糖化麦汁麦芽糖含量预测模型的建立

Establishment of Predictive Model for Maltose in Wort

采用正交实验设计法研究了糖化条件对麦汁中麦芽糖含量的影响。通过实验发现，所选的4个糖化条件，即麦芽选择，辅料比例、糖化温度和保温时间都对麦芽糖含量有显著影响。在此基础上，建立了麦汁麦芽糖含量预测神经网络模型，采用10组数据进行网络训练和仿真，并且用3组未参加训练的数据进行预测。模型的仿真最大相对误差为1.87％，预测最大相对误差为7.85％。

The effect of mashing factors on maltose in wort was studied by orthogonal test. And conclusions were achieved that the choice of malt, the proportion of rice, the temperature and the time of mashing have signifi- cant effect on maltose in wort. Then a model for maltose in wort was established based on 10 group of data. For simulation, the model of maltose has relative error of 1.87 %, and for prediction, the relative error is 7.85 %.