基于动态神经网络的轧制液温控系统的辨识

Identification for Control System of Rolling Liquid Temperature Based on Dynamic Recurrent Neural Network

轧制液是铝箔轧制中的一种对稳定性要求很高的工艺润滑液,它不能局部受高热只能采用间接均加热,因此其控制量与温度之间是一种十分复杂的非线性关系,采用传统的建模方法难以建立其精确的数学模型.人工神经网络能够以任意精度逼近连续的非线性关系,并对复杂不确定问题具有自适应和自学习能力,为解决这一类非线性系统的辨识建模提供了新的途径.通过比较选用了一种动态递归网络来建立轧制液温控系统的辨识模型.

The rolling liquid is a highly stable technic lubricate, it is indirectly heated by exchange. It is difficult to establish precision model by traditional way. The neural network with self-adaptive and selflearning about complexities, uncertainties can dispose the connection of nonlinear at highly precision. It provides a new way to settle the identification of connection nonlinear and control. The model of temperature control system is established by selecting an dynamic recurrent neural network