shu 人工神经网络和响应面法优化黑曲霉发酵产淀粉酶.

王媛; 王永红

The medium and cultivation conditions are critical determinants in the fermentation process of Aspergillus niger for α-amylase production. To optimize α-amylase yield from A. niger F223, three significant factors influencing α-amylase activity were initially identified through single-factor and Plackett-Burman experiments, i.e. soybean meal, corn steep liquor, and soluble starch. Subsequently, datasets for regression modeling were generated using steepest ascent methodology and central composite design. Two models were developed to elucidate the relationship between medium components and α-amylase activity employing traditional polynomial regression as well as artificial neural network regression fitting techniques. The optimal medium composition and cultivation parameters for enhanced α-amylase production were successfully determined. The results indicated that the integration of artificial neural networks with genetic algorithms outperformed polynomial regression in terms of data fitting accuracy and predictive capability. Ultimately, under ideal concentrations of soybean meal (36 g/L or 38 g/L), corn steep liquor (33 g/L or 31 g/L), and soluble starch (56 g/L or 49 g/L), the achieved α-amylase activity reached 5 566.79 U/mL, reflecting a remarkable 92.6% increase compared to unoptimized culture conditions.

ARTIFICIAL neural networks; REGRESSION analysis; SOYBEAN meal; ASPERGILLUS niger; GENETIC algorithms; AMYLASES

Journal of East China University of Science & Technology, 2024, Vol 50, Issue 6, p840

1006-3080

Academic Journal

10.14135/j.cnki.1006-3080.20231209001