Font Size: a A A

Application Of Artificial Neural Network And Genetic Algorithm Combine To The Simultaneous Determination Of Trp, Tyr And Phe

Posted on:2015-07-04Degree:MasterType:Thesis
Country:ChinaCandidate:M HeFull Text:PDF
GTID:2298330434965240Subject:Inorganic Chemistry
Abstract/Summary:PDF Full Text Request
Ultraviolet spectra of the polycomponent severely overlap and interferewith each other, can’t be measured by conventional chemical methods. Usingchemometrics method to distinguish the overlapping signals,realized on thesimultaneous determination of unknown samples.In this dissertation,application of chemometric methods,such as Error-back Propagation ArtificialNeural Network(BP-ANN) and Genetic Algorithms (GA), have beenresearched in polycomponent analysis,especially in UV spectrophotometricanalysis.Three-components system of tryptophan, tyrosine and phenylalaninehave been developed in Spectrophotometric Analysis,the main work is asfollows:Firstly, BP-ANN has been applied to the tryptophan, tyrosine andphenylalanine without separation by spectrophotometry and analyzed theoverlapping spectra of amino acid mixture. This method has been applied todetermine amino acid mixture in legume and the responder is well.Secondly, GA has been used to optimize the structure and parameter ofANN-BP.A new genetic neural network model GA-BP has beendeveloped,which avoids BP algorithm by means of repeating the experimentto determine the number of hidden layer and solves over fitting and overtraining problems of BP-ANN,determining the BP neural network structurerapidly.GA-BP effectively compensates BP network for the defects thatrandom selection to the network structure weight and threshold value, takesadvantage of GA global search capability and BP networks local searchcapabilities, enhances the network intelligent search capability.Reducingnetwork convergent time, the network output is fitted after training and theresults is more relevant.What’s more, GA-BP-ANN has been applied to the simultaneousdetermination of tryptophan, tyrosine and phenylalanine without separation byspectrophotometry and analyzed the overlapping spectra of amino acid mixture,a new model has been developed.This method has been applied todetermine amino acids in legume,the experimental result is more satisfactoryand relevant.In this article a predictive stability GA-BP model is established by thecombination of two chemometrics methods.There are some theoreticalsignificance and practical value of simultaneous determination of unknownsamples,chemical synthesis, pharmaceuticals and other areas.
Keywords/Search Tags:Artificial neural networks, Genetic Algorithms, Genetic neural networks, Spectrophotometry, amino acidmixture
PDF Full Text Request
Related items