PCA-SVM优化WO_(3)气敏薄膜的VOC选择性研究  

VOC selectivity of PCA-SVM optimized WO_(3) gas sensing film

作　　者：唐潇 王明松[1] 柏凌 刘桂武[1] 乔冠军[1] TANG Xiao;WANG Mingsong;BAI Ling;LIU Guiwu;QIAO Guanjun(School of Materials Science and Engineering,Jiangsu University,Zhenjiang 212013,Jiangsu Province,China)

机构地区：[1]江苏大学材料科学与工程学院,江苏镇江212013

出　　处：《电子元件与材料》2022年第5期463-472,共10页Electronic Components And Materials

基　　金：国家自然科学基金(22002051)。

摘　　要：挥发性有机物(VOC)对人体健康的危害日益加重。为有效地解决电阻式气体传感器因交叉敏感导致对气体选择性普遍较差的问题,实现仅通过一种气敏材料结合机器学习算法对VOC类型进行预测分类的目的,提出了一种基于主成分分析(PCA)和支持向量机(SVM)优化WO_(3)气敏薄膜对VOC选择性的方法。采用化学浴沉积法制备WO_(3)气敏薄膜,在250~400℃工作温度下对不同浓度的六种VOC气体进行测试得到多维响应矩阵;通过PCA降低特征量之间的相关性,实现原始数据的降维,进行定性识别;再将优化后的数据代入SVM模型中对VOC类型进行预测,结果达到100%的分类准确率。相比于传统SVM模型,PCA-SVM在达到相同准确率的同时运行耗时减少40%,更适合处理具有多维特征量的数据样本,基于PCA-SVM的方法可以有效且快速地优化WO_(3)气敏薄膜对VOC选择性。Volatile organic compounds(VOC)endangers the human health increasingly.In order to effectively solve the problem of generally poor gas selectivity of resistance-type gas sensors due to cross-sensitivity,this work aimed at the predictive classification of VOC types by single gas sensing material in combination with machine learning algorithms,and the method based on principal component analysis(PCA)and support vector machine(SVM)was proposed to optimize the selectivity of WO_(3) gas sensing films to VOC.The WO_(3) gas sensing films were prepared by chemical bath deposition,and a multidimensional response matrix was generated by testing six VOCs with different concentrations at the working temperature of 250-400℃.The correlation between the feature quantities was reduced by using PCA,and dimensionality reduction of the original data was achieved,then the qualitative identification was performed.The optimized data were input in the SVM model to predict the VOC type,and 100%classification accuracy was achieved.Compared with the traditional SVM model,PCA-SVM achieves the same accuracy rate within 40%less running time,indicating that PCA-SVM is more suitable for processing data samples with multi-dimensional features.So the method based on PCA-SVM can effectively and quickly optimize the VOC selectivity of WO_(3) gas sensing films.

关 键 词：挥发性有机物 气体传感器 选择性 主成分分析 支持向量机 

分 类 号：TN212.[电子电信—物理电子学]