Fault Detection and Estimation of Battery Strings without Voltage Measurements to Individual Battery Cells using a Fractional-Order Learning Observer

This paper is the first of its kind to present a fractional-order learning observer for diagnosing individual battery faults in a battery string. There are no voltage measurements to each individual battery cell on this string; instead a sensor is mounted at the two ends of the string to measure the total of the voltage values for all cells on the same string. To accurately model a battery cell, the constant phase element is used in the electric circuit model. Because it has the fractional-order characteristic, the battery cell is modeled as a fractional-order system. Based on the fractional-order model of the battery string, a fractional-order learning observer is designed to diagnose the faults from battery controller signals and internal short-circuit faults. Simulation studies are conducted to verify the effectiveness of the proposed fractional-order learning observer.