A BATTERY MANAGEMENT SYSTEM WITH FAST CHARGING STRATEGIES AND STATE OF CHARGE ESTIMATION BASED ON QUANTUM NEURAL NETWORK FOR LEADACID BATTERIES

Electric vehicles and portable devices use batteries as a portable power source. Moreover, a battery as a storage device plays an important role in providing stable power, especially for operating with renewable energy sources. Therefore, it is very important to design an appropriate battery managem...

全面介紹

Saved in:
書目詳細資料
Main Authors: , Kevin Gausultan H.M, , Dr. Eng. F. Danang Wijaya, ST., MT.
格式: Theses and Dissertations NonPeerReviewed
出版: [Yogyakarta] : Universitas Gadjah Mada 2014
主題:
ETD
在線閱讀:https://repository.ugm.ac.id/133716/
http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=74503
標簽: 添加標簽
沒有標簽, 成為第一個標記此記錄!
實物特徵
總結:Electric vehicles and portable devices use batteries as a portable power source. Moreover, a battery as a storage device plays an important role in providing stable power, especially for operating with renewable energy sources. Therefore, it is very important to design an appropriate battery management system (BMS) for maintaining optimum battery performance. Charging-discharging strategy, State-of-Charge (SoC) estimation, and battery Voltage Balancer are implemented to manage 8 units of lead-acid batteries in series. A scale-down experimental battery system is tested by the proposed BMS. For charging strategies, both two-step and multi-stage charging methods are described. Regarding the state-of-charge estimation, direct open circuit voltage (OCV), coulometric, OCV prediction, and neural network (NN) estimation methods are delineated and compared. To improve the performance of NN, a new estimation method based on Quantum Neural Network (QNN) is proposed for battery SoC estimation. Finally, battery voltage balancer using fixed-resistor method is employed to reduce the voltage difference between each cell of the battery. Experimental results show that, multistage charging has faster charging process, which is 58 seconds less charging time compared with two-step charging. The NN estimation provides good SoC estimation with maximum average error no more than 1.03%. The proposed QNN method has further improved the NN performance, and yields more accurate results.