Battery Cell Model fitting based on Enhanced Self-Correcting Cell Model

Current trends of full or partial electrification of commercially available vehicles have highlighted the need for advanced onboard energy storage systems, capable of providing both long driving ranges and limited performances degradation derived from use. Great strides in this regard have been made through the adoption of battery packs composed of lithium-ion cells. It is precisely the latter that have been analyzed in this project, with the aim of building a numerical simulation model known as Enhanced Self Correcting cell Model, based on an equivalent circuit structure similar to that predicted by the more widely used Thevenin model, but with the additional capability of predicting and modeling the hysteresis phenomenon peculiar to lithium-ion cells, which modifies their actual relaxation voltage with respect to the theoretically predicted Open Circuit Voltage value. Thus, starting from the SoC-independent model proposed by Prof. George Plett, a procedure was developed, based on nonlinear curve-fitting in least-squares sense performed through MATLAB, to define an ESC model with parameters varying according to the State of Charge (SoC) of the target cell, so as to achieve better predictive performances, in terms of reducing the error evaluated between real and simulated cell ends voltages, verified during generic current profiles delivered by the cell in usage. The results obtained with the SoC-dependent ESC Model thus developed showed great improvements in this regard, compared with what was obtained through the SoC-independent one, especially in the simulation of very dynamic operating scenarios of the cell, i.e., which require very irregular current profiles; in parallel with this performance improvement, an increase in the complexity of the model occurs, the employment of which requires the extrapolation of the instantaneous values of each parameter from proper Look-up tables, leading to an increase in the computational cost during fitting and simulation phases, so that the choice of its use must be made according to the accuracy required by the final application.