An approach to determine the state of charge of a lithium iron phosphate cell using classification methods based on frequency domain data

The exact knowledge of the state of charge (SOC) of a battery is essential for automotive applications. Common time domain based methods such as the state space observer and the Kalman Filter are limited in their range of functionality regarding LFP cells. Those common methods depend on an open circuit voltage (OCV) curve to correct the basic Ah counting structure. Therefore an approach to determine the SOC of a lithium iron phosphate (LFP) cell using classification methods is presented. In order to improve the SOC determination of a LFP cell information from the frequency domain data, basically the impedance spectra for different specific SOC, is used to determine the SOC under in situ conditions. Classification methods such as the Support Vector Machine, Nearest Neighbor Decision and Artificial Neural Networks are in the scope of this investigation. The proposed new approach gains an advantage because of its in-dependency on the OCV curve. Every SOC has a specific impedance spectra representation. Classifying measured impedances to SOC specific impedance spectra, using the afore mentioned classification methods is used to determine the SOC, comparing the different used classification methods, a proposal for future in operando applications and a hybrid algorithm conclude the analysis.