Electronic flavour assessment techniques for Orthosiphon stamineus tea from different manufacturers

The development of electronic nose (e-nose) and electronic tongue (e-tongue) to recognize simple or complex solutions has been a great success to overcome the drawbacks of conventional analytical instrument and human organoleptic profiling panels. The fusion of these sensors is believed to be able to assess flavours. To grab the advantage of the promising achievements and its broad prospect, this research paper focuses on the discrimination of herbal tea flavour. Multiple analyses based on low level data fusion (LLDF) and intermediate level data fusion (ILDF) in assessing herbal drink from different manufacturers were demonstrated in this research. Classification using Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), Probabilistic Neural Network (PNN) and k-Nearest Neighbour (KNN) analysis were illustrated. Experimental results show better classification was achieved for fusion data using the KNN and LDA with zero error rate. The findings demonstrate that the combination of e-nose and e-tongue with either LDA or KNN as the classification methods are suitable for discrimination of herbal tea flavour.