Rheological modelling of apple puree based on machine learning combined Monte Carlo simulation: Insight into the fundamental light- particle structure interaction processes

In this work, apple purees from different particle concentration and verifying in size were reconstituted to investigate their impacts on rheological behaviors, optical properties and light interaction at 900–1650 nm. The optical scattering of different puree particle concentration varied more intensively than particle size. Based on Monte Carol simulation (MCX), the highest proportion of diffuse reflection energy and absorption were observed in 1050 nm and 1480 nm, respectively. The averaged light propagation of all the photons in purees varied from 259.29 mm to 199.17 mm with the particle concentration from 25 % to 45 %. Finally, support vector machine regression based on optical scattering properties fused with MCX parameters at 1050 nm effectively evaluated puree apparent viscosity parameters (RPD > 2.39), viscous and elastic modulus (RPD > 2.47). These extended our knowledge of the fundamental light - particle structure interaction processes and provided a new solution for rheological modelling in pureed food. [Display omitted] •Particle concentration impacts puree rheology stronger than particle size.•Obvious optical scattering changes related to particle concentrations than sizes.•Monte Carol simulation can describe the light attenuation process in puree matrix.•SVM based on optical scattering and MCX parameter can predict puree viscoelasticity.