Estudo e proposição de técnicas para a avaliação do isolamento em motores de indução trifásicos de baixa e média tensão

• Ano de defesa: 2018
• Autor(a) principal: Armando Souza Guedes
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Resistência de isolamento; Inteligência computacional; Motor de indução trifásico; Motores eletricos de indução; Engenharia elétrica
• Link de acesso: http://hdl.handle.net/1843/BUBD-AYTLCJ

StudiesindicatethatthesecondmostimportantcauseoffailuresinInductionMotors(IMs) is the low Resistance to Ground (RG) of the stator insulation. To identify this problem in Low and Medium Voltage (LV, MV) IMs, most of the departments of industrial electrical maintenance uses only predictive tools that are applied when the motor is out of service, as during a plant maintenance shutdown. The evaluation techniques commonly used are the measurement of the RG using a megohmmeter and the calculation of the Polarization and Absorption Index (PI, AI), whose results are limited the status of approved or not approved, without any inference about stress that the motor insulation may be subjected to. In this context, this work presents promising predictive techniques for the evaluation of the insulation of LV and MV IMs when the motor is out of service (o-line) or in operation (on-line), whose approach is not aected by no-execution of maintenance plans that require of the equipment shutdown. The use of Computational Intelligence (IC) to diagnose failures in IMs is becoming increasingly popular, presenting excellent results when compared to traditional methodologies. Inthisway,anArticialNeuralNetworks(ANN)areproposedtoclassifythestressfactors, along with a predictor for the Time to Failure (TF) of the stator insulation. The results of the o-line classier show that it is possible to identify with good accuracy the stress factor and evaluate the motor operating condition from RG and PI and AI indexes. Using as inputs for on-line classier the measuring of the resistive and capacitive components of the leakage current IL, shows that it is also possible to monitor and evaluate the insulation state on-line for a reliable motor operation. The on-line experimental tests show good results for the classication of the stress factor and for the prediction of TF of the motor insulation. Based on these results, managers, technicians and maintenance engineers will have sucient support for decision about the operational continuity of the motor and its proper maintenance. Thus, the new techniques presented in this work will contribute signicantly to increasing the reliability and operational availability of these equipments.