We develop highly accurate Battery Management Analytics that can accurately estimate both the State of Charge of Lithium-Ion batteries and the State of Health degradation.
Combined with industry leading prognosis of End of Discharge, State of Maximum Power and End of Usable Life.
We are currently developing our systems for primarily the Electric Vehicle market but, our technology can be applied equally to:
Our area of expertise is in the Estimation and PHM (Prognostics & Health Management) algorithms and hardware for batteries (mostly Lithium-Ion). These include state of the art:
SoC refers to the amount of energy instantaneously available in a battery or battery pack as a fraction of the total capacity
We have developed and published cutting edge SoC estimation and prediction algorithms based on Particle Filtering Estimation & Prognosis
EoD refers to the prediction of the SoC below a certain threshold.
Based on our state of the art Particle Filter based Prognosis we are able to infer the probability distribution of the EoD given a predicted usage profile.
SoH refers to the capacity degradation of the battery due to ageing (and the SEI growth internally)
Our Particle Filter based Prognosis models are applicable to both the SoC and SoH, which combined with our inference models of the polarising impedance we are able to accurately characterise and predict the SoH degradation of batteries.
The Polarising Impedance refers to the lumped parameter model of the internal resistance and capacitance of the battery.
We have developed variational inference methods for characterising the polarising impedance model of a given battery which allows us to accurately estimate, among other things, the state of maximum power and the SoH degredation.
Fault detection and prognosis refers to the early detection of anomalies (unexpected behaviour, including catastrophic) and prediction of the time when the event will have the greatest impact e.g. total failure of the battery.
We use both phenomenological and statistic models to characterise and detect failures in systems and using Particle Filter based Prognosis we can infer the probability distribution of the likely point of failure.
The integration of algorithms for the detection of capacity regeneration phenomena can significantly improve the estimation and prediction of both the SoC and SoH of batteries.
In Lithium-Ion batteries, the polarising impedance is an important characteristic that has been shown to be a complex function of, among others, both the state of charge and the demanded current. Therefore within a prognostic framework, which typically solely relies on the a prior modelling of the hidden state evolution, the correct characterisation of the functional surface with respect to state of charge and current impacts the accuracy of the predicted end-of-discharge probability density function. This is important in critical systems that rely solely on the Lithium-Ion as a power source and require an unbiased prediction of the end-of-discharge time. This paper demonstrates how a correctly modelled polarising surface can improve prediction accuracy over the state of the art models found in literature.