Projection To Latent Structures Model Research Articles

Tuning the mechanical properties of pectin films with polyphenol-rich plant extracts

The mechanical properties of pectin films enhanced with polyphenol-rich fruit extracts were investigated. The scavenging and reducing activity of plant extracts incorporated into the pectin films were determined using bench assays, and their antioxidant activity was correlated with a high presence of polyphenols, which were predominantly comprised of flavonoids and anthocyanins. The pectin films generated from the extracts exhibited a range of mechanical properties; tensile strength (4.99 MPa – 6.91 MPa), elongation at break (45.8 % - 52.3 %), and stiffness (1835 g mm−1 - 2765 g mm−1). To investigate the underlying relationships between plant extract composition and mechanical properties, Projection to Latent Structures (PLS) models were developed. The PLS models revealed that extracts containing high sugar and polyphenol content increase the tensile strength and moisture content of films. The elongation at break of the films was improved or diminished depending on the profile of sugar, acids, and polyphenols in the fruit extracts. Furthermore, the structures and concentration of anthocyanins and flavonoids were identified to strongly influenced the elongation at break differences. By modifying the concentration of sugars, organic acids, and polyphenols, the mechanical properties of pectin-based films can be tuned for tailored applications as food packaging materials.

Robust soft sensor systems for industry: Evaluated through real-time case study

A challenge for “Big Data” in the chemical production industry is not only to evaluate file storage but also to use online information to improve process performance. It should be spectral, vibration, thermal, and other sensors are more and more widely available. In today's harsh industrial conditions, accurate and reliable reviews or product quality assessments are critical. To predict important attribute factors utilizing quantifiable signals, information soft sensors dependent on Projection to Latent Structure (PLS) techniques are frequently used. However, due to changes in equipment, raw material, sensors, or management, most operations are carried out under real and stable conditions. The structure of the flexible sensors must be maintained at regular intervals. Reconstruction of the method using more recent sensor primary data focus of current design maintenance techniques, such as mobile window updates and recursive updates within the enterprise. In situations where data were collected with extremes, downtimes, and other transients in the non-stationary phase, this strategy was not sufficiently resilient. An alternative model update strategy was reviewed as part of this study. To assess the effectiveness of the current soft sensor approach, they modified two Key Performance Indicators (KPIs). The residue-dependent forecast KPI identifies long-term forecast damping models using a filtered estimation error. The KPI dependent on T2 would be a forecast KPI that checks the system's speculations to the expected original data. This updated strategy is effective in improving predictive accuracy without completely reconstructing the PLS model using research papers using industrial operations information. Finally, the KPI attributes and model upgrade mechanism could be used together. The researchers demonstrated that this update technique significantly improved the accuracy of the PLS soft detector predictions through the emulation of live behavior using industrial data. The configuration technique also made it possible to quickly identify underlying issues in situations where the original sample was ideal as well as informed engineers that a new method needed to be built.

Establishing Multivariate Specification Regions for Incoming Raw Materials Using Projection to Latent Structure Models: Comparison Between Direct Mapping and Model Inversion

Increasing raw material variability is challenging for many industries since it adversely impacts final product quality. Establishing multivariate specification regions for selecting incoming lot of raw materials is a key solution to mitigate this issue. Two data-driven approaches emerge from the literature for defining these specifications in the latent space of Projection to Latent Structure (PLS) models. The first is based on a direct mapping of good quality final product and associated lots of raw materials in the latent space, followed by selection of boundaries that minimize or best balance type I and II errors. The second rather defines specification regions by inverting the PLS model for each point lying on final product acceptance limits. The objective of this paper is to compare both methods to determine their advantages and drawbacks, and to assess their classification performance in presence of different levels of correlation between the quality attributes. The comparative analysis is performed using simulated raw materials and product quality data generated under multiple scenarios where product quality attributes have different degrees of collinearity. First, a simple case is proposed using one quality attribute to illustrate the methods. Then, the impact of collinearity is studied. It is shown that in most cases, correlation between the quality variable does not seem to influence classification performance except when the variables are highly correlated. A summary of the main advantages and disadvantages of both approaches is provided to guide the selection of the most appropriate approach for establishing multivariate specification regions for a given application.

Effect of Silane A-174 Modifications in the Structure, Chemistry, and Compressive Strength of PLA-HAP and PLA-β-TCP Biocomposites: Toward the Design of Polymer–Ceramic Implants with High Performance

The effect of silane A-174 (3-(trimethoxysilyl)propyl methacrylate (methacryloxypropyl trimethoxysilane)) modifications of the ceramic surface on the structural characteristics and mechanical performance of hydroxyapatite (HAP)–polylactic acid (PLA) and β-tricalcium phosphate (β-TCP)–PLA composite systems was investigated employing a combination of Fourier transform infrared (FTIR) spectroscopy, X-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), compressive strength measurements, and statistical models constructed via projection to latent structures (PLS) analysis. The composites were synthesized by solvent casting method at ambient conditions employing ceramic/PLA ratios of 60/40 and 50/50, and silane A-174 percentages of 0, 1, and 3 wt %. Peak positions of O 1s, Ca 2p, P 2p, Si 2p, and C 1s in XP spectra of ceramics, surface-modified ceramics, and the corresponding composites and compressive strength values were used as the predictors and response variables in PLS models. It was found that Ca 2p, P 2p, and Si 2p binding energies of the ceramic component employed in composite synthesis played the most significant role in determining compressive strength. PLA-HAP composite with a ceramic/PLA ratio of 50/50, synthesized employing 3 wt % silane A-174 treated HAP (HAP50PLA50Si3), resulted in enhancement of the compressive strength up to 365 MPa, which is the highest compressive strength value reported in the literature to the best of our knowledge. Additional P 2p, Ca 2p, and C 1s features positioned at 129.64, 350.08, and 280.74 eV in the XP spectra of this specimen together with significant shifts of Si 2p and C 1s bands toward lower binding energies and polymer-bridge-like surface characteristics signaled improved polymer–ceramic interaction in this composite. Attaining mechanical performance comparable with and even higher than that of cortical bone upon silane A-174 modifications might provide opportunities for clinical applications toward developing bioresorbable PLA-HAP-based implants that can be employed in load-bearing applications such as knee and hip replacements.

Projection to Latent Spaces Disentangles Pathological Effects on Brain Morphology in the Asymptomatic Phase of Alzheimer's Disease

Alzheimer's disease (AD) continuum is defined as a cascade of several neuropathological processes that can be measured using biomarkers, such as cerebrospinal fluid (CSF) levels of Aβ, p-tau, and t-tau. In parallel, brain anatomy can be characterized through imaging techniques, such as magnetic resonance imaging (MRI). In this work we relate both sets of measurements and seek associations between biomarkers and the brain structure that can be indicative of AD progression. The goal is to uncover underlying multivariate effects of AD pathology on regional brain morphological information. For this purpose, we used the projection to latent structures (PLS) method. Using PLS, we found a low dimensional latent space that best describes the covariance between both sets of measurements on the same subjects. Possible confounder effects (age and sex) on brain morphology are included in the model and regressed out using an orthogonal PLS model. We looked for statistically significant correlations between brain morphology and CSF biomarkers that explain part of the volumetric variance at each region-of-interest (ROI). Furthermore, we used a clustering technique to discover a small set of CSF-related patterns describing the AD continuum. We applied this technique to the study of subjects in the whole AD continuum, from the pre-clinical asymptomatic stages all the way through to the symptomatic groups. Subsequent analyses involved splitting the course of the disease into diagnostic categories: cognitively unimpaired subjects (CU), mild cognitively impaired subjects (MCI), and subjects with dementia (AD-dementia), where all symptoms were due to AD.

Parallel projection to latent structures for quality-relevant process monitoring

Abstract Projection to latent structures(PLS) is widely applied in quality-related process monitoring. Since it is a data-based method, the noise in data will affect PLS model parameter, which may lead to inappropriate space division. In this paper, an approach called parallel projection to latent structures (P-PLS) is proposed. To remove negative effect of the noise, P-PLS firstly updates original correlation matrix obtained by PLS through l0 minimization. Then process input (i.e., process variables) are divided into two orthogonal subspaces – the output-relevant and the input-relevant subspace by singular value decomposition (SVD) on updated correlation matrix. With respect to process output (i.e., quality variables), the unpredictable part is further decomposed by principal component analysis (PCA). Thus, P-PLS decomposes the whole input and output data space into only 4 subspaces, in which correlation subspace, unpredictable output principal subspace and unpredictable output residual subspace are output-relevant and input-residual subspace is input-relevant. Four fault detection indices are designed for complete monitoring the faults in respective subspaces. Furthermore, the relationships among P-PLS and concurrent PLS models are analyzed in detail. Numerical examples and Tennessee Eastman Process (TEP) are utilized to verify the effectiveness and superiority of proposed P-PLS algorithm.

Semi-supervised online soft sensor maintenance experiences in the chemical industry

With the increasing availability of spectral, vibrational, thermal and other sensors, the challenge of “Big Data” in chemical processing industry is not only to analyze the data offline, but also to make use of the data online to improve process operation. To this end, accurate and reliable measurements or estimations of product quality are critical in today’s demanding manufacturing environments. Data-driven soft sensors based on Projection to Latent Structure (PLS) methods are often used to model key quality variables using measureable inputs. However, most processes do not operate around a true steady state due to changes in equipment, feedstock, sensor and operating strategy. Therefore, soft sensor models need to be updated periodically. Current model maintenance approaches such as moving window update, recursive update in industry center around rebuilding the model using more recent process data. This approach is not robust enough in scenarios where process data is contaminated with outliers, downtime and other non-steady state transients. In this study, an alternative model update approach is developed. First, we adapted two key performance indicators (KPIs) for assessing the performance of the current soft sensor model. The Hotelling’s T2 based KPI is a predictive KPI that monitors for model extrapolations against future process data; the prediction residual based KPI then detects long term prediction degradation trends using a filtered prediction error. Second, we developed an update strategy using the robust mean and variance estimators of the inputs and outputs. Through case studies using industrial process data, this update method was demonstrated to be effective in improving prediction performance without rebuilding the PLS model from scratch. Lastly, the model update mechanism can be combined with both KPIs indicators. Through simulation of online behavior using industrial data, we showed that this update strategy effectively improved the prediction performance of the PLS soft sensor. In cases where the initial model was suboptimal, the update strategy also allowed for timely identification of underlying problems and alerted engineers of the need to rebuild the model.

Liquid chromatography coupled to quadrupole-time of flight tandem mass spectrometry based quantitative structure–retention relationships of amino acid analogues derivatized via n-propyl chloroformate mediated reaction

In the current study, quantitative structure-retention relationships (QSRR) were constructed based on data obtained by a LC–(ESI)-QTOF-MS/MS method for the determination of amino acid analogues, following their derivatization via chloroformate esters. Molecules were derivatized via n-propyl chloroformate/n-propanol mediated reaction. Derivatives were acquired through a liquid–liquid extraction procedure. Chromatographic separation is based on gradient elution using methanol/water mixtures from a 70/30% composition to an 85/15% final one, maintaining a constant rate of change. The group of examined molecules was diverse, including mainly α-amino acids, yet also β- and γ-amino acids, γ-amino acid analogues, decarboxylated and phosphorylated analogues and dipeptides. Projection to latent structures (PLS) method was selected for the formation of QSRRs, resulting in a total of three PLS models with high cross-validated coefficients of determination Q2Y. For this reason, molecular structures were previously described through the use of descriptors. Through stratified random sampling procedures, 57 compounds were split to a training set and a test set. Model creation was based on multiple criteria including principal component significance and eigenvalue, variable importance, form of residuals, etc. Validation was based on statistical metrics Rpred2, QextF22,QextF32 for the test set and Roy's metrics rm(Av)2 and rm(δ)2, assessing both predictive stability and internal validity. Based on aforementioned models, simplified equivalent were then created using a multi-linear regression (MLR) method. MLR models were also validated with the same metrics. The suggested models are considered useful for the estimation of retention times of amino acid analogues for a series of applications.

Screening of variable importance for optimizing electrodialytic remediation of heavy metals from polluted harbour sediments

Using multivariate design and modelling, the optimal conditions for electrodialytic remediation (EDR) of heavy metals were determined for polluted harbour sediments from Hammerfest harbour located in the geographic Arctic region of Norway. The comparative importance of the variables, current density, remediation time, light/no light, the liquid–solid ratio and stirring rate of the sediment suspension, was determined in 15 laboratory-scale EDR experiments by projection to latent structures (PLS). The relation between the X matrix (experimental variables) and the Y matrix (removal efficiencies) was computed and variable importance in the projection was used to assess the influence of the experimental variables. Current density and remediation time proved to have the highest influence on the remediation of the heavy metals Cr, Cu, Ni, Pb and Zn in the studied experimental domain. In addition, it was shown that excluding the acidification time improved the PLS model, indicating the importance of applying a limited experimental domain that covers the removal phases of each heavy metal in the specific sediment. Based on PLS modelling, the optimal conditions for remediating the Hammerfest sediment were determined; operating in the experimental domain of 0.5–0.8 mA/cm2 and a remediation time after acidification of 450–570 h met acceptable levels according to Norwegian sediment quality guidelines.

Adaptive total PLS based quality-relevant process monitoring with application to the Tennessee Eastman process

Increasing global competition is setting even higher demands for the safety, quality and operating efficiency of industrial processes. The traditional projection to latent structures (PLS) based methods for quality-relevant fault detection has appeared in several industrial applications, while total PLS that performs more completely has been used as a better tool for monitoring associated with the product quality. However, the running/operating states for the process variables are often non-stationary, time-varying. Thus, the static PLS or TPLS for these processes will reduce the efficiency of the monitoring, unreliable monitoring results will affect the engineers׳ decision-making. Under this background, an adaptive modification on total PLS model named as recursive TPLS will be proposed to adapt the monitoring model on line. The new recursive version is achieved via a far more computation-efficient manner and the operating cost is significantly lowered. The simulation on TE process illustrates the effectiveness of the new adaptive fault monitoring approach based on RTPLS.

An error-in-variable projection to latent structure framework for monitoring technical systems with orthogonal signal components

This article introduces an error-in-variable (EIV) data structure for process monitoring models, that assumes the presence of source variables that can be correlated and uncorrelated to the system response variables. To identify such models, the paper proposes a different objective function for the projection to latent structure (PLS) approach. Compared to existing work, this modified PLS formulation does not remove uncorrelated, or orthogonal, components from the predictor set prior to the identification of a PLS model nor identify an initial PLS model and carry out a subsequent extraction of various correlated and orthogonal components from the predictor set. The proposed PLS algorithms extract latent components to predict the system response variables as accurately as possible after estimating the error covariance matrices using a maximum likelihood algorithm that is also introduced in this article. A detailed analysis of this extended PLS framework yields that the objective function is an iterative maximum redundancy formulation. The article finally shows that the developed algorithms here are computationally more efficient than existing PLS-based approaches.

Prediction of polar oil and grease contamination levels in refinery wastewater through multivariate statistical modeling

A dynamic mass balance was implemented for polar oil and grease contamination in a petroleum refinery wastewater circuit through developing a predictive model for a kerosene caustic washing unit, identified as the major source of polar substances into the industrial wastewater effluent. The model was developed following the Projection to Latent Structures (PLS) approach. Comparison between analytical data for polar oil and grease concentrations in refinery wastewater in the Dissolved Air Flotation (DAF) effluent and the predictions of the dynamic mass balance calculations are in a very good agreement and highlights the dominant impact of spent caustic in increasing the oil and grease contamination levels. The results of this work have led to implementation of the dynamic mass balance in the refinery as an online tool, which allows for prevision of possible contamination situations and taking the required preventive actions, and can serve as a basis for establishing relationships between the oil and grease contamination in the refinery wastewater with the properties of the refined crude oils and the process operating conditions. The PLS model developed could be of great asset in both optimizing the existing and designing new refinery wastewater treatment units or reuse schemes.

Assessment of phenomena underlying the removal of micropollutants during water treatment by nanofiltration using multivariate statistical analysis

Multivariate statistically-based models were developed for the comprehension of the phenomena underlying the removal of several micropollutants during drinking water treatment. Projection to latent structures (PLS) modeling was used to describe the apparent rejection as well as adsorption using specific descriptors of the molecules (physico-chemical properties and molecular size parameters), descriptors of the membrane used and of the water matrices, and operating related conditions. Multilinear PLS models with good descriptive capability towards rejection were developed. Alkalinity, molecular size descriptors, molecular weight, molar volume, and LogD were found to be the most relevant contributors for the rejection of the selected micropollutants, showing the impact of size exclusion and electrostatic interactions with minor contributions of hydrophobic interactions. The incorporation of molecular size descriptors demonstrated that the geometry of the molecule is important to determine rejection when molecules have very different geometry. Conversely, PLS modeling of adsorption required also the inclusion of quadratic and interaction terms to achieve good description. Adsorption of the selected compounds seems to be mainly determined by polar and electrostatic interactions rather than hydrophobic and may be described by polarizability as well as several specific interactions. Additionally, the incorporation of molecular size descriptors did not improve modeling of adsorption, showing that molecular geometry is not so important to describe this process as for rejection.

Contribution rate plot for nonlinear quality-related fault diagnosis with application to the hot strip mill process

In this paper, a nonlinear fault diagnosis scheme is established for the hot strip mill process (HSMP). In HSMP, the faults affecting quality index are denoted as quality-related faults, which should be taken care as soon as possible. Projection to latent structures (PLS) is a basic model for quality-related fault detection in linear processes. In the presented work, a total kernel PLS (T-KPLS) model is utilized for modeling and monitoring HSMP, which is a typical nonlinear process. However, diagnosis tools have not been developed aiming at the nonlinear case based on T-KPLS model. Motivated by the successful use of contribution plot for the linear case, a contribution rate plot is proposed to extend contribution plots to the nonlinear case. In the end of this paper, the proposed method is applied to the hot strip mill process effectively.

Quality-Related Process Monitoring Based on Total Kernel PLS Model and Its Industrial Application

Projection to latent structures (PLS) model has been widely used in quality-related process monitoring, as it can establish a mapping relationship between process variables and quality index variables. To enhance the adaptivity of PLS, kernel PLS (KPLS) as an advanced version has been proposed for nonlinear processes. In this paper, we discuss a new total kernel PLS (T-KPLS) for nonlinear quality-related process monitoring. The new model divides the input spaces into four parts instead of two parts in KPLS, where an individual subspace is responsible in predicting quality output, and two parts are utilized for monitoring the quality-related variations. In addition, fault detection policy is developed based on the T-KPLS model, which is more well suited for nonlinear quality-related process monitoring. In the case study, a nonlinear numerical case, the typical Tennessee Eastman Process (TEP) and a real industrial hot strip mill process (HSMP) are employed to access the utility of the present scheme.

Associations between complex OHC mixtures and thyroid and cortisol hormone levels in East Greenland polar bears

The multivariate relationship between hair cortisol, whole blood thyroid hormones, and the complex mixtures of organohalogen contaminant (OHC) levels measured in subcutaneous adipose of 23 East Greenland polar bears (eight males and 15 females, all sampled between the years 1999 and 2001) was analyzed using projection to latent structure (PLS) regression modeling. In the resulting PLS model, most important variables with a negative influence on cortisol levels were particularly BDE-99, but also CB-180, -201, BDE-153, and CB-170/190. The most important variables with a positive influence on cortisol were CB-66/95, α-HCH, TT3, as well as heptachlor epoxide, dieldrin, BDE-47, p,p′-DDD. Although statistical modeling does not necessarily fully explain biological cause–effect relationships, relationships indicate that (1) the hypothalamic–pituitary–adrenal (HPA) axis in East Greenland polar bears is likely to be affected by OHC-contaminants and (2) the association between OHCs and cortisol may be linked with the hypothalamus–pituitary–thyroid (HPT) axis.

Monitoring and Control of a Continuous Grignard Reaction for the Synthesis of an Active Pharmaceutical Ingredient Intermediate Using Inline NIR spectroscopy

Inline near-infrared (NIR) spectroscopy has been used to monitor a continuous synthesis of an active pharmaceutical ingredient (API) intermediate by a Grignard alkylation reaction. The reaction between a ketone substrate and allylmagnesium chloride may form significant impurities with excess feeding of the Grignard reagent beyond the stoichiometric ratio. On the other hand, limiting this reagent would imply a loss in yield. Therefore, accurate dosing of the two reactants is essential. A feedforward–feedback control loop was conceived in order to maintain the reaction as closely as possible to the stoichiometric ratio, leading the path to full process automation. The feedback control loop relies on NIR transmission measurements performed in a flow cell where, in contrast to labor-intensive offline HPLC analytical methods, the whole reaction product can be scanned in real time without sample dilution. A robust PLS (projection to latent structures) model was developed with a satisfactory standard error of pr...

Hybrid modeling of counterion mass transfer in a membrane-supported biofilm reactor

This paper presents a hybrid mechanistic/statistical model for predicting counterion fluxes across an ion-exchange membrane in a membrane-supported biofilm reactor. The model was calibrated with operating data for the removal of nitrate and perchlorate from a simulated contaminated drinking water stream. Two different modeling strategies were tested: a cooperative parallel hybrid model and a competitive mixture-of-experts (MOE) structure both joining a mechanistic Donnan-dialytic transport model and a multivariate projection to latent structures (PLS) model. The MOE structure proved to be a better predictive tool since it combines the two hybrid model elements in a mediated network. The PLS model was used to identify the process variables that are responsible for the mechanistic model inaccuracy. The results showed that biocompartment physicochemical data need to be considered in the modeling of the transport of counterions across the membrane, especially in situations in which the target counterion (e.g., perchlorate or nitrate) is metabolically reduced in the biocompartment. By using this strategy, the complex biofilm contribution to the transport was accounted for, without the need of developing mechanistic models built on simplified and/or inaccurate assumptions.

Quality Relevant Data-Driven Modeling and Monitoring of Multivariate Dynamic Processes: The Dynamic T-PLS Approach

In data-based monitoring field, the nonlinear iterative partial least squares procedure has been a useful tool for process data modeling, which is also the foundation of projection to latent structures (PLS) models. To describe the dynamic processes properly, a dynamic PLS algorithm is proposed in this paper for dynamic process modeling, which captures the dynamic correlation between the measurement block and quality data block. For the purpose of process monitoring, a dynamic total PLS (T-PLS) model is presented to decompose the measurement block into four subspaces. The new model is the dynamic extension of the T-PLS model, which is efficient for detecting quality-related abnormal situation. Several examples are given to show the effectiveness of dynamic T-PLS models and the corresponding fault detection methods.

Multivariate statistical modelling of mass transfer in a membrane-supported biofilm reactor

The main objective of this work is to develop an overall mass transfer model applicable to a particular case of membrane supported biofilm, the ion-exchange membrane bioreactor (IEMB). A multivariate projection to latent structures (PLS) model of the anionic membrane transport in an IEMB was developed and analyzed to establish the mass transfer limiting variables for the removal of anionic pollutants (nitrate and perchlorate) from drinking water. The proposed PLS model accounts for the biological contribution to the mass transfer and predicts the anionic fluxes across the ion-exchange membrane with a prediction improvement of at least 50% when compared with a simplified mechanistic Donnan dialysis-based transport model. The PLS model allowed for predicting the transport of target anions using only operational physicochemical data, therefore, the use of several assumptions as in mechanistic model building was avoided as well as the need for biofilm characterization. To decrease the model complexity, several techniques which select the most informative predictors were also successfully used. The analyses of important predictors to each anionic transport model show that transport driving force related variables were the most important. Moreover, at least 30% of the model information is related with biocompartment bulk variables.