A novel data fusion strategy based on multiple intelligent sensory technologies and its application in the quality evaluation of Jinhua dry-cured hams

Multiple intelligent sensory technologies including the electronic nose (E-nose), electronic tongue (E-tongue), and computer vision system have been developed for mimicking the functioning of olfaction, taste, and vision. The traditional data fusion relies on the data-level and feature-level fusion strategies, which may result in the unsatisfactory pattern recognition performance. In this study, a novel distance-probability classification (DPC) method was proposed for the pattern recognition of multiple intelligent sensory technologies. The proposed method was tested by data of multiple intelligent sensory technologies obtained from Jinhua dry-cured hams with different aging time. For qualitative classification, the DPC method can classify different hams with an accuracy rate of 100 %. To further use the fused data, the back propagation neural network (BPNN) models were built to predict the aging time and simultaneously predict 12 sensory attributes. The BPNN models exhibited satisfying performance on predicting aging time (R2 > 0.972) and sensory attributes (R2 > 0.935). This study suggests that data fusion of multiple intelligent sensory technologies is a promising method for food evaluation.