On-board state-of-health estimation based on charging current analysis for LiFePO<inf>4</inf> batteries

Predicting the Batteries' State of Health in Wireless Sensor Networks Applications

“…Fig. 7 shows that the cycles of the first battery taken at 20ºC fit the linear model and are mostly concentrated within an error bound of ±1.5%, as it was expected from results presented in [14]. However, the cycles taken at 45ºC do not match the same linear model because the temperature affects the charging time and this parameter is not considered in the model generation, according the authors, it is a future work.…”

“…The application of this method starts with the acquisition of battery parameters during the charging stage at constant voltage. With these data traces, it can be obtained an experimental estimation of the equivalent time constant of the battery, which is highly correlated with the SoH, as it is demonstrated in [14]. In our case, we have determined these values from the experimental battery data traces, using the charging time of every charge/discharge cycle and approximating the equivalent time constant to the 63% of this time.…”

“…In this subsection we compare the previous models with an additional model obtained using the method proposed in reference [14]. The application of this method starts with the acquisition of battery parameters during the charging stage at constant voltage.…”

“…In our case, we have determined these values from the experimental battery data traces, using the charging time of every charge/discharge cycle and approximating the equivalent time constant to the 63% of this time. We have implemented the model applying the least squares regression, as it is proposed in [14], to the data cycles of the second battery taken at 20ºC. With this model, we have estimated the error of predicting the SoH using the data coming from the first battery at 20ºC and 40ºC.…”

“…In some cases, the parameter identification can be performed using least squares algorithms [14][15][16]. In this group, we can highlight the work presented in reference [14] that is focused on the development of SoH models using linear regression to estimate the equivalent constant time of the battery. This parameter is highly correlated with the normalized capacity that provides direct information of the battery SoH.…”