Simultaneous Measurement of Solution Concentration and Slurry Density by Raman Spectroscopy with Artificial Neural Network
Mengxing Lin, Yuanyi Wu, Sohrab Rohani
Abstract
In this work, the capability of Raman spectroscopy to measure the solution concentration and slurry density simultaneously and quantitatively was studied. Paracetamol–ethanol and l-glutamic acid–water systems were chosen as model systems. Different preprocessing methods (spectra range selection, baseline removal, direct orthogonal signal correction (DOSC), or no processing) and multivariable analysis techniques (characteristic peaks regression (CPR), principal component regression (PCR), partial least-squares regression (PLSR), and artificial neural network (ANN)) were applied and compared based on the root mean squared error (RMSE). It was demonstrated that the solution and solids concentration can be extracted separately from Raman spectroscopy. On the one hand, it is found that DOSC preprocessing can improve the fitting performance of the linear regression models (CPR, PCR, and PLSR) but not for ANN model. On the other hand, the ANN method, owing to its nonlinear prediction ability, better predicted the results than the linear models when the signal was weak.