Wavelet Function Based Spectral Model Calibration for Measuring Crystallization Solution via ATR-FTIR Spectroscopy

For measuring the solution concentration of crystallization process by ATR-FTIR spectroscopy, this paper proposes an improved spectral model calibration method to guarantee in-situ measurement accuracy, based on wavelet function regression. The spectral model structure is composed of wavelet basis functions, which could address the nonlinear properties and high-dimensional problem of infrared spectral variables. Moreover, the mean centralization strategy is taken to reduce the influence from measurement noise. In addition, a comparative study on measuring the solution concentration of the L-glutamic acid (LGA) crystallization process is conducted in terms of using different modeling methods and spectral variables. It is found that the wavelet function regression with all spectral variables could obtain good measurement accuracy compared to the traditional partial least-squares (PLS) method. Experimental results well demonstrate the measurement accuracy of the proposed spectral calibration method.