Neural Network Modeling for Bio-enzymatic Degumming on Kenaf

In order to reduce chemical residues on kenaf fiber and solve the environmental pollution caused by traditional degumming methods, bio-enzymatic degumming on kenaf was proposed, and a neural network model was built to optimize the parameters during industrialized production. Pre-treatment of laboratory data was conducted through changes of scale and training samples were determined by equilibrium selection. The optimal process parameters were finally fixed by genetic algorithm: temperature is 34.11, pH value is 8, liquor ratio is 1: 37.61, time is 60 h, and enzyme consumption is 1.00 mL/10 g kenaf bast fiber. After verification by degumming experiments, the residual gum content and the weight-loss ratio was 10.9% and 19.8% respectively, obviously superior to the results of orthogonal experiments, and it could provide theoretical basis for industrialized degumming.