二级模型预报的神经元网络自适应改进  

The Neural Network Adaptive Improvement of Level-two Model Prediction

作　　者：任志淼[1] Ren Zhimiao(Shanxi Water Technical ＆ Professional College,Taiyuan Shanxi 030027,China)

机构地区：[1]山西水利职业技术学院,山西太原030027

出　　处：《山西电子技术》2018年第4期42-44,共3页Shanxi Electronic Technology

摘　　要：轧制模型是对现实世界的模拟,所以理论和现实存在差距是不可避免的,本文目的就是缩小这种差距。系统研究二级模型弱化因子对模型精度的影响,同时就二级建模精细化特性进行了分析并提出改进方案,此方案使设定精度及活套稳定性得到实质性提高,对弱化因子的系统研究并应用到二级建模、最大程度改善带钢产品质量是本文一次成功的实践。The rolling model is built to simulate the real world, so the gap existence between theory and reality is inevitable, the study in this paper aims to narrow the gap. In the present study, the influence of weakening factors of level-two model on the precision of the model is systematically researched; meanwhile, the subtle characteristics of the level-two modeling are analyzed and an improved scheme is proposed. The scheme effectively improves precision of the setting and the stability of the loop. It is the successful practice in this study to research systematically the weakening factors and apply them to the level-two modeling, which enhance the quality of strip steel products to the greatest extent.

关 键 词：轧钢 神经元网络 长遗传设定 自适应学习 精细建模 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]