Novel alternative use of near-infrared spectroscopy to indirectly forecast 3D printability of purple sweet potato pastes

Abstract Conventional assessment of whether a food material is suitable for 3D printing requires time-consuming rheological properties measurement. Here, a novel alternative use of near-infrared spectroscopy to forecast 3D printability of a food material is proposed; purple sweet potato pastes of different formulations were used as test materials. According to actual 3D printing experiments as well as principle component and Fisher discriminant analyses of measured and fitted power-law based rheological parameters, the pastes could be divided into four categories, namely, supportable but non-flowable, well flowable and supportable, flowable and poorly supportable and flowable but non-supportable. NIR spectra within wavelength range of 921-1361 nm exhibited strong correlations with rheological property data. Models based on partial least square, principal component regression and artificial neural network employing suitable NIR spectroscopic parameters could well predict rheological properties of the pastes and could therefore indirectly but rapidly predict 3D printability of the pastes.