RESUMO
During the production process of beer, it is of utmost importance to guarantee a high consistency of the beer quality. For instance, the bitterness is an essential quality parameter which has to be controlled within the specifications at the beginning of the production process in the unfermented beer (wort) as well as in final products such as beer and beer mix beverages. Nowadays, analytical techniques for quality control in beer production are mainly based on manual supervision, i.e., samples are taken from the process and analyzed in the laboratory. This typically requires significant lab technicians efforts for only a small fraction of samples to be analyzed, which leads to significant costs for beer breweries and companies. Fourier transform mid-infrared (FT-MIR) spectroscopy was used in combination with nonlinear multivariate calibration techniques to overcome (i) the time consuming off-line analyses in beer production and (ii) already known limitations of standard linear chemometric methods, like partial least squares (PLS), for important quality parameters Speers et al. (J I Brewing. 2003;109(3):229-235), Zhang et al. (J I Brewing. 2012;118(4):361-367) such as bitterness, citric acid, total acids, free amino nitrogen, final attenuation, or foam stability. The calibration models are established with enhanced nonlinear techniques based (i) on a new piece-wise linear version of PLS by employing fuzzy rules for local partitioning the latent variable space and (ii) on extensions of support vector regression variants (ð-PLSSVR and ν-PLSSVR), for overcoming high computation times in high-dimensional problems and time-intensive and inappropriate settings of the kernel parameters. Furthermore, we introduce a new model selection scheme based on bagged ensembles in order to improve robustness and thus predictive quality of the final models. The approaches are tested on real-world calibration data sets for wort and beer mix beverages, and successfully compared to linear methods, showing a clear out-performance in most cases and being able to meet the model quality requirements defined by the experts at the beer company. Figure Workflow for calibration of non-Linear model ensembles from FT-MIR spectra in beer production .
Assuntos
Cerveja/análise , Cerveja/normas , Análise de Alimentos/métodos , Espectroscopia de Infravermelho com Transformada de Fourier , CalibragemRESUMO
A Near Infrared (NIR) method was developed using a small benchtop feed frame system to quantify Saccharin potency in a powder blend during continuous manufacturing process. A 15-point Design of Experiments (DoE) was created based on the NIR spectral response and compositions of the formulation to develop a calibration set. The calibration set was designed to create compositional and raw material lots variation using minimum resources. The calibration experiments utilized around 0.5 kg Saccharin (Active Pharmaceutical Ingredient (API) surrogate) and 1.8 kg of excipients. Partial Least Square (PLS) modeling was used to develop a quantitative NIR method from the calibration data. The NIR method was implemented during 5 test batches in two different manufacturing sites across different potency levels at a continuous manufacturing platform for direction compression. Acceptable prediction performance was achieved from the NIR method at both sites. The NIR method was robust against changes in process scale and NIR instruments. The variance information built into the calibration set was found to be critical to successful model performance. This study shows a benchtop feed frame can be used for material sparing calibration method development without operating at a full-scale process line and applied across multiple sites, instruments at different potency levels.
Assuntos
Excipientes , Espectroscopia de Luz Próxima ao Infravermelho , Calibragem , Composição de Medicamentos , Análise dos Mínimos Quadrados , Pós , Comprimidos , Tecnologia FarmacêuticaRESUMO
A model recalibration method based on additive Partial Least Squares (PLS) regression is generalized for multi-adjustment scenarios of independent variance sources (referred to as additive PLS - aPLS). aPLS allows for effortless model readjustment under changing measurement conditions and the combination of independent variance sources with the initial model by means of additive modelling. We demonstrate these distinguishing features on two NIR spectroscopic case-studies. In case study 1 aPLS was used as a readjustment method for an emerging offset. The achieved RMS error of prediction (1.91 a.u.) was of similar level as before the offset occurred (2.11 a.u.). In case-study 2 a calibration combining different variance sources was conducted. The achieved performance was of sufficient level with an absolute error being better than 0.8% of the mean concentration, therefore being able to compensate negative effects of two independent variance sources. The presented results show the applicability of the aPLS approach. The main advantages of the method are that the original model stays unadjusted and that the modelling is conducted on concrete changes in the spectra thus supporting efficient (in most cases straightforward) modelling. Additionally, the method is put into context of existing machine learning algorithms.
RESUMO
The physico-chemical properties of Melamine Formaldehyde (MF) based thermosets are largely influenced by the degree of polymerization (DP) in the underlying resin. On-line supervision of the turbidity point by means of vibrational spectroscopy has recently emerged as a promising technique to monitor the DP of MF resins. However, spectroscopic determination of the DP relies on chemometric models, which are usually sensitive to drifts caused by instrumental and/or sample-associated changes occurring over time. In order to detect the time point when drifts start causing prediction bias, we here explore a universal drift detector based on a faded version of the Page-Hinkley (PH) statistic, which we test in three data streams from an industrial MF resin production process. We employ committee disagreement (CD), computed as the variance of model predictions from an ensemble of partial least squares (PLS) models, as a measure for sample-wise prediction uncertainty and use the PH statistic to detect changes in this quantity. We further explore supervised and unsupervised strategies for (semi-)automatic model adaptation upon detection of a drift. For the former, manual reference measurements are requested whenever statistical thresholds on Hotelling's T2 and/or Q-Residuals are violated. Models are subsequently re-calibrated using weighted partial least squares in order to increase the influence of newer samples, which increases the flexibility when adapting to new (drifted) states. Unsupervised model adaptation is carried out exploiting the dual antecedent-consequent structure of a recently developed fuzzy systems variant of PLS termed FLEXFIS-PLS. In particular, antecedent parts are updated while maintaining the internal structure of the local linear predictors (i.e. the consequents). We found improved drift detection capability of the CD compared to Hotelling's T2 and Q-Residuals when used in combination with the proposed PH test. Furthermore, we found that active selection of samples by active learning (AL) used for subsequent model adaptation is advantageous compared to passive (random) selection in case that a drift leads to persistent prediction bias allowing more rapid adaptation at lower reference measurement rates. Fully unsupervised adaptation using FLEXFIS-PLS could improve predictive accuracy significantly for light drifts but was not able to fully compensate for prediction bias in case of significant lack of fit w.r.t. the latent variable space.