Charles E. Rohrs, Lena Valavani, Michael Athans, Gunter Stein. / "June 1983" / Bibliography: p. 14. / "NASA/NGL-22-009-124" "ONR/N00014-82-K-0582 (NR 606-003) "NSF/ECS-8210960" "NSF/ECS-8206495"

A dissertation submitted in fulfilment of the requirements for the degree of Masters of Science in Engineering (Electrical) to the Faculty of Engineering and the Built Environment, University of the Witwatersrand, Johannesburg, 2017 / The Travelling Salesman Problem (TSP) is an important NP-hard combinatorial optimisation problem that forms the foundation of many modern-day, practical problems such as logistics or network route planning. It is often used to benchmark discrete optimisation algorithms since it is a fundamental problem that has been widely researched. The Flower Pollination Algorithm (FPA) is a continuous optimisation algorithm that demonstrates promising results in comparison to other well-known algorithms. This research proposes the design, implementation and testing of two new algorithms based on the FPA for solving discrete optimisation problems, more specifically the TSP, namely the Discrete Flower Pollination Algorithm (DFPA) and the iterative Discrete Flower Pollination Algorithm (iDFPA). The iDFPA uses two proposed update methods, namely the Best Tour Update (BTU) and the Rejection Update (RU), to perform the iterative update process. The two algorithms are compared to the Ant Colony Optimisation’s (ACO) MAX−MIN Ant System (MMAS) as well as the Genetic Algorithm (GA) since they are well studied and developed. The DFPA and iDFPA results are significantly better than the GA and the iDFPA is able to outperform the ACO in all tested instances. The iDFPA with 300 iterations was able to achieve the optimal solution in the Berlin52 benchmark TSP problem as well as have improvements of up to 4.56% and 41.87% compared to the ACO and GA respectively. An analysis of how the RU and the annealing schedule used in the RU impacts on the overall results of the iDFPA is given. The RU analysis demonstrates how the annealing schedule can be manipulated to achieve certain results from the iDFPA such as faster convergence or better overall results. A parameter analysis is performed on both the DFPA and iDFPA for different TSP problem sizes and the suggested initial parameters for these algorithms are outlined. / XL2018

No description available.

No description available.

Due to the complexity, multi-scale, and multi-physics nature of the mathematical models for fires, current numerical models require too much computational effort to be useful in design and real-time decision making, especially when dealing with fires over large domains. To reduce the computational time while retaining the complexity of the domain and physics, our research has focused on several reduced-order modeling techniques. Our contributions are improving wildland fire reduced-order models (ROMs), creating new ROM techniques for nonlinear systems, and preserving optimality when discretizing a continuous-time ROM. Currently, proper orthogonal decomposition (POD) is being used to reduce wildland fire-spread models with limited success. We use a technique known as the discrete empirical interpolation method (DEIM) to address the slowness due to the nonlinearity. We create new methods to reduce nonlinear models, such as the Burgers' equation, that perform better than POD over a wider range of input conditions. Further, these ROMs can often be constructed without needing to capture full-order solutions a priori. This significantly reduces the off-line costs associated with creating the ROM. Finally, we investigate methods of time-discretization that preserve the optimality conditions in a certain norm associated with the input to output mapping of a dynamical system. In particular, we are able to show that the Crank-Nicholson method preserves the optimality conditions, but other single-step methods do not. We further clarify the need for these discrete-time ROMs to match at infinity in order to ensure local optimality. / Ph. D.

La compétitivité des entreprises manufacturières dépend fortement de la productivité des machines etdes moyens de production. Pour garantir un haut niveau de productivité il est indispensable de minimiser lestemps d'arrêt dus aux fautes ou dysfonctionnements. Cela nécessite des méthodes efficaces pour détecter et isolerles fautes apparues dans un système (FDI). Dans cette thèse, une méthode FDI à base de modèles est proposée.La méthode est conçue pour la classe des systèmes à événements discrets industriels composés d’une bouclefermée du contrôleur et du processus. En comparant les comportements observés et attendus par le modèle, il estpossible de détecter et d’isoler des fautes. A la différence de la plupart des approches FDI des systèmes àévénements discrets, une méthode basée sur des modèles du comportement normal au lieu de modèles descomportements fautifs est proposée. Inspiré par le concept des résidus bien connu pour le diagnostic dessystèmes continus, une nouvelle approche pour l’isolation des fautes dans les systèmes à événements discrets aété développée. La clé pour l’application des méthodes FDI basées sur des modèles est d’avoir un modèle justedu système considéré. Comme une modélisation manuelle peut être très laborieuse et coûteuse pour dessystèmes à l’échelle industrielle, une approche d’identification pour les systèmes à événements discrets enboucle fermée est développée. Basée sur un algorithme connu pour l’identification des modèles monolithiques,une adaptation distribuée est proposée. Elle permet de traiter de grands systèmes comportant un haut degré deparallélisme. La base de cette approche est une décomposition du système en sous systèmes. Cettedécomposition est automatisée en utilisant un algorithme d’optimisation analysant le comportement observé dusystème. Les méthodes conçues dans cette thèse ont été mises en oeuvre sur une étude de cas et sur uneapplication d’échelle industrielle. / The competitiveness of manufacturing companies strongly depends on the productivity of machinesand production processes. To guarantee a high level of productivity, downtimes occurring due to faults have tobe kept as short as possible. This necessitates efficient fault detection and isolation (FDI) methods. In this work,a model-based FDI method for the widely used class of industrial closed-loop Discrete Event Systems isproposed. The considered systems consist of the closed-loop of plant and controller. Based on the comparison ofobserved and modeled system behavior, it is possible to detect and to isolate faults. Unlike most known methodsfor FDI in Discrete Event Systems, this work proposes working with a model of the fault-free behavior ratherthan working fault models. Inspired by the concept of residuals known from FDI in continuous systems, a newapproach for fault isolation based on fault-free Discrete Event System models is developed. The key of anymodel-based diagnosis method is to have an accurate model of the considered system. Since manual modelbuildingcan be very difficult for large industrial systems, an identification approach for this class of systems isintroduced. Based on an already existing monolithic identification algorithm, a distributed adaptation isdeveloped which allows treating large, concurrent systems. The key of the proposed approach is an automaticdecomposition of a given closed-loop Discrete Event System using an optimization approach which analyzesobserved system behavior. The methods developed in this thesis are applied to a mid-sized laboratory system andto an industrial winder to show their scalability.

Discrete-time survival analysis as an educational research technique has focused on analysing and interpretating parameter estimates. The purpose of this study was to examine the effects of certain data characteristics on the hazard estimates and goodness of fit statistics. Fifty-four simulated data sets were crossed with four conditions in a 2 (time period) by 3 (distribution of Y = 1) by 3 (distribution of Y = 0) by 3 (sample size) design.

Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1980. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING. / Bibliography: leaf 99. / by Michael Andrew Gennert. / M.S.

Thesis. 1976. Elec.E.--Massachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science. / Microfiche copy available in Archives and Engineering. / Bibliography: leaves 138-140. / Elec.E.

Thesis. 1978. Ph.D.--Massachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING. / Includes bibliographical references. / Ph.D.