http://hdl.handle.net/2160/3044 2013-05-15T16:33:31Z http://hdl.handle.net/2160/7583 A multi-level, multi-purpose qualitative network flow analysis technique Shipman, Richard Charles Research at Aberystwyth into qualitative simulation of electrical circuits has been word-leading for nearly two decades. During that time, the Advanced Reasoning Group has developed a variety of electrical circuit simulators. The most successful, CIRQ, has been deployed for more than ten years in industrially useful design analysis systems, and has undergone a range of improvements over that time. MCIRQ is a more advanced version of CIRQ that was developed on the Design for the Whole Vehicle, Whole Lifecycle EPSRC Project at Aberystwyth. It is able to reason about circuits with multiple levels of qualitative resistance, and to indicate the different levels of current activity in the circuit. It showed great promise of being able to assist the design analysis systems in producing improved analysis reports, but was never deployed industrially because it was not able to analyse as wide a set of circuits as the original CIRQ software. This thesis describes a reimplementation of MCIRQ for use with both electrical and fluid flow systems. It extends the theory of electrical qualitative analysis in several ways, and makes several improvements to the original version of MCIRQ: It replaces the Forward-Reverse labelling system of path finding with Series- Parallel reduction system to determine current magnitude and direction. It improves the Star-Delta Reduction algorithm to deal with problem cases that do not occur in well-formed electrical circuits, but which do occur in the circuits with failure that are dealt with in failure analysis work. It extends MCIRQ to deal with many bridge circuits that were impossible to resolve in the previous version of MCIRQ. It introduces a principled strategy for resolving the direction of current flow within the circuit. The overall effect of these improvements is to transform MCIRQ into a qualitative analysis tool that can be used for real world applications. The thesis illustrates the kinds of analysis that can now be done by MCIRQ that was impossible before. The new MCIRQ has been employed on the ASTRAEA project. It has made possible applications of design analysis for engineering systems that could not have been addressed by any of the previous circuit analysers, notably in the domain of systems with both electrical and hydraulic aspects. 2011-09-01T00:00:00Z http://hdl.handle.net/2160/5805 Routing for logical ring topology networks : a comparison of two methods Giagkos, Alexandros This project looks at the design, implementation and comparison of routing protocols for self-healing, bidirectional, logical ring topology networks. The main goal is to propose two simple protocols by exploring two different techniques of searching and routing optimization. The first one uses a set of straight-forward, static rules and is inspired by a number of already existing protocols for both intradomain and interdomain routing, whereas the second one explores the capabilities of a biological inspired searching method; the use of genetic algorithms. The secondary goal of this work is to compare the two routing protocols by experimenting in an as realistic as possible network environment. For this reason, a widely known network simulator, ns-2, was studied and used during the implementation and testing phases of the project. The valuable knowledge taken from the experiments and their results lead to a number of interesting outcomes regarding the efficiency, speed and scalability of the two protocols. Although this project succeeded in creating two prototypes, collecting experimentation data and comparing their results, the protocols could have been improved and built to provide additional capabilities and sophistication. Possible ways to achieve further enhancements and improvements are included in this work, as the author presents his critical evaluation. 2008-01-01T00:00:00Z http://hdl.handle.net/2160/5297 Development of a Control System and Simulator for a Skid-Steer Amphibious Vehicle Clarke, Michael With autonomous robotic systems increasingly being used for various tasks including research projects, their ability to operate away from their ideal laboratory environment comes into question. Whilst long term autonomous robotic platforms do exist, few are based on vehicles that provide all terrain manoeuvrability, and certainly none combine this with skid-steering mechanisms. Inspired by the robotic requirements of the Rees Scan project in New Zealand, this dissertation aims to show the development of an all terrain robotic platform based on an ARGO 6x6 Amphibious Petrol Powered Skid-Steer Vehicle. The design and development of a control system, including software and hardware, is discussed. Further, a simplistic simulation environment is developed in an attempt to model the performance of the robotic platform on different surfaces. It is shown that the control of such a vehicle is possible with a reasonably high level of precision with relatively low cost components, suggesting that with more expensive hardware or different techniques, high-precision control would be possible. Further, it is shown that based on the various characteristics of the robotic platform such as the throttle values, wheel rotations and turn rates it is possible to classify, with some level of accuracy, the varying surfaces allowing for future work towards adaptive control systems able to adjust their performance as the robot moves from one unknown environment to another. 2010-08-27T17:26:21Z http://hdl.handle.net/2160/3841 Rich Internet Applications - A comparison of two technologies that can be used for their development Davies, Geraint Huw Through this project I aim to identify and articulate the relative capabilities of two technologies (Adobe Flex and Microsoft Silverlight) in their support for the development of Rich Internet Applications. Chapters 2 to 7 provide what is essentially background or supporting information for the key comparisons that I make in Chapters 8 to 12 (Note: Chapter 5 provides the key ideas that were encompassed in my original proposal together with supporting addenda). The later chapters each contain sub-sections in which I compare the two technologies with respect to the specific findings that I relate in those chapters. The overall conclusions which bring these all together are detailed in Chapter 13, with a critical evaluation of my work appearing in Chapter 14. Underpinning the investigation are two applications which I have developed in order to both assist my investigations and to use where necessary to demonstrate my conclusions and assertions. Whilst not being able to address all of the aspects of the project that I had originally wished to, let alone all the facilities that the two technologies support, I believe that this project shows that there are enough differences between them such that for those parties who have yet to make any financial, intellectual or needs-driven commitment to either, careful consideration of their relative capabilities should be taken. Within this, I also contend that there is a philosophical difference between the two technologies, i.e. that Microsoft is aiming its Silverlight offering towards web designers, and Adobe’s target for Flex is the application developer, and that this drive by both companies to poach a share of the market traditionally seen as the province of the other has resulted in some interesting differences between the implementations of their respective product platforms. Through a demonstration of the results of my investigations I conclude that, all other external factors being equal, the Adobe Flex / Flex Builder 3 approach is overall more comprehensive, efficient and therefore I believe productive. However, with the exception of two very key issues (i.e. The lack of formal support by Adobe for an implementation of multi-scale image technology and problems with running Silverlight applications on Linux / Unix-based systems), I also conclude that the differences I have identified and have declared as important from my perspective could well be outweighed by other issues that are specifically pertinent to another individual, organisation or academic institution. I also conclude that whilst I believe my arguments are sound in respect of the factors and criteria investigated, I also acknowledge that further work may identify other differences that may be influential. In addition, both Microsoft and Adobe announced beta release developments of their respective products during the course of this project and although I took the, what I believe was justifiable, decision not to include them in my investigations (primarily for technical reasons), I must also acknowledge that it is possible that some of these developments may highlight a divergence (or desirable convergence) between the two technologies in one or more important measures. 2009-09-01T00:00:00Z