Technology And Its Application In The Compound Spectral Data Processing Data Mining

Energetic materials (namely energetic compounds) is a high-energy density materials. As few chemical changes may complete elementary reaction in one step, and most of them are accompanied by multi-step or complicated process, the synthesis of energetic materials is rather complex. Reaction mechanism is the key of energetic materials synthesis and process optimization. We utilize online infrared detector throughout the reaction to fully understand the complex chemical changes in the process of energetic materials synthesis. By studying intellisense of the spectral data collected, we may understand the complexity of the synthesis changes and the reaction mechanism in energetic materials.This thesis is based on data mining theory and technology, which are employed to investigate the synthesis process of the mid-infrared spectral data collected by online detection. The thesis is focused on spectral data dimension reduction, spectral intellisense and the indeterminacies of independent component analysis. On this basis, this paper presents three views. The first one, focused on the problem of mid-infrared spectral high dimension and the requirement of independent component analysis, principal component analysis dimension reduction algorithm based on subspace comparison method is proposed and applied the algorithm in the mid-infrared spectral analysis pre-treat. The simulation results proved that the new algorithm can accurate dimension reduction to achieve the smaller target subset and improve the quality and efficiency of mining. The second one, for the intellisense of the mid-infrared spectral, the new algorithm of fixed-point algorithm mixed alternating least squares algorithm is proposed and applied the algorithm in the mid-infrared spectral analysis. Through simulation experiment, concentration curve derived from the new algorithm is optimized and more in line with the physical and chemical significance. The last one is to proposed a method to eliminate the indeterminacies of independent component analysis, and applied the proposed algorithm to the mid-infrared spectral analysis. Through simulation experiments, it can be detected that the spectrum of pure substances anomalies has subsided and the curve becomes smooth, which eliminate the indeterminacies in some extent.The research results are useful to analyse the reaction mechanism in energetic materials. It also can provide decision support for the synthesis of new compounds and reaction optimization.