论文部分内容阅读
采用摇床培养,测试了20种食品防腐剂在实验条件下对大肠杆菌的最低抑菌浓度;计算了所有供试防腐剂的10种结构、电子、理化性质参数,随机抽取其中17种防腐剂作为训练样本,另外3种防腐剂作为预测样本,构造并训练得到能较好预测“未知”食品防腐剂在实验条件下对大肠杆菌的最低抑菌浓度的BP人工神经网络,建立了能较准确预测食品防腐剂抗菌活性的QSAR模型,该模型对防腐剂抗菌活性的预测值和实测值相对误差不超过±5%。
The minimum inhibitory concentrations (MICs) of 20 kinds of food preservatives against Escherichia coli were tested by shaker culture. Ten kinds of structural, electronic and physicochemical properties of all tested preservatives were calculated. Among them, 17 kinds of preservatives As training samples, the other three preservatives were used as predictive samples to construct and train a BP artificial neural network that can better predict the minimum inhibitory concentration (MIC) of Escherichia coli under the experimental conditions of “unknown” food preservatives. The QSAR model, which accurately predicts the antimicrobial activity of food preservatives, has a relative error of ± 5% between the predicted and measured values of the antimicrobial activity of the preservative.