论文部分内容阅读
该文以平菇平板培养菌丝总蛋白含量为指标,在1 000-1 799 nm近红外光谱区域采集光谱信息,采用化学计量学法建立菌株各参数的偏最小二乘法(PLS)定量预测模型。结果表明:最佳光谱预处理方法为SavitzkyGolay平滑+Savitzky-Golay导数+多元散射校正(MSC)+均值中心化,所建定量模型的校正集相关系数、校正标准差(SEC)、验证集相关系数、预测标准差(SEP)、主因子数、SEP/SEC均在合理范围,模型真实值与预测值的相关系数为0.672 63,模型可靠性、稳健性和预测效果较好,可用于菌丝蛋白质含量检测。
In this paper, the total protein content of flat mycelium of flat mushrooms was taken as an index to collect the spectral information in the near-infrared spectral region of 1 000-1 799 nm and the PLS quantitative prediction model of each strain was established by stoichiometry . The results showed that the optimal spectral preprocessing methods were SavitzkyGolay smoothing + Savitzky-Golay derivative + MSS + mean centralization, Correlation Set Correlation Coefficient (SEC) and Correlation Set Correlation Coefficient , The standard deviation of prediction (SEP), the number of major factors and the SEP / SEC were in a reasonable range. The correlation coefficient between the true value of the model and the predicted value was 0.672 63. The model was more reliable, robust and predictive and could be used for mycelial protein Content detection.