Research On Chemometrics Alignment Algorithms And Development Of Computer Intelligently Aided Flavoring System

Abstract:As a subject that very hard to kown well, chemometrics met its problem in directly application on analytical works. Many algorithms proposed very year, but just a small part of them can be used to slove problem. The main reasons led to this kind of situation are:there was seldom instructive study to guide analysis work, and there was lack of easy and useful softwares to help analytical workers. If some guiding instructions of chemometrics algorithms and some simple and easy chemometrics softwares are avaliable to us, the efficience of our analytical work can be greatly improved.Specific to the problems mentioned above, a particular study is carried out in three factors, they are:Firstly, compared many different kinds of chemometrics alignment algorithms, and chose five of them to make a detail comparison from many aspects, compared them with the help of several criteria, the results of comparisons have been discussed and suggestions have been presented to different kind of datasets. The reasonability of those criteria is also investigated, a scientific method to compare the chemometrics alignment algorithms is proposed. The result of this study can be a valuable reference and be used for development of chemometrics software.Secondly, integrated the chemometrics multi-element resolution algorithm and non-negative least squares algorithm to the updated software, those two algorithms can solve the qualitative and quantitative problems in tobacco flavor analysis respectively. Flavor GC-MS data were used to test the new computer intelligently aided flavoring system, for the quantitative problem we met in test, the best alignment algorithm we got in comparison was introduced to this system. The parameters are optimized to improve qualitative result and the data are aligned to get a better quantitative result. Thirdly, based on the accumulation of flavor database and the maturation of flavor analytical method, a new flavor analysis software named computed intelligent assistant flavor analysis system is improved by our research group. A new database and some better qualitative and quantitative algorithms are added to the new flavor analysis system, and the frame of the whole system is rebuilt. The targets to develop this software are guiding flavor analytical work and making full use of the flavor data.