Improved Process Monitoring Using Nonlinear Principal Component Models

This paper presents two new approaches for use in complete process monitoring. The firstconcerns the identification of nonlinear principal component models. This involves the application of linear
principal component analysis (PCA), prior to the identification of a modified autoassociative neural network (AAN) as the required nonlinear PCA (NLPCA) model. The benefits are that (i) the number of the reduced set of linear principal components (PCs) is smaller than the number of recorded process variables, and (ii) the set of PCs is better conditioned as redundant information is removed. The result is a new set of input data for a modified neural representation, referred to as a T2T network. The T2T NLPCA model is then used for complete process monitoring, involving fault detection, identification and isolation. The second approach introduces a new variable reconstruction algorithm, developed from the T2T NLPCA model. Variable reconstruction can enhance the findings of the contribution charts still widely used in industry by reconstructing the outputs from faulty sensors to produce more accurate fault isolation. These ideas are illustrated using recorded industrial data relating to developing cracks in an industrial glass melter process. A comparison of linear and nonlinear models, together with the combined use of contribution charts and variable reconstruction, is presented.

Process monitoring approach using fast moving window PCA

This paper introduces a fast algorithm for moving window principal component analysis (MWPCA) which will adapt a principal component model. This incorporates the concept of recursive adaptation within a moving window to (i) adapt the mean and variance of the process variables, (ii) adapt the correlation matrix, and (iii) adjust the PCA model by recomputing the decomposition. This paper shows that the new algorithm is computationally faster than conventional moving window techniques, if the window size exceeds 3 times the number of variables, and is not affected by the window size. A further contribution is the introduction of an N-step-ahead horizon into the process monitoring. This implies that the PCA model, identified N-steps earlier, is used to analyze the current observation. For monitoring complex chemical systems, this work shows that the use of the horizon improves the ability to detect slowly developing drifts.

Recent applications of Chemical Imaging to pharmaceutical process monitoring and quality control

Chemical Imaging (CI) is an emerging platform technology that integrates conventional imaging and spectroscopy to attain both spatial and spectral information from an object. Vibrational spectroscopic methods, such as Near Infrared (NIR) and Raman spectroscopy, combined with imaging are particularly useful for analysis of biological/pharmaceutical forms. The rapid, non-destructive and non-invasive features of CI mark its potential suitability as a process analytical tool for the pharmaceutical industry, for both process monitoring and quality control in the many stages of drug production. This paper provides an overview of CI principles, instrumentation and analysis. Recent applications of Raman and NIR-CI to pharmaceutical quality and process control are presented; challenges facing Cl implementation and likely future developments in the technology are also discussed. (C) 2007 Elsevier B.V. All rights reserved.

Improved Nonlinear PCA for Process Monitoring Using Support Vector Data Description

Nonlinear principal component analysis (PCA) based on neural networks has drawn significant attention as a monitoring tool for complex nonlinear processes, but there remains a difficulty with determining the optimal network topology. This paper exploits the advantages of the Fast Recursive Algorithm, where the number of nodes, the location of centres, and the weights between the hidden layer and the output layer can be identified simultaneously for the radial basis function (RBF) networks. The topology problem for the nonlinear PCA based on neural networks can thus be solved. Another problem with nonlinear PCA is that the derived nonlinear scores may not be statistically independent or follow a simple parametric distribution. This hinders its applications in process monitoring since the simplicity of applying predetermined probability distribution functions is lost. This paper proposes the use of a support vector data description and shows that transforming the nonlinear principal components into a feature space allows a simple statistical inference. Results from both simulated and industrial data confirm the efficacy of the proposed method for solving nonlinear principal component problems, compared with linear PCA and kernel PCA.

Low-cost Process monitoring for polymer extrusion

Polymer extrusion is regarded as an energy-intensive production process, and the real-time monitoring of both energy consumption and melt quality has become necessary to meet new carbon regulations and survive in the highly competitive plastics market. The use of a power meter is a simple and easy way to monitor energy, but the cost can sometimes be high. On the other hand, viscosity is regarded as one of the key indicators of melt quality in the polymer extrusion process. Unfortunately, viscosity cannot be measured directly using current sensory technology. The employment of on-line, in-line or off-line rheometers is sometimes useful, but these instruments either involve signal delay or cause flow restrictions to the extrusion process, which is obviously not suitable for real-time monitoring and control in practice. In this paper, simple and accurate real-time energy monitoring methods are developed. This is achieved by looking inside the controller, and using control variables to calculate the power consumption. For viscosity monitoring, a ‘soft-sensor’ approach based on an RBF neural network model is developed. The model is obtained through a two-stage selection and differential evolution, enabling compact and accurate solutions for viscosity monitoring. The proposed monitoring methods were tested and validated on a Killion KTS-100 extruder, and the experimental results show high accuracy compared with traditional monitoring approaches.

Characterisation of Bioresorbable Composite Performance for Validation of New On-Line Process Monitoring Technology

There is an increasing interest in the biomedical field to create implantable medical devices to provide a temporary mechanical function for use inside the human body. In many of these applications bioresorbable polymer composites using PLLA with β-TCP , are increasingly being used due to their biocompatability, biodegradability and mechanical strength.1,3 These medical devices can be manufactured using conventional plastics processing methods such as injection moulding and extrusion, however there is great need to understand and control the process due to a lack of knowledge on the influence of processing on material properties. With the addition of biocompatible additives there is also a requirement to be able to predict the quality and level of dispersion within the polymer matrix. On-line UV-Vis spectroscopy has been shown to monitor the quality of fillers in polymers. This can eliminate time consuming and costly post-process evaluation of additive dispersion. The aim of this work was to identify process and performance relationships of PLLA/β-TCP composites with respect to melt-extrusion conditions. This is part of a wider study into on-line process monitoring of bioresorbable polymers as used in the medical industry.
These results show that final properties of the PLLA/ β-TCP composite are highly influenced by the particle size and loading. UV-Vis spectroscopy can be used on-line to monitor the final product and this can be utilised as a valuable tool for quality control in an application where consistent performance is of paramount importance.

The inferential monitoring of the screw disturbance torque to predict process fluctuations in polymer extrusion

Polymer extrusion is one of the major methods of processing polymer materials and advanced process monitoring is important to ensure good product quality. However, commonly used process monitoring devices, e.g. temperature and pressure sensors, are limited in providing information on process dynamics inside an extruder barrel. Screw load torque dynamics, which may occur due to changes in solids conveying, melting, mixing, melt conveying, etc., are believed to be a useful indicator of process fluctuations inside the extruder barrel. However, practical measurement of the screw load torque is difficult to achieve. In this work, inferential monitoring of the screw load torque signal in an extruder was shown to be possible by monitoring the motor current (armature and/or field) and simulation studies were used to check the accuracy of the proposed method. The ability of this signal to aid identification and diagnosis of process issues was explored through an experimental investigation. Power spectral density and wavelet frequency analysis were implemented together with a covariance analysis. It was shown that the torque signal is dominated by the solid friction in the extruder and hence it did not correlate well with melting fluctuations. However, it is useful for online identification of solids conveying issues.