【作者】 邹天来； 王秀梅； 张晓峰； 王国栋； 刘相华；

【Author】 Li Hongbin Zhang Gong Zou Tianlai (Benxi Iron and Steel Co.Hot Strip Mill) Wang Xiumei Zhang Xiaofeng Wang Guodong Liu Xianghua (The State Key Laboratory of Rolling and Automation,Northeastern University Shen Yang)

【机构】 本溪钢铁公司热轧厂； 东北大学轧制技术及连轧自动化国家重点实验室；

【摘要】 为了提高轧制力预设定精度,在热连轧精轧机组轧制力预报系统中引入模糊推理和神经网络技术。以神经网络作为学习机构,引入模糊运算节点及乘法运算节点。模糊神经网络的学习算法采用BP网络的6学习规则,对权系数即控制规则和隶属函数进行修正。采用现场实测数据训练模糊神经网络。离线仿真表明,采用模糊神经网络方法预报轧制力,预报精度比传统数学模型方法有很大提高。更多还原

【Abstract】 In order to improve the precision of rolling load in a finishing train,neural networks and fuzzy reasoning are introduced to construct predicting system for rolling load.Fuzzy operation nodes and multiplication nodes are introduced to the neural network, which is regarded as learning mechanism.The learning algorithm of the Fuzzy Neural Network(FNN) is via theδlearning rule of back-propagation methodology.The FNN has the advantages that it permits automatic identification of fuzzy rules and tunes the membership functions.The data collected from the rolling process are used to train the FNN.Off-line simulation indicates that comparing the mathematical models,the FNN improves the prediction accuracy of rolling load significantly.更多还原