Search

This work proposes a Markov Decision Process (MDP) model for identifying windows of opportunities to perform preventive maintenance for multi-unit parallel systems subject to a varying demand. The main contribution lies in proposing: (i) a reward function that does not depend on maintenance costs, which are typically difficult to assess and classify; and (ii) a new metric for prevention.By optimizing the capacity utilization rate and the decision flexibility, which is denoted in terms of standby units, for a set of typical operational scenarios, the optimal opportunities for preventive interventions are identified within the respective prevention ranges, in relation to an offshore power plant (case study). The sequential decision problem is solved using the Value Iteration algorithm to obtain the optimal long-term policies.As a result, a backlog management decision-support solution is developed, using a low-cost computational model, which provides scenario-dependent preventive policies and promotes the integration of operations with maintenance, being easy to implement, maintain and communicate with stakeholders.

Submitted by Maria Cristina (library@lncc.br) on 2018-06-27T12:31:36Z No. of bitstreams: 1 Dissertacao_final_Fortia.pdf: 758629 bytes, checksum: 6b31d1df1ed8f464b298cce7e1ee4180 (MD5) Approved for entry into archive by Maria Cristina (library@lncc.br) on 2018-06-27T12:31:54Z (GMT) No. of bitstreams: 1 Dissertacao_final_Fortia.pdf: 758629 bytes, checksum: 6b31d1df1ed8f464b298cce7e1ee4180 (MD5) Made available in DSpace on 2018-06-27T12:32:06Z (GMT). No. of bitstreams: 1 Dissertacao_final_Fortia.pdf: 758629 bytes, checksum: 6b31d1df1ed8f464b298cce7e1ee4180 (MD5) Previous issue date: 2017-02-24 Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) Stochastic differential equations with Markovian jump parameters constitute one of the most important class of hybrid dynamical systems, which has been extensively used for the modeling of dynamical systems which are subject to abrupt changes in their structure. The abrupt changes can be due, for instance, to abrupt environmental disturbances, component failure, volatility in economic systems, changes in subsystems interconnections, abrupt changes in the operation of a nonlinear plant, etc. This can be found, for instance, in aircraft control systems, robot systems, large flexible structure for space station, etc. We shall be particularly interested in the linear class which is dubbed in the literature as the class of Markov jump linear systems (MJLS). The jump mechanism is modeled by a Markov process, which is also known in the literature as the operation mode. The dissertation address the filtering problem of the operation mode for the class of MJLS. Previous result in the literature on this problem has been obtained by Wonham, which has shown the existence of an optimal nonlinear filter for this problem. The main hindrance with Wonham’s result, in the context of the control problem with partial observation of operation mode, is that it introduces a great deal of nonlinearity in the Hamilton-Jacobi- Belman equation, which makes it difficult to get an explicit closed solution for the control problem. Motivated by this, the main contribution of this dissertation is to devise an optimal linear filter for the mode operation, which we believe could be more favorable in the solution of the control problem with partial observations. In addition, relying on Murayama’s stochastic numerical method and the results of Yuan and Mao, we carry out simulation of Wonham’s filter, and the one devised in the dissertation, in order to compare their performances. As equações diferenciais estocáticas com salto Markoviano constituem uma das clases de sistemas dinâmicos híbridos mais importantes, e tem sido muito usados para modelar sistemas sujeitos a mudanças abruptas na sua estructura. Essas mudanças podem ser devido a, por exemplo, perturbações ambientais, falhas em componentes, volatilidade em sistemas econômicos, mudanças em interconexões de subsistemas, mudanças abruptas em operações de plantas não lineares, etc. Estas falhas podem ser encontradas em sistemas de controle para aeronaves, sistemas robóticos, estructuras grandes e flexíveis em estações espaciais, etc. Nós estamos especialmente interessados na clase de sistemas lineares que é referenciada na literatura como sistemas lineares com salto Markoviano (SLSM). O mecanismo de salto é modelado por um processo de Markov, que é conhecido na literatura como modo de operação do sistema. Essa dissertação visa o problema de filtragem para o modo de operação do sistema linear com salto. Na literatura pode-se encontrar resultados já obtidos para esse problema como é o caso do filtro ótimo não linear deduzido por Wonham. Mas no contexto de controle ótimo com observações parciais do modo de operação, o filtro de Wonham introduz não linearidades na equação de Hamilton-Jacobi-Belman, fazendo com que seja muito complexo obter uma solução fechada para o problema de controle. A principal motivação desta dissertação é deduzir o filtro ótimo linear para o modo de operação, já que esta pode ser uma solução mais favorável para o problema de controle ótimo. Finalmente, usando o método numérico para equações diferenciais estocásticas de Euler-Murayama e o resultado de Yuan e Mao, realizamos a simulação do filtro de Wonham tal como o filtro deduzido neste trabalho, com o objetivo de comparar as respectivas performances.