Improving the quantitative analysis accuracy of bagged liquid components with strong scattering by multi-pathlength data fusion

This work proposes the application of VIS-NIR spectroscopy as a potential analytical method to rapid and non-destructive quantitative analysis of bagged liquid products. Rapid and non-destructive quantitative analysis of bagged liquid products with strong scattering properties has been widely demanded in many fields, but due to various interference factors such as differences in packaging bags and strong scattering characteristics of products, the model is unstable and the model is less accurate. In order to improve the measurement accuracy and the robustness of the model, this paper utilized the “multi-pathlength data fusion” to increase the information of the measured substance carried by the spectrum, which is based on the nonlinear relationship between spectrum and optical pathlength caused by strong scattering. A non-destructive analysis of hemoglobin (Hb) concentration in blood bags was used to test the effectiveness of the method. The experimental results showed that the data fusion of three optical pathlengths can make the correlation coefficient of Hb in the prediction set up to 0.9915 and make the RMSEP less than 2 g/L. This results demonstrated that the proposed method could significantly improves the analysis accuracy of Hb in blood bags and reduces the requirements for measurement condition, it also provides ideas for the non- destructive measurement of other bagged liquid products.