An automated non-destructive prediction of peroxide value and free fatty acid level in mixed nut samples

This study aimed to develop an automated technique, which is rapid, non-destructive and inexpensive, test for rancidity of nuts. A visible near infrared benchtop hyperspectral camera was used capture images from blanched canarium, unblanched canarium macadamia samples. Support vector machine classification (SVC) PLSR models were developed segregate the pooled spectra nuts predict their peroxide values (PV) free fatty acid (FFA) concentrations. The SVC then in a hierarchical model system predicting PV FFA. tested using set providing accuracy 87% R2 between 0.60 0.76 RPD 1.6 2.7 FFA prediction. Overall, has potential commercial application nut processing detect mixed samples non-destructively real-time. It suggested train other learning with more improve predictions.