Xue Z. Wang

A new approach to near-infrared spectral data analysis using independent component analysis.

This paper presents a new approach to near-infrared spectral (NIR) data analysis that is based on independent component analysis (ICA). The main advantage of the new method is that it is able to separate the spectra of the constituent components from the spectra of their mixtures. The separation is a blind operation, since the constituent components of mixtures can be unknown. The ICA based method is therefore particularly useful in identifying the unknown components in a mixture as well as in estimating their concentrations. The approach is introduced by reference to case studies and compared to other techniques for NIR analysis including principal component regression (PCR), multiple linear regression (MLR), and partial least squares (PLS) as well as Fourier and wavelet transforms.

Application of wavelets and neural networks to diagnostic system development, 1, feature extraction

An integrated framework for process monitoring and diagnosis is presented which combines wavelets for feature extraction from dynamic transient signals and an unsupervised neural network for identification of operational states. Multiscale wavelet analysis is used to determine the singularities of transient signals which represent the features characterising the transients. This simultaneously reduces the dimensionality of the data and removes noise components. A modified version of the adaptive resonance theory is developed, which is designated ARTnet and uses wavelet feature extraction as the substitute of the data pre-processing unit. ARTnet is proved to be more effective in dealing with noise contained in the transient signals while retains being an unsupervised and recursive clustering approach. The work is reported in two parts. The first part is focused on feature extraction using wavelets. The second part describes ARTnet and its application to a case study of a refinery fluid catalytic cracking process.

Dimension reduction of process dynamic trends using independent component analysis

A new approach is presented which removes the dependencies of variables through separating a smaller number of latent variables called independent components which are the constituent elements of the observed (monitored) variables. The method is introduced by reference to a case study of two continuous stirred tank reactors, which demonstrates that the method can effectively reduce the dimension. Comparison of the method with the well established principal component analysis is also made.

Knowledge discovery from process operational data using PCA and fuzzy clustering

An industrial case study is presented which uses principal component analysis and fuzzy c-means clustering to identify operational spaces and develop operational strategies for manufacturing desired products. Analysis of 303 data cases collected from a refinery fluid catalytic cracking process revealed that the data can be projected to four operational zones in the reduced two-dimensional plane. Three zones were found to correspond to three different product grades and the fourth is a zone corresponding to product changeover. Variable contribution analysis was also carried out to identify the most important variables that are responsible for the observed operational spaces and consequently strategies were developed for monitoring and operating the process in order to be able to move the operation from producing one product grade to another, with minimum time delays.

Agent-based information flow for process industries' supply chain modelling

Abstract With the every increasingly competitive and changing markets, the dependencies and relationships between market components are becoming more and more important. In the past, organisations have focused on making effective decisions within a department or section of a company because their functions could be easily de-coupled and this made analysis simpler. However, gnoring the component dependencies for the sake of simplicity can have costly consequences for the company and other organisations outside on the market. Today there are still a number of reasons that hinder the development of fully integrated business processes, as for example the lack of understanding of the complexity of the organisations and the high cost of acquiring and translating organisational and engineering data. The economic advantage of studying and improving these issues is considerable. This contribution describes an on-going effort in developing an integrated framework for supporting supply chain management of process industries. Retailers, warehouses, plants and raw material are modelled as a network of co-operative agents, each performing one or more supply chain functions. Interactions between agents are made through the common agent communication language knowledge query message language (KQML) and data is modelled using standard exchange of product model data (STEP).

A multi-agent system for chemical supply chain simulation and management support

Abstract. Modern chemical production is customer-driven and the desired delivery time for the products is often shorter than their campaign length. In addition, the raw materials supplying time is often long. These features make it desirable to provide tools to support collaborative supply chain decision making, preferably over the Internet, and where there are conflicts, compromise decisions can be quickly reached and the effects of the decisions can be quantitatively simulated. This paper des cribes such a multi-agent system (MAS) that can be used to simulate the dynamic behaviour and support the management of chemical supply chains over the Internet. Geographically distributed retailers, logistics, warehouses, plants and raw material suppliers are modelled as an open and re-configurable network of co-operative agents, each performing one or more supply chain functions. Communication between agents is made through the common agent communication language KQML (knowledge query message language). A t the simulation layer, the MAS allows distributed simulation of the chain behaviour dynamically, so that compromise decisions can be rapidly and quantitatively evaluated. Because in a chemical supply chain the scheduling of the plant often dominates the chain performance, an optimum scheduling system for batch plants is integrated into the MAS. The functions of the system are illustrated by reference to a case study for the supply and manufacture using a multi-purpose batch plant of paints and coatings.

Neural network based fault diagnosis using unmeasurable inputs

Abstract Much of the earlier work presented in the area of on-line fault diagnosis focuses on knowledge based and qualitatively reasoning principles and attempts to present possible root causes and consequences in terms of various measured data. However, there are many unmeasurable operating variables in chemical processes that define the state of the system. Such variables essentially characterise the efficiency and really need to be known in order to diagnose possible malfunction and provide a basis for deciding on appropriate action to be taken by operators. This paper is concerned with developing a soft sensor to assist in on-line fault diagnosis by providing information on the critical variable that is not directly accessible. The features of dynamic trends of the process are extracted using a wavelet transform and a qualitative interpretation, and then are used as inputs in the neural network based fault diagnosis model. The procedure is illustrated by reference to a refinery fluid catalytic cracking reactor.

Multi-dimensional population balance modeling of the growth of rod-like L-glutamic acid crystals using growth rates estimated from in-process imaging

Abstract —Traditionally, population balance (PB) modeling of crystal growthin crystallizers has been based on a single scalar parameter for particle size, typically the volume equivalent diameter. This misses important information about particle shape, especially for crystals of high aspect ratios. In recent years attempts have been made to extend PB to two or more size dimensions by taking into consideration of the crystal shape. A key step in multi-dimensional PB (M-PB) modeling is the estimation of the growth rates of individual faces as a function of temporal operating conditions, e.g. the supersaturation. In this paper, we propose to carry out M-PB modeling based on real-time experimentally derived growth rates for different faces using in-process imaging and image analysis. Results are presented for the seeded cooling crystallisation of rod-like β-form L-glutamic acid in a 0.5-l batch reactor.

Principal component and causal analysis of structural and acute in vitro toxicity data for nanoparticles.

Abstract Structure toxicity relationship analysis was conducted using principal component analysis (PCA) for a panel of nanoparticles that included dry powders of oxides of titanium, zinc, cerium and silicon, dry powders of silvers, suspensions of polystyrene latex beads and dry particles of carbon black, nanotubes and fullerene, as well as diesel exhaust particles. Acute in vitro toxicity was assessed by different measures of cell viability, apoptosis and necrosis, haemolytic effects and the impact on cell morphology, while structural properties were characterised by particle size and size distribution, surface area, morphology, metal content, reactivity, free radical generation and zeta potential. Different acute toxicity measures were processed using PCA that classified the particles and identified four materials with an acute toxicity profile: zinc oxide, polystyrene latex amine, nanotubes and nickel oxide. PCA and contribution plot analysis then focused on identifying the structural properties that could determine the acute cytotoxicity of these four materials. It was found that metal content was an explanatory variable for acute toxicity associated with zinc oxide and nickel oxide, while high aspect ratio appeared the most important feature in nanotubes. Particle charge was considered as a determinant for high toxicity of polystyrene latex amine.

Application of wavelets and neural networks to diagnostic system development, 2, an integrated framework and its application

Abstract A method for feature extraction from process dynamic transient signals using wavelet multiscale analysis was introduced in part 1 of this paper. In part 2 we describe an integrated framework combining wavelet feature extraction and an unsupervised neural network for identification of operational states. Application of the system to a refinery residual fluid catalytic cracking process is also presented.