模糊神经网络在冷连轧厚度控制中的应用

摘要

由于冷连轧厚度控制系统具有非线性、大时滞的特点,在冷连轧厚度的常规PID控制中,PID控制器的参数往往针对某一种情况进行整定,很难控制冷连轧厚度始终处于一个好的状态.为此,在分析了厚度控制原理的基础上,设计了用一个2-5-1结构的BP网络实现的模糊神经网络控制器(FNNC),并将该FNNC控制器与积分作用相结合构成一个FNNC-I控制器.仿真结果表明,该FNNC-I控制器提高了系统的动态和稳态性能、抗干扰性以及鲁棒性,其控制效果优于常规PID控制器.%There are nonlinear, large time delay characteristics of tandem cold mill thickness control,so it is difficult to keep thickness within a small tolerance using PID controller, whose parameters are set only for one stable situation. Based on the analysis of thickness control theory, a fuzzy neural network controller (FNNC) with simple structure was designed, which was realized by a BP network with 2-5-1 structure. On the basis of this controller, an intergral action was added to constitute FNNC-I controller. Simulation results show that the dynamic, static, anti-interference performance and the robusness of the system were all improved by this FNNC-I controller, so it is better than the conventional PID controller.