论文部分内容阅读
目的建立近红外快速预测片剂硬度的方法。方法采用硬度仪获得片剂真实硬度,运用偏最小二乘回归法(PLSR)和反向人工神经网络(BP-ANN)法建立近红外光谱与硬度之间的校正模型。结果偏最小二乘回归模型的相关系数r=0.9778,内部交叉验证均方根误差(RMSECV)为0.742 kg,预测均方根误差(RMSEP)为0.815 kg;反向人工神经网络训练集、监控集和测试集的相关系数r分别为0.987 3、0.985 6、0.986 8,各数据集的相对标准偏差(RSE%)分别为6.83%、8.77%、6.69%。结论反向人工神经网络非线性模型预测准确度要优于偏最小二乘回归线性模型。
Objective To establish a method of rapid prediction of tablet hardness by near infrared spectroscopy. Methods The real hardness of tablets was obtained by hardness tester. The calibration model between near-infrared spectra and hardness was established by using partial least squares regression (PLSR) and reverse artificial neural network (BP-ANN). Results The correlation coefficient of partial least-squares regression model was 0.9778, RMSECV was 0.742 kg and root mean square error of prediction (RMSEP) was 0.815 kg. The artificial neural network training set, monitoring set And test set r were 0.987 3,0.985 6,0.986 8 respectively. The relative standard deviations (RSE%) of each dataset were 6.83%, 8.77% and 6.69% respectively. Conclusion The ANN artificial neural network nonlinear model prediction accuracy is better than partial least squares regression linear model.