基于电子鼻的不同香型白酒快速识别

Rapid Identification of Liquors of Different Flavor Types by Use of Electronic Nose

为了探索电子鼻对不同香型白酒的识别,以STM32为系统核心,筛选TGS2600、TGS2602、TGS2610、TGS2611和TGS2620共5个TGS传感器组成传感器阵列,设计了白酒检测电子鼻。利用该系统对浓香型、酱香型、清香型、米香型4种香型的代表酒样进行了气味数据采集。对数据进行平滑处理后提取其稳态响应值,并分别利用主成分分析、线性判别分析和概率神经网络建立了识别模型。实验数据显示:主成分分析的前2个主元累计贡献率达93.55%,线性判别分析的前2个主元累计贡献率为97.33%,概率神经网络模型识别率达到100%。结果表明,设计的电子鼻可以应用于对不同香型白酒的快速识别。

In order to identify liquors of different flavor types rapidly, the electronic nose had been designed with STM32 as the system nucleus and sensor array composed of five sensors including TGS2600,TGS2602,TGS2610,TGS2611 and TGS2620. In the experiment, the aroma data of liquor samples of four different flavor types including Nong-flavor, Jiang-flavor, Qing-flavor and Rice-flavor were collected by such eletronic nose, then after data smooth process, the steady-state response data were obtained, and principal component analysis, linear discriminant analysis, and probabilistic neural network were utilized respectively to establish the identification models.The experimental results showed that the contri- bution rate of the first two principal components of PCA reached 93.55 %, the contribution rate of the first two principal components of LDA reached 97.33 %, and the identification rate of probabilistic neural network model reached up to 100 %, which suggested that the desiged elec- tronic nose could be applied for rapid identification of liquors of different flavor types.