In the search for increased productivity and efficiency in the industrial sector, a new industrial revolution, called Industry 4.0, was promoted. In the electric sector, power plants seek to adapt these new concepts to optimize electric power generation processes, as well as to reduce operating costs and unscheduled downtime intervals. In these plants, PID control strategies are commonly used in water cooling systems, which use fans to perform the thermal exchange between water and the ambient air. However, as the nonlinearities of these systems affect the performance of the drivers, sometimes a greater number of fans than necessary are activated to ensure water temperature control which, consequently, increases energy expenditure. In this work, our objective is to develop digital twins for a water cooling system with auxiliary equipment, in terms of the decision making of the operator, to determine the correct number of fans. This model was developed based on the algorithm of automatic extraction of fuzzy rules, derived from the SCADA of a power plant located in the capital of Paraíba, Brazil. The digital twins can update the fuzzy rules in the case of new events, such as steady-state operation, starting and stopping ramps, and instability. The results from experimental tests using data from 11 h of plant operations demonstrate the robustness of the proposed digital twin model. Furthermore, in all scenarios, the average percentage error was less than 5% and the average absolute temperature error was below 3 °C.
Os estudos de previsão de demanda têm grande importância para empresas do ramo de energia elétrica, pois existe a necessidade de alocação de recursos com antecedência, exigindo um planejamento a curto, médio e longo prazo. Tais recursos incluem a compra de equipamentos, aquisição e construção de linhas de transmissão, manutenções preventivas e comércio de energia. Diante disso, foi desenvolvida uma ferramenta computacional de apoio aos especialistas da área de planejamento estratégico em sistemas de distribuição elétrica, utilizando redes neurais artificiais para previsão de demanda, e incluindo a temperatura como fator externo. Na metodologia proposta, foi implementado um sistema de previsão a curto prazo para a subestação de uma cidade da Paraíba, utilizando técnicas computacionais de inteligência artificial baseadas em redes neurais artificiais (RNA), com auxílio do software MATLAB R. Para isso, foram utilizados dados de potência ativa, fornecidos pela concessionária de energia, e o histórico dos valores de temperatura locais foram obtidos via site do INMET, para o ano de 2008. A janela de previsão utilizada foi de 12 valores atrasados para fornecer um horizonte de 4 dias. Finalmente, a acurácia das redes obtidas via treinamento foi medida considerando o MAPE e erro relativo percentual.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.