Publication: Battery state of charge estimation using machine learning
Loading...
Date
2022-07-01
Authors
Gaffer, Razan Tarig
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Currently, Electric Vehicles (EVs) have emerged as an alternative to reduce the usage of fossil fuels and the production of hazardous pollutants, with a far cleaner power source instead, which is batteries. Since batteries are the core energy sources that need to be ensured, they can run securely under a variety of conditions and circumstances. In this instance, a battery management system is necessary to monitor the status of charge, avoiding overcharging and over discharging of the battery, which shortens its life duration. Thus, this paper purposes a method for State of Charge (SOC) estimation using an artificial neural network (ANN) based on three types of batteries (Nickel-metal hydride, Lithium-Polymer, and Lithium-Ion). The developed neural network model is trained and optimized, to estimate the SOC based on three types of batteries through MATLAB simulation. The acquired results demonstrate a good level of accuracy for SOC estimation of the three batteries. Based on the Root Mean Square Error (RMSE) value for the Lithium-Ion battery, the difference between the actual and predicted values reduced from 0.0774 to 0.0053 for the training set after optimization, and for the Lithium Polymer it is 0.0029, and the Nickel-metal hydride it is 0.0030, all of which are very low values, indicating that the neural network model have been using resulted in better estimation approach.