Exploration of using acoustic vibration technology to non-destructively detect moldy kernels of in-shell hickory nuts (Carya cathayensis Sarg.)

Abstract

The feasibility of acoustic vibration signals acquired by a self-made device was assessed as a non-destructive technology for identifying moldy kernels of in-shell hickory nuts (MK-IHNs). Twenty-two acoustic features extracted from time-domain and frequency-domain signals were used to establish a precious model for classifying MK-IHNs and healthy kernels of in-shell hickory nuts (HK-IHNs). Based on a combination of time-domain and frequency-domain features, an optimized LIB support vector machine (Lib-SVM) model was obtained with accuracy of 91.67% for identifying MK-IHNs. Some physical parameters of hickory nuts such as weight and bulk density were significantly correlated with most of the acoustic features (P < 0.05). Moreover, the microstructures of HK-IHN and MK-IHN shells, whatever outside surfaces or cross-sections, were obviously different. It follows then that the physical forms and structures of hickory nuts would play important roles in building a reliable model for identifying MK-IHNs based on acoustic vibration technology. Results of satisfactory proved that the acoustic vibration technology combined with Lib-SVM method could serve as a feasible approach for non-destructive damage identification of MK-IHNs.