中文摘要：

以鳕鱼骨为原料制备胶原蛋白并对其酶解，利用人工神经网络建立鳕鱼骨胶原蛋白酶解反应的预测模型。结果显示，胶原蛋白及水解物的粘度随温度的升高而下降，随着胶原蛋白的水解程度加深，肽链长度与分子量减小，特征粘度呈现下降趋势，水解度与特征粘度呈一一映射关系。以水解产物的特征粘度、温度为输入参数，水解度作为输出参数，建立神经网络预测模型。样本值与仿真值的月。值为0．9916，平均相对误差为2．5％，所建网络预测模型性能好、预测精度高。验证试验的相对误差为1．06％～4．32％，实现了鳕鱼骨胶原蛋白酶解反应的仿真预测与监控。

英文摘要：

Collagen was extracted from pollock bones and hydrolyzed with trypsin. An enzymatic hydrolysis prediction model of pollock bone collagen was established based on artificial neural network. It shows that the viscosity of collagen and its hydrolysates is decreased with the increasing of temperature. With the hydrolysis degree increasing, the peptide chain length, the molecular weight decrease, and the intrinsic viscosity decrease. In addition, it exhibited a one-to-one mapping relationship between the hydrolysis degree and the intrinsic viscosity. Using the intrinsic viscosity and the temperature as input parameters and the degree of hydrolysis as output parameter, a neural network was trained and simulated by 51 samples. The value of R2 is O. 991 6 and the average relative error is only 2.5% , which indicates a good relevance between the sample actual value and the simulation value. Then three verification tests were performed using the prediction model, and the theoretical value is in agreement with the experimental value. The relative error is in range of 1.06% -4.32%. Therefore, the model can predict and monitor the hydrolysis of pollock bone collagen.