Development of charge management system of lithium-ion battery with using fractional order model str

 The amount of energy needed in daily life and industrial areas is increasing day by day, and the fossil fuels used more to meet the need are polluting the environment. Thus, it has become necessary to search for alternative energy sources. One of the first solutions found in this regard is to store the electrical energy and to present the stored energy to use directly in the area where it is needed. The most well-known specific usage is electric vehicles developed instead of vehicles consuming fossil fuels. One of the most important elements of this transformation is batteries which are energy storage units. Among the different types of batteries, Li-ion batteries are the most preferred type due to the performance and safety features they provide. Battery management systems are needed to get convenient performance from this kind of batteries and to extend the life of batteries as much as possible. In this study, an approach that will enable Li-ion battery management systems to operate more effectively and with higher accuracy has been proposed with the use of fractional order battery model and the improvements made to the calculations for estimation of state of charge in the battery management systems. In this method, firstly, PNGV (The Partnership for a New Generation of Vehicles) equivalent circuit battery model, which represents open circuit voltage changes, is taken as a basis and integer order and fractional order batteries models are created based on PNGV equivalent circuit battery model. One of the most important subjects after creating the battery model is the determination of model parameters. At this stage, in order to make the problem simpler, the synthetic battery data is divided into sub-layers with a layered approach and the parameters are determined with analysis and data extractions on each sub-layer reflecting different state of charge level. In the parameter extraction phase, all parameters are allowed to change. This situation contributes to increase the accuracy of other parameters although the value of some parameters are slightly deviating from the required value. When the voltage values occurring with current applying to battery models whose parameters are determined is compared with the voltage values corresponding to the same current values in the data set, it has been observed that fractional order battery models are closed to the reference value in the data set. The second most emphasized subject in this study is to develop a method for estimating of state of charge parameter which is one of the most observed parameters by battery management systems. For this purpose, Luenberger observer, which is one of the most preferred observers in the literature, is adopted in the method. An important step in the calculations for state of charge estimation is the linearization of the open circuit voltage curve of battery model. Although this process is performed in both integer order and fractional order battery model, errors occur as a result of ignoring the state transitions in the open circuit voltage curve of fractional order battery model, which causes problems in the calculations. At this point, a new method has been proposed to eliminate the error occurring in linearization and this increases the accuracy of state of charge estimation. In conclusion, state of charge estimation of the battery using fractional order battery model gives approximately 0.4% better results than the estimation using integer order battery model. In order to observe if the first results obtained with synthetic dataset in this study are also valid for different Li-ion batteries, the parameters of the integer order and fractional order PNGV battery models were determined using a second dataset obtained in experimental environmental with similar method used in the first study. In parallel with the results obtained in the first study, when the voltage values occurring with current applying to battery models whose parameters are determined is compared with the voltage values corresponding to the same current values in the data set, it is clearly seen that fractional order battery model is superior to integer order battery model again. In addition, the highest error rate in the results obtained using integer order PNGV battery model in the state of charge (SoC) estimation is around 9% while the highest error rate in the state of charge (SoC) estimation made with fractional order PNGV battery model is less than 0,2%. The obtained results show that the results obtained with fractional order PNGV battery model are better in these studies.