Parallelization of the Complex Nonlinear Regression procedure applied in electrochemical impedance data for a wide potential range

Electrochemical impedance spectroscopy (EIS) is a widely used technique in electrochemical systems characterization. Modeling this data is usually done using equivalent electrical circuits. These circuits have parameters that need to be fitted correctly, in order to enable the simulation of impedance data. Furthermore, the circuit fitting can be made for a wide potential range, allowing a characterization of the circuit elements evolution according to potential. At first, this work presents a sequential fitting methodology with high computational cost, using the optimization method Differential Evolution in each applied potential. The fitted parameters obtained for each potential step are used in the next, accelerating the fitting process and ensuring the smoothness necessary for the evolution of the circuit. Then, a parallelized algorithm is proposed for the problem, in order to reduce the fitting runtime, keeping the dependency relationship among applied potentials. Finally, results show that the parallelized algorithm is almost 50 times faster than the original and reaches the correct fitted values with the same accuracy.