论文部分内容阅读
对嗜酸乳杆菌、变异链球菌和保加利亚乳杆菌这三种菌的荧光光谱进行研究,发现在紫外光的激励下,益生菌溶液发出荧光.在最佳激发波长290nm的激励下,荧光峰值在300—650nm范围内.采用小波变换对测得的150组光谱数据进行压缩,压缩后每组数据由原来的1341个点减少为168个点,既保留了原图谱的特征,又提高了神经网络的处理速度.径向基函数神经网络方法对压缩后的数据进行研究,对每种菌的40组实验数据进行训练,在此基础上对30组未知数据进行识别.结果表明经过训练之后,径向基函数神经网络能够准确预测未知菌种.
Fluorescence spectra of three kinds of bacteria, Lactobacillus acidophilus, Streptococcus mutans and Lactobacillus bulgaricus were studied and found that under the excitation of ultraviolet light, the probiotic solution fluoresced.Under the optimal excitation wavelength of 290nm, the fluorescence peak was in the range of 300-650nm.Using wavelet transform to compress the measured 150 sets of spectral data, each set of data reduced from the original 1341 points to 168 points after compression, not only retains the characteristics of the original spectrum, but also improves the neural network The radial basis function neural network method was used to study the compressed data and train 40 sets of experimental data of each bacterium on the basis of which 30 sets of unknown data were identified.The results showed that after training, Basis function neural networks can accurately predict unknown species.