Real time estimation of battery impedance

Electrochemical impedance measurement systems use the Bode analysis technique to characterize the impedance of an electrochemical process. It is a well established and proven technique. The battery being evaluated is excited with a current that is single frequency and its response is measured. The process is repeated over a range of frequencies of interest until the spectrum of the impedance is obtained. The method is effective but time consuming, as the process is serial. A parallel approach using band width limited noise as an excitation current can obtain the same information in less time. The system response to the noise is processed via correlation and fast Fourier transform (FFT) algorithms and many such responses are averaged. The result is the spectrum of response over the desired frequency range. The averaging of many responses also makes this process somewhat serial. Another technique assembles the current noise waveform from a sum of sinusoids each at a different frequency. The system response as a time record is acquired and processed with the FFT algorithm. To reduce noise multiple time records of waveforms are processed and their resultant spectra averaged. This process is also serial. There is interest in real time acquisition of battery impedance for control and diagnostics over a limited frequency range. To support this need a true parallel approach is proposed that uses a single acquired time record of the battery response with duration compatible to a real time control process. The time record duration is the battery impedance sample period. If that sample period can be as short as 10's of seconds then, many control system algorithms can use battery impedance as a sensed variable