Design, development and evaluation of a single-task electronic nose rig for assessing adulterated hydrosols

Abstract

The application of electronic noses (e-nose) in assessing food quality and authenticity is becoming more popular due to their excellent performance. Commercial e-noses are emerging in the market, however, they are diverse in function and still expensive, which economically limit their potential single-task food applications. This study concentrates on the design, development and evaluation of an e-nose rig capable of detecting adulterated hydrosols with the case study of rose water. The developed rig consists of eight gas sensors, an 18F4550 PIC microcontroller programmed in PICBasic Pro, and a mass flow controller. In addition, a computer based graphical user interface (GUI) programmed in MATLAB was developed. The ability of the developed e-nose to discriminate varying levels of adulterated rose water was evaluated. To this, eight different levels of adulterated rose water were prepared using rose geranium hydrosol and pure rose water. Radar-like graphs were established to visualize any differences in the response patterns. Principal component analysis (PCA) was performed to confirm the existing classes and also to reduce data dimensionality. This resulted in a data-point to variable ratio of 92:6 which is acceptable for the sake of this study to avoid overfitting problems. The PCA was followed by linear discriminant analysis (LDA) to evaluate the ability of the e-nose to indiscriminate adulterated rose water samples. This analysis revealed that the developed e-nose can perfectly discriminate various levels of adulteration found in our samples. LDA results were also compared to those of artificial neural networks (ANN), k-nearest neighbors and classification and regression trees among which the ANN showed the best performance. Hence, the rig successfully served as a cost-effective nondestructive tool in assessing adulterated rose water.