Browsing by Author "Aldrich, Christiaan"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Application of data analytics and knowledge-based systems in mineral processing
    (Stellenbosch : Stellenbosch University, 2015-12) Aldrich, Christiaan; Burger, A. J.; Stellenbosch University. Faculty of Engineering. Dept. of Process Engineering.
    ENGLISH ABSTRACT: This dissertation covers research carried out over the past 20 years in the area of knowledge engineering in mineral processing, specifically with regard to process data as a form of knowledge. This focus on data-driven plant automation includes the acquisition, interpretation and application of data in the development of decision support systems in mineral processing, as well as the development of data analytical methodologies required to accomplish this. The following subthemes have been covered: o Inferential sensors - predominantly the development of computer vision systems for froth flotation and the analysis of particulate systems, but also acoustic sensors and the interpretation of electrochemical noise. My research into inferential sensors has centred on the development of methodologies and algorithms to interpret image data and not the development of hardware, such as camera systems or other types of sensing devices. A major part of this pioneering research has focused on the interpretation of froth flotation images. Instead of attempting to identify individual objects (bubbles) in these images, we have treated the froth images as statistical patterns. These patterns could be interpreted by suitable feature extraction algorithms and models that could relate these features to meaningful process indicators. The novelty and impact of my research in this area can be inferred not only from the corpus of highly cited papers that associated with the technology, but also from the commercialization of the technology. o Exploratory data analysis - Focusing on unsupervised learning, such as applied in data visualization, cluster analysis and feature extraction. In exploratory data analysis, the main issue is attempting to make sense of many measurements of large sets of variables. Standard multivariate statistical methods have their limitations when dealing with complex data, and a significant part of my research has concentrated on the extension of linear methods to their nonlinear variants by use of neural networks or other machine learning approaches. Work in this area has formed the basis of a sizeable number of industrial workshops and has significantly influenced the development of commercial process systems software. o Data-based process modelling - Machine learning approaches to predictive and diagnostic modelling. The construction of process models plays a key role in process systems engineering. This is the case in advanced control systems, where the ability to predict future process states is critical. Models also play an important role in the interpretation of process data and hence the acquisition of insight into process behaviour and mechanisms. Such models can be developed from first principles, but this is costly and with the abundance of process data, often not necessary. The primary impact of this research has been in the development and application of methods to predict process states or key performance indicators for mineral processing systems. o Process monitoring and fault diagnosis - Multivariate statistical process control from a machine learning perspective. Process monitoring and fault diagnosis has evolved into a key element of process control over the last couple of decades, and is currently experiencing strong growth, with commercial application still lagging significantly behind the advances in academia. My research in this area has centred on the application of neural networks, kernel-based systems, random forests and other machine learning methods to extend current approaches. It has led to the foundation of the Anglo American Platinum Centre for Process Monitoring at Stellenbosch University and the development of algorithms that were adopted by industry on a proprietary basis. o Intelligent decision support and advanced control - Fuzzy decision support systems and neurocontrol based on the use of reinforcement learning. Apart from data that are generated by instruments, tacit knowledge in the form of plant operator experience and theoretical knowledge is also a valuable resource that can be used in the automation of plant operations. This is the domain of knowledge-based or expert systems and research was undertaken in the development and application of these systems in mineral processing. The novelty of this research has mainly been in the proof-of-concept studies published in academic journals and conference proceedings. It goes without saying that in my research, I have been assisted by many colleagues, industrial collaborators, students and assistants. The contributions of these co-workers were often critical to the investigations indicated in this thesis and are indicated as such, hopefully without omission, where appropriate.
  • Thumbnail Image
    Item
    The simulation and optimization of steady state process circuits by means of artificial neural networks
    (Stellenbosch : Stellenbosch University, 1993) Aldrich, Christiaan; van Deventer, J. S. J.; Stellenbosch University. Faculty of Engineering. Department of Electrical and Electronic Engineering.
    ENGLISH ABASTRACT: Since the advent of modern process industries engineers engaged in the modelling and simulation of chemical and metallurgical processes have had to contend with two important dilemmas. The first concerns the ill-defined nature of the processes they have to describe, while the second relates to the limitations of prevailing computational resources. Current process simulation procedures are based on explicit process models in one form or another. Many chemical and metallurgical processes are not amenable to this kind of modelling however, and can not be incorporated effectively into current commercial process simulators. As a result many process operations do not benefit from the use of predictive models and simulation routines and plants are often poorly designed and run, ultimately leading to considerable losses in revenue. In addition to this dilemma, process simulation is in a very real way constrained by available computing resources. The construction of adequate process models is essentially meaningless if these models can not be solved efficiently - a situation occurring all too often. In the light of these problems, it is thus not surprising that connectionist systems or neural network methods are singularly attractive to process engineers, since they provide a powerful means of addressing both these dilemmas. These nets can form implicit process models through learning by example, and also serve as a vehicle for parallel supercomputing devices. In this dissertation the use of artificial neural networks for the steady state modelling and optimization of chemical and metallurgical process circuits is consequently investigated. The first chapter is devoted to a brief overview of the simulation of chemical and metallurgical plants by conventional methods, as well as the evolution and impact of computer technology and artificial intelligence on the process industries. Knowledge of the variance covariance matrices of process data is of paramount importance to data reconciliation and gross error detection problems, and although various methods can be employed to estimate these often unknown variances, it is shown in the second chapter that the use of feedforward neural nets can be more efficient than conventional strategies. In the following chapter the important problem of gross error detection in process data is addressed. Existing procedures are statistical and work well for systems subject to linear constraints. Non-linear constraints are not handled well by these methods and it is shown that back propagation neural nets can be trained to detect errors in process systems, regardless of the nature of the constraints. In the fourth chapter the exploitation of the massively parallel information processing structures of feedback neural nets in the optimization of process data reconciliation problems is investigated. Although effective and sophisticated algorithms are available for these procedures, there is an ever present demand for computational devices or routines that can accommodate progressively larger or more complex problems. Simulations indicate that neural nets can be efficient instruments for the implementation of parallel strategies for the optimization of such problems. In the penultimate chapter a gold reduction plant and a leach plant are modelled with neural nets and the models shown to be considerably better than the linear regression models used in practice. The same technique is also demonstrated with the modelling of an apatite flotation plant. Neural nets can also be used in conjunction with other methods and in the same chapter the steady state simulation and optimization of a gravity separation circuit with the use of two linear programming models and a neural net are described.