Computational Model of Taste Pathways: A Biomimetic Algorithm for Electronic Tongue Based on Nerve Conduction Mechanism

被引:8
作者
Zheng, Wenbo [1 ,2 ]
Men, Hong [1 ,2 ]
Shi, Yan [1 ,2 ]
Ying, Yuxiang [1 ,2 ,3 ]
Liu, Jingjing [1 ,2 ]
Liu, Qingjun [4 ]
机构
[1] Northeast Elect Power Univ, Sch Automat Engn, Jilin 132012, Jilin, Peoples R China
[2] Northeast Elect Power Univ, Adv Sensor Technol Inst, Jilin 132012, Jilin, Peoples R China
[3] Kunsan Natl Univ, Sch Elect & Informat Engn, Gunsan 541150, South Korea
[4] Zhejiang Univ, Coll Biomed Engn & Instrument Sci, Hangzhou 310027, Peoples R China
基金
中国国家自然科学基金;
关键词
Taste substance identification; e-tongue; computational model; nerve conduction; bionic performance; NUCLEUS; CLASSIFICATION; MODULATION; FUSION; NOISE;
D O I
10.1109/JSEN.2022.3152057
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Research on the taste conduction mechanisms of data processing by the electronic tongue (e-tongue) is lacking. These mechanisms affect taste substance identification by the e-tongue system. This work proposed a computational model of taste pathways based on nerve conduction mechanisms to identify taste substances. This work modelled human taste pathways based on the Katchalsky models and validated the rationality of the modelling forms of the critical pathway structures. Next, this work simulated the response status and power spectral density of the model nodes and verified the bionic performance of the computational model. The results showed, first, that the computational model described the dynamic characteristics of the taste pathways. Second, the phase diagrams of the pathway structures revealed chaotic characteristics. This result showed that the modelling forms were rational. Third, when stimulated, the model nodes exhibited fast responses. During stimulation, the power spectral density of the model nodes demonstrated 1/f characteristics. The fast-response ability and 1/f characteristics reflected the bionic performance of the computational model. Finally, the computational model was used to identify beer, tea, and apple samples. Compared with other classification methods, the computational model achieved better classification results of 93.33%, 94.00%, and 76.67%, the best specificity of 0.9833, 0.9850, and 0.9417, and the best Kappa coefficients of 0.9180, 0.9261, and 0.7245 for the identification of the beer, tea, and apple samples, respectively. In conclusion, the computational model can effectively identify taste substances.
引用
收藏
页码:6859 / 6870
页数:12
相关论文
empty
未找到相关数据