基于传感器阵列多特征优化融合的茶叶品质检测研究

为提高电子鼻对不同品质茶叶的识别能力,分别提取电子鼻传感器信号的总体平均值、上升阶段斜率平均值和相对稳态平均值作为特征值,对电子鼻传感器阵列进行多特征数据融合优化.首先对原始数据进行归一化处理,统一值的量纲和数量级;通过因子载荷分析,去除各个象限内主成分投影较小和投影重叠的因子,对多特征向量矩阵进行优化;最后采用单因素方差分析,缩小不同品质茶叶组内间距,增大组间间距,更利于实现茶叶品质的区分.结果显示,主成分分析(PCA)可有效区分3种不同等级茶叶,因子载荷优化使各品质茶叶组内间距减小,单因素方差优化使一级与二级茶叶区分效果更明显;线性判别分析(LDA)效果要优于PCA分析,3个不同等级的茶叶可得到极为明显的区分.研究结果表明,用多特征优化融合可有效提取电子鼻对茶叶的响应信息,有利于对不同品质茶叶进行识别.

Detection method for tea quality using sensor array coupled with multi-feature optimization fusion

In order to improve correct rate of discrimination result of different grades of tea using the electronic nose (E-nose),the overall average value,rising slope average value and relative steady-state average value were extracted from the sensor array as the feature values to optimize and fusion the sensor array of E-nose. The original feature values were normalized,the dimensional and magnitudes of feature values were unified. The overlap factor were removed through the factor loading analysis,the multi feature vector matrix was optimized.Reducing the quality of tea in intra group spacing,increasing the quality of the tea in spacing between groups by one-way analysis of vari-ance. Through the multi-feature optimization fusion of sensor array,E-nose can perform better in the detection of tea quality and reduce the inter group spacing of different grades of tea. The results of Linear Discriminant Analysis (LDA)is better than Principal Component Analysis(PCA),and three different grades of tea can be distinguished very clearly. The research provides an efficient method for E-nose's application in various fields. The results shows that the response signal of E-nose to tea can be more effectively represented using multi-feature optimization fusion, and the discrimination result can be improved.