Implementation of Lithium-Ion Battery Management System with an Efficient SOC Estimation Algorithm

As fossil fuel sources are depleting day by day across the globe, the development of battery-based hybrid vehicles are gaining importance among automobile manufacturers. Battery Management System (BMS) plays an important role in EVs and HEVs by protecting the battery from operating outside its operating region and helps in monitoring the life of the battery by tracking its State of Charge (SOC) and State of Health (SOH). The battery life can be prolonged by efficiently managing the charging and discharging process. The currently deployed BMS in vehicles is costly, difficult for implementation, and not so accurate in predicting battery SOC. Hence there is lots of research being carried out across the world for monitoring battery health. Recently some successes have been reported using Kalman filter for battery SOC estimation in HEV application, without increasing the complexity of the battery model. This work is focused on the design and development of a battery management system with an efficient SOC estimation algorithm. First, the Resistive-Capacitive (RC) battery model was developed by deriving mathematical state-space variable equations. Considering the battery parameters to be time-invariant quantities, the recursive Kalman filter algorithm has been implemented on the equivalent battery model developed in MATLAB. The integrated model is tested for voltage tracking. A BMS was constructed using sensors, a data acquisition system, an electronic switching circuit, and connected to a load. A lithium-ion battery was used to test with the developed BMS for both OFF-line and ON-line implementation. A filter such as moving average, linear predictive coding, and Kalman filter was implemented on the developed BMS to estimate the battery SOC. Over the other implemented filters, the Kalman filter was able to track the battery SOC with at least twenty percentage lesser Mean Square Error [MSE] than other filters. The implemented Kalman filter on the designed BMS was able to predict the change in battery parameter with approximately thirty seconds faster than the other filter algorithms