论文部分内容阅读
针对铝电解生产过程的复杂性,建立了基于网格共享近邻聚类(GNN)最小二乘支持向量机(LS-SVM)的铝电解生产过程极距软测量模型.该模型采用GNN算法将训练集分成具有不同聚类中心的子集,对各子集分别采用LS-SVM进行训练并建立子模型,同时通过参数转化实现模型对新数据样本的动态学习.仿真结果表明,基于GNN最小二乘方法建立的铝电解极距软测量模型具有精度高、泛化性能好等特点,能够为铝电解生产过程操作优化提供实时准确的信息.
Aiming at the complexity of the aluminum electrolysis process, a polar distance soft sensor model based on the grid shared neighbors clustering (GNN) least square support vector machine (LS-SVM) was developed. This model uses the GNN algorithm to train Which is divided into subsets with different clustering centers, LS-SVM is used to train each sub-set separately and a sub-model is established, and the dynamic learning of the new data samples is realized through parameter transformation.The simulation results show that, based on GNN least-square The aluminum electrodeposition pitch soft-sensing model established by the method has the characteristics of high precision and good generalization performance, and can provide real-time and accurate information for the operation optimization of the aluminum electrolysis production process.