strip, aiming at the characteristics of hydraulic APC system with multi-variable, higher order and time-varying, a RBF neural network with memory factor （MF） is proposed. To improve the accuracy of the network, improved shuffled frog leaping algorithm （ISFLA） is used to fully optimize MF-RBF neural network and making normal and optimized neural networks pack the CFC （Continuous Functional Diagram） modules, then an intelligent controller that ISFLA fully optimized the MF-RBF neural network is obtained. Through analyzing principles and components of the cold rolling thickness control APC system, the hydraulic APC mathematical model is established using mechanism method and an analog circuit is also built for the load, then making the above controllers for experimental study in the platform of Siemens FM458. By comparing experiment result of each controller, it is found MF-RBFNN controller optimized by ISFLA has quick response speed, high precision and adaptability, which has high practical value.