Resolution of Chromatographic Peaks byRadial Basis Function Neural Network Based onPlate Model Based on a Developed Sorting Genetic Algorithm / 分析化学

ABSTRACT

Radial Basis Function Neural Network Based on Plate Model （P-RBFNN） is constructed for resolution of chromatographic peaks of unknown components number. Then a two-phase sorting genetic algorithm (TP-SGA)-training structure and evolving is intruduced to train the network so that it has the ability of re-constructed structure. TP-SGA has robustness and random globe optimization.　The alternate use of gradient descent and TP-SGA makes the network have the ability to learn structure,　therefore makes itself adaptable to resolution of the chromatographic peaks of unknown components number.　The method proposed here needs no artificial interference,　not only has it robustness and globalism. With its characteristics related above and its ability of decomposing and analysing, this method has obvious advantages comparing with others.