节点文献
近红外光谱结合径向基神经网络无损分析松茸菌丝体
Application of Near Infrared Reflectance Spectroscopy-Radial Basis Function Neural Network for Non-Destructive Determination of Tricholoma Matsutake
【作者】 逯家辉; 张益波; 孟庆繁; 类婷婷; 高波; 滕利荣;
【Author】 LU Jia-hui ZHANG Yi-bo MENG Qing-fan LEI Ting-ting GAO Bo TENG Li-rong College of Life Science,Jilin University,Jilin Changchun 130012,China
【机构】 吉林大学生命科学学院;
【摘要】 将近红外光谱(NIRS)与三层径向基神经网络(RBFNN)结合,建立药用真菌松茸中活性成分蛋白和多糖的快速无损分析模型(NIRS-RBFNN)。采用卷积平滑,傅里叶变换,一阶变换,二阶变换,多尺度小波和小波包变换对原始光谱进行预处理。对处理后的光谱进行主成分的提取,以前15个主成份得分作为径向基神经网络的输入节点选择范围。对与网络相关的参数(输入节点数,中间神经原数,径向基宽度常数)进行了优选。得到了松茸蛋白的最佳分析模型为:小波包变换5尺度重构光谱,模型参数为WPT-NIRS- RBFNN(10-18-1,3.0),此时模型的RMSECV为0.00543,模型的Rcv为0.99335,此模型对预测集的RM- SEP为0.00943,Rp为0.98292;松茸多糖的最佳分析模型为:小波包变换5尺度重构光谱,模型参数为WPT-NLRS-RBFNN(12-10-1,2.6),此时模型的RMSECV为0.00825,模型的Rcv为0.99005,此模型对预测集的RMSEP为0.009 56,Rp为0.98638。结果表明了模型有很好的稳健性和精确度。可以作为常规方法的替代方法,对实现药用真菌的绿色、无损、快速分析有重要的意义。
【Abstract】 A calibration model(NIRS-RBFNN)combination of near infrared reflectance spectroscopy and radial basis function neural network(RBFNN)has been proposed for synchronous and rapid non-destructive determination of Tricholoma matsutake. savitzky-golay smoothing(SGS),fast fourier transform(FFT),Derivative,of wavelet transformation(WT)and wavelet packet transformation(WPT)with multi-scale analysis were used to disposal the origin NIRS,then we used principal component analy- sis(PCA)method to obtain the principal components(PC)scores.The anterior 15 PC scores were used as in put data.These developed RBFNN have been optimized by selecting suitable parameters of in put data,numbers of hidden layer neurons and spread constant though different pretreated spectra of calibration.The optimal quantitative analysis(NIRS-RBFNN)model for polysaccharide and protein of Tricholoma matsutake:for Protein the optimal model is 5 scales decompose of WPT,WPT-NIRS- RBFNN(10-18-1,3.0),and RMSECV is 0.005 43,Rcvis 0.99335;RMSEP is 0.009 43;Rp is 0.982 92.for polysaccharide the optimal model is 5 scales decompose of WPT,WPT-NIRS-RBFNN(12-10-1,2.6),and RMSECV is 0.00825,Rcvis 0.990 05;RMSEP is 0.009 56;Rp is 0.986 38.These results are so satisfied and NIRS technology is convenient,rapid,no pretreat- merit and no pollution that this method could be popularized in the in situ measurement and the on-line quality control for Tri- choloma matsutake.
【Key words】 Near infrared reflectance spectroscopy (NIRS); Radial basis function neural network (RBFNN); Tricholoma matsutake; Non-destructive determination;
- 【会议录名称】 第十五届全国分子光谱学术报告会论文集
- 【会议名称】第十五届全国分子光谱学术报告会
- 【会议时间】2008-10
- 【会议地点】中国北京
- 【分类号】TQ461
- 【主办单位】中国光学学会、中国化学学会