Abstract:

A quantitative analysis method for Ranitidine in Ranitidine powder was established by nearinfrared diffuse reflectance spectroscopy combined with movable window partial least squares (MWPLS). The model was optimized by selecting, such as, first order derivative, second order derivative, savitzky golay smoothing and FFT pretreatment methods, and  the hidden number of variables of model was optimized. The degree of the approaching (Da) was employed as the evaluation criterium of selecting effective parameters of  building models. Lastly, the optimum model of determination of Ranitidine powder drugs was obtained. The model was used to predict the correlation values. The results show that the correlation coefficient (Rc) between  predictive value and the true value of the calibration set and that of validation set were 0.984 5 and 0.977 5, respectively. The root mean square errors of the calibration value and validation value were 0.003 1 and 0003 3, respectively. Compared with that of the measurement results of HPLC, its relative error ≤2.422%.

Key words: near infrared reflectance spectrum, Ranitidine pharmaceutical powder, movable window partial least square method, quantitative analysis