This thesis presents two new algorithms for doing constrained Magnetotelluric (MT) inversion based on an existing Occam 2D inversion program. The first algorithm includes borehole resistivity logs as prior information to constrain resistivity directly in the vicinity of boreholes. The second algorithm uses reflection seismic data as prior constraints to transfer structural information from seismic images to 2D resistivity models. These two algorithms are efficient (proved through tests of synthetic examples) and widely applicable. In this thesis, they have been successfully applied to the COSC (Collisional Orogeny in the Scandinavian Caledonides) MT data. The COSC project aims to study the mountain belt dynamics in central Sweden by drilling two 2.5 km deep boreholes. MT data were collected to locate the main décollement that separates the overlying Caledonian allochthons and the underlying Precambrian basement, as the main décollement is associated with very conductive Alum shale. The previous interpretation based on part of the COSC seismic profile (CSP) was that the main décollement was located along a reflection with depth of 4.5 km underneath Åre and ~3 km underneath Mörsil, in central Jämtland. The MT resistivity model reveals a very conductive layer in the central and western parts of the profile, the top of which coincides with the first seismic reflection. This means that the first conductive alum shale layer occurs at less than 1 km depth, supporting a new interpretation of the main décollement at shallower depth. In a re-interpretation of the CSP data based on the MT model, the main décollement occurs a few hundred metres below the top of the conductor and is coincident with a laterally continuous seismic reflection. Further, the overlying seismic reflections resemble imbricated alum shale of the Lower Allochthon. MT inversion using seismic constraints from CSP gives further support to the new interpretation. Moreover, MT investigations were conducted in the Alnö alkaline and carbonatite ring-intrusion complex in Sweden. 2D and 3D resistivity models inverted from MT data together with resistivity and porosity laboratory measurements delineate a fossil magma chamber as a resistive anomaly surrounded by electrically conductive up-doming and ring-shaped faults and fractures.

IoT devices are   increasingly used in the process of gathering scientific data. In   environmental monitoring IoT devices can be used as remote sensing devices to   collect information about e.g. temperature. To keep data reliable, various   security aspects have to be considered. Constrained devices are limited by   memory size and battery life, a security solution has to be developed with   this in mind. In this study an IoT security solution was developed in collaboration   with a research group in environmental science at Umeå University. We   selected commonly used algorithms and compared them with the goal to provide   authentication and integrity for an IoT application, while minimizing energy   use running on an Atmega 1284P. The results showed that the encryption   algorithm AES-256-GCM is a good choice for a total security solution.   AES-256-GCM provides authenticated encryption with additional data while, in   relation to the other tested algorithms, using energy at a low level and   leaving a small program size footprint.

In this thesis we propose four new methods for solving constrained global optimization problems. The first proposed algorithm is a differential evolution (DE) algorithm using penalty functions for constraint handling. The second algorithm is based on the first DE algorithm but also incorporates a filter set as a diversification mechanism. The third algorithm is also based on DE but includes an additional local refinement process in the form of the pattern search (PS) technique. The last algorithm incorporates both the filter set and PS into the DE algorithm for constrained global optimization. The superiority of feasible points (SFP) and the parameter free penalty (PFP) schemes are used as constraint handling mechanisms. The new algorithms were numerically tested using two sets of test problems and the results where compared with those of the genetic algorithm (GA). The comparison shows that the new algorithms outperformed GA. When the new methods are compared to each other, the last three methods performed better than the first method i.e. the DE algorithm. The new algorithms show promising results with potential for further research. Keywords: constrained global optimization, differential evolution, pattern search, filter method, penalty function, superiority of feasible points, parameter free penalty. ii

Este trabalho apresenta uma nova estrutura matemática para a álgebra de caminhos, que permite analisar a convergência dos algoritmos de roteamento multi-restritivos hop-by-hop e, sob o ponto de vista da engenharia de tráfego e da Qualidade de Serviço (QoS) na arquitetura Generalized Multiprotocol Label Switching (GMPLS), garantir de maneira confiável a incorporação de novas métricas de roteamento aos algoritmos de roteamento baseados em múltiplas restrições. Baseando-se nessa nova álgebra de caminhos, são analisadas as propriedades de monotonicidade, isotonicidade e liberdade, conhecidas por garantir a convergência dos algoritmos de roteamento e, ao contrário do indicado na literatura até o momento, verifica-se que a propriedade de monotonicidade não e condição necessária e nem suficiente para garantir a convergência dos algoritmos de roteamento multi-restritivos hop-by-hop. Sendo assim, este trabalho propõe uma nova propriedade, denominada coerência, para a garantia da convergência do roteamento hop-by-hop e um novo algoritmo de roteamento hop-by-hop com convergência garantida. Para avaliar os resultados teóricos obtidos, s~ao analisados dois estudos de casos de aplicação do roteamento multi-restritivos hop-by-hop com o uso de uma ferramenta de simulação desenvolvida em MATLAB e baseada no algoritmo Eliminação de Loop pelo Nó de Destino (ELND) também proposto. Como resultado das simulações desses estudos de casos, verifica-se que as diferentes estratégias de otimização, necessárias as redes (GMPLS), impõem a necessidade de trabalhar com algoritmos de roteamento que permitam a definição de mais de duas métricas de roteamento com diferentes critérios de otimização para cada uma delas, comprovando, portanto, a necessidade do desenvolvimento e da continuação deste trabalho. / This work presents a new mathematical structure for paths algebra that allows the convergence analysis of hop-by-hop multi-constrained routing algorithms and, under the traffic engineering and quality of service perspectives in the Generalized Multiprotocol Label Switching (GMPLS) architecture, trustily ensures the aggregation of new routing metrics in a constrained-based routing. Based on this new paths algebra, we analyze the monotonicity, isotonicity and freeness properties, known as ensuring routing algorithms convergence, and despite of what has been indicated in the literature, we verified that the monotonicity property is not sufficient to ensure the hop-by-hop routing convergence. Therefore, this work proposes a new property, called coherence, as a necessary and sufficient condition to ensure it, as well as, a new multi-constrained hop-by-hop routing algorithm with ensured convergence. In order to evaluate the theoretical results obtained, two study cases of the hop-by-hop multi-constrained routing applications are analyzed in the present thesis by using the Eliminação de Loop pelo Nó de Destino (ELND) simulation tool, developed in MATLAB and also presented as a product of this work. As result of these study cases simulations, we verified that different optimization strategies, requested by the (GMPLS) networks, compel the use of routing algorithms that allow the specification of more than two routing metrics with different optimization criteria for each one of them, thus proving the necessity of this work and its continuation.

Os sistemas baseados em redes neurais artificiais e algoritmos genéticos oferecem um método alternativo para solucionar problemas relacionados à otimização de sistemas. Os algoritmos genéticos devem a sua popularidade à possibilidade de percorrer espaços de busca não-lineares e extensos. As redes neurais artificiais possuem altas taxas de processamento por utilizarem um número elevado de elementos processadores simples com alta conectividade entre si. Redes neurais com conexões realimentadas fornecem um modelo computacional capaz de resolver vários tipos de problemas de otimização, os quais consistem, geralmente, da otimização de uma função objetivo que pode estar sujeita ou não a um conjunto de restrições. Esta tese apresenta uma abordagem inovadora para resolver problemas de otimização não-linear restrita utilizando uma arquitetura neuro-genética. Mais especificamente, uma rede neural de Hopfield modificada é associada a um algoritmo genético visando garantir a convergência da rede em direção aos pontos de equilíbrio factíveis que representam as soluções para o problema de otimização não-linear restrita. / Systems based on artificial neural networks and genetic algorithms are an alternative method for solving systems optimization problems. The genetic algorithms must its popularity to make possible cover nonlinear and extensive search spaces. Artificial neural networks have high processing rates due to the use of a massive number of simple processing elements and the high degree of connectivity between these elements. Neural networks with feedback connections provide a computing model capable of solving a large class of optimization problems, which refer to optimization of an objective function that can be subject to constraints. This thesis presents a novel approach for solving constrained nonlinear optimization problems using a neuro-genetic approach. More specifically, a modified Hopfield neural network is associated with a genetic algorithm in order to guarantee the convergence of the network to the equilibrium points, which represent feasible solutions for the constraint nonlinear optimization problem.

Mesh generation is a fundamental precursor to finite element implementations for solution of partial differential equations in engineering and science. This dissertation advances the field in three distinct but coupled areas. A robust and fast three dimensional mesh generator for arbitrarily shaped geometries was developed. It deploys nodes throughout the domain based upon user-specified mesh density requirements. The system is integer and pixel based which eliminates round off errors, substantial memory requirements and cpu intensive calculations. Linked, but fully detachable, to the mesh generation system is a physical boundary recovery routine. Frequently, the original boundary topology is required for specific boundary condition applications or multiple material constraints. Historically, this boundary preservation was not available. An algorithm was developed, refined and optimized that recovers the original boundaries, internal and external, with fidelity. Finally, a node repositioning algorithm was developed that maximizes the minimum solid angle of tetrahedral meshes. The highly coveted 2D Delaunay property that maximizes the minimum interior angle of a triangle mesh does not extend to its 3D counterpart, to maximize the minimum solid angle of a tetrahedron mesh. As a consequence, 3D Delaunay created meshes have unacceptable sliver tetrahedral elements albeit composed of 4 high quality triangle sides. These compromised elements are virtually unavoidable and can foil an otherwise intact mesh. The numerical optimization routine developed takes any preexisting tetrahedral mesh and repositions the nodes without changing the mesh topology so that the minimum solid angle of the tetrahedrons is maximized. The overall quality enhancement of the volume mesh might be small, depending upon the initial mesh. However, highly distorted elements that create ill-conditioned global matrices and foil a finite element solver are enhanced significantly.

The research conducted in this dissertation is divided into two main parts. The first part provides further improvements in power system state estimation and the second part implements Contingency Constrained Optimal Power Flow (CCOPF) in a stochastic multiple contingency framework. As a real-time application in modern power systems, the existing Newton-QR state estimation algorithms are too slow and too fragile numerically. This dissertation presents a new and more robust method that is based on trust region techniques. A faster method was found among the class of Krylov subspace iterative methods, a robust implementation of the conjugate gradient method, called the LSQR method. Both algorithms have been tested against the widely used Newton-QR state estimator on the standard IEEE test networks. The trust region method-based state estimator was found to be very reliable under severe conditions (bad data, topological and parameter errors). This enhanced reliability justifies the additional time and computational effort required for its execution. The numerical simulations indicate that the iterative Newton-LSQR method is competitive in robustness with classical direct Newton-QR. The gain in computational efficiency has not come at the cost of solution reliability. The second part of the dissertation combines Sequential Quadratic Programming (SQP)-based CCOPF with Monte Carlo importance sampling to estimate the operating cost of multiple contingencies. We also developed an LP-based formulation for the CCOPF that can efficiently calculate Locational Marginal Prices (LMPs) under multiple contingencies. Based on Monte Carlo importance sampling idea, the proposed algorithm can stochastically assess the impact of multiple contingencies on LMP-congestion prices.

Un problème fondamental à résoudre en Automatique réside dans la commande des systèmes incertains qui présentent des contraintes sur les variables de l’entrée, de l’état ou la sortie. Ce problème peut être théoriquement résolu au moyen d’une commande optimale. Cependant la commande optimale par principe n’est pas une commande par retour d’état ou retour de sortie et offre seulement une trajectoire optimale le plus souvent par le biais d’une solution numérique.Par conséquent, dans la pratique, le problème peut être approché par de nombreuses méthodes, tels que”commande over-ride” et ”anti-windup”. Une autre solution, devenu populaire au cours des dernières décennies est la commande prédictive. Selon cette méthode, un problème de la commande optimale est résolu à chaque instant d’échantillonnage, et le composant du vecteur de commande destiné à l’échelon curant est appliquée. En dépit de la montée en puissance des architecture de calcul temps-réel, la commande prédictive est à l’heure actuelle principalement approprié lorsque l’ordre est faible, bien connu, et souvent pour des systèmes linéaires. La version robuste de la commande prédictive est conservatrice et compliquée à mettre en œuvre, tandis que la version explicite de la commande prédictive donnant une solution affine par morceaux implique une compartimentation de l’état-espace en cellules polyédrales, très compliquée.Dans cette thèse, une solution élégante et peu coûteuse en temps de calcul est présentée pour des systèmes linéaire, variant dans le temps ou incertains. Les développements se concentre sur les dynamiques en temps discret avec contraintes polyédriques sur l’entrée et l’état (ou la sortie) des vecteurs, dont les perturbations sont bornées. Cette solution est basée sur l’interpolation entre un correcteur pour la région extérieure qui respecte les contraintes sur l’entrée et de l’état, et un autre pour la région intérieure, ce dernier plus agressif, conçue par n’importe quelle méthode classique, ayant un ensemble robuste positivement invariant associé à l’intérieur des contraintes. Une simple fonction de Lyapunov est utilisée afin d’apporter la preuve de la stabilité en boucle fermée. / A fundamental problem in automatic control is the control of uncertain plants in the presence of input and state or output constraints. An elegant and theoretically most satisfying framework is represented by optimal control policies which, however, rarely gives an analytical feedback solution, and oftentimes builds on numerical solutions (approximations).Therefore, in practice, the problem has seen many ad-hoc solutions, such as override control, anti-windup, as well as modern techniques developed during the last decades usually based on state space models. One of the popular example is Model Predictive Control (MPC) where an optimal control problem is solved at each sampling instant, and the element of the control vector meant for the nearest sampling interval is applied. In spite of the increased computational power of control computers, MPC is at present mainly suitable for low-order, nominally linear systems. The robust version of MPC is conservative and computationally complicated, while the explicit version of MPC that gives an affine state feedback solution involves a very complicated division of the state space into polyhedral cells.In this thesis a novel and computationally cheap solution is presented for linear, time-varying or uncertain, discrete-time systems with polytopic bounded control and state (or output) vectors, with bounded disturbances. The approach is based on the interpolation between a stabilizing, outer controller that respects the control and state constraints, and an inner, more aggressive controller, designed by any method that has a robustly positively invariant set within the constraints. A simple Lyapunov function is used for the proof of closed loop stability.In contrast to MPC, the new interpolation based controller is not necessarily employing an optimization criterion inspired by performance. In its explicit form, the cell partitioning is simpler that the MPC counterpart. For the implicit version, the on-line computational demand can be restricted to the solution of one linear program or quadratic program. Several simulation examples are given, including uncertain linear systems with output feedback and disturbances. Some examples are compared with MPC. The control of a laboratory ball-and-plate system is also demonstrated. It is believed that the new controller might see wide-spread use in industry, including the automotive industry, also for the control of fast, high-order systems with constraints.

Os sistemas baseados em redes neurais artificiais e algoritmos genéticos oferecem um método alternativo para solucionar problemas relacionados à otimização de sistemas. Os algoritmos genéticos devem a sua popularidade à possibilidade de percorrer espaços de busca não-lineares e extensos. As redes neurais artificiais possuem altas taxas de processamento por utilizarem um número elevado de elementos processadores simples com alta conectividade entre si. Redes neurais com conexões realimentadas fornecem um modelo computacional capaz de resolver vários tipos de problemas de otimização, os quais consistem, geralmente, da otimização de uma função objetivo que pode estar sujeita ou não a um conjunto de restrições. Esta tese apresenta uma abordagem inovadora para resolver problemas de otimização não-linear restrita utilizando uma arquitetura neuro-genética. Mais especificamente, uma rede neural de Hopfield modificada é associada a um algoritmo genético visando garantir a convergência da rede em direção aos pontos de equilíbrio factíveis que representam as soluções para o problema de otimização não-linear restrita. / Systems based on artificial neural networks and genetic algorithms are an alternative method for solving systems optimization problems. The genetic algorithms must its popularity to make possible cover nonlinear and extensive search spaces. Artificial neural networks have high processing rates due to the use of a massive number of simple processing elements and the high degree of connectivity between these elements. Neural networks with feedback connections provide a computing model capable of solving a large class of optimization problems, which refer to optimization of an objective function that can be subject to constraints. This thesis presents a novel approach for solving constrained nonlinear optimization problems using a neuro-genetic approach. More specifically, a modified Hopfield neural network is associated with a genetic algorithm in order to guarantee the convergence of the network to the equilibrium points, which represent feasible solutions for the constraint nonlinear optimization problem.

Este trabalho apresenta uma nova estrutura matemática para a álgebra de caminhos, que permite analisar a convergência dos algoritmos de roteamento multi-restritivos hop-by-hop e, sob o ponto de vista da engenharia de tráfego e da Qualidade de Serviço (QoS) na arquitetura Generalized Multiprotocol Label Switching (GMPLS), garantir de maneira confiável a incorporação de novas métricas de roteamento aos algoritmos de roteamento baseados em múltiplas restrições. Baseando-se nessa nova álgebra de caminhos, são analisadas as propriedades de monotonicidade, isotonicidade e liberdade, conhecidas por garantir a convergência dos algoritmos de roteamento e, ao contrário do indicado na literatura até o momento, verifica-se que a propriedade de monotonicidade não e condição necessária e nem suficiente para garantir a convergência dos algoritmos de roteamento multi-restritivos hop-by-hop. Sendo assim, este trabalho propõe uma nova propriedade, denominada coerência, para a garantia da convergência do roteamento hop-by-hop e um novo algoritmo de roteamento hop-by-hop com convergência garantida. Para avaliar os resultados teóricos obtidos, s~ao analisados dois estudos de casos de aplicação do roteamento multi-restritivos hop-by-hop com o uso de uma ferramenta de simulação desenvolvida em MATLAB e baseada no algoritmo Eliminação de Loop pelo Nó de Destino (ELND) também proposto. Como resultado das simulações desses estudos de casos, verifica-se que as diferentes estratégias de otimização, necessárias as redes (GMPLS), impõem a necessidade de trabalhar com algoritmos de roteamento que permitam a definição de mais de duas métricas de roteamento com diferentes critérios de otimização para cada uma delas, comprovando, portanto, a necessidade do desenvolvimento e da continuação deste trabalho. / This work presents a new mathematical structure for paths algebra that allows the convergence analysis of hop-by-hop multi-constrained routing algorithms and, under the traffic engineering and quality of service perspectives in the Generalized Multiprotocol Label Switching (GMPLS) architecture, trustily ensures the aggregation of new routing metrics in a constrained-based routing. Based on this new paths algebra, we analyze the monotonicity, isotonicity and freeness properties, known as ensuring routing algorithms convergence, and despite of what has been indicated in the literature, we verified that the monotonicity property is not sufficient to ensure the hop-by-hop routing convergence. Therefore, this work proposes a new property, called coherence, as a necessary and sufficient condition to ensure it, as well as, a new multi-constrained hop-by-hop routing algorithm with ensured convergence. In order to evaluate the theoretical results obtained, two study cases of the hop-by-hop multi-constrained routing applications are analyzed in the present thesis by using the Eliminação de Loop pelo Nó de Destino (ELND) simulation tool, developed in MATLAB and also presented as a product of this work. As result of these study cases simulations, we verified that different optimization strategies, requested by the (GMPLS) networks, compel the use of routing algorithms that allow the specification of more than two routing metrics with different optimization criteria for each one of them, thus proving the necessity of this work and its continuation.