Selective ensemble modeling based on nonlinear frequency spectral feature extraction for predicting load parameter in ball mills

Strong mechanical vibration and acoustical signals of grinding process contain useful information related to load parameters in ball mills. It is a challenge to extract latent features and construct soft sensor model with high dimensional frequency spectra of these signals. This paper aims to develop a selective ensemble modeling approach based on nonlinear latent frequency spectral feature extraction for accurate measurement of material to ball volume ratio. Latent features are first extracted from different vibrations and acoustic spectral segments by kernel partial least squares. Algorithms of bootstrap and least squares support vector machines are employed to produce candidate sub-models using these latent features as inputs. Ensemble sub-models are selected based on genetic algorithm optimization toolbox. Partial least squares regression is used to combine these sub-models to eliminate collinearity among their prediction outputs. Results indicate that the proposed modeling approach has better prediction performance than previous ones.

Simultaneous characterization of multiple properties of solid and liquid phases in crystallization processes using NIR

Near infrared spectroscopy （NIR） is now probably the most popular process analytical technology （PAT） for pharmaceutical and some other industries. However, unlike mid-IR, NIR is known to have difficulties in monitoring crystallization or precipitation processes because the existence of solids could cause distortion of the spectra. This phenomenon, seen as unfavorable previously, is however an indication that NIR spectra contain rich information about both solids and liquids, giving the possibility of using the same instrument for multiple property characterization. In this study, transflectance NIR calibration data was obtained using solutions and slurries of varied solution concentration, particle size, solid concentration and temperature. The data was used to build calibration models for prediction of the multiple properties of both phases. Predictive models were developed for this challenging application using an approach that combines genetic algorithm （GA） and support vector machine （SVM）. GA is used for wavelength selection and SVM for mode building. The new GA-SVM approach is shown to outperform other methods including GA-PLS （partial least squares） and traditional SVM. NIR is thus successfully applied to monitoring seeded and unseeded cooling crystallization processes of L-glutamic acid.