基于灰色关联分析的神经网络轧制力预报模型的研究

Neural Network Prediction Model of Rolling Force Based on Grey Incidence Analysis

为了进一步提高采用神经网络对热轧机轧制力的预报精度及建模速度,在分析研究目前已有的各种轧制力预报模型的基础上提出:在采用神经网络进行轧制力预报时,必须首先根据对象合理确定神经网络的输入变量。本文采用灰色关联分析法,利用生产现场实际数据,对影响轧制力设定值计算的多种因素与轧制力进行了相关性分析,最终简化了神经网络的结构,提高了模型的在线应用能力。

To further improve the precision of the rolling force prediction and the modeling speed with neural network,it is necessary to chose the input variables of neural network reasonably,thus the grey incidence analysis between the rolling force and the relevant factors is carried out to determine the most important factors on the rolling force and the structure of neural network is simplified to improve on-line applications of the model.