The Resolution of Overlapping Autofluorescence Spectrum Using A Modified Particle Swarm Optimization Algorithm

• Main Authors: Cheng-Hsueh Chuang, 莊承學
• Other Authors: Chuen-Yau Chen
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/76022545424235903050

碩士 === 國立高雄大學 === 電機工程學系碩士班 === 99 === In this thesis, we take an improved particle swarm optimization algorithm for the separation of overlapping spectra where the speed update is the major improvement. Taking advantage of the concepts of inertia factor and constriction coefficient in the proposed algorithm, the convergence rate and the accuracy are significant improved. In the experiments, we used an optical Y-type fiber to conduct excitation light from the 337-nm nitrogen laser (VSL-337ND-S, LSI) to irradiate on the skin surface so that we could get an excitation fluorescence signal with the wavelength ranging from 400 nm to 600 nm. We collected the data from the skin to the spectrometer (SP-150, Princeton Instrument). We designed three experiments and compared the results with other algorithms. For the fitness function values, this algorithm achieves a higher fitness value about 1. For the convergence rate, this algorithm can successfully separate two fluorescent wrist skin materials in the signal of human skin autofluorescence experiment with 67% of individuals (particles) and 15 times of speed as compared with the genetic algorithm. The experimental results show that the content ratios of reduced-nicotinamide-adenine-dinucleotide and flavin-adenine-dinucleotide are 96.3533% and 3.6467%, respectively.