Optimization of Preparation of Octenylsuccinic Sub-gelatinized Cassava Starch Based on RBFNN-GA

何国庆,石闪闪,赵勤勇
DOI: https://doi.org/10.16429/j.1009-7848.2013.11.017
2013-01-01
Abstract:On the basis of the results of orthogonal design L16(45), an optimized preparation of OSA-modified cassa- va starch was forecasted using radius basis function neural network and genetic algorithm(RBFNN-GA). The optimal combinations of modification were as follows: starch concentration 27.985787%(in proportion to aqueous solution of ethanol, w/w); esterification temperature 33.207857 ℃; esterification time 161.586120min; esterification pH 7.111757; the volume ratio of alcohol and octenyl succinic anhydride was 3.897390:1. Under these conditions, the degree of sub- stitution(DS) was 0.017994 and the reaction efficiency(RE) was 77.558148%. We did verification test under these condi- tions as closely as possible, the DS obtained was 0.01650195, the RE was 71.14593%. Compared with the predicted value, the DS and RE errors were 8.29193% and 8.267631%, which described that RBFNN-GA was a better accuracy optimization.
What problem does this paper attempt to address?