中文摘要：

通过建立酶传感器-人工神经网络模型,实现对大马哈鱼皮明胶酶解进程的在线监控。采用大马哈鱼皮为原料制备明胶并用胰蛋白酶和碱性蛋白酶组成的复合酶对其进行酶解,分析不同的复合酶质量浓度和比例对水解度和水解液中游离氨基酸质量浓度的影响。以胰蛋白酶质量浓度、碱性蛋白酶质量浓度、底物质量浓度、游离谷氨酸和赖氨酸作为输入参数,水解度作为输出参数,建立酶传感器-人工神经网络预测模型。结果表明：在一定的复合酶酶解条件下,酶解液中游离谷氨酸和赖氨酸含量与其水解度具有显著的相关关系。因此,建立了复合酶酶解体系下游离谷氨酸和赖氨酸质量浓度的酶生物传感器-人工神经网络预测模型。该模型水解度样本值与拟合值的R^2值为0.998,拟合误差为0～4%,平均误差值为0.56%,对样本值拟合度很高。验证实验的相对误差为2.24%～4.66%,实现了大马哈鱼皮明胶在复合酶酶解体系下水解反应进程的预测。

英文摘要：

To achieve the on-line monitoring of salmon skin gelatin enzymatic hydrolysis process,biosensors-artificial neural networks（BIO-ANNs）model was established.The gelatin derived from salmon skin was extracted using hot water and hydrolyzed by trypsin and alkaline protease.The effects of different concentrations and proportions of compound enzyme on the content of free amino acid and the hydrolysis degree were analyzed.Then,BIO-Glu-Lys-ANNs predicting model was established using the concentrations of enzyme,substrate and free amino acid as input parameters and degree of hydrolysis as output parameter.The results show that there is a significant relationship between the contents of free glutamic acid and lysine and the hydrolysis degree under the condition of compound enzyme.The values of R^2,fitting error range and the average error of this model are 0.998,0-4% and 0.56%,respectively,which indicate that a good reliability and the theoretical value is in agreement with experimental value.The relative error range is 2.24%-4.66%in verification tests.Therefore,the model can predict the hydrolysis of salmon skin gelatin under the compound enzyme conditions.