Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/B1423462-ACB3-4FA2-8575-FCBA1367ABEF | ||
| Year | 2020 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Ioannis Roditis, "Application of neuro fuzzy methods for the optimal management of the charging and discharging of lithium-ion batteries", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2020 https://doi.org/10.26233/heallink.tuc.86453 | ||
| Appears in Collections |
Over the last decade, energy storage has continued to evolve and adapt to energy requirements. The battery is a widely used electrical energy storage system. To achieve an efficient battery storage system an online and correct estimation of the state of charge is essential. Furthermore, state-of-the-art batteries can be characterized as a complex technological system. In order to model and/or simulate such systems, neural networks, fuzzy logic and Adaptive Neuro Fuzzy Inference Systems are often utilized. In this thesis, firstly, three systems (charging, discharging and hybrid electric vehicle operation) using batteries are introduced. Battery system data were produced through simulations for the training and evaluation of the proposed algorithm, based on a modified neuro fuzzy logic system. By using this algorithm the state of charge can be predicted for the three different operations of the lithium-ion battery (charging, discharging and hybrid electric vehicle operation). Specifically, in this work an Adaptive Neuro Fuzzy Inference System is implemented in order to predict the state of charge of the lithium battery. Consequently, the estimated state of charge is compared with the state of charge from the experimental data for validation (charging, discharging and hybrid electric vehicle operation). All the simulated systems and the adaptive neuro fuzzy inference system were implemented in Matlab/Simulink. The simulation results when compared to relevant studies validated the model developed in this project, as they achieve better performance in satisfactory time. For a variety of different input data sets, the prediction error (root mean square error) for the battery state of charge ranged from 0.061 to 0.064 for charging system, from 0.275 to 0.061 for discharging and from 2.81 to 2.85 for hybrid electric vehicle. In addition, the proposed algorithm has an average runtime of some milliseconds (2msec) for the charging and discharging systems and a few seconds (50sec) for hybrid electric vehicle operation.
Copyright © DIAS 2013
