摘要
本文提出了一种用电子鼻来区分富士、花牛、姬娜3种不同品种苹果气味的方法。首先利用多尺度小波分析对气体传感器的数据进行去噪处理,再用支持向量机建立识别模型,最后通过优化2个支持向量机模型的核函数及其参数,将重叠的苹果气味数据进行高维空间变换用SVM回归模型识别。实验结果表明,第一个支持向量机模型对花牛苹果的识别正确率达到100%,第二个支持向量机模型对姬娜和富士的识别率大于90%。
Electronic nose was used to distinguish three different kinds of cultivar apples - ＂fuji＂, ＂huaniu＂ and ＂jina＂. In the data processing, multi - resolution decomposition was used to de - noise the gas sensor data, and support vector machine （SVM） was used to develop recognition models. Through optimizing the kernel functions of two SVM models and their parameters, the overlapped gas sensor data were mapped into high dimension space, so that, the three kinds of apples can be distinguished with two SVM models. Experiment results show that the first SVM model gives recognition rate of 100% for ＂huaniu＂ apple; and the distinguishing rate between ＂jina＂ and ＂fuji＂ from second SVM model is better than above 90% .
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2007年第3期534-538,共5页
Chinese Journal of Scientific Instrument
基金
国家863计划(2002AA248051)
国家自然科学基金(30370813)
教育部博士点基金(20040299009)
江苏省青年人才启动基金(BK2006552)资助项目。
