Broadband Identification of Battery Electrical Impedance for HEVs

Electrical impedance measurements provide useful information about the characteristics of a Li-ion battery. The classical method of measurement consists of performing electrochemical impedance spectroscopy (EIS) that relies on offline records of the response of the battery to a controlled current or voltage test signal. Although robust, it is considered as an expensive, complex, and very time-consuming method, particularly when used for an embedded system where a new whole measurement set should be carried once the impedance is to be tracked. To overcome these problems and to address embedded applications, we propose applying broadband excitation signals to perform such impedance measurements. Spectral coherence is an advanced parameter estimated to define the frequency bands where the transfer function of a system is accurately identified. The calculation of this parameter can also assess the normalized random errors on both the magnitude and the phase of the identified impedance since they are related by an explicit mathematical expression. After a brief review of some signal processing tools for identification with broadband excitation signals, we apply this method to identify the impedance of a Li-ion battery and to compare performances of various identification patterns on noisy simulated and experimental data.