Feed forward neural network and its reverse mapping aspects for the simulation of ginger drying kinetics

Abstract

In the present study, simulation and modeling features of hot air based ginger drying kinetics were investigated by applying artificial neural system (ANN) and its reverse mapping aspects. Mapping of moisture ratio (MR) of the drying process as a function of temperature of drying (TD), slice thickness (ST) and drying time (DT) was accomplished based on the ANN architecture. A tale strategy of reverse neural system was built up to anticipate the drying process of ginger slices under given TD and ST for desired moisture content. Further, proportional odd displaying (POM) approach was applied for the tangible assessment of the dried samples. The ANN architecture, 3-5-1 was chosen as the best for modeling drying behavior of the ginger slices. Simulation of the ginger drying process was assured from the sensitivity analysis implemented based on the inversion of 3-5-1 ANN architecture. Effective diffusivity of the drying process as evaluated by the use of ANN (3-5-1) incorporated Fick's law approach varied from 6.92×10^-11 to 2.87×10^-11 m²/s. Color analysis of the dried products indicated best quality for the ginger samples treated at TD of 60^oC and ST of 7 mm (60-7). Besides, the results of POM demonstrated most elevated adequacy for the 60-7 ginger dried samples. Hence, the current investigation will be helpful for enhancing viable simulation and control during the hot air drying of ginger.