No abstract

A comparative study of the springback effect, otherwise known as elastic return, on Dual Phase DP600 and DP800 steel sheets is reported herein, which have been widely used in the automotive industry due to having good mechanical properties, such as high strength and ductility if compared to conventional steels. For such a purpose, it is proposed to analyze whether anisotropy and varying forming parameters interfere with the springback effect or not. The parameters selected to compare DP600 with DP800 dual phase steels were descending speed of 4 mm min −1 along the vertical axis of sheets until reaching internal bending angle of 30 degrees during bending tests at two rolling angles (0 and 90 degrees), thus forming a U-shaped steel sheet. In addition to bending tests, tensile tensting and Vickers microhardness tests have been performed at three rolling angles (0, 45 and 90 degrees). It was concluded that punching rate, internal bending angle, observation time and rolling angle exert an influence on the springback effect. Thereby, there is an important contribution to areas that require quality in formability, such as vehicle structure, which must have high impact strength and energy absorption. Steel sheets with increasingly smaller thicknesses are able to reduce density, product cost and greenhouse gases emissions from automobiles. MethodologyIn order to measure the springback effect, it was necessary to perform every metallography step and mechanical tests, such as bending, tension and Vickers microhardness testing. Metallography stepsThe metallography steps were performed in accordance with the ASTM E3-11 standard.

The Settlement Price of Differences is the short-term price published weekly by the Electric Energy Trading Chamber, which is a relevant variable for the free electricity market. Monitoring predicting its behavior allows market participants to be able to define their strategies assertively, resulting in successful energy contracting with the right amounts at the right time and at lower prices. These prices are influenced by several factors, mainly related to uncertainties of demand and hydrology, resulting in their excessive variability as dynamic systems. The present work proposes the use of Recurrent Artificial Neural Networks in order to assist the decision process of the purchase of energy in the Free Energy Market. The artificial neural network of the proposed model is trained through the resilient backpropagation algorithm and is applied to the Brazilian energy market. The network showed high performance, being able to perform a weekly prediction, with results presented up to 24 weeks ahead, for the Southeast / Center-West submarket, with a relevant hit level, facilitating the decision making for short term market agents. Resumo: O Preço de Liquidação das Diferenças é o preço de curto prazo divulgado semanalmente pela Câmara de Comercialização de Energia Elétrica, sendo uma variável relevante para o mercado livre de energia elétrica. Acompanhá-lo e prever seu comportamento permite que os participantes do mercado consigam definir suas estratégias de maneira assertiva, resultando em contratações energéticas bemsucedidas, com os montantes adequados, no momento certo e com menores preços. Estes preços são influenciados por diversos fatores, principalmente associados a incertezas sobre demanda e hidrologia, resultando em sua excessiva variabilidade como sistemas dinâmicos. O presente trabalho propõe a utilização de Redes Neurais Artificiais Recorrentes visando auxiliar o processo de decisão da compra de energia no Mercado Livre de Energia. A rede neural artificial do modelo proposto é treinada por meio do algoritmo resilient backpropagation e é aplicada ao mercado brasileiro de energia. A rede mostrou desempenho elevado, sendo capaz de realizar uma predição semanal, com resultados apresentados de até 24 semanas à frente, para o submercado Sudeste/Centro-Oeste, com nível de acerto relevante, facilitando a tomada de decisão para agentes do mercado de curto prazo.

The Settlement Price of Differences (SPD) is one of the main variables in the Electric Energy Free Market, and is disclosed by the Electric Energy Trading Chamber. Its behavior is nonlinear and contaminated by random variables. For any market participant, knowing their behavior is imperative to define their energy strategies. This work consists in the development of a model for prediction of SPD in the Brazilian Short-Term Market. The model is based on Artificial Neural Networks delayed in time with Levenberg-Marquardt method with Momentum. As input variables, SPD and its explanatory variables were used, Subsystem Induced Natural Energy, Energy Exchange and Stored Energy. As an output, the future SPD is obtained one week ahead. The results show an efficient methodology for the prediction, with the monitoring of its tendencies. Resumo: O Preço de Liquidação das Diferenças (PLD) é uma das principais variáveis no Mercado Livre de Energia Elétrica, e é divulgada pela Câmara de Comercialização de Energia Elétrica. Seu comportamento é não-linear e contaminado por variáveis aleatórias. Para qualquer participante do mercado, conhecer seu comportamento é imperativo para definir suas estratégias energéticas. Este trabalho consiste no desenvolvimento de um modelo para predição do PLD no Mercado de Curto Prazo Brasileiro. O modelo é baseado em Redes Neurais Artificiais atrasadas no tempo com método de Levenberg-Marquardt com Momentum. Como variáveis de entrada foram utilizados o PLD e suas variáveis explanatórias, a Energia Natural Afluente por Subsistema, o Intercâmbio de Energia e a Energia Armazenada. Como saída, obtêm-se o PLD futuro uma semana afrente. Os resultados mostram uma eficiente metodologia para a predição, com o acompanhamento de suas tendências.

No abstract