No description available.

No description available.

No description available.

This work describes a framework for simultaneous estimation and modeling (SEAM) of dynamic systems using non-Gaussian belief fusion by first presenting the relevant fundamental formulations, then building upon these formulations incrementally towards a more general and ubiquitous framework. Multi-Gaussian belief fusion (MBF) is introduced as a natural and effective method of fusing non-Gaussian probability distribution functions (PDFs) in arbitrary dimensions efficiently and with no loss of accuracy. Construction of some multi-Gaussian structures for potential use in MBF is addressed. Furthermore, recursive Bayesian estimation (RBE) is developed for linearized systems with uncertainty in model parameters, and a rudimentary motion model correction stage is introduced. A subsequent improvement to motion model correction for arbitrarily non-Gaussian belief is developed, followed by application to observation models. Finally, SEAM is generalized to fully nonlinear and non-Gaussian systems. Several parametric studies were performed on simulated experiments in order to assess the various dependencies of the SEAM framework and validate its effectiveness in both estimation and modeling. The results of these studies show that SEAM is capable of improving estimation when uncertainty is present in motion and observation models as compared to existing methods. Furthermore, uncertainty in model parameters is consistently reduced as these parameters are updated throughout the estimation process. SEAM and its constituents have potential uses in robotics, target tracking and localization, state estimation, and more. / Doctor of Philosophy / The simultaneous estimation and modeling (SEAM) framework and its constituents described in this dissertation aim to improve estimation of signals where significant uncertainty would normally introduce error. Such signals could be electrical (e.g. voltages, currents, etc.), mechanical (e.g. accelerations, forces, etc.), or the like. Estimation is accomplished by addressing the problem probabilistically through information fusion. The proposed techniques not only improve state estimation, but also effectively "learn" about the system of interest in order to further refine estimation. Potential uses of such methods could be found in search-and-rescue robotics, robust control algorithms, and the like. The proposed framework is well-suited for any context where traditional estimation methods have difficulty handling heightened uncertainty.

The purpose of the thesis was to investigate the influence communication architecture decisions have on the performance of a simulation system with distributed components. In particular, the objective was to assess the relative importance of factors affecting reliability and variability of an external data interface to the performance of the simulation, as compared to factor within the simulation itself.

The optimal estimation theory of the Wiener-Kalman filter is extended to cover the situation in which the number of memory elements in the estimator is restricted. A method, based on the simultaneous diagonalisation of two symmetric positive definite matrices, is given which allows the weighted least square estimation error to be minimised. A control system design method is developed utilising this estimator, and this allows the dynamic controller in the feedback path to have a low order. A 12-order once-through boiler model is constructed and the performance of controllers of various orders generated by the design method is investigated. Little cost penalty is found even for the one-order controller when compared with the optimal Kalman filter system. Whereas in the Kalman filter all information from past observations is stored, the given method results in an estimate of the state variables which is a weighted sum of the selected information held in the storage elements. For the once-through boiler these weighting coefficients are found to be smooth functions of position, their form illustrating the implicit model reduction properties of the design method. Minimal-order estimators of the Luenberger type also generate low order controllers and the relation between the two design methods is examined. It is concluded that the design method developed in this thesis gives better plant estimates than the Luenberger system and, more fundamentally, allows a lower order control system to be constructed. Finally some possible extensions of the theory are indicated. An immediate application is to multivariable control systems, while the existence of a plant state estimate even in control systems of very low order allows a certain adaptive structure to be considered for systems with time-varying parameters.

Discrete event simulation is an important tool for modeling and analysis. Some of the simulation applications such as telecommunication network performance, VLSI logic circuits design, battlefield simulation, require enormous amount of computing resources. One way to satisfy this demand for computing power is to decompose the simulation system into several logical processes (Ip) and run them concurrently. In any parallel discrete event simulation (PDES) system, the events are ordered according to their time of occurrence. In order for the simulation to be correct, this ordering has to be preserved. There are three approaches to maintain this ordering. In a conservative system, no lp executes an event unless it is certain that all events with earlier time-stamps have been executed. Such systems are prone to deadlock. In an optimistic system on the other hand, simulation progresses disregarding this ordering and saves the system states regularly. Whenever a causality violation is detected, the system rolls back to a state saved earlier and restarts processing after correcting the error. There is another approach in which all the lps participate in the computation of a safe time-window and all events with time-stamps within this window are processed concurrently. In optimistic simulation systems, there is a global virtual time (GVT), which is the minimum of the time-stamps of all the events existing in the system. The system can not rollback to a state prior to GVT and hence all such states can be discarded. GVT is used for memory management, load balancing, termination detection and committing of events. However, GVT computation introduces additional overhead. In optimistic systems, large number of rollbacks can degrade the system performance considerably. We have studied the effect of load balancing in reducing the number of rollbacks in such systems. We have designed three load balancing algorithms and implemented two of them on a network of workstations. The other algorithm has been analyzed probabilistically. The reason for choosing network of workstations is their low cost and the availability of efficient message passing softwares like PVM and MPI. All of these load balancing algorithms piggyback on the existing GVT computation algorithms and try to balance the speed of simulation in different lps. We have also designed an optimal GVT computation algorithm for the hypercubes and studied its performance with respect to the other GVT computation algorithms by simulating a hypercube in our network cluster. We use the topological properties of a star network in order to design an algorithm for computing a safe time-window for parallel discrete event simulation. We have analyzed and simulated the behavior of an open queuing network resembling such an architecture. Our algorithm is also extended for hierarchical stars and for recursive window computation.

Esta tese trata do problema de controle ótimo estocástico de sistemas com saltos Markovianos e ruído multiplicativo a tempo discreto, com horizontes de tempo finito e infinito. A função custo é composta de termos quadráticos e lineares nas variáveis de estado e de controle, com matrizes peso indefinidas. Como resultado principal do problema com horizonte finito, é apresentada uma condição necessária e suficiente para que o problema de controle seja bem posto, a partir da qual uma solução ótima é derivada. A condição e a lei de controle são escritas em termos de um conjunto acoplado de equações de Riccati interconectadas a um conjunto acoplado de equações lineares recursivas. Para o caso de horizonte infinito, são apresentadas as soluções ótimas para os problemas de custo médio a longo prazo e com desconto, derivadas a partir de uma solução estabilizante de um conjunto de equações algébricas de Riccati acopladas generalizadas (GCARE). A existência da solução estabilizante é uma condição suficiente para que tais problemas sejam do tipo bem posto. Além disso, são apresentadas condições para a existência das soluções maximal e estabilizante do sistema GCARE. Como aplicações dos resultados obtidos, são apresentadas as soluções de um problema de otimização de carteiras de investimento com benchmark e de um problema de gestão de ativos e passivos de fundos de pensão do tipo benefício definido, ambos os casos com mudanças de regime nas variáreis de mercado. / This thesis considers the finite-horizon and infinite-horizon stochastic optimal control problem for discrete-time Markov jump with multiplicative noise linear systems. The performance criterion is assumed to be formed by a linear combination of a quadratic part and a linear part in the state and control variables. The weighting matrices of the state and control for the quadratic part are allowed to be indefinite. For the finite-horizon problem the main results consist of deriving a necessary and sufficient condition under which the problem is well posed and a optimal control law is derived. This condition and the optimal control law are written in terms of a set of coupled generalized Riccati difference equations interconnected with a set of coupled linear recursive equations. For the infinite-horizon problem a set of generalized coupled algebraic Riccati equations (GCARE) is studied. In this case, a sufficient condition under which there exists the maximal solution and a necessary and sufficient condition under which there exists the mean square stabilizing solution for the GCARE are presented. Moreover, a solution for the discounted and long run average cost problems is presented. The results obtained are applied to solver a portfolio optimization problem with benchmark and a pension fund problem with regime switching.

Bibliography : leaves 196-200. / Thesis (Ph. D.)--Massachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science, 1977. / Microfiche copy available in the Institute Archives and Barker Engineering Library. / by Loren Kerry Platzman. / Ph.D.

Thesis (Sc.D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1982. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING / Bibliography: leaves 898-903. / by Howard Jay Chizeck. / Sc.D.