流浆箱神经网络解耦控制系统的设计

The design of a practical headbox neural network decoupling control system

利用BP神经网络原理结合流浆箱控制系统实际情况设计了神经网络解耦控制器,根据事先采集的大量实验数据训练出网络参数,得到与系统相匹配的静态解耦控制器,然后通过编程实现神经网络控制器输入输出的拓扑关系。该控制器既解除了气垫式流浆箱压力和液位的耦合关系,又具有实现简单的特点。

A neural network decoupling controller is designed according to the principle of Back Propagation Neural Network as well as theknowledge and experience in headbox control of paper machine, and by means of a large amount of data acquired in the on-spot experiment inadvance, the parameter of the network and the static Decoupling Controller are gained, the topological relationship of NN controllers outputand input being implemented by programming.Besides the feature of the decoupling between pressure and level of the headbox, the system hasother advantages such as convenience in operation and easiness in implementation.