Estimação do estado de carga em baterias de Lítio-Íon baseada em filtro de Kalman Unscented

• Autor(a) principal: Tertuliano, Filippe José Gadelha
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UFPB
• Texto Completo: https://repositorio.ufpb.br/jspui/handle/123456789/20750
• topic: UKF; Estimação de SoC; Modelagem de bateria; Bateria de Lítio-Íon; SoC estimation; Battery modelling; Lithium-Ion batteries; CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA

The batteries have become a subject widely studied in different scientific areas after the increase of applications dependent on energy systems in different sectors of society and industry. In order to guarantee the safety and efficiency of energy storage, it is necessary to monitor and control the batteries continuously with robust and accurate algorithms, based on a model of the system. The State of Charge (SoC) is one of the most important parameters of the battery, as it represents its remaining capacity in relation to its nominal capacity. Among the most widely used algorithms there is the Extended Kalman Filter (EKF) method, which estimates parameters of the nonlinear dynamic system model using a complex but effective linearization process. However, there is the Unscented Kalman Filter (UKF) method capable of estimating the model parameters from the Unscented transform, having greater computational efficiency than the linearization performed by EKF. Thus, in this work, SoC lithium-ion battery estimation is performed based on UKF to identify the implementation complexity, considering the influence of temperature variation during the system operation. A battery model was also developed based on a second-order equivalent circuit, capable of representing the system behaviour without compromising the use of the UKF. The SoC estimation using UKF showed better results than the SoC estimation via EKF, under all the different temperature conditions considered in this work, being, for the UKF, the maximum Root-Mean Square Error equals to 4,51 % and maximum Mean Absolute Error equals to 3,69 %. To assist in the implementation of the algorithms and to develop the battery model, the tools available in Matlab/Simulink® were used.