Estimating the state of charge of lithium-ion batteries in electric vehicles with statistical methods depending on the effect of temperature and current

Battery state of charge estimation is a critical issue in terms of both safety and efficiency. Considering the explosion and fire risks that occur in the event of overcharging lithium-ion batteries with high energy density and the fact that lithium-ion batteries are used in almost all electric vehicles, it is clear that how serious problems can be caused by incorrect state of charge estimation. Because of that, state of charge estimation, which has existed for years and has become more important with electric vehicles, has attracted the attention of researchers in recent years. Traditional state of charge estimation algorithms is preferred because they are cheap and easy to apply in simple battery applications (battery applications that do not endanger human life and where accuracy is not very important). Researchers have developed many complex methods with the demand to increase the state of charge estimation accuracy rate in electric vehicles. One of the popular method is Kalman filter-based methods. Kalman filter and its extensions require a battery model for state of charge estimation. The accuracy rate of Kalman filter algorithms, which are usually used together with the equivalent circuit model, is proportional to the accuracy rate of the battery model. Therefore, researchers have tried to increase the model accuracy rate by improving battery models. To do this, the environmental and chemical factors that affect the battery are required to be considered. In this study, a current and temperature dependent battery state of charge estimation algorithm is proposed for electric vehicles. In the proposed method, the effect of temperature and current change in the equivalent circuit model is considered in order to increase the SoC estimation accuracy rate. A third-order equivalent circuit model was used to simulate a real battery characteristic in the battery model. The UKF algorithm, which is widely used in SoC estimation and has a high accuracy rate, was used in the proposed method. In order to estimate the equivalent circuit parameters, lithium-ion battery samples were tested by pulse discharge method under certain current and temperature conditions. The test results obtained for each temperature and current value were collected and the parameters of the third-order equivalent circuit model were estimated by the parameter estimation algorithm. The estimated parameters were combined into three-dimensional lookup tables and used in the battery model. MATLAB Simulink software was used to estimate the battery parameters, establish the battery equivalent circuit model and create the UKF SoC estimation algorithm. The equivalent circuit model and the SoC estimation algorithm were combined in the Simulink environment and the proposed algorithm model was created. After the proposed algorithm was matured in the simulation environment, it was loaded onto the STM32F401RE development board using embedded system tools. The success of the proposed method was observed by testing it with both simulation and experimental assessment methods. The implementation of the proposed method on a commercial development board is promising in terms of commercialization of the method. When the simulation and experimental verification tests are examined, the proposed method surpasses traditional methods. Absolute error of the proposed method is below 3%. This result prove that this method can be applied in electric vehicle applications.