钢坯拉速模糊信息粒化及钢坯定重切割的极限学习机预报

为提高钢坯定重切割精度，分析了钢坯质量与钢坯平均拉速间的关联性。建立了钢坯拉速数据的模糊信息粒化模型，将每5根钢坯的平均拉速数据变换为一个三角型模糊粒，得到模糊粒子中的3个参数：钢坯平均拉速变化的最小值vLow、均值vmid和最大值vup，降低钢坯拉速数据的复杂度，得到含不同信息的拉速数据粒化子集。建立了基于信息粒化数据的支持向量机（SVM）回归模型，以模糊粒子参数为输入向量对钢坯平均拉速进行回归预测，得到下一根钢坯的平均拉速预测值。综合考虑钢坯截面积、钢坯平均拉速、定尺长度、下一根钢坯平均拉速预测值等影响因素，建立了极限学习机（ELM）神经网络预报模型，实现了钢坯定重预报。

Fuzzy Information Granulation for Casting Speed of Billets and Limited Weight Cutting Forecast Based on ELM

In order to improve the precision of limited weight cutting of billets, the relationship between the quality of billets and the average casting speed of billets was analyzed. The fuzzy information granulation model of billet casting speed was established, the average casting speed data of each 5 billets were changed into a triangular fuzzy granule, 3 parameters in fuzzy particle: the minimum vLow, average vmid and maximum vup values of variations of average casting speed of billets were obtained, the complexity of the casting speed data of billet was reduceds, the granular subset of casting speed data containing different informations was obtained. SVM regression model was established based on information granulation data, the fuzzy granule parameters were used as input vectors to predict the average casting speed of the billets, the prediction value of average casting speed of the next billet was obtained. Some factors were considered comprehensively, such as cross-sectional area of billets, average casting speed, fixed length, the prediction value of average casting speed of the next billet and so on, a neural network prediction model was established based on ELM and the prediction of limited weight of billets was realized.