【摘要】 基于神经网络原理 ,对微合金钢热轧控制参数的选取进行了研究 .首先 ,制定了一套获取样本数据的实验方案 .该方案利用Gleeble - 15 0 0热力模拟机提取了轧制温度、应变量、应变速率和相应的应力应变曲线 ,并通过显微观察获取了实验后样品断面的奥氏体晶粒尺寸 .通过归一化把实验所得数据进行必要的处理 .采用BP算法训练网络 ,对热轧控制参数 (轧制温度、应变量、应变速率 )和描述微合金钢组织性能的参数(奥氏体晶粒尺寸及流变应力 )之间的映射关系进行了函数逼近 ,建立了奥氏体晶粒尺寸及流变应力神经网络模型 .实践证明 ,将该神经网络模型运用于热轧控制预报 ,提高了预测精度并取得较好的效果更多还原

【Abstract】 Presents a new approach based on Artificial Neural Network to estimate hot rolling quality of micro alloyed steel, which obtains rolling temperature, strain, strain rate and curve of stress and strain using Gleeble 1500 thermal mechanical machine and then grain sizes and flow stress measured and gets them processed by ANN, and then builds the artificial neural networks originated BP arithmetic through training, and concludes that the predicted austenite grain size and flow stress of hot rolled micro alloyed steel is achievable and the mathematics model generated by ANN has higher accuracy a that by the regression method.更多还原