基于气体传感器阵列的地沟油识别方法

Identification Method of Waste Oil Based on Gas Sensor Array

目前有许多检测地沟油的方法,例如电导法、光谱法、色谱法等,但这些检测方法均需要在实验室环境下对油样进行处理,因此在地沟油检测方面仍缺乏快速、实时、对油样无损的检测方法。在此基础上,提出利用气体传感器阵列采集油样气味特征信息,使用支持向量机(SVM)算法对油样进行识别的方法。对75组正常食用油及45组地沟油建立了基于SVM的地沟油鉴别模型,并对15组正常食用油和15组地沟油进行识别,使用Matlab的SVM库及LIBSVM时的正确率均为27/30。实验证明了此方法的可行性,将算法在嵌入式平台上实现后可对油样进行快速、实时、无损鉴别。

At present, there are many methods to detect waste oil, such as conductivity, spectroscopy and chromatography, etc. However, these methods require the processing of oil samples in the laboratory environment, so there is still lack of rapid, real-time and nondestructive detection methods. On this basis, a method of using gas sensor array to collect the characteristic information of the oil samples odor, and using support vector machine(SVM) algorithm to identify oil samples is proposed. A discriminative model of waste oil based on SVM is established for 75 groups of normal cooking oil and 45 groups of waste oil. 15 groups of normal cooking oil and 15 groups of waste oil are identified. When using SVM library and LIBSVM, the correct rate is both 27/30. The experiment proves the feasibility of this method. After the algorithm is implemented on the embedded platform,the oil samples can be identified quickly, real-time and nondestructive.