基于光寻址(LAPS)型电子舌的基本结构及检测原理,提出了一种基于主元分析(PCA)的味觉识别算法.通过对LAPS型电子舌提取到的信号特征进行主元分析,在其第一主元和第二主元得分图上,5种基本味道样本溶液的落点在各自区域内互不干扰,5种基本味道能够很好地区分.进一步采用最小距离法对其分类,表明该方法优势突出,对5种基本味道分类的平均识别准确率达到96.54%.试验表明,将主元分析应用于LAPS型电子舌的味觉检测分析方面有很好的效果,为研制小型化LAPS型电子舌提供了科学依据.

According to the fundamental structure and the detection principle of the Light Addressable Potentiometric Sensor (LAPS) type electronic tongue, a taste recognition algorithm based on Principal Component Analysis (PCA) is designed. The signal detected by LAPS type electronic tongue is processed by PCA. The scores pictures of the first and the second principal components show that the spots of five basic tastes do not interfere with each other. Therefore, the five basic tastes could be more accurately recognized by PCA. Meanwhile, minimum distance method is employed to classify the principal components and the results of the experiments show that the method is prominent and the average rate of identification could reach at 96.54%. Final test results indicate that applying PCA to LAPS type electronic tongue has the huge potential for analyzing the taste of liquid, and also provides a scientific proof for developing the miniature LAPS type electronic tongue.