Partial Least Squares Framework Research Articles

Partial least squares enhance multi-trait genomic prediction of potato cultivars in new environments

It is of paramount importance in plant breeding to have methods dealing with large numbers of predictor variables and few sample observations, as well as efficient methods for dealing with high correlation in predictors and measured traits. This paper explores in terms of prediction performance the partial least squares (PLS) method under single-trait (ST) and multi-trait (MT) prediction of potato traits. The first prediction was for tested lines in tested environments under a five-fold cross-validation (5FCV) strategy and the second prediction was for tested lines in untested environments (herein denoted as leave one environment out cross validation, LOEO). There was a good performance in terms of predictions (with accuracy mostly > 0.5 for Pearson’s correlation) the accuracy of 5FCV was better than LOEO. Hence, we have empirical evidence that the ST and MT PLS framework is a very valuable tool for prediction in the context of potato breeding data.

Marketing or methodology? Exposing the fallacies of PLS with simple demonstrations

PurposeOver the past 20 years, partial least squares (PLS) has become a popular method in marketing research. At the same time, several methodological studies have demonstrated problems with the technique but have had little impact on its use in marketing research practice. This study aims to present some of these criticisms in a reader-friendly way for non-methodologists.Design/methodology/approachKey critiques of PLS are summarized and demonstrated using existing data sets in easily replicated ways. Recommendations are made for assessing whether PLS is a useful method for a given research problem.FindingsPLS is fundamentally just a way of constructing scale scores for regression. PLS provides no clear benefits for marketing researchers and has disadvantages that are features of the original design and cannot be solved within the PLS framework itself. Unweighted sums of item scores provide a more robust way of creating scale scores.Research limitations/implicationsThe findings strongly suggest that researchers abandon the use of PLS in typical marketing studies.Practical implicationsThis paper provides concrete examples and techniques to practicing marketing and social science researchers regarding how to incorporate composites into their work, and how to make decisions regarding such.Originality/valueThis work presents a novel perspective on PLS critiques by showing how researchers can use their own data to assess whether PLS (or another composite method) can provide any advantage over simple sum scores. A composite equivalence index is introduced for this purpose.

A novel anomaly detection scheme for high dimensional systems using Kantorovich distance statistic

The partial least squares (PLS) is a commonly applied multi-variate method in anomaly detection problems. The PLS strategy has been amalgamated with T^{2} and squared prediction error (SPE) based statistical indicators to detect anomalies in process. These traditional indicators have few setbacks that has made them ineffective in monitoring applications. Hence, a statistical indicator based on Kantorovich distance (KD) is proposed for detecting sensor anomalies in this study. The proposed strategy integrates anomaly indicator based on KD metric with PLS based multi-variate method. The KD metric computes difference between the residuals of anomaly-free data and the data with anomaly and uses this distance as an indicator of anomaly. The proposed strategy’s critical feature is that a single anomaly indicator is sufficient to be integrated with PLS modeling framework. The Tennessee Eastman process benchmark and experimental distillation column processes data are used for assessing the performance of the proposed strategy. Further, comparisons have been provided between KD, T^{2}, SPE and Generalized Likelihood Ratio indicators. The results demonstrate the superiority of the KD statistical indicator in detecting sensor anomalies in comparison to the traditional indicators of PLS based strategy. The KD indicator integrated with PLS framework also enhances the detection of small magnitude anomalies.

A global surrogate model for high-dimensional structural systems based on partial least squares and Kriging

Surrogate model techniques have been widely used in uncertainty propagation (UP) of structural systems. However, establishing surrogate models for structural systems with high-dimensional input and output variables remains a challenging task. The computational burden of constructing surrogate model for such systems can be alleviated by simultaneously reducing the dimensionality of the high-dimensional input–output variables in advance. Partial least squares (PLS) technique can achieve this purpose by searching for the latent structures of the high-dimensional structural systems, i.e., the principal components (PCs) of the input and output variables and their functional relationships. Kriging is a surrogate model with analytical properties and good fitting effects for nonlinear functions, which has been widely used in UP and quantification of engineering structural systems. Therefore, this paper combines Kriging and PLS to develop a global surrogate model for high-dimensional structural systems, named as PLS-K. In the proposed method, PLS is employed to identify the input–output PCs, wherein Kriging model is used to establish the relationship between each pair of PCs. In this way, establishing Kriging model of the structural system with high-dimensional inputs and outputs is decomposed into a series of one-dimensional Kriging construction of input–output PCs, which can significantly alleviate the burden of surrogate model construction. In addition, in order to better incorporate Kriging model into PLS framework, the error-based weights updating (EBWU) procedure of PLS is analytically derived based on Kriging. Several examples demonstrate the efficiency and accuracy of the proposed method.

The impact of lean manufacturing practices on operational and business performances at SMES in the wooden furniture industry

PurposeFirst, this study aims to examine the level of implementation of lean manufacturing (LM) practices by the small- and medium-sized enterprises (SMEs) in the wooden furniture industry in Jepara. Second is to examine the impact of LM practices on operational performance (OP) and business performance (BP). Third is to examine the impact of LM practices on BP with OP as a mediating and moderating variable. OP is the quantifiable aspect of the process of an organisation, such as production cycle time, reliability and inventory turnover, whereas BP is usually related to the responsibility of the firms to their shareholders and has the goal of profit maximization.Design/methodology/approachThis study used primary data collected through an offline questionnaire. The questionnaire was intended to identify the extent of the implementation of LM practices and the level of OP and BP achieved by SMEs. The LM practices consist of 6 dimensions with 26 indicators; OP, 4 dimensions with 11 indicators; and BP, 3 dimensions with 9 indicators. The data obtained from the questionnaire were processed via partial least squares (PLS) regression using the SmartPLS software. PLS is generally used to assess the extent to which LM practices predict values in OP and BP.FindingsA total of 100 questionnaires were administered, of which 81% were properly filled, completed and returned. The result of the study confirms that LM practices should be applied collectively and comprehensively as each practice is interdependent. Moreover, it confirms that these practices have a direct effect on OP and BP and that OP can drive broader BP measures. Finally, the result confirms that these practices can improve BP indirectly with OP as a mediator variable. While the empirical evidence in this study supported the role of OP as a mediating variable between the implementation of LM practices and OB, it did not support the role of OP as a moderating variable.Research limitations/implicationsFirst, it is arguable that LM practices, OP and BP are only measured by the Likert scale, which would likely create bias and inconsistency from the owners or managers of SMEs in expressing the level of LM practices and performances achieved by the SMEs in the wooden furniture industry. Future research may help establish qualitative approaches to better measure the LM practices as well as the OP and BP through observation and probing. Second, this study was limited geographically (limited to only the SMEs in the wooden furniture industry in Jepara and did not cover all regions in Central Java as well as in Indonesia). Therefore, a wider geographical area could be considered, including the other regions in Indonesia which also produce wooden furniture.Practical implicationsThis study practically contributes to the LM body of knowledge by identifying the relationships among the LM practices, OP and BP in SMEs. Understanding these relationships will help the owner or managers of SMEs make better decisions in achieving the OP and BP. The owner or managers of SMEs who implement the LM practices individually may experience disappointing performance results as these practices should be applied collectively and comprehensively.Social implicationsThe LM practices may help the owners or managers of the SMEs to be competitive and achieve the optimum result.Originality/valueThis is the first known study that adopts the PLS framework to examine how OP measurement affects the relationship between the LM practices and BP in the case of the SMEs in Indonesia.

A Partial Least Squares Aided Intelligent Model Predictive Control Approach

A data-driven model predictive control (MPC) that combines modified partial least squares (PLSs) and MPC is proposed in this paper. A theoretical comparison among traditional MPC, MPC in PLS framework and in modified PLS framework is presented, which demonstrates that the proposed MPC approach has high prediction precision and the ability in coping with dynamics in the process compared to MPC in traditional PLS framework. Furthermore, the proposed MPC requires no prior knowledge, and the simplicity in computation makes it possible to update the prediction model online. The model validity and intelligence of the control strategy are guaranteed by the model updating strategy to a certain degree. Steady-state performance and dynamic response of the proposed MPC is testified through a tracking control simulation of the benchmark of a continuous stirred tank heater system, which illustrates that the advantages of the proposed MPC.

PLS-based multi-loop robust H2 control for improvement of operating efficiency of waste heat energy conversion systems with organic Rankine cycle

Organic Rankine cycle (ORC) is a system suitable for generating electric power by recovering energy from a low-temperature heat source. The system is multivariate, coupled and ill-conditioned. In this paper, a multi-loop robust H2 control strategy based on the partial least squares (PLS), called PLS-H2, is proposed to improve the operating efficiency of a 100 kW waste heat recovery process with ORC. Using the PLS framework, the coupled ORC system can be decomposed into multiple single-loop systems in the PLS subspace to facilitate the controller design. The PLS framework compresses the original input and output variables into latent subspaces of the lower dimension, in which the collinearity of the original variables and the noise is eliminated. Taking into account model mismatch and disturbances from the heat source, the robust H2 controller is designed in each PLS subspace independently. The design is based on minimizing the 2-norm of the transfer function matrix from the exogenous inputs to the controlled outputs. The simulation results demonstrate the effectiveness of the proposed PLS-H2 and its superiority over PLS-PID control. The proposed control method can achieve better performance than PLS-PID control as the latter has more input efforts by1,800 times.

Multiloop robust H∞ control design based on the dynamic PLS approach to chemical processes

In this paper, a multiloop robust control strategy is proposed based on H∞ control and a partial least squares (PLS) model (H∞_PLS) for multivariable chemical processes. It is developed especially for multivariable systems in ill-conditioned plants and non-square systems. The advantage of PLS is to extract the strongest relationship between the input and the output variables in the reduced space of the latent variable model rather than in the original space of the highly dimensional variables. Without conventional decouplers, the dynamic PLS framework automatically decomposes the MIMO process into multiple single-loop systems in the PLS subspace so that the controller design can be simplified. Since plant/model mismatch is almost inevitable in practical applications, to enhance the robustness of this control system, the controllers based on the H∞ mixed sensitivity problem are designed in the PLS latent subspace. The feasibility and the effectiveness of the proposed approach are illustrated by the simulation results of a distillation column and a mixing tank process. Comparisons between H∞_PLS control and conventional individual control (either H∞ control or PLS control only) are also made.

An Improved Generalized Predictive Control in a Robust Dynamic Partial Least Square Framework

To tackle the sensitivity to outliers in system identification, a new robust dynamic partial least squares (PLS) model based on an outliers detection method is proposed in this paper. An improved radial basis function network (RBFN) is adopted to construct the predictive model from inputs and outputs dataset, and a hidden Markov model (HMM) is applied to detect the outliers. After outliers are removed away, a more robust dynamic PLS model is obtained. In addition, an improved generalized predictive control (GPC) with the tuning weights under dynamic PLS framework is proposed to deal with the interaction which is caused by the model mismatch. The results of two simulations demonstrate the effectiveness of proposed method.

A multiple model predictive control strategy in the PLS framework

For chemical processes with a wide range of operating conditions, a switched multiple model predictive control (MMPC) strategy in the partial least squares (PLS) framework is proposed. Interactive MIMO systems can be automatically decoupled with inputs and outputs paired in their dynamic PLS models. Based on the identified PLS models, companion controllers are designed to form the MMPC strategy. A novel switching criterion based on output statistics is proposed to assure each model/control pair works in its operating region spanned by the identification data sets. The control results of disturbance rejection and setpoint tracking in a two-phase chemical reactor process are presented to demonstrate the capability and effectiveness of the proposed MMPC strategy.

A model predictive control approach with relevant identification in dynamic PLS framework

In this paper, a generalized predictive control (GPC) scheme under a dynamic partial least squares (PLS) framework is proposed. At the modeling stage, a model predictive control relevant identification (MRI) approach is used to improve the identification of the model. Within PLS framework, the MIMO system can be automatically decomposed into several SISO subsystems in the latent space. For each subsystem, MRI is implemented and GPC is designed independently. With the advantage of MRI and PLS framework, fewer parameters are needed to be estimated in the identification stage, nonsquare and ill-conditioned system can be handled naturally, control parameters tuning is easier and better control performance can be obtained. Furthermore, the computing time of control action which is very crucial for GPC on-line application decreases since each GPC is running in SISO subsystem in parallel. The results of two simulation examples and a laboratory experiment demonstrate the merit of the proposed method.

Nonlinear PLS Integrated with Error-Based LSSVM and Its Application to NOx Modeling

This article presents a novel nonlinear partial least-squares (PLS) method to address a nonlinear problem with input collinearity in an industrial process. The proposed method integrates an inner least-squares support vector machine (LSSVM) function with an outer linear PLS framework. First, the input and output latent variables are extracted to eliminate the collinearity through PLS projection, and then LSSVM is used to construct nonlinear relation between each pair of latent variables. Moreover, a weight-updating procedure is incorporated to enhance the accuracy of prediction. Then, training and predicting algorithms based on modified nonlinear iterative partial least-squares (NIPALS) steps are also described in detail. The performance of the new method is also investigated with a benchmark data set. Finally, this approach is applied to a real industrial process to predict the NO x emissions of a coal-fired boiler. The root-mean-square errors (RMSEs) on the training and testing data decreased to only 12.6632 and 37.6609, respectively. Compared with the original linear PLS and other kinds of nonlinear PLS methods, a reduction of approximately 40.8―47.4% in the prediction errors is attained. The results reveal that the new approach is capable of modeling the nonlinear relation of NO x emissions with other process parameters and improving the prediction performance.

Multi-loop nonlinear internal model controller design under nonlinear dynamic PLS framework using ARX-neural network model

In this paper, a novel multi-loop nonlinear internal model control (IMC) strategy for multiple-input multiple-output (MIMO) systems is presented under the partial least squares (PLS) framework, which automatically decomposes the system into several univariate subsystems in the latent space. To formulate a nonlinear dynamic PLS framework, we propose an ARX-neural network (ARX-NN) cascaded structure, and incorporate it into PLS inner model. A gradient-based optimization approach is then provided to identify the parameter sets of the ARX-NN PLS model so that the plant-model mismatch is minimized. Furthermore, with perfect model, we show that the response of the closed loop system can be reduced to a simple linear IMC filter with the original system delay. The simulation results of a methylcyclohexane (MCH) distillation column from Aspen Dynamic Module, demonstrate the effectiveness of our approach in terms of disturbance rejection and tracking performance.

Diagnosis of process faults in chemical systems using a local partial least squares approach

AbstractThis article discusses the application of partial least squares (PLS) for monitoring complex chemical systems. In relation to existing work, this article proposes the integration of the statistical local approach into the PLS framework to monitor changes in the underlying model rather than analyzing the recorded input/output data directly. As discussed in the literature, monitoring changes in model parameters addresses the problems of nonstationary behavior and presents an analogy to model‐based approaches. The benefits of the proposed technique are that (i) a detailed mechanistic plant model is not required, (ii) nonstationary process behavior does not produce false alarms, (iii) parameter changes can be non‐Gaussian, (iv) Gaussian monitoring statistics can be established to simplify the monitoring task, and (v) fault magnitude and signatures can be estimated. This is demonstrated by a simulation example and the analysis of recorded data from two chemical processes. © 2008 American Institute of Chemical Engineers AIChE J, 2008

Nonlinear Projection to Latent Structures Method and Its Applications

Projection to latent structures or partial least squares (PLS) is one of the most powerful linear regression techniques to deal with noisy and highly correlated data. To address the inherent nonlinearity present in the data from many industrial applications, a number of approaches have been proposed to extend it to the nonlinear case, and most of them rely on the nonlinear approximation capability of neural networks. However, these methods either merely introduce nonlinearities to the inner relationship model within the linear PLS framework or suffer from training a complicated network. In this paper, starting from an equivalent presentation of PLS, both nonlinear latent structures and nonlinear reconstruction are obtained straightforwardly through two consecutive steps. First, a radial basis function (RBF) network is utilized to extract the latent structures through linear algebra methods without the need of nonlinear optimization. This is followed by developing two feed-forward networks to reconstruct t...

Non-Linear Model Based Predictive Control Through Dynamic Non-Linear Partial Least Squares

The extension of model predictive control (MPC) to non-linear systems is proposed through dynamic non-linear Partial Least Squares (PLS) models. PLS has been shown to be an appropriate multivariate regression methodology for modelling noisy, correlated and/or collinear data. It has been applied extensively, within a ‘static’ framework, for the modelling and analysis of industrial process data. The contribution of this paper is the development of a non-linear dynamic PLS framework for applications in MPC. The non-linear dynamic PLS models make use of an error based non-linear partial least squares algorithm where the non-linear inner models are built within an AutoRegressive with eXogeneous inputs (ARX) framework. In particular, quadratic and feedforward neural network inner models are considered. The application of a dynamic PLS model within a MPC framework opens up the potential of using multivariate statistical projection based methods not only for process modelling, inferential estimation and performance monitoring, but also for model predictive control. A benchmark simulation of a pH neutralization system is used to demonstrate the application of a non-linear dynamic PLS framework for model predictive control.

Extended PLS approach for enhanced condition monitoring of industrial processes

AbstractAn extended partial‐least squares (EPLS) algorithm is introduced to correct a deficiency of conventional partial least squares (PLS) when used as a tool to detect abnormal operating conditions in industrial processes. In the absence of feedback control, an abnormal operating condition that affects only process response variables will not be propagated back to the process predictor (or input) variables. Thus monitoring tools developed under the conventional PLS framework and based only on the predictor matrix will fail to detect the abnormal condition. The EPLS algorithm described removes this deficiency by defining new scores that are based on both predictor and response variables. The EPLS approach provides two monitoring charts to detect abnormal process behavior, as well as contribution charts to diagnose this behavior. To demonstrate the utility of the new approach, the extended algorithm and monitoring tools are applied to a realistic simulation of a fluid catalytic cracking unit and to a real industrial process that involves a complex chemical reaction.