基于金属氧化物传感器阵列的小麦霉变程度检测

研制了一套由8个金属氧化物传感器组成、用于检测小麦霉变的电子鼻系统.使用该电子鼻对不同霉变程度和掺入不同百分比含量霉麦的小麦样品进行检测.通过方差分析和主成分分析优化传感器阵列并去掉冗余传感器,对优化后的数据进行主成分分析(PCA)和线性判别分析(LDA),其中PCA的前两个主成分对两类实验结果分析的总贡献率为98.30％和99.27％,LDA前两个判别因子对两类实验结果分析的总贡献率为99.68％和93.30％,且由得分图可知两种方法均能很好地区分不同的小麦样品.利用BP神经网络建立预测模型,对样品菌落总数和掺入样品中霉麦的百分比进行预测.两种预测模型的预测值和测量值之间的相关系数分别为0.91和0.94,表明预测模型具有较好预测性能.

Detection of moldy wheat using MOS sensor array in electronic nose

In this study, an electronic nose was developed to detect the moldy wheat. Different wheat samples which had different degree of mildew and were mixed with different percentage of moldy wheat were detected by this electronic nose system. Analysis of variance and principal component analysis (PCA) were applied to optimize the sensor array and 2 redundant sensors were removed. PCA and linear discriminant analysis (LDA) methods were used to qualitatively analysis the optimized data of electronic nose. The total contributions of first two principal components in PCA were 98.30% and 99.27%, and the total contributions of first two discriminant functions in LDA were 99.68% and 93.30%, respectively. The analysis of PCA and LDA indicated that both of the two methods had good classification performance. BP neural network were applied to build the predictive models to predict the aerobic bacterial count in the samples and the percentage of moldy wheat mixed into the samples. The correlation coefficient between the predictive values and virtual values of the two predictive models were 0.91 and 0.94 which indicated that the predictive models had good prediction performance.