Machine-Learning Assisted Global Optimization of Equivalent Circuit Parameters in the Field of Electrochemical Impedance Spectroscopy

In this paper, we introduce a machine-learning regression system for initial-parameter guessing for optimization of Electrochemical impedance spectroscopy equivalent circuit parameters.<br/>In Electrochemical impedance spectroscopy, the fitting of equivalent circuits models to experimental data is the standard technique for data interpretation. The currently used techniques and tools for solving this global optimization problem require the user to guess starting parameters, which degenerates the problem into a local optimization problem, or apply an expensive stochastic optimization algorithm with performance characteristics that are incompatible with large scale automated data analysis. To help alleviate this deficiency, a Machine-learning regression stage for initial-parameter guessing is proposed and implemented.