Application Of Artificial Neural Network And Genetic Algorithms In Multi-Components Analysis

In this dissertation, application of mathematical methods, such as Artificial Neural Network (ANN) and Genetic Algorithms (GAs) , have been researched in multi-components analysis, especially in Spectrophotometric analysis, new chemometric methods have been developed. Six papers have been accomplished on one's own and in collaboration with others.Two new multi-components coloring systems that tungsten (molybdenum)-dibromoalizarin violet-CTMAB and molybdenum (titanium)-dibromoalizarin violet-CTMAB have been developed respectively in Spectrophotometric Analysis.BP-ANN has been applied to the simultaneous determination of Tungsten and Molybdenum without separation by spectrophotometry with above first coloring system, a new chemometric method has been developed. This method has been applied to determine Tungsten and Molybdenum in steel, the results were satisfactory.BP-ANN has been applied to the simultaneous determination of Molybdenum and Titanium without separation by spectrophotometry with above second coloring system, a new chemometric method has been developed. This method has been applied to determine Molybdenum and Titanium in steel, the results were satisfactory.GAs has been used to optimize the structure and parameters of ANN-BP, a new genetic neural network model GA-ANN-BP has been developed, which optimized the node number in the hidden layer, the learning rate (η) and the momentum factor (α),solved overfitting and overtraining problems of BP-ANN. The problems that analytical results of training sets was better, prediction sets was worse, and the output errors of two elements were disagree have been overcome.GA-BP-ANN has been applied to the simultaneous determination of Tungsten and Molybdenum without separation by spectrophotometry with above first coloring system, a new chemometric method has been developed. This method has been applied to determine Tungsten and Molybdenum in steel, the results were superior to BP-ANN.GA-BP-ANN has been applied to the simultaneous determination of Molybdenum and Titanium without separation by spectrophotometry with above second coloring system, a new chemometric method has been developed. This method has been applied to determine Molybdenum and Titanium in steel, the results were superior to BP-ANN.The programs of BP-ANN and GA-BP-ANN were written with Turbo C.By citing 123 literatures, the author reviewed the late development of application of ANN and GAs in chemistry.