近红外光谱法在红曲菌固态发酵过程参数检测中的应用

Application of NIR Spectroscopy in Detection of Process Parameters for Solid-state Fermentation of Monascus

研究了近红外光谱技术快速检测红曲菌固态发酵过程参数水分含量和pH值的可行性。针对传统基于间隔策略波长选择方法忽略非线性因素的缺点,采用一种基于最小二乘支持向量机(Least squares support vector machines,LS-SVM)非线性模型的波长筛选算法:联合区间最小二乘支持向量机(Synergy interval least squares support vector machines,siLS-SVM),并将新算法与相关系数法、iPLS算法、siPLS算法对比。实验结果显示,联合siLS-SVM算法和LS-SVM模型取得了最好的预测效果,水分含量、pH值的预测集相关系数(R p)分别为0.962 1、0.976 1,预测均方根误差(RMSEP)分别为0.012 9、0.145 2,表明模型具有较好的拟合度和预测性能。应用近红外光谱法进行红曲菌固态发酵过程的水分含量和pH值的快速检测可行,该方法为进一步实现其过程参数的在线检测及发酵条件优化提供了技术基础。

The feasibility of rapid detection of the process parameters such as moisture content and pH value in solid-state fermentation of Monascus was studied by near infrared technique.In view of the traditional wavelength selection methods based on interval strategy ignoring the disadvantages of the nonlinear factors,this paper adopted a new wavelength selection algorithm based on LS-SVM nonlinear model which was named synergy interval least squares support vector machines （siLS-SVM）,and the new algorithm was compared with correlation coefficient method,iPLS algorithm and siPLS algorithm.The results showed that the combination of siLS-SVM and LS-SVM model achieved the best prediction result.The prediction correlation coefficients （Rp） of moisture content and pH value were 0.962 1 and 0.976 1,respectively.The root mean square errors for prediction set （RMSEP） were 0.012 9 and 0.145 2,respectively.The obtained results demonstrated that the fitting and the predictive accuracy were satisfactory,and it was feasible to apply NIRS method to fast determination of moisture contents and pH value in solid-state fermentation of Monascus.This study provided a technical foundation for the further online detection of process parameters in solid-state fermentation of Monascus and the optimization of fermentation conditions.