Data Processing Based on Glomerular Microcircuits for Electronic Noses

There are some complicated steps in a conventional data processing procedure, including signal preprocessing, feature extraction and classification. This increases the complexity and difficulty of selecting suitable approaches. Thus, the simplification of processing procedure has become a crucial part for electronic noses (e-noses) applications. To overcome the shortcomings, we present a simplified e-nose data processing method based on glomerular microcircuits. Inspired by the structure and function of olfactory glomeruli, a glomerular microcircuit model is established by the synaptic interactions of olfactory receptor neurons, mitral cells, external tufted cells, periglomerular cells and short axon cells. The continuous sensor data is transformed into spike time sequences by the microcircuit model to enhance computational efficiency. To refine the olfactory information contained in spike time, a SpikeProp neural network is designed for feature extraction and classification. The proposed method, which can automatically implement feature learning without signal preprocessing, significantly improves data processing procedure as well as detection performance. The glomerular microcircuits based method was validated for different brands of Chinese liquors with conventional data processing methods. Experimental results show that the proposed method has higher classification rate and simpler procedure than conventional methods.