Yes / In this paper, we will consider the selective maintenance problem for multistate series systems with stochastic dependent components. In multistate systems, the health state of a component may vary from perfect functioning to complete failure. The stochastic dependence (S-dependence) between components is discussed and categorized into two types in multistate context. First, the failure of a component can immediately cause complete failures of some other components in the system. Second, as components deteriorate, the reduced working performance rate of a multistate component affects the state as well as the degradation rate of its subsequent components in series structure. The system reliability is evaluated using an approach based on stochastic process. A cost-based selective maintenance model is developed for the multistate system with S-dependent components to maximize the total system profit, which includes the production gain and loss in the next mission as well as possible maintenance costs for the system. Analyses of systems with independent and dependent components are provided. It is observed that ignoring S-dependence in the system may lead to alternative maintenance decision making and an optimistic estimation of the system performance.

Made available in DSpace on 2016-06-02T20:06:09Z (GMT). No. of bitstreams: 1 5762.pdf: 2315381 bytes, checksum: 2d23013b02c4b33dcbf1b10405b613b9 (MD5) Previous issue date: 2014-02-12 / Financiadora de Estudos e Projetos / In this work, we present and discuss the operational risk in the financial institutions, Basel Accord II, the structure of dependence between cumulative operational losses, a tool for modeling this dependence (theory of copula) and the allocation of a capital, called regulatory capital. The usual method for calculation of regulatory capital for operational risk, suggested by Basel Committee, overestimates the final capital because it is considered that the losses are perfectly positively dependents. Then, we propose a new method for this calculation based on theory of copula for the bivariate case. Such method models the dependence between two losses and considers a index (representing the expert opinion). We discuss also a method studied on Alexander (2003) and perform a simulation study in order to compare all methods, the usual, the proposed and the convolution one. / Nesse trabalho, abordamos o risco operacional nas instituições financeiras sob o ponto de vista do Acordo de Basileia II, a característica da presença de dependência estocástica entre as variáveis aleatórias em questão, a ferramenta para modelagem de tal dependência (teoria de cópulas) e a alocação de capital regulatório. Como o método usual para alocação de capital regulatório sugerido pelo Acordo de Basileia II superestima tal capital por considerar que as variáveis perdas são perfeitamente dependentes, propomos neste trabalho uma metodologia alternativa, baseada em teoria de cópulas, para o caso bivariado. Tal metodologia modela a dependência entre duas perdas e ainda inclui a opinião de especialistas da área no modelo final. Também discutimos uma metodologia existente na literatura (método da convolução) e fazemos um estudo de simulação para analisar o comportamento dos métodos abordados: método usual, proposto e da convolução.