Development of pistachio sorting system using principal component analysis (PCA) assisted artificial neural network (ANN) of impact acoustics

This paper discusses the design and evaluation of an intelligent sorting system for open and closed-shell pistachio nuts. The system includes a feeder, an acoustical part, an electronic control unit, a pneumatic air-rejection mechanism and ANN classifier. A prototype system was set up to detect closed-shell pistachio nuts by dropping them onto a steel plate and recording the acoustic signal that was generated when a kernel hit the plate. The recognition is based on combined PCA of impact acoustics and ANN classifier. To generate useful features, both time and frequency-domain analysis of recorded sound signals were performed. Through PCA the original multiple variables were represented by seven principal components (PCs) and, hence, led to more than 99% reduction of input parameters in the ANN model. To find the optimal classifier, both off-line and online modeling was carried out. In the off-line phase 3200 nuts were used for training ANN models. Various ANN topologies with varying number of PCs were designed. The best classifier had a 7-12-2 structure. This optimal model was selected after several evaluations based on minimizing of mean square error (MSE), correct classification rate (CCR) and coefficient of correlation (r). Testing with 300 each of closed-shell, open-shell and thin-split nuts resulted in an overall error of 4.3%. The results indicated the CCR of thin-split nuts obtained from this system is higher than prior systems. A further experiment was conducted to find out the accuracy of the system in sorting open and closed-shell pistachio nuts of various sizes. The results indicated the size of pistachio nuts has no effect on the accuracy of the sorter. Experimental results establish the superior performance of the proposed approach when compared to prior techniques reported in the literature or used in the field.