板形板厚综合调节神经模糊智能方法的研究  

Research on Nerve-fuzzy Intelligent Method of Strip Shape and Thickness Combined Governing

作　　者：贾春玉[1] 王英华[2] 周会锋[1] 

机构地区：[1]燕山大学机械工程学院 [2]新疆特变电工股份有限公司

出　　处：《中国机械工程》2003年第20期1741-1744,共4页China Mechanical Engineering

摘　　要：针对板形板厚综合调节系统的复杂性及其数学模型难以建立的特点 ,提出了基于神经模糊智能建模方法的四辊冷轧机板形板厚综合调节方案 ,研究建立了基于动态 BP神经网络的自组织模糊控制模型。它在 BP网络模型基础上 ,对网络的自身结构进行了动态优化。网络能自组织和自学习自己的结构 ,即在学习过程中 ,网络可根据具体问题自动调整本身的结构 ,从而使结构达到最优。网络的输入输出均为模糊集 ,训练后的网络能完成合成关系 ,即模糊推理。为了减少 BP网络的离线训练时间 ,对模糊集进行了“编码”。For the complexity of shape and thickness combined governing system and the characteristics that the mathematical model of the system is established difficultly, the shape and thickness combined governing scheme of 4-high cold mill, based on the nerve-fuzzy intelligent modeling method, is put forward, and the self-organizing fuzzy control model based on dynamic BP neural network is studied and established. On the basis of BP network model, the self-structure of the network is optimized dynamically. The network can self-organize and self-study its structure, namely in the study process, the network can adjust automatically its structure based on specific problems, which optimizes its structure. The input and output of the network are fuzzy sets, and the network can complete composite relation after training, namely fuzzy reasoning. For decreasing the off-line training time of BP network, the fuzzy sets are 'encoded'. The simulation results indicate that the method is effective, and provides a bran-new method of non-analytial modeling method for complex systems.

关 键 词：板形板厚 综合调节 动态神经网络 自组织模糊控制 编码 

分 类 号：TG334.9[金属学及工艺—金属压力加工]