[pt] Oleodutos têm um papel importante no transporte de petróleo e de seus derivados, pois são a maneira mais eficaz de transportar grandes volumes por longas distâncias. A motivação deste trabalho é que uma parte não negligenciável do preço final de um derivado de petróleo é influenciada pelo custo de transporte. Apesar disso, até onde sabemos, apenas alguns autores trabalharam neste problema específico, a maioria utilizando técnicas de programação inteira. Este trabalho analisa a utilização de técnicas de inteligência artificial, arcabouços de software e simulação discreta orientada a eventos para a construção de planejadores automáticos capazes de lidar com instâncias reais de problemas de transporte em oleodutos. A primeira contribuição dessa tese é a especificação de um novo domínio para problemas de planejamento, denominado PIPESWORLD. Este domínio é inspirado no problema de transporte em oleodutos e especificado em PDDL. Por sua estrutura original, ele foi incorporado ao benchmark oficial da 4th International Planner Competition, evento bi-anual que compara o desempenho de diversos planejadores automáticos de propósito geral. Mesmo sendo uma simplificação do problema original, o PIPESWORLD se mostra um domínio bastante desafiador para o estado da arte dos planejadores. É demonstrado também que problemas de decisão derivados de diversas configurações do Pipesworld são NP-Completos. A segunda contribuição dessa tese é o arcabouço de software PLANSIM. Este framework incorpora uma máquina de busca que pode utilizar diversas estratégias, e define uma estrutura que facilita a construção de planejadores automáticos baseados em busca heurística direta que utilizam como modelo do processo a ser planejado simuladores orientados a eventos discretos. São apresentadas instanciações do PLANSIM para a construção de planejadores para problemas clássicos de como o das Torres de Hanoi e Blocksworld. A terceira contribuição da tese é a instanciação do PLANSIM para a construção de um planejador automático capaz de tratar instâncias reais de planejamento de transporte em oleodutos, denominado PLUMBER 05. A utilização de técnicas de simulação discreta orientada a eventos para a representação do modelo do sistema a ser planejado permite que este seja bastante fiel ao problema original. Isto somado ao uso do PLANSIM facilita a construção de planejadores capazes de lidar com instâncias reais. / [en] Pipelines have an important role in oil and its derivatives transportation, since they are the most effective way to transport high volumes through long distances. The motivation for this work is that a non negligible part of the final price for those products are due to transportation costs. Few authors have addressed this problem, with most of the previous work using integer programming techniques. This work analyses the use of Artificial Intelligence techniques, discrete event simulators and software frameworks for building automated planners that are able to deal with real-world oil pipeline transportation instances. The first contribution of this thesis is the specification of a new planning domain called PIPESWORLD. This domain is inspired by the oil pipeline transportation problem, and is defined in PDDL. Due to its original structure, the PIPESWORLD domain has been incorporated to the 4th International Planning Competition benchmark. Even being a simplification of the original problem, PIPESWORLD instances in the benchmark are challenging to state of art solvers. It is also shown that decision problems based on PIPESWORLD configurations are NP-Hard. The second contribution of this thesis is the PLANSIM opensource framework. This framework incorporates a search engine that may use several different strategies, and defines a structure that facilitates the construction of automated planners based on heuristic forward search that use discrete event simulators as the model for the process to be planned. The third contribution of this thesis is a PLANSIM instantiation that results in an automated planner able to deal with real-world oil pipeline transportation instances, called PLUMBER 2. The use of discrete event simulation techniques for the model of the system to be planned allows this model to be very close to the original problem. This, in conjunction with PLANSIM usage, facilitates the construction of planners that are able to cope with real-world instances.

Although high-level synthesis has been researched for many years, synthesizing minimum hardware implementations under a throughput constraint for computationally intensive algorithms remains a challenge. In this thesis, three important techniques are studied carefully and applied in an integrated way to meet this challenging synthesis requirement. The first is pipeline scheduling, which generates a pipelined schedule that meets the throughput requirement. The second is module selection, which decides the most appropriate circuit module for each operation. The third is resource sharing, which reuses a circuit module by sharing it between multiple operations. This work shows that combining module selection and resource sharing while performing pipeline scheduling can significantly reduce the hardware area, by either using slower, more area-efficient circuit modules or by time-multiplexing faster, larger circuit modules, while meeting the throughput constraint. The results of this work show that the combined approach can generate on average 43% smaller hardware than possible when a single technique (resource sharing or module selection) is applied. There are four major contributions of this work. First, given a fixed throughput constraint, it explores all feasible frequency and data introduction interval design points that meet this throughput constraint. This enlarged pipelining design space exploration results in superior hardware architectures than previous pipeline synthesis work because of the larger sapce. Second, the module selection algorithm in this work considers different module architectures, as well as different pipelining options for each architecture. This not only addresses the unique architecture of most FPGA circuit modules, it also performs retiming at the high-level synthesis level. Third, this work proposes a novel approach that integrates the three inter-related synthesis techniques of pipeline scheduling, module selection and resource sharing. To the author's best knowledge, this is the first attempt to do this. The integrated approach is able to identify more efficient hardware implementations than when only one or two of the three techniques are applied. Fourth, this work proposes and implements several algorithms that explore the combined pipeline scheduling, module selection and resource sharing design space, and identifies the most efficient hardware architecture under the synthesis constraint. These algorithms explore the combined design space in different ways which represents the trade off between algorithm execution time and the size of the explored design space.