Optimal arithmetic for hierarchical cluster and pattern recognition applied in coal dust sensor

Abstract

In order to resolve the conflict between coal dust sensor measurement precision and real time demands, an optimal arithmetic for dust pattern hierarchical cluster and pattern recognition was presented according to the dust sample diffraction angular distribution. The coal dust distribution of different mines had marked difference because there was the diversity of geological structure and degree of mechanization. The hierarchical cluster method had been successfully applied in the analysis of large numbers of dust patterns. The arithmetic first clustered the coal dust distribution type from the diffraction angular spectrum, and then the hierarchical cluster is performed to get the pattern sequence within the corresponding type. Then patterns could be recognized easily and rapidly with the principle of minimum variance sum between pattern and dust sample eigenvectors. Simulation proves the maximal recognition speed improves evidently. Based on clustering number of transitional patterns was increased reasonably and the errors declined markedly. Results indicate the precision achieves 96%. It is concluded that the optimal arithmetic of hierarchical cluster and pattern recognition enhances the precision and real-time capability of the sensor effectively.