Optimized Battery Modeling Using Adaptive Forgetting Factor in Recursive Least Squares Algorithm

Pursuant to the more critical modelling criteria for accuracy, setup effort, computationally complexities, and simplicity of adoption, the Equivalent Circuit Model (ECM) is the recommended option for online applications. Variable Forgetting Factor Recursive Least Square (VFFRLS) is the approach that is presented for recognising the model variables. This is due to the reality that the values of the model variables are variable. Not linear Kalman filtering approaches may be utilised for determining the state of charge (SOC) of a battery, and the VFFRLS approach that was proposed can input its recognised findings into ECM. On the other hand, the execution of the Kalman filter is highly dependent on the parameters of the procedure's noise covariance matrix (Q) and the noise level covariance matrices (R). The incorrect quantity for such vectors raises the estimate error while simultaneously lowering the pace of convergence rate. We use the trial-and-error approach to investigate the effect that such matrices have on the estimate of SOC in this study. To guarantee the reliability and fast convergence achieved by the filtering for the purpose of estimating the SOC of the lithium-ion battery, filtering modification is thus important. In terms of minimising mistakes in voltage estimating, the suggested VFFRLS approach outperforms the FFRLS method by a wide margin. When compared to FFRLS, VFFRLS obtains an accuracy that is about 90% higher, demonstrating how well it maintains accurate voltage estimate during the discharge procedure.