Enhancements of online adaptive lyapunov-based observer for state of charge estimation of lithium-ion batteries

Abstract

Owing to the rapid growth of electric vehicles (EV), temporary energy storage and mobile applications, the battery management system (BMS) plays an indispensable role in ensuring the safety, efficiency, and longevity of the battery. To achieve these features, the state of charge (SOC) estimation algorithm must be enhanced. Since the BMS processor repeatedly executes the SOC for a massive number of cells, the algorithm must be computationally simple, efficient, and accurate. The online estimation of lithium-ion SOC using the recently published adaptive Lyapunov-based observer is an attractive proposition due to the stability, adaptability, and reduced computing requirements. However, the observer requires the presence of persistent excitation (PE) to guarantee the convergence of the battery model parameters to their correct values. Although several important works have utilized this observer, they only apply dc excitation—which implies that the PE condition was never met. Thus, one objective of this thesis is to modify the observer so that it can be used to estimate the SOC for the dc and low excitation signals. Furthermore, there is insufficient work in the literature that demonstrates the application of the observer to estimate the SOC for EV. The motivation is the possibility of capitalizing on the EV driving profiles' inherent sufficiently rich (SR) signals to satisfy the PE condition. The performance of the SOC algorithm based on the proposed online observer is simulated on MATLAB/Simulink. Furthermore, the experimental validations are done at room temperature for a 3 Ah single cell of type Lithium Nickel Manganese Cobalt oxide (NMC). The algorithm is tested using dynamic stress test (DST) and real EV driving profiles, namely the supplemental federal test procedure (US06) and the federal urban driving schedule (FUDS). The performance of the observer is compared to the extended Kalman filter-recursive least squares (EKF-RLS). The proposed scheme requires 2.5 times less computational effort while retaining similar degree of accuracy to the latter. In addition, to fulfil the PE condition at low current excitation, a method called forced excitation is proposed. The SR signals are generated by chopping the battery current at a certain rate for a specific interval. The simulation and experimental results showed that the forced excitation method enables the observer to estimate the SOC reliably under dc condition. In addition, a simple scheme using a supercapacitor to compensate for the interruption in battery current and deliver continuous current to the load is suggested. It is envisaged that the proposed observer can contribute to the design of a customized and high performance BMS for many applications.