Improved method for prediction of milled rice moisture content based on Weibull distribution

The loss on drying method, which is regarded as the standard method of rice moisture content analysis, provides the most reliable results but is both labor intensive and time consuming. In order to improve the detection efficiency of the loss on drying method, this study investigated the drying characteristics of milled rice and developed an information fusion algorithm with which to predict milled rice moisture content based on the Weibull distribution and Levenberg-Marquardt (LM) algorithm. Application of the Weibull distribution model was investigated regarding its description of the drying kinetics of milled rice during infrared drying. An adaptive mechanism was applied to algorithm design, with the starting point of the estimation algorithm determined by calculating the drying rate at each measuring point, and the end-point distinguished using a two-level threshold algorithm. The calculated results were then compared with the measured data regarding the infrared drying of milled rice. For milled rice samples varying in moisture content from 14.44%-17.67% (dry basis), the relative error between predicted and observed values ranged 0.0037-0.0589, with a reduction in test time of 50.71%-67.87%.