Research And Optimization Of Technological Process Based On Fermentation For Production Of Seaweed Feed

In recent years, with the increasing numbers and varieties of seaweed chemicalproducts, amount of their processing wastes is also raising, resulted in seriousenvironmental pollution. Therefore, disposal and utilization of seaweed processingwaste is an urgent problem in the field of ocean chemical engineering need to besolved.In this paper, a new method was provided for utilization of algae industrialwaste, which can convert algae industrial waste into animal feed by interaction ofenzyme and microorganism. Single factor experiments were carried out first todetermine the level of each fermentation parameters; also optimum conditions forfermentation were obtained by response surface analysis, then prediction andverification of experimental results was taken by using neural network model.According to Box-Behnkenb central composite experimental design principles,quadratic polynomial mathematical model of the process of fermentation wasestablished to make five factors-five levels response surface analysis based on singlefactor experiment. Content of crude protein and crude fiber in fermentation productswas taken as response value for analysis of response surface and contour lines,accordingly, effects of effective microorganesms, the cellulase, fermentationtemperature and fermentation time on fermentation process were examined. Underthe optimal conditions by above analysis, content of crude protein and crude fiber infermentation product of content was19.35%,12.53%respectively, ideal result.By response surface analysis, influence of compound agents and fermentationtime on crude protein content in the fermented products is significant, as well asinfluence of interaction of composite agents and cellulase on content of crude fiberin fermentation product. Effect of fermentation temperature and material-water ratioon fermentation process is relatively less significant, this showed that amount ofcompound agents and fermentation time has a very significant impact on the entirefermentation process.By neural network prediction model, R2of crude protein and crude fiber is94.41%and97.25%respectively, illustrated model predictive ability is excellent,relative error between predicted value and true value of validation experiment is0.82%and0.56%respectively, proved that prediction model was able to reflect theprocess of fermentation excellently, also content results, therefore, this method couldprovide a reference for further study of processing optimization.