Multivariate monitoring of fermentation processes with non-linear modelling methods

Multiway principal components analysis (MPCA) and parallel factor analysis (PARAFAC) are widely used in exploratory data analysis and multivariate statistical process control (MSPC). These models are linear in nature, thus, limited when non-linear relations are present in the data. Principal component analysis (PCA) can be extended to non-linear principal components analysis using autoassociative neural networks. In this paper, the network’s bottleneck layer outputs (non-linear components) were made orthogonal. A method to estimate confidence limits based on a kernel probability density function was proposed since these limits do not assume that the non-linear scores are normally distributed. A measure for the non-linear scores (D_NL) was presented here to monitor on-line the process replacing the well known Hotelling’s T² statistic. One hundred and two industrial fermentation runs were used to evaluate the performance of a non-linear technique for multivariate process statistical monitoring. Three process runs with faults were used to compare the error detection performance using a statistic for the non-linear scores and the residuals statistic (SPE).     

Study of the application of multiway multivariate techniques to model data from an industrial fermentation process

Several multivariate statistical techniques have been extensively proposed for monitoring industrial processes. In this paper, multiway extensions of two such techniques: multiway principal component analysis (MPCA) and multiway partial least squares regression (MPLS) were applied to a large data set from an industrial pilot-scale fermentation process to improve process knowledge. The MPCA model is able to diagnose faults occurring in the process whether they affect or not process productivity while the MPLS model enables the prediction of final product concentration and the detection of faults that will influence the fermentation productivity.     

Enhanced monitoring of biopharmaceutical product purity using liquid chromatography-mass spectrometry

LC-MS is a widely used technique for impurity detection and identification. It is very informative and generates huge amounts of data. However, the relevant chemical information may not be directly accessible from the raw data map, particularly in reference to applications where unknown impurities are to be detected. This study demonstrates that multivariate statistical process control (MSPC) based on principal component analysis (PCA) in conjunction with multiple testing is very powerful for comprehensive monitoring and detection of an unknown and co-eluting impurity measured with liquid chromatography-mass spectrometry (LC-MS). It is demonstrated how a spiked impurity present at low concentrations (0.05% (w/w)) is detected and further how the contribution plot provides clear diagnostics of the unknown impurity. This tool makes a fully automatic monitoring of LC-MS data possible, where only relevant areas in the LC-MS data are highlighted for further interpretation.     

Monitoring of pigmented and wooden surfaces in accelerated ageing processes by FT-Raman spectroscopy and multivariate control charts

Two of the most suitable analytical techniques used in the field of cultural heritage are NIR (near-infrared) and Raman spectroscopy. FT-Raman spectroscopy coupled to multivariate control charts is applied here for the development of a new method for monitoring the conservation state of pigmented and wooden surfaces. These materials were exposed to different accelerated ageing processes in order to evaluate the effect of the applied treatments on the goods surfaces. In this work, a new approach based on the principles of statistical process control (SPC) to the monitoring of cultural heritage, has been developed: the conservation state of samples simulating works-of-art has been treated like an industrial process, monitored with multivariate control charts, owing to the complexity of the spectroscopic data collected.The Raman spectra were analysed by principal component analysis (PCA) and the relevant principal components (PCs) were used for constructing multivariate Shewhart and cumulative sum (CUSUM) control charts. These tools were successfully applied for the identification of the presence of relevant modifications occurring on the surfaces. CUSUM charts however proved to be more effective in the identification of the exact beginning of the applied treatment. In the case of wooden boards, where a sufficient number of PCs were available, simultaneous scores monitoring and residuals tracking (SMART) charts were also investigated. The exposure to a basic attack and to high temperatures produced deep changes on the wooden samples, clearly identified by the multivariate Shewhart, CUSUM and SMART charts. A change on the pigment surface was detected after exposure to an acidic solution and to the UV light, while no effect was identified on the painted surface after the exposure to natural atmospheric events.     

Construction of stable multivariate calibration models using unsupervised segmented principal component regression

In multivariate spectral calibration by principal component regression (PCR), the principal components (PCs) are calculated from the response data measured at all employed instrument channels; however some channels are redundant and their responses do not possess useful information. Thus, the extracted PCs possess mixed information from both useful and redundant channels. In this work, we propose a segmentation approach based on unsupervised pattern recognition to identify the most informative spectral region and then to construct a stable multivariate calibration model by PCR. In this method, the instrument channels are clustered into different segments via Kohonen self‐organization map. The spectral data of each segment are then subjected to PCA and the derived PCs are used as input variables for an inverse least square (ILS) regression model employing stepwise selection of the informative PCs. The proposed method was evaluated by the analysis of four simulated and six experimental data sets. It was found that our proposed method can model the above data sets with prediction errors lower than conventional partial least squares (PLS) and PCR methods. In addition, the prediction ability of our method was better than the previously reported models for these data sets. Copyright © 2011 John Wiley & Sons, Ltd.     

Multivariate data analysis of UV spectra in monitoring elution and determining endpoint of chromatography using polyamide column

Chromatography using polyamide column is one of the most critical purification operations that affect the safety and efficacy of traditional Chinese medicine (TCM) products. To ensure successful operation and reduce time and solvent consumption, UV spectroscopy combined with multivariate data analysis as an online analytical tool was developed for monitoring the polyamide column chromatography of the TCM Danshen. The process trajectories were established by principal component analysis of the UV spectra and used to determine the endpoint of the washing stage and investigate the impacts of the process conditions. The online analysis method developed determined the concentration of salvianolic acid B (an important compound in Danshen) in the effluent rapidly and precisely with a coefficient of determination of 0.9963 and helped to collect salvianolic acid B quantitatively for determining the endpoint of elution. The methodology proposed is an effective approach applicable in guiding successful operations in the chromatographic separation.     

Classification and Prediction of Geological Samples by Principal Component Analysis

The principal component analysis is an ancient multivariate statistical method[1]. It is extensively used in spectrometry with the popularization of computer and development of the method of chemometrics. It is regarded as an effective method of multivariate statistical analysis. The principal component analysis is universally included in common program package of multivariate statistical analysis. The method, as well as other multivariate calibration methods, combined with artificial neural networks forms the foundation of the chemometrics.     

Bilinear modeling of batch processes. Part III: parameter stability

A paramount aspect in the development of a model for a monitoring system is the so‐called parameter stability. This is inversely related to the uncertainty, i.e., the variance in the parameters estimates. Noise affects the performance of the monitoring system, reducing its fault detection capability. Low parameters uncertainty is desired to ensure a reduced amount of noise in the model. Nonetheless, there is no sound study on the parameter stability in batch multivariate statistical process control (BMSPC). The aim of this paper is to investigate the parameter stability associated to the most used synchronization and principal component analysis‐based BMSPC methods. The synchronization methods included in this study are the following: indicator variable, dynamic time warping, relaxed greedy time warping, and time linear expanding/compressing‐based. In addition, different arrangements of the three‐way batch data into two‐way matrices are considered, namely single‐model, K‐models, and hierarchical‐model approaches. Results are discussed in connection with previous conclusions in the first two papers of the series. Copyright © 2013 John Wiley & Sons, Ltd.     

A novel multi‐mode data processing method and its application in industrial process monitoring

Multi‐mode process monitoring is a key issue often raised in industrial process control. Most multivariate statistical process monitoring strategies, such as principal component analysis (PCA) and partial least squares, make an essential assumption that the collected data follow a unimodal or Gaussian distribution. However, owing to the complexity and the multi‐mode feature of industrial processes, the collected data usually follow different distributions. This paper proposes a novel multi‐mode data processing method called weighted k neighbourhood standardisation (WKNS) to address the multi‐mode data problem. This method can transform multi‐mode data into an approximately unimodal or Gaussian distribution. The results of theoretical analysis and discussion suggest that the WKNS strategy is more suitable for multi‐mode data normalisation than the z‐score method is. Furthermore, a new fault detection approach called WKNS‐PCA is developed and applied to detect process outliers. This method does not require process knowledge and multi‐mode modelling; only a single model is required for multi‐mode process monitoring. The proposed method is tested on a numerical example and the Tennessee Eastman process. Finally, the results demonstrate that the proposed data preprocessing and process monitoring methods are particularly suitable and effective in multi‐mode data normalisation and industrial process fault detection. Copyright © 2014 John Wiley & Sons, Ltd.     

Spectroscopic on-line monitoring of reactions in dispersed medium: chemometric challenges

Emulsion and suspension polymerizations are important industrial processes for polymer production. The end-user properties of polymers depend strongly on how the polymerization reactions proceed in time (i.e. a batch or semicontinuous, rate of reagents feeding, etc.). In other words, these reactions are process dependent, which makes the successful process control a key point to ensure high-quality products. In several process control strategies the on-line monitoring of reaction performance is required. Due to the multiphase nature of the emulsion and suspension processes, there is a lack of sensors to perform successful on-line monitoring. Near infrared and Raman spectroscopies have been pointed out as useful approaches for monitoring emulsion and suspension polymerizations and several applications have been described. In such instance, the chemometric approach on relating near infrared and Raman spectra to polymer properties is widely used and has proven to be useful. Nevertheless, the multiphase nature of emulsion and suspension polymerizations also represents a challenge for the chemometric approach based on multivariate calibration models and demands the development of new methods. In this work, a set novel results is presented from the monitoring of 15 batch emulsion reactions that show the chemometric challenge to be faced on development of new methods for successful monitoring of processes taken under dispersed medium. In order to discuss these results, several chemometric approaches were revised. It is shown that Raman and NIR spectroscopic techniques are suitable for on-line monitoring of monomer concentration and polymer content during the polymerizations, as well as medium heterogeneity properties, i.e. average particle size. It is also shown that Hotteling and Q statistics, widely used in chemometrics, might fail in monitoring these reactions, while an approach based on principal curves is able to overcome such restriction.     

Classification of secondary ion mass spectrometry (SIMS) micrographs to characterize chemical phases

This work demonstrates the potential of multivariate image analysis methods in the extraction of useful, problem dependent information from SIMS images. Specific algorithms have been developed to classify SIMS micrographs manually as well as automatically. A feature selection has been achieved by means of principal component analysis with a subsequent image classification.As an application example for these improved digital image processing tools chemical phases within a soldered industrial metal sample have been identified. This is of highly practical value as it was assumed that during the soldering process inhomogeneities occur along the joint site which cause a cracking of the brazed material under mechanical strain conditions.     

Multimode process monitoring based on Bayesian method

Multimode process monitoring has recently attracted much attention both in academy and industry. Conventional methods assume that either the process data are Gaussian in each operation mode, or some process knowledge should be incorporated, thus making the methods supervised. In this paper, a new unsupervised method is developed for multimode process monitoring, which is based on Bayesian inference and two‐step independent component analysis–principal component analysis (ICA–PCA) feature extraction strategy. ICA–PCA is first introduced for feature extraction and dimension reduction. By transferring the traditional monitoring statistic to fault probability in each operation mode, monitoring results in different operation modes can be easily combined by the Bayesian inference. Another contribution of the present paper is the development of a new fault identification method. Through analyses of the posterior probability and the joint probability for the monitored data sample, the correct operation mode or fault scenario can be identified. Three case studies are demonstrated to evaluate the feasibility and efficiency of the proposed method. Copyright © 2009 John Wiley & Sons, Ltd.     

Resolution of ternary mixtures of salicylic, salicyluric and gentisic acids by partial least squares and principal component regression: Optimization of the scanning path in the excitation-emission matrices

The resolution of ternary mixtures of salicylic, salicyluric and gentisic acids has been accomplished by partial least squares (PLS) and principal component regression (PCR) multivariate calibration. The total luminescence information of the compounds has been used to optimize the spectral data set to perform the calibration. A comparison between the predictive ability of the three multivariate calibration methods, PLS-1, PLS-2 and PCR, on three spectral data sets, excitation, emission and synchronous spectra, has been performed. The excitation spectrum has been the best scanning path for salicylic and salicyluric acid determinations, while the emission spectrum has been the best for the gentisic acid determination. The convenience of analysing the total luminescence spectrum information when using multivariate calibration methods on fluorescence data is demonstrated.     

Parallel factor analysis combined with PLS regression applied to the on-line monitoring of Pichia pastoris cultures

Various key variables (biomass, substrate and product) of bioprocesses should be monitored in order to retrieve useful information on the system, with the biomass (the cell density) the principal target. Although several analytical methods have been adapted and used to monitor the evolution of cell density evolution in cultures, a general method for performing this determination has not yet been established, as each technique has its own advantages and drawbacks. In the present work, noninduced glycerol batch cultures (for which biomass and substrate are the key variables) were monitored using multiwavelength fluorescence spectroscopy. The data gathered were modelled via PARAFAC-PLS chemometric methodologies, resulting in important qualitative and quantitative information about the behaviours of different biogenic fluorophors in batch cultures of the yeast Pichia pastoris. This information was used to predict the target process variables in such cultures; this permitted the applicability of this combined technique to bioprocess monitoring to be assessed.     

Fourier transform infrared spectroscopy supported by multivariate statistics in compatibility study of atenolol with excipients

The objective of this study was to use Fourier transform infrared spectroscopy (FTIR) and multivariate statistics to investigate compatibility/incompatibility of atenolol as a representative of active pharmaceutical ingredients and excipients, such as β-cyclodextrin, methylcellulose, starch and chitosan, when used in solid dosage formulations. Two-component physical mixtures consisting of atenolol and selected excipients were studied by FTIR spectroscopy and two methods of multivariate statistical analysis – principal component analysis (PCA) and cluster analysis (CA), which were used as a supplementary tool for interpretation of the FTIR spectra. Taking into account variability explained by the first two principal components, the results of PCA were visualized in the form of a bi-dimensional scatterplot. A lack of interaction was confirmed by two distinct clusters created by both atenolol and a particular excipient with their mixtures. In the case of CA, lack of interaction between both ingredients was also indicated by two large clusters at a level of 33 or 66% of the maximum distance. The results of the investigations show that with the exception of β-cyclodextrin, the remaining excipients are compatible with atenolol. These findings were confirmed by complementary methods, such as differential scanning calorimetry, thermogravimetry and X-ray powder diffraction.     

Daubechies小波主成分回归法机理及算法研究

将小波变换与主成分回归相结合,提出一种新型多元校正算法---小波基主成分回归法。理论分析和仿真实验表明,该法可更有效地去除噪声,提取有用信息。将其用于氯霉素及甲硝唑实际药物体系分析,与主成分回归法(PCR)相比,得到的回收率总平均相对误差由1.70%下降到0.90%。此外,通过将统计判据和小波多尺度分析相结合,发展了一种新的因子数判定方法。理论和实验研究表明,该法比传统因子数判定法具有更高的可靠性。     

OPLS in batch monitoring – Opens up new opportunities

In batch statistical process control (BSPC), data from a number of “good” batches are used to model the evolution (trajectory) of the process and they also define model control limits, against which new batches may be compared. The benchmark methods used in BSPC include partial least squares (PLS) and principal component analysis (PCA).     

Fault detection and classification for a two‐stage batch process

Multiway principal component analysis (MPCA) has been extensively applied to batch process monitoring. In the case of monitoring a two‐stage batch process, the inter‐stage variation is neglected if MPCA models for each individual stage are used. On the other hand, if two stages of reference data are combined into a large dataset that MPCA is applied to, the dimensions of the unfolded matrix will increase dramatically. In addition, when an abnormal event is detected, it is difficult to identify which stage's operation induced this alarm. In this paper, partial least squares (PLS) is applied to monitor the inter‐stage relation of a two‐stage batch process. In post‐analysis of abnormalities, PLS can clarify whether root causes are from previous stage operations or due to the changes of the inter‐stage correlations. This approach was successfully applied to a semiconductor manufacturing process. Copyright © 2008 John Wiley & Sons, Ltd.     

Monitoring of pigmented surfaces in accelerated ageing process by ATR-FT-IR spectroscopy and multivariate control charts

This work is an extension of a method for monitoring the conservation state of pigmented surfaces presented in a previous paper. A cotton canvas painted with an organic pigment (Alizarin) was exposed to UV light in order to evaluate the effects of the applied treatment on the surface of the sample. The conservation state of the pigmented surface was monitored with ATR–FT-IR spectroscopy and multivariate control charts.

The IR spectra were analysed by principal component analysis (PCA) and the relevant principal components (PCs) were used for constructing multivariate Shewhart, cumulative sums (CUSUM) and simultaneous scores monitoring and residuals tracking (SMART) control charts.

These tools were successfully applied for the identification of the presence of relevant modifications occurring on the surface of the sample.

Finally, with the aim to more deeply investigate what happened to the sample surface during the UV exposure, a PCA of the residuals matrix of degradation analyses only, not present in the previous paper, was performed. This analysis produced interesting results concerning the identification of the processes taking place on the irradiated surface.