Abstract
Lead-Acid batteries continue to be the preferred choice for backup energy storage systems. However, the inherent variability in the manufacturing and component design processes affect the performance of the manufactured battery. Therefore, the developed Lead-Acid battery models are not very flexible to model this type of variability. In this paper, a new and flexible modeling of a Lead-Acid battery is presented. Using curve fitting techniques, the model parameters were derived as a function of the battery’s state of charge based on a modified Thevenin equivalent model. In addition, the charge and discharge characteristics of the derived model were investigated and validated using a real NP4-12 YUASA battery manufacturer's data sheet to match performance at different capacity rates. Furthermore, an artificial neural network based learning system with back-propagation technique was used for estimating the model parameters using MATLAB software. The proposed neural model had the ability to predict values and interpolate between the learning curves data at various characteristics without the need of training. Finally, a closed-form analytical model that connects between inputs and outputs for neural networks was presented. It was validated by comparing the target and output and resulted in excellent regression factors.
Original language | American English |
---|---|
Pages (from-to) | 17-24 |
Number of pages | 8 |
Journal | Journal of Electrical Engineering |
Volume | 15 |
Issue number | 2 |
State | Published - Jun 1 2015 |
Keywords
- Lead-Acid Battery
- Storage
- Model
- Neural Network
- Estimation
- And Estimation
DC Disciplines
- Electrical and Computer Engineering