Masters Degrees (Faculty of Engineering (former Departments))

Permanent URI for this collection

https://scholar.sun.ac.za/handle/10019.1/3925

Browse

Recent Submissions

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Modelling and control of an autogenous mill using a state space methodology and neural networks
    (Stellenbosch : Stellenbosch University, 2002-12) Groenewald, Jacobus Willem de Villiers; Aldrich, C.; Lorenzen, L.; Eksteen, J. J.; Stellenbosch University. Faculty of Engineering. Dept. of Process Engineering.
    ENGLISH ABSTRACT: Metallurgical processes are often high dimensional and non-linear making them difficult to understand, model and control. Whereas the human eye has extensively been used in discerning temporal patterns in historical process data from these processes, the systematic study of such data has only recently come to the forefront. This resulted predominantly from the inadequacy of previously used linear techniques and the computational power required when analysing the non-linear dynamics underlying these systems. Furthermore, owing to the recent progress made with regard to the identification of non-linear systems and the increased availability of computational power, the application of non-linear modelling techniques for the development of neural network models to be used in advanced control systems has become a potential alternative to operator experience. The objective of this study was the development ofa non-linear, dynamic model of an autogenous mill for use in an advanced control system. This was accomplished through system identification, modelling and prediction, and application to control. For system identification, the attractor was reconstructed based on Taken's theorem making use of both the Method Of Delays and singular spectrum analysis. Modelling consisted of the development of multi-layer perceptron neural network, radial basis function neural network, and support vector machine models for the prediction of the power drawn by an autogenous mill. The best model was subsequently selected and validated through its application to control. This was accomplished by means of developing a neurocontroller, which was tested under simulation. Initial inspection of the process data to be modelled indicated that it contained a considerable amount noise. However, using the method of surrogate data, it was found that the time series representing the power drawn by the autogenous mill clearly exhibited deterministic character, making it suitable for predictive modelling. It was subsequently found that, when using the data for attractor reconstruction, a connection existed between the embedding strategy used, the quality of the reconstructed attractor, and the quality of the resulting model. Owing to the high degree of noise in the data it was found that the singular spectrum analysis embeddings resulted in better quality reconstructed attractors that covered a larger part of the state space when compared to the method of delays embeddings; the data embedded using singular spectrum analysis also resulting in the development of better quality models. From a modelling perspective it was found that the multi-layer perceptron neural network models generally performed the best; a multi-layer perceptron neural network model having an appropriately embedded multi-dimensional input space outperforming all the other developed models with regard to free-run prediction success. However, none of the non-linear models performed significantly better than the ARX model with regard to one-step prediction results (based on the R2 statistic); the one-step predictions having a prediction interval of 30 seconds. In general the best model was a multi-layer perceptron neural network model having an input space consisting of the FAG mill power (XI), the FAG mill load (X2), the FAG mill coarse ore feed rate (X3), the FAG mill fine ore feed rate (X4), the FAG mill inlet water flow rate (X7) and the FAG mill discharge flow rates (X9, XIO). Since the accuracy of any neural network model is highly dependent on its training data, a process model diagnostic system was developed to accompany the process model. Linear principal component analysis was used for this purposes and the resulting diagnostic system was successfully used for data validation. One of the models developed during this research was also successfully used for the development of a neurocontroller, proving its possible use in an advanced control system.
  • Thumbnail Image
    Item
    Evaluation of the constant current angle controlled reluctance synchronous machine drive
    (Stellenbosch : Stellenbosch University, 2002-03) Fick, Pieter D.; Kamper, M. J.; Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering.
    ENGLISH ABSTRACT: This thesis describes. the design and evaluation of a constant current angle controller for a variable speed reluctance synchronous machine (RSM) drive, as an energy efficient high performance drive. An accurate model of the RSM, with the use of finite element analysis, is derived and implemented in simulation software. The current- and speed controllers are designed and evaluated using a complete simulation model of the whole drive. The controller is implemented on a TMS320F240 DSPbased digital controller, which was developed. The dynamic performance of the constant-current-angle control is compared with that of the conventional constant-daxis- current control method. The results obtained from the RSM drive confirm the simulation results. In the comparison of the two control methods it is shown that the constant-current-angle controlled RSM drive is an energy-efficient drive with good dynamic performance.
  • Thumbnail Image
    Item
    Verification of patient position for proton therapy using portal X-Rays and digitally reconstructed radiographs
    (Stellenbosch : University of Stellenbosch, 2006-12) Van der Bijl, Leendert; Muller, N.; University of Stellenbosch. Faculty of Science. Dept. of Mathematical Sciences. Applied Mathematics.
    This thesis investigates the various components required for the development of a patient position verification system to replace the existing system used by the proton facilities of iThemba LABS1. The existing system is based on the visual comparison of a portal radiograph (PR) of the patient in the current treatment position and a digitally reconstructed radiograph (DRR) of the patient in the correct treatment position. This system is not only of limited accuracy, but labour intensive and time-consuming. Inaccuracies in patient position are detrimental to the effectiveness of proton therapy, and elongated treatment times add to patient trauma. A new system is needed that is accurate, fast, robust and automatic. Automatic verification is achieved by using image registration techniques to compare the PR and DRRs. The registration process finds a rigid body transformation which estimates the difference between the current position and the correct position by minimizing the measure which compares the two images. The image registration process therefore consists of four main components: the DRR, the PR, the measure for comparing the two images and the minimization method. The ray-tracing algorithm by Jacobs was implemented to generate the DRRs, with the option to use X-ray attenuation calibration curves and beam hardening correction curves to generate DRRs that approximate the PRs acquired with iThemba LABS’s digital portal radiographic system (DPRS) better. Investigations were performed mostly on simulated PRs generated from DRRs, but also on real PRs acquired with iThemba LABS’s DPRS. The use of the Correlation Coefficient (CC) and Mutual Information (MI) similarity measures to compare the two images was investigated. Similarity curves were constructed using simulated PRs to investigate how the various components of the registration process influence the performance. These included the use of the appropriate XACC and BHCC, the sizes of the DRRs and the PRs, the slice thickness of the CT data, the amount of noise contained by the PR and the focal spot size of the DPRS’s X-ray tube. It was found that the Mutual Information similarity measure used to compare 10242 pixel PRs with 2562 pixel DRRs interpolated to 10242 pixels performed the best. It was also found that the CT data with the smallest slice thickness available should be used. If only CT data with thick slices is available, the CT data should be interpolated to have thinner slices. Five minimization algorithms were implemented and investigated. It was found that the unit vector direction set minimization method can be used to register the simulated PRs robustly and very accurately in a respectable amount of time. Investigations with limited real PRs showed that the behaviour of the registration process is not significantly different than for simulated PRs.