Normalized least mean squares observer for battery parameter estimation

Energy storage systems in Smart Grid applications can provide key services to transform the current power system through large-scale integration of renewable energy sources. They can assist in stabilizing the intermittent energy production, improve power quality and mitigate system peak loads. With the integration of energy storage systems into the grid, accurate and adaptive modeling becomes a necessity, in order to gain robust real-time control, in terms of network stability and energy supply forecasting. In this context, we propose an adaptive observer technique to identify the values of battery model parameters for the design of robust, low-maintenance battery management systems and integration alongside models of energy sources and electric loads into a real-time Smart Grid management system. The adaptive parameter estimation is based on a normalized recursive least mean squares algorithm and state-space mapping with a low computational burden which can accurately track parameter variations due to changing operating conditions and battery aging. Experimental data from commercial Li-Ion battery cells are used to validate the observer design and test results are reported.