基于神经网络的单步和多步电弧炉模型的研究与应用  

Research and Application of One-step or Multis-tep Pre-estimating Neural Network Model of Electric arc Furnace

作　　者：陈兵[1] 王华[2] 

机构地区：[1]西安理工大学,陕西西安710048 [2]西安工业学院,陕西西安710032

出　　处：《现代电子技术》2003年第2期57-60,共4页Modern Electronics Technique

摘　　要：介绍了电弧炉计算机智能控制系统的特点与方法 ,并着重讨论了神经网络预估模型的设计 ,包括遗传算法在建立神经网络模型的应用和神经网络模型实时学习的要求 ,基于此基础对于多步预估进行了讨论和研究。通过实际应用证实该系统具有效率高 ,电能消耗低 。Introduces the characteristics and method of the intelligencetype computer control system of electric arc furnace, and then focus on the designing of preestimating neural network model include the application of genetic algorithm in establishing of neural network model and the requirement of real time study for neural network model, and then do some research of multistep EAF current prediction It is verified that this system has a series of advantagessuch as high efficient, low consumption of energy, electrodes and refractories through practical operation

关 键 词：电弧炉 神经网络预估模型 遗传算法 多步预估 

分 类 号：TM924.4[电气工程—电力电子与电力传动] TP183[自动化与计算机技术—控制理论与控制工程]