近红外光谱结合遗传算法快速检测红曲菌固态发酵生物量

Rapid Determination of Biomass in Solid-state Fermentation of Monascus Using NIR Spectroscopy and Genetic Algorithm

该文研究了近红外光谱技术在红曲菌固态发酵生物量快速检测方面的应用。共采集了4个批次80个样本的光谱,采用氨基葡萄糖法测定生物量。为降低模型的复杂度和提高模型的预测性能,研究了遗传算法(Genetic Algorithm,GA)的光谱谱区选择方法,并建立所优选光谱变量的预测红曲菌固态发酵生物量的PLS模型。为说明遗传算法优选光谱变量的可行性,另外分别建立了全谱和相关系数法两种波长选择方法下的PLS定量模型,比较分析了3种方法所获模型的预测能力,并对GA方法优选的光谱波段信息与菌体成分中含氢基团的对应吸收进行分析。结果表明,遗传算法能在降低模型复杂度的同时提高模型的预测性能,其建模结果为Rc=0.998 3,RMSECV=3.580 2,Rp=0.993 1,RMSEP=3.643 7,参与建模的数据点由全谱的1 457个减少到585个,且模型预测精度相比FS-PLS模型提高了11.55%。研究表明近红外光谱技术结合遗传算法所建的PLS预测模型能够实现红曲菌固态发酵生物量的快速检测,从而为进一步实现在线发酵过程优化控制提供依据。

The application of near infrared spectrum technology in rapid detection of Monascus bio mass in solid-state fermentation was studied in this paper. Four batches of 80 samples＇s spectra were collected, and glucosamine method was used to estimate the biomass. In order to reduce the complex ity of the model and improve the prediction performance, the application of the genetic algorithm （GA） to selected wavelenths region was studied, and partial least squares regression was constructed for the prediction of biomass value in solid state fermentation of Monascus with effective wavelengths selected by GA. To illustrate the feasibility of GA to optimize spectral variables, partial least squares regression （PLSR） was constructed with full-spectrum and the wavelengths were selected by the corre- lation coefficient method, respectively. The prediction ability of the three models were comparatively analyzed, and the correlation between the spectral bands information selected by GA algorithm and the corresponding absorption generated by hydrogen groups of bacteria＇s composition was explained. The results showed that GA could reduce the complexity of the model and improve the model＇s prediction performance, with Rc = 0. 998 3, RMSECV = 3.580 2, Rp = 0. 993 1, RMSEP = 3.643 7, data points participate in modeling decreasing from the original 1 457 to 585, and the model predic-tion accuracy is improved by 11.55% compared with that of the full spectrum＇s PLS model. The re- sult showed that the PLS model built by using near infrared spectroscopy combined with GA could realize the rapid detection of biomass of Monascus in solid state fermentation. The method provided the technical foundation to further realize online fermentation process optimization control.