摘要
针对传统数学模型预测精轧出口厚度逐渐不能满足精度要求的现象,提出一种基于粗糙集和神经网络的热轧薄带钢厚度预测方法。该方法通过对精轧出口厚度影响因素进行属性约简,找出主要影响因素,以此作为径向基函数神经网络预测模型的输入,从而提高网络的收敛速度与预测精度。仿真结果显示,粗糙集神经网络预测模型优于传统数学模型,并且样本的预测误差基本控制在-0.05~0.05 mm以内。
Traditional mathematical thickness prediction model does not gradually satisfy the current accuracy requirements, therefore, a modeling method based on rough sets and neural network is presented. In this method, rough sets were used to reduce influence factors so that the significant factors for thin strip thickness prediction could be found and were used as the input variables of RBF neural network to improve the network convergence speed and prediction accuracy. The simulation results show that the rough set neural network forecasting model is superior to the traditional mathematical model and the prediction error is between plus and minus 0. 05 mm.
出处
《济南大学学报（自然科学版）》
CAS
北大核心
2014年第2期110-113,共4页
Journal of University of Jinan(Science and Technology)
基金
国家自然科学基金(60973042)
山东省自然科学基金(Y2008G20
2008F61)
关键词
热连轧
粗糙集
RBF神经网络
薄带钢
hot rolling
rough set
radial basis function neural network
thin strip
