A data processing method for electronic tongue based on computational model of taste pathways and convolutional neural network

Research on the bionic degree of the electronic tongue (e-tongue) is limited. Therefore, a computational model of taste pathways and convolutional neural network (CMTP-CNN) is proposed for performance improvement while enhancing the bionic degree of the e-tongue. In this study, the enhancement effects of CMTP-CNN on the bionic degree are shown by simulation results. The simulation results demonstrate that the bionic degree of the e-tongue is enhanced by the fast-response ability and chaotic characteristics of CMTP nodes. Next, CMTP-CNN is used to identify tea and beer samples. Compared with the identification results of multiclass classification methods, the best accuracy of 96.00% and 96.67%, the best Kappa coefficients of 0.9495 and 0.9577, and the best area under the curve values of 0.9750 and 0.9792 in the tea and beer recognition, respectively, are acquired by CMTP-CNN. In conclusion, an improved identification performance for taste substances with the e-tongue is achieved using CMTP-CNN.