Investigation of different methods of online impedance spectroscopy of batteries

Abstract

A key challenge in a battery energy storage system is understanding the availability and reliability of the system from the perspective of the end customer. A key task in this process is recognising when a battery or a module within a system starts to degrade and then mitigating against this using the control system or battery management system. Battery characterisation parameters such as internal impedance and state of health and state of charge of the battery are a useful representation of the battery conditions. This thesis investigates the feasibility of undertaking Electrochemical Impedance Spectroscopy (EIS) methods online to generate an understanding of battery impedance. In order to perform an EIS measurement, an excitation signal of fixed frequency must be generated and the voltage and current measured and used to calculate the impedance. This thesis proposed different methods of generating a low-frequency excitation signal using hardware found in most battery systems to extract the harmonic impedance of a battery cell to aim towards a low cost on-line impedance estimation. This work focuses on producing impedance spectroscopy measurements through the power electronics system, a battery balancing system and the earth leakage monitoring system to attempt to get comparable results to off-line EIS measurements under similar conditions. To generate an excitation signal through the power electronic circuit, different control methods were used including varying; the duty cycle, the switching frequency and the starting position of the switched wave and the addition of an impulse type function. Although utilising a variable duty cycle to generate a harmonic impedance has been previously published in literature, the other techniques analysed within this these have not previously been considered. The thesis looks at the theoretical analysis of the circuits and control techniques and then follows this up with simulation and experimental studies. The results showed that all the methods investigated have the capability to generate a low frequency perturbation signal to undertake online EIS measurement. However, there are potential trade-offs, for example increased inductor ripple current. Not all of the methods produce sufficiently accurate results experimentally .However, five of the methods were used to generate EIS plots similar to those undertaken offline.