Calibration Model Building for Online Monitoring of the Granule Moisture Content during Fluidized Bed Drying by NIR Spectroscopy

Abstract

For monitoring the granule moisture content during a fluidized bed drying (FBD) process, a calibration model building method is proposed for in situ measurement using the near-infrared (NIR) spectroscopy. It is found that the FBD operating conditions such as the chamber temperature and heating power have a nonnegligible impact on the NIR model prediction of granule moisture. By combining these operating variables with the measured NIR spectra for model calibration, the prediction accuracy for online measurement of the granule moisture content under different process conditions could be evidently improved compared to only using the measured NIR spectra for model calibration. To determine the optimal number of factors for establishing a partial-least-squares (PLS) regression model for predicting the granule moisture content, it is proposed to combine the leave-one-out cross validation (LOOCV) approach with the median absolute percentage error (MdAPE) index to deal with measurement outliers often involved with practical applications, based on a comparative study with the well-known K-fold cross validation (KCV) and Monte Carlo cross validation (MCCV) methods. Experimental results on monitoring the silica gel granule moisture under different FBD operating conditions demonstrate the effectiveness of the proposed spectral calibration method.

110th Anniversary: Real-Time End Point Detection of Fluidized Bed Drying Process Based on a Switching Model of Near-Infrared Spectroscopy

“…A brief introduction of the working mechanism and offline measurement by the losson-drying (LOD) method for validation can be found in a previous work. 4 2.2. Drying Materials.…”

“…More details about the experimental conditions and data collection are provided in the previous work. 4 Figure 2 illustrates the sampled NIR spectra and granule moisture content from batch 1. It is seen from the plot of granule moisture content that the FBD process basically consists of two phases, that is, underdrying and postdrying phases.…”

Review of portable near infrared spectrometers: Current status and new techniques

“…10 They can also reduce variables that do not correctly characterize the chemical properties of measured objects; this is called outlier removal. 11,12 Other pre-processing methods can also reduce the dimensionality of spectral data. 13 Based on the pre-processed data, classification and regression techniques are employed for qualitative and quantitative analyses.…”

Functional Soft Sensor Based on Spectra Data for Predicting Multiple Quality Variables

“…It was found that PLS is better than the other methods as the former has the minimum prediction error. With the prior process knowledge, some process variables were found to have a great effect on the final product quality, which were used for model building by combination with NIR spectroscopy [5] [15]. However, most existing calibration modeling methods are linear models.…”

“…Based on the above analysis, the dimension of spectral data should be reduced before modeling. Meanwhile, the nonlinear relationship should be taken into consideration in order to obtain more accurate calibration performance [4] [15]. In most published calibration methods, the modeling data are analyzed discretely and discretely by the traditional multivariate statistical methods, without considering the implied continuous property within the variables.…”