Thesis (MScEng (Electrical and Electronic Engineering))--Stellenbosch University, 2008. / A navigation performance sensitivity study with respect to the quality variation of sensors and sensor combinations is presented in this thesis. Navigation simulation software capable of using real-time and simulated sensor data is developed in this project. The simulation software is used to simulate different sensor combinations and therefore evaluate the best configuration for each AUV mission. A sensor module is also developed to capture real-time sensor data. The sensor module includes a low-cost 6-degreeof- freedom inertial measurement module (rate gyroscopes and accelerometers), a three-axes magnetometer and other sensor interfaces. The real-time sensor data are used to test and calibrate the navigation simulation software. Different sensor combinations are evaluated by using the navigation simulation software with simulated sensor data as input. The quality of each sensor is varied by changing its noise characteristics. The performance study, together with the developed simulation tools, simplifies the process of selecting a sensor combination to fulfill a specific navigation accuracy requirement.

Thesis (MScEng (Electrical and Electronic Engineering))--Stellenbosch University, 2008. / The nite element method (FEM) is a powerful tool for the computational analysis of a wide range of electromagnetic problems. As the complexity of the problems is increased so are the demands in terms of the computational resources required to obtain a su ciently accurate solution. In an attempt to obtain a desired accuracy at a lower computational cost adaptive and higher order methods are often employed. These methods generally entail re ning the solution only in the areas where greater complexity is required, thus decreasing the total computational demand. The adaptive nite element method is implemented and used to analyse the transverse electric cuto eigenmodes of 2D waveguiding structures. The higher order hierarchical vector basis functions that form part of this implementation are automatically generated to very high orders, with the results presented in excellent agreement to analytical ones where applicable. Accuracy to the order of numeric precision is attained. Using these adaptive methods, it is also possible to achieve improved cost e ciency of the error metrics considered with respect to storage requirements and computational cost.

Thesis (MScEng (Electrical and Electronic Engineering))--University of Stellenbosch, 2009. / This thesis investigates aggressive all-attitude flight control systems. These are flight controllers capable of controlling an aircraft at any attitude and will enable the autonomous execution of manoeuvres such as high bank angle turns, steep climbs and aerobatic flight manoeuvres. This class of autopilot could be applied to carry out evasive combat manoeuvres or to create more efficient and realistic target drones. A model for the aircraft’s dynamics is developed in such a way that its high bandwidth specific force and moment model is split from its lower bandwidth kinematic model. This split is done at the aircraft’s specific acceleration and roll rate, which enables the design of simple, decoupled, linear attitude independent inner loop controllers to regulate these states. Two outer loop kinematic controllers are then designed to interface with these inner loop controllers to guide the aircraft through predefined reference trajectories. The first method involves the design of a linear quadratic regulator (LQR) based on the successively linearised kinematics, to optimally control the system. The second method involves specific acceleration matching (SAM) and results in a linear guidance controller that makes use of position based trajectories. These position based trajectories allow the aircraft’s velocity magnitude to be regulated independently of the trajectory tracking. To this end, two velocity regulation algorithms were developed. These involved methods of optimal control, implemented using dynamic programming, and energy analysis to regulate the aircraft’s velocity in a predictive manner and thereby providing significantly improved velocity regulation during aggressive aerobatic type manoeuvres. Hardware in the loop simulations and practical flight test data verify the theoretical results of all controllers presented

Thesis (MScEng (Electrical and Electronic Engineering))--University of Stellenbosch, 2009. / In this thesis we studied large-vocabulary continuous speech recognition. We considered the components necessary to realise a large-vocabulary speech recogniser and how systems such as Sphinx and HTK solved the problems facing such a system. Hidden Markov Models (HMMs) have been a common approach to acoustic modelling in speech recognition in the past. HMMs are well suited to modelling speech, since they are able to model both its stationary nature and temporal e ects. We studied HMMs and the algorithms associated with them. Since incorporating all knowledge sources as e ciently as possible is of the utmost importance, the N-Best paradigm was explored along with some more advanced HMM algorithms. The way in which sounds and words are constructed has been studied extensively in the past. Context dependency on the acoustic level and on the linguistic level can be exploited to improve the performance of a speech recogniser. We considered some of the techniques used in the past to solve the associated problems. We implemented and combined some chosen algorithms to form our system and reported the recognition results. Our nal system performed reasonably well and will form an ideal framework for future studies on large-vocabulary speech recognition at the University of Stellenbosch. Many avenues of research for future versions of the system were considered.

Thesis (MScEng (Electrical and Electronic Engineering))--University of Stellenbosch, 2009. / ENGLISH ABSTRACT: Phoneme recognition systems typically suffer from low classification accuracy. Recognition for South African English is especially difficult, due to the variety of vastly different accent groups. This thesis investigates whether a fusion of classifiers, each trained on a specific accent group, can outperform a single general classifier trained on all. We implemented basic voting and score fusion techniques from which a small increase in classifier accuracy could be seen. To ensure that similarly-valued output scores from different classifiers imply the same opinion, these classifiers need to be calibrated before fusion. The main focus point of this thesis is calibration with the Pool Adjacent Violators algorithm. We achieved impressive gains in accuracy with this method and an in-depth investigation was made into the role of the prior and the connection with the proportion of target to non-target scores. Calibration and fusion using the information metric Cllr was showed to perform impressively with synthetic data, but minor increases in accuracy was found for our phoneme recognition system. The best results for this technique was achieved by calibrating each classifier individually, fusing these calibrated classifiers and then finally calibrating the fused system. Boosting and Bagging classifiers were also briefly investigated as possible phoneme recognisers. Our attempt did not achieve the target accuracy of the classifier trained on all the accent groups. The inherent difficulties typical of phoneme recognition were highlighted. Low per-class accuracies, a large number of classes and an unbalanced speech corpus all had a negative influence on the effectivity of the tested calibration and fusion techniques. / AFRIKAANSE OPSOMMING: Foneemherkenningstelsels het tipies lae klassifikasie akkuraatheid. As gevolg van die verskeidenheid verskillende aksent groepe is herkenning vir Suid-Afrikaanse Engels veral moeilik. Hierdie tesis ondersoek of ’n fusie van klassifiseerders, elk afgerig op ’n spesifieke aksent groep, beter kan doen as ’n enkele klassifiseerder wat op alle groepe afgerig is. Ons het basiese stem- en tellingfusie tegnieke ge¨ımplementeer, wat tot ’n klein verbetering in klassifiseerder akkuraatheid gelei het. Om te verseker dat soortgelyke uittreetellings van verskillende klassifiseerders dieselfde opinie impliseer, moet hierdie klassifiseerders gekalibreer word voor fusie. Die hoof fokuspunt van hierdie tesis is kalibrasie met die Pool Adja- cent Violators algoritme. Indrukwekkende toenames in akkuraatheid is behaal met hierdie metode en ’n in-diepte ondersoek is ingestel oor die rol van die aanneemlikheidswaarskynlikhede en die verwantskap met die verhouding van teiken tot nie-teiken tellings. Kalibrasie en fusie met behulp van die informasie maatstaf Cllr lewer indrukwekkende resultate met sintetiese data, maar slegs klein verbeterings in akkuraatheid is gevind vir ons foneemherkenningstelsel. Die beste resultate vir hierdie tegniek is verkry deur elke klassifiseerder afsonderlik te kalibreer, hierdie gekalibreerde klassifiseerders dan te kombineer en dan die finale gekombineerde stelsel weer te kalibreer. Boosting en Bagging klassifiseerders is ook kortliks ondersoek as moontlike foneem herkenners. Ons poging het nie die akkuraatheid van ons basislyn klassifiseerder (wat op alle data afgerig is) bereik nie. Die inherente probleme wat tipies is tot foneemherkenning is uitgewys. Lae per-klas akkuraatheid, ’n groot hoeveelheid klasse en ’n ongebalanseerde spraak korpus het almal ’n negatiewe invloed op die effektiwiteit van die getoetsde kalibrasie en fusie tegnieke gehad.

Thesis (MScEng (Electrical and Electronic Engineering))--Stellenbosch University, 2008. / ENGLISH ABSTRACT: This thesis discusses the development of a 5 kVA single phase AC voltage regulator, specifically designed to assist in the reduction of electrification costs in sparsely populated rural areas. The voltage regulator is based on a solid state auto-transformer tap changer, designed to be robust and maintenance free. Electrification cost savings can be realized if the length of the LV network can be extended to reach more households. To accomplish this, a voltage regulator can be installed onto the extended LV feeder at the point where the LV voltage will drop below the minimum valid voltage, thereby boosting the voltage downstream and enabling more customers to be connected. A variety of voltage regulator topologies were investigated to obtain the best topology for the application. The voltage regulator design is discussed in detail with careful attention given to the power loss incurred, surge voltage protection requirements, protection coordination with the existing LV network and the thermal design requirements. An electronic controller based on a digital signal processor together with an appropriate power supply is designed and built. The software to control the voltage regulator is developed, integrated with the hardware and debugged. The complete voltage regulator is evaluated through extensive laboratory testing and field trials are performed to verify the performance of the device. / AFRIKAANSE OPSOMMING: Hierdie tesis bespreek die ontwikkeling van ‘n 5 kVA, enkelfase, wisselstroom spannings reguleerder, spesifiek ontwerp om koste besparings te bewerkstellig in die elektrifisiëring van yl bevolkte landelike gemeenskappe. Die spannings reguleerder se ontwerp is gebaseer op ‘n outo-transformator tap wisselaar met vaste toestand skakelaars, om sodoende robuust en instandhoudings vry te funksioneer. Elektrifisiërings koste besparings is moontlik indien die laag spannings distribusie kabel verleng kan word, om sodoende elektrisiteit aan meer huishoudings te voorsien. Om dit mootlik te maak kan ‘n spannings reguleerder geinstalleer word op die punt waar die kabel spanning onder die minimum toegelate spanning daal. Sodoende word die spanning weer verhoog aan al die huishoudings wat aan die verlengde gedeelte van die kabel verbind is. ‘n Verskeidenheid uiteenlopende spannings reguleerder topologieë is ondersoek om die beste topologie vir die toepassing te identifiseer. Die ontwerp van die spannings reguleerder is baie deeglik bespreek en spesifieke aandag is gegee aan die verliese, spits opwelling spannings beveiliging, sinkronisasie met die huidige laag spannings netwerk se beveiligings meganismes en die termiese ontwerp van die stelsel. ‘n Elektroniese beheerder, gebaseer op ‘n digitale sein verwerker, tesame met ‘n toepaslike kragbron is ontwerp en gebou. Die nodige sagteware om die spannings reguleerder te beheer is ontwikkel, geintegreer met die hardeware en ontfout. Die volledige spannings reguleerder is ontleed deur intensiewe toetse in die laboratorium en toets installasies op laag spannings netwerke, om sodoende die nakoming van die werks verrigting vereistes van die toestel te bevestig.

Thesis (PhD (Electrical and Electronic Engineering))--University of Stellenbosch, 2007. / Most speech recognition and language identification engines are based on hidden Markov models (HMMs). Higher-order HMMs are known to be more powerful than first-order HMMs, but have not been widely used because of their complexity and computational demands. The main objective of this dissertation was to develop a more time-efficient method of decoding high-order HMMs than the standard Viterbi decoding algorithm currently in use. We proposed, implemented and evaluated two decoders based on the Forward-Backward Search (FBS) paradigm, which incorporate information obtained from low-order HMMs. The first decoder is based on time-synchronous Viterbi-beam decoding where we wish to base our state pruning on the complete observation sequence. The second decoder is based on time-asynchronous A* search. The choice of heuristic is critical to the A* search algorithms and a novel, task-independent heuristic function is presented. The experimental results show that both these proposed decoders result in more time-efficient decoding of the fully-connected, high-order HMMs that were investigated. Three significant facts have been uncovered. The first is that conventional forward Viterbi-beam decoding of high-order HMMs is not as computationally expensive as is commonly thought. The second (and somewhat surprising) fact is that backward decoding of conventional, high-order left-context HMMs is significantly more expensive than the conventional forward decoding. By developing the right-context HMM, we showed that the backward decoding of a mathematically equivalent right-context HMM is as expensive as the forward decoding of the left-context HMM. The third fact is that the use of information obtained from low-order HMMs significantly reduces the computational expense of decoding high-order HMMs. The comparison of the two new decoders indicate that the FBS-Viterbi-beam decoder is more time-efficient than the A* decoder. The FBS-Viterbi-beam decoder is not only simpler to implement, it also requires less memory than the A* decoder. We suspect that the broader research community regards the Viterbi-beam algorithm as the most efficient method of decoding HMMs. We hope that the research presented in this dissertation will result in renewed investigation into decoding algorithms that are applicable to high-order HMMs.

Thesis (PhD (Electrical and Electronic Engioneering))--Stellenbosch University, 2008. / Unmanned Aerial Vehicle (UAV) technology is classified as being disruptive since it has the potential to radically change the utilization of airspace. Most unmanned vehicles are aimed at military applications, yet civilian applications of unmanned aerial vehicle technology could benefit South Africa considerably. At present, the lack of UAV regulations and standards precludes UAVs from being certified to operate on a file and fly basis in un-segregated civilian airspace. The inability for UAVs to be certified because of a lack of standards creates a “chicken and egg” – “stale mate” situation. If principles such as “equivalence”, initially proposed by Eurocontrol are adopted in South Africa, it then follows that equivalent standards used by manned aircraft could be used by UAVs. UAVs must therefore be tested and evaluated in order to prove compliance with equivalent existing manned aircraft regulations in the foreseeable future until UAV regulations and standards become available. It has been suggested that specific UAV missions such as maritime patrol, border control, search & rescue, and cargo transport could fulfil current requirements. Design considerations and possible concepts of UAV operations, maintenance and training that will enable UAVs to satisfy the immediate South African strategic requirements whilst complying with existing airspace and airworthiness regulations have been proposed in this document while further UAV specific standards and regulations are being developed. UAV testing is an essential part of proving the enabling technology, and part of the process of gaining acceptance into wider airspace. Fortunately, flight test methods and procedures applicable to manned aircraft are directly applicable to UAVs, while systems unique to UAVs can be adapted from existing procedures applied to missiles and military UAVs. Once UAVs are developed and tested, it will be necessary to start full scale operations. Some considerations will be necessary during mission planning. Air traffic management regulations however will prohibit some UAVs from operating in all airspace until enabling technology is developed and tested, while some existing UAVs will never be permitted to “file and fly”. This study also analyses existing airspace and UAV platforms in order to identify the airspace and platforms that will have the most chance of being successfully permitted to “file and fly” in civil airspace. For South Africa to advance as a UAV operating and manufacturing nation, it is therefore essential to compile a roadmap that will guide the process of developing, certifying and operating UAVs. The roadmap must include an interim process, as well as stating the end objective, which is “file and fly”. This South African UAV Roadmap proposal is based on international research that uses documentation and lessons learned from elsewhere to guide the process for creating UAV regulations and standards, while allowing existing UAV operations to expand into the existing airspace in order for further UAV research to take place. This roadmap proposal is the conclusion of a 3 year study, and references to the applicable literature are made throughout the document.

Thesis (PhD)--Stellenbosch University, 2000. / ENGLISH ABSTRACT: The engineering of dependable real-time systems for mission critical applications is a resource intensive and error prone process. Achieving dependability requires a general consensus on the correctness of a system with regard to its intended function. For a consensus to be achieved, the properties of the system must be well understood which, in turn, requires consensus on a rigorously defined computational architecture. There is currently no single agreed upon computational architecture at the application level which can serve as a common denominator for the design and implementation of real-time systems. It is the thesis of this dissertation that a rigorous computational architecture, applicable from design to implementation, enables engineers to better understand software for real-time systems. To substantiate this claim, the real-time data flow architecture RDF with its notation allowing the description of complete systems from design to implementation will be explored. Four distinct research areas for improving the engineering process of real-time systems are dealt with in the dissertation: 1) the development of an architecture for real-time systems being suitable for design and implementation in software and hardware, 2) the consolidation of a number of graphical languages into a graphical notation for functional specification, design and construction of real-time systems, 3) the development of a simple processor architecture for the execution of real-time applications, and 4) and the evaluation of the architecture in the framework of a microsatellite case study. In particular, the following original contributions are made: 1) the firing semantics of data flow systems are expanded to include disjunctive firing semantics in a novel way in addition to the classical conjunctive firing semantics, 2) the inherent real-time data flow property, Le. that a receiving task must be ready to receive the next incoming message when it is sent, is extended to the synchronous data flow model, 3) a notation for describing all properties of real-time systems is defined 'with the real-time data flow language RDF as base language, 4) two hardware processor architectures are introduced that offer one-to-one correspondence between design and implementation and, thus, reduce the semantic gap between design language and program execution, and 5) the class of systems that can be modelled with data flow architectures is shown to include control systems and data flow systems. The language set and processor architecture were applied to certain aspects of the SUNSAT microsatellite project. / AFRIKAANSE OPSOMMING:Die skep van betroubare intydsestelsels vir missie kritiese toepassings is 'n proses wat baie hulpbronne verg en waarin maklik 'n fout gemaak kan word. Om 'n betroubare stelsel te skep vereis 'n konsensus oor die korrektheid van 'n stelsel, wat bereik word wanneer die eienskappe van die stelsel goed verstaan word. Dit vereis op sy beurt weer 'n konsensus oor 'n goed gedefinieerde berekenings argitektuur. Daar is tans geen enkel ooreengekome berekenings argitektuur op die toepassingsvlak wat kan dien as 'n gemeenskaplike voertuig vir die ontwerp en implementering van intydsestelsels nie. Dit is die hipotese van die proefskrif dat 'n berekenings argitektuur met 'n streng basis, wat toegepas kan word vanaf ontwerp tot implementering, ingenieurs in staat sal stel om intydsestelsels beter te kan verstaan. Die hipotese word ondersoek deur die intydse datavloei argitektuur, RDF, te gebruik om 'n stelsel vanaf ontwerp tot implementering te beskryf. Daar is vier spesifieke navorsings areas ter verbetering van die ingenieurswese proses vir intydsestelsels, wat in die proefskrif aangespreek word: 1) die ontwikkelling van 'n argitektuur vir intydsestelsels wat geskik is vir die ontwerp en implementering in programmatuur en apparatuur, 2) . die konsolidering van 'n aantal grafiese tale in 'n grafiese notasie vir die funksionele spesifikasie, ontwerp en implementering van intydsestelsels, 3) die ontwikkelling van 'n eenvoudige verwerker argitektuur vir die uitvoering van intydse toepassings en 4) die evaluering van die argitektuur in die konteks van 'n mikrosatelliet gevallestudie. Die volgende oorspronklike bydraes word gemaak: 1) die sneller voorwaardes vir datavloei stelsels word uitgebrei met 'n disjunktiewe patroon saam met die tradisionele konjunktiewe patroon, 2) die inherente intydse datavloei eienskap, n.l. dat'n taak wat boodskappe ontvang, alle verwerking wat met 'n vorige boodskap gepaard gegaan het moet afhandel, voordat 'n volgende boodskap ontvang word, word uitgebrei na die sinkrone datavloei model, 3) 'n notasie om al die eienskappe van 'n intydsestelsel te beskryf word gedefinieer met RDF as die basis taal, 4) twee apparatuur verwerker argitekture word beskryf wat 'n een-tot-een kartering aanbied tussen die ontwerp en die implementering, en wat gevolglik die semantiese gaping verklein tussen ontwerpstaal en die uitvoeringsargitektuur en 5) die klasse van stelsels wat gemodelleer kan word met RDF sluit beheerstelsels en datavloeistelsels in. Die grafiese notasie en verwerker argitektuur was toegepas op sekere aspekte van die SUNSAT mikrosatelliet projek.

Thesis (MScEng (Electrical and Electronic Engineering))--University of Stellenbosch, 2007. / The aim of this project was to design and test a partial attitude determination and control system for a nanosatellite. The reaction wheel system was designed and tested as an actuator for the nanosatellite. This reaction wheel system consists of four reaction wheels mounted in a tetrahedral formation. A rate sensor system was also designed and its viability for this space application was examined. The rate sensor system consists of 3 orthogonally mounted planes, each with three rate sensors mounted on it. Hardware-inthe- loop tests were used along with an air bearing experimentational setup, which created near frictionless circumstances, to prove the effectiveness of the designed reaction wheel setup. The results following from this project were the following: The reaction wheel system proved to be an adequate actuator for this nanosatellite application and the rate sensor systemwhich was analysed proved to be inadequate for a nanosatellite application.