Prediction System of Burning Through Point (BTP) Based on Adaptive Pattern Clustering and Feature Map

In this paper, due to the property of the long time delay, time varying and multimode of sintering process, an adaptive pattern clustering and feature map (APCFM) is proposed to solve the challenging problem for building a predictive system of burning through point. By using the density clustering and learning vector quantization (LVQ), the whole vector is divided automatically into subclasses which have similar clustering center and labeled fitting number, then these labeled subclasses samples are token into genetic neural network (GNN) to train. The training set consists 707 groups of actual process data and GNN are trained with APCFM algorithm, these experiments proved that this system is stronger robust and generality in clustering analysis and feature extraction