Thesis, M.Phil., University of Hong Kong, 1984. / Also available in print.

Thesis (doctoral)--Stockholm School of Economics, 1989. / Thesis statement in Swedish inserted. Includes bibliographical references.

Thesis (M. Sc.)--University of Canterbury, 2008. / Typescript (photocopy). Includes bibliographical references (p. 71-75). Also available via the World Wide Web.

Thesis (M.S.)--Rochester Institute of Technology, 2007. / Typescript. Includes bibliographical references (leaves 161-164).

[pt] Este trabalho trata do problema de equacionamento de grafos de ligação. Inicialmente, é proposta uma representação padrão que será utilizada na eliminação dos campos dissipador implícito e armazenador dependente. A seguir, são apresentados dois procedimentos; o primeiro permite obter a função de transferência e o segundo fornece as equações de estado. Em ambos casos o resultado final é obtido por inspeção, não sendo necessário resolver as equações do grafo. As limitações dos procedimentos são discutidas e futuros trabalhos de pesquisa são propostos. / [en] This work deals with the problem of bond graph resolution. First, a standard representation is proposed which will be used for the implicit dissipation and dependent storage fields elimination. Two procedures are presented. The first generates the graph transfer function and second gives the state equations. In Both procedures the final result is obtained by inspection, avoiding the need for solving the bond graph equation. The procedures limitations are discussed and topics for future research are proposed.

In this work, we present the implementation and evaluation of a novel parameterizable transfer function for use in artificial neural networks. It allows the continuous change between summation and multiplication for the operation performed by a neuron. The transfer function is based on continuously differentiable fractional iterates of the exponential function and introduces an additional parameter per neuron and layer. This parameter can be determined along weights and biases during standard, gradient-based training. We evaluate the proposed transfer function within neural networks by comparing its performance to conventional transfer functions for various regression problems. Interpolation between summation and multiplication achieves comparable or even slightly better results, outperforming the latter on a task involving missing data and multiplicative interactions between inputs. / I detta arbete presenterar vi implementationen och utvärderingen av en ny överföringsfunktion till användning i artificiella neuronnät. Den tillåter en kontinuerlig förändring mellan summering och multiplikation för operationen som utförs av en neuron. Överföringsfunktionen är baserad på kontinuerligt deriverbara bråkiterationer av exponentialfunktionen och introducerar ytterligare en parameter för varje neuron och lager. Denna parameter kan bestämmas längs vikter och avvikelser under vanlig lutningsbaserad träning. Vi utvärderar den föreslagna överföringsfunktionen inom neurala nätverk genom att jämföra dess prestanda med konventionella överföringsfunktioner för olika regressionsproblem. Interpolering mellan summering och multiplikation uppnår jämförbara eller något bättre resultat, till exempel för en uppgift som gäller saknade data och multiplikativ interaktion mellan indata.

No description available.

Increasingly stringent emissions regulations have led combustion system designers to look for more environmentally combustion strategies. For gas turbine combustion, one promising technology is lean premixed combustion, which results in lower flame temperatures and therefore the possibility of significantly reduced nitric oxide emissions. While lean premixed combustion offers reduced environmental impacts, it has been observed to experience increased possibility of the occurrence of combustion instabilities, which may damage hardware and reduce efficiency. Thermoacoustic combustion instabilities occur when oscillations in the combustor acoustics and oscillations in the flame heat release rate form a closed feedback loop, through one of two possible mechanisms. The first is direct coupling which occurs due to the mean mass flow oscillations induced by the acoustic velocity. Secondly, the acoustics may couple with the flame due to acoustic interactions with fuel/air mixing, resulting in an oscillating equivalence ratio. Only velocity coupling was considered in this study. The methodology used in this study is analysis of instabilities through linear systems theory, requiring knowledge of the individual transfer functions making up the closed-loop system. Methods already exist by which combustor acoustics may be found. However, significant gaps still remain in knowledge of the nature of flame dynamics. Prior knowledge in literature about the flame transfer function suggests that the flame behaves as a low-pass filter, with cutoff frequency on the order of hundreds of hertz. Nondimensionalization of the frequency by flame length scales has been observed to result in a convenient scaling for the flame transfer function, suggesting that the flame dynamics may be dominated by spatial effects. This work was proposed in two parts to extend and apply the body of knowledge on flame dynamics. The phase one goal of this study was to further understand this relationship between the flame heat release rate dynamics and the dynamics of the reaction zone size. The second goal of this work was to apply this flame transfer function knowledge to predictions of instability, validated against measurements in an unstable combustor. Both of these goals meet an existing practical need, providing a design tool for prediction of potential thermoacoustic instabilities in a combustor at the design stage.Measurements of the flame transfer function were made in a swirl-stabilized, lean-premixed combustor. The novel portion of these measurements was the inclusion of spatial resolution of the heat release rate dynamics. By using a speaker, a sine dwell excitation to the velocity was introduced over the range of 10-400Hz. Measurements were then made of the input (inlet velocity) and output (heat release rate, or flame size) resulting in the flame transfer function. The spatial dynamics measurement was approached through several measures of the flame size: the volume and offset distance to the center of the heat release. Each was obtained from deconvoluted, phase averaged images of the flame, referenced to the speaker excitation signal. The results of these measurements showed that the spatial dynamics for each of these three measures were virtually identical to the heat release rate dynamics. This suggests a quite important result, namely that the flame heat release rate dynamics are completely determined by the dynamics of the flame structure. Therefore, prediction of flow structure interaction with the flame distribution is crucial to predict the dynamics of the flame. These spatially resolved transfer function measurements were used in conjunction with the linear closed-loop model to make predictions of instability. These predictions were made by applying the Bode stability criterion to the open-loop system transfer function. This criterion states that instabilities may occur at frequencies where the heat release rate and acoustic oscillations occur in phase and the system gain has a value greater than unity. Performing this analysis on the combined system transfer function yielded results that agreed quite well with actual instability measurements made in the combustor. Closed-loop predictions identified two possible modes for instability, both of which were observed experimentally. One mode resulted from an acoustic peak around 160 Hz, and occurred at lean equivalence ratios. A second mode occurred at lower frequencies (100-150 Hz) and was associated with the increase in flame transfer function gain at increasing equivalence ratios. These are some of the first successful predictions of combustion instability based on linear systems theory. When multiple modes were predicted, it was assumed that if non-linear effects were to be considered, the lower frequency mode would become the dominant mode at these operating conditions due to its higher gain margin. Also of note is that in the practical system, high frequency oscillations are observed, but not predicted, associated with harmonics of the low frequency mode due to the linear nature of the predictions. While these non-linear effects are not captured, the linear predictive capability is thought to be most important, as from a practical perspective, instabilities should be avoided altogether. The primary findings of this study have significant applications to modeling and prediction of combustion dynamics. The classic heat release rate flame transfer function was observed to coincide almost exactly with the flame size transfer functions. The time scales observed in these transfer functions correspond to convective length scales in the combustor, suggesting a fluid mechanical basis of the heat release rate response. Additionally, linear systems theory predictions of instability based on the measured flame transfer functions were proved capable of capturing the stability of the actual combustor with a reasonable degree of accuracy. These predictions should have considerable application to design level avoidance of combustion instability in practical systems. / Ph. D.

Problems of cascade connections (synthesis) and decomposition (analysis) are analyzed for two classes of linear systems with infinite dimensional state spaces, namely, 1) admissible systems in the sense of Bart, Gohberg and Kaashoek and 2) regular systems as recently introduced by Weiss. For the class of BGK-admissible systems, it is shown that the product of two admissible systems is again admissible and that a Wiener-Hopf factorization problem can be solved just as in the finite-dimensional case. For the class of regular systems, it is shown that the cascade connection of a rational stable and antistable system has an additive stable-antistable decomposition; this involves giving a distribution interpretation to the solution of a linear Sylvester equation involving unbounded operator coefficients. As an application, some preliminary work is presented toward obtaining a state space solution of the sensitivity minimization problem for a pure delay plant. / Ph. D.

Simulation, whether in-flight or ground-based, is an invaluable tool for testing and evaluating aircraft. Classically, a simulation model is specific to a single particular airframe, only able to model those flying characteristics. Vast information can be gained from a simulation that is able to model a wide range of aircraft, through a comparison of the performance of these aircraft. Such a variable stability simulation model was created based on 46 stability parameters, including natural frequencies, damping ratios, time constants, and gains. The simulation was obtained using transfer functions representing the aircraft state responses to control inputs. These transfer functions were converted into state space systems used to create the linear equations for the model. The model was first developed as a desktop simulation and then converted for use with the Virginia Tech's 2F122A flight simulator. This conversion required a simple dynamic inversion of the body axis force and moment terms. To reduce the error in these terms, a model following scheme was incorporated. A series of canned inputs and real-time pilot-in-the-loop tests were flown to evaluate the variable stability model. Results in this paper have demonstrated the successful creation of a variable stability simulation model. / Master of Science