Context:

Battery modeling can be a complex activity if techniques based on chemical behavior are employed. To facilitate this, inverse modeling techniques have been used which are based on experimental curves and adjustments of circuit models. Different techniques are used for parameterization according to their complexity, accuracy, and convergence time.

Method:

This paper uses a particle swarm optimization algorithm to parameterize a dual-polarization model for a 18650-type lithium cell. The proposed methodology divides the problem into different optimization cases and proposes a localized search strategy based on the experience of the previous case.

Results:

The PSO algorithm allows adjusting the model parameters for each case analyzed. Dividing the problem by stages allows improving the global precision while reducing the convergence times of the algorithm. Based on the possible cases, it is possible to find the dynamics of each of the parameters as a function of the charge state.

Conclusions:

The proposed methodology allows reducing the parameterization times of the dual-polarization model. Due to the approximation generated by previous experiences, it is possible to reduce the swarm population and further decrease the convergence time of the process. Additionally, the methodology can be used with different optimization algorithms.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )