基于电子舌检测的橙汁贮藏品质研究

为实现对不同储存时间的鲜榨橙汁品质进行客观、快速的评价,采用基于虚拟仪器技术的电子舌系统对6种不同储存时间下的鲜榨橙汁样本进行定性和定量分析。针对电子舌输出信号特点,分别采用主成分分析(Principal Component Analysis,PCA)和离散小波变换(Discrete Wavelet Transform,DWT)方法对输出信号进行预处理,以分类效果为依据,确定离散小波变换作为较佳特征提取方法。在此基础上,采用线性判别分析(Linear Discriminant Analysis,LDA)方法对不同储存时间鲜榨橙汁样本进行定性分析,然后采用粒子群优化最小二乘支持向量机(Particle Swarm Optimization Least Squared-Support Vector Machines,PSOLSSVM)对鲜榨橙汁的不同储存时间进行定量预测。结果表明:LDA结果中第一判别式(LD1)和第二判别式(LD2)的综合贡献率为95.7%,6种储存时间下的鲜榨橙汁样本均得到有效定性辨别;而PSO-LSSVM预测模型对鲜榨橙汁的不同储存时间具有较高的定量预测精度,其相关系数(R~2)、均方根误差、平均绝对误差分别为0.999 1,0.287 7,0.232 8。

Research on detection for the storage quality of orange juice based on the electronic tongue

The aim of this work was to fulfill the objective and rapid assessment of quality and flavor of fresh orange juice with different storage time. An electronic tongue system that based on virtual instrument technology was developed and used to the qualitative and quantitative analysis of fresh orange juice samples with six kinds of storage time. According to the characteristics of electronic tongue respond signal, it was first preprocessed by the principal component analysis (PCA) method and discrete wavelet transform (DWT) method, respectively. According to the classification result, the DWT was selected as a recommended feature extraction method. Then the linear discriminant analysis (LDA) was used to the qualitative analysis of fresh orange juice samples with different storage time. Moreover, the least squared-support vector machines based on particle swarm optimization method (PSO-LSSVM) was applied to quantitative forecast the different storage time. The results showed that the cumulative contribution rate of LD1 and LD2 was reached 95.7% when the linear discriminant analysis was employed, and the fresh orange juice samples with the six kinds of storage time were effectively discriminated; The PSO-LSSVM prediction model had high prediction precision for different storage time of fresh orange juice, the correlation coefficient (R²) root mean square error (RMSE), mean absolute error (MAE) were 0.999 1, 0.287 7, and 0.232 8, respectively. This study could provide technical reference for quality evaluation and monitoring of fresh fruit juice.