基于KPCA与MRVM的二元混合气体成分识别算法研究

A Binary Mixed Gas Component Identification Algorithm Based on KPCA and MRVM

混合气体成分识别是电子鼻系统进行气体检测与分析的关键技术。为了提高二元混合气体成分识别准确率,本文提出了一种基于KPCA与MRVM相结合的二元混合气体成分识别算法。该算法利用KPCA的非线性特征提取能力对传感器阵列的响应信号进行特征提取;再利用多分类相关向量机(MRVM)分类器对二元混合气体成分进行识别。通过自主设计的实验系统获得的气体样本集对算法的有效性进行了验证,实验结果说明二元混合气体成分识别准确率达到99.83%。

Gas component identification is a key technology for gas detection and analysis of electronic nose system.In order to improve the identification accuracy of binary mixed gas components,a binary mixed gas component identification algorithm based on KPCA coupled with MRVM is proposed in this paper. This algorithm uses the nonlinear feature extraction capability of KPCA to extract the features of the response signal of sensor array. The multiple classification correlation vector machines( MRVM) classifier is used to identify the binary mixed gas component. The availability of the proposed algorithm is verified by the gas sample set obtained by the self-designed experiment system. The experimental results show that the identification accuracy of binary mixed gas component is 99.83%.