论文部分内容阅读
针对轧钢生产中大批过程数据没有被用于提高厚度质量的现象,提出了一种基于减法聚类的带钢厚度数据驱动在线建模方法.首先通过减法聚类将输入空间划分为一些小的局部空间,在每个局部空间中用最小二乘支持向量机建立子模型,子模型加权输出作为带钢厚度的离线模型;然后当在线数据不断增加时,通过在线减法聚类算法实时调整局部空间,子模型的参数采用最小二乘支持向量机的递推算法进行相应的在线辨识,子模型的预测输出作为模型的最后输出.实验结果表明,该方法具有良好的预测精度和较强的在线学习能力.
Aiming at the phenomenon that a large amount of process data in the rolling production is not used to improve the thickness quality, a strip-thickness data driven on-line modeling method based on subtractive clustering is proposed.First, the input space is divided into some small parts by subtractive clustering Space, the sub-model is set up in each local space by least-squares support vector machine, and the sub-model is weighted and output as the offline model of the strip thickness. Then, when the online data increases continuously, the local space is adjusted by online subtraction clustering algorithm in real time, The parameters of the sub-model are identified by on-line recursion algorithm of Least Squares Support Vector Machine (SVM), and the predicted output of the sub-model is the final output of the model.The experimental results show that this method has good predictive accuracy and strong online learning ability .