Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

Submitted by Neusa Fagundes (neusa.fagundes@unioeste.br) on 2018-06-15T17:07:21Z No. of bitstreams: 2 Ana Claudia_Silva2018.pdf: 1741410 bytes, checksum: 83384ab7c02835c3d776f862defc84c1 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2018-06-15T17:07:21Z (GMT). No. of bitstreams: 2 Ana Claudia_Silva2018.pdf: 1741410 bytes, checksum: 83384ab7c02835c3d776f862defc84c1 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2018-02-07 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / The design of hydrologically homogeneous regions (RHH) is an essential procedure to provide information essential to the modeling, planning, and management of water resources, especially when it is necessary to perform the regionalization of flows, aiming to estimate the water availability in sections without measurements. The definition of strategies for the management and conservation of natural resources depends on information obtained through the identification of RHH, also being one of the steps of a study of regionalization of flows. Thus, this work has the objective of identifying the RHH in the state of Paraná through the grouping method Fuzzy C-Means. A total of 9 variables were used for the 114 fluviometric stations, with 4 dependent variables related to the characteristic flows (annual average long-term flow (Qmld), minimum annual flow with seven days duration and 10-year return period (Q7,10), flow rates associated to the 95% (Q95) and 90% (Q90) permanencies) and 5 independent variables related to the morphometric characteristics of the station (drainage area (AD - m²), sum of drainage (SD - m) (LA - Lat and longitude - Long). From the principal components analysis (PCA), the variables Qmld, DD, Lat and Long were identified as the least representative, being discarded from the study, proceeding with the analysis using only the variables AD, SD, Q90, Q95, and Q7,10. The results were obtained using the Fuzzy C-Means for the chosen variables, and the smallest objective function was found for 4 Clusters in the study group, with index of and fuzzification (m) 1.7. Separating the fluviometric stations by clusters through degrees of pertinence, the largest number of stations were obtained in Cluster 3 (83 stations), followed by Cluster 4 (13 stations) and Clusters 1 and 2 (7 stations in each cluster), and only 4 stations were not inserted in any cluster, being classified as nebulae, where the groups were determined practically by the distribution of the AD and SD variables. The smaller areas of coverage, analyzed flows and the smaller amount of drainage in the coverage area of the stations were found in Cluster 3, considering they were well spread in the state of Paraná. Clusters 1 and 4 were intermediate among the other clusters in all parameters evaluated. The Fuzzy C-Means algorithm proved to be efficient for the grouping of fluviometric stations in the state of Paraná, where it was possible to find the characteristics of each cluster formed, without overlapping of data in the analyzed variables. / O delineamento de regiões hidrologicamente homogêneas (RHH) é um procedimento essencial para provimento de informações indispensáveis aos trabalhos de modelagem, planejamento e gestão de recursos hídricos, principalmente quando se tem a necessidade de realizar a regionalização de vazões, visando estimar a disponibilidade hídrica em seções desprovidas de medições. A definição de estratégias de manejo e conservação dos recursos naturais depende de informações obtidas por meio da identificação de RHH, sendo também um dos passos de um estudo de regionalização de vazões. Assim, este trabalho tem como objetivo a identificação das RHH no estado do Paraná através do método de agrupamento Fuzzy C-Means. Foram utilizadas 9 variáveis, individualizadas para as 114 estações fluviométricas adotadas, sendo 4 variáveis dependentes referentes às vazões características (vazão média anual de longa duração (Qmld), vazão mínima anual com sete dias de duração e período de retorno de 10 anos (Q7,10), vazões associadas às permanências de 95% (Q95) e 90% (Q90)) e 5 independentes referentes às características morfometrias da estação (área de drenagem (AD – m²), soma das drenagens (SD - m), densidade de drenagem (DD – 1/m) e a localização geográfica (latitude - Lat e longitude - Long). A partir da análise de componentes principais (ACP) identificou-se as variáveis Qmld, DD, Lat e Long como as menos representativas, sendo excluídas do estudo, dando procedência à análise de agrupamentos apenas com as variáveis AD, SD, Q90, Q95 e Q7,10. Aplicou-se o Fuzzy C-Means para as variáveis escolhidas, sendo que a menor função objetiva encontrada foi para 4 Clusters no índice de fuzzificação (m) 1,7. Separando as estações fluviométricas por clusters através dos graus de pertinência, obtivemos o maior número de estações no Cluster 3 (83 estações), seguidos do Cluster 4 (13 estações) e dos Clusters 1 e 2 (7 estações em cada cluster), e apenas 4 estações não foram inseridas em nenhum cluster, sendo classificadas como nebulosas, sendo que os grupos foram determinados praticamente pela distribuição das variáveis AD e SD. As menores áreas de abrangência, vazões analisadas e as menores quantidade de drenagens na área de cobertura das estações foram encontras no Cluster 3, que estão bem espalhadas no estado do Paraná. Já os Clusters 1 e 4 ficaram intermediários entre os demais clusters em todos os parâmetros avaliados. O algoritmo Fuzzy C-Means se mostrou eficiente para o agrupamento das estações fluviométricas no estado do Paraná, onde foi possível encontrar as características de cada cluster formado, sem haver sobreposição de dados nos intervalos das variáveis analisadas.

The uniform interval normalization technique is proposed as an approach to handle sparse data and to handle noise in the data. The technique is evaluated transforming and normalizing the MoodMapper and Safebase data sets, the predictive capabilities are compared by forecasting the data set with aLSTM model. The results are compared to both the commonly used MinMax normalization technique and MinMax normalization with a time2vec layer. It was found the uniform interval normalization performed better on the sparse MoodMapper data set, and the denser Safebase data set. Future works consist of studying the performance of uniform interval normalization on other data sets and with other machine learning models.

This thesis deals with the design of fuzzy system that can evaluate supplier of spare parts for service. The result should be applicable to a company Iveta Šťastníková - car and tire service. Primarily it should simplify operations associated with the selection of appropriate spare parts, tools and other equipment needed to operate with car service station. First, we introduce the theoretical basis for the paper, and then we go to the present state and the analysis itself. The result is a proposed solution which should correspond to the needs of the owner.

The diploma thesis deals with the draft of fuzzy model used for decisions of choosing the suppliers of the tool steel for the company S.CH.W.SERVICE, s.r.o. In the introduction of the thesis the theoretical basis for the process are summarized and the company is introduced. The main part consists of the actual suggestions for the evalutaion of the company´s suppliers. The deciosion making models are created in MS Excel and MATLAB. The last part of the thesis is dedicated to the comparison of the results from both suggested models.

The master’s thesis deals with Balanced Scorecard implementation within a company environment. The object of the thesis is to increase the effectiveness of strategic management which can be achieved through proper implementation of the method. The theoretical section explains the basic concepts of strategic management, followed by the description of method BSC, its implementation and risks. The practical part is focused on creating a business strategy and the model BSC implementation in particular company. As an auxiliary tool for management decision making a fuzzy model was developed and then a dialogue is held with fuzzy expert system.