Research on electrochemical impedance spectroscopy analysis method based on sparse MLP

EIS data are typically presented in the form of complex impedance spectra, which require interpretation through equivalent circuit models. Traditional methods for selecting equivalent circuits often rely on manual expertise, leading to inefficiencies and subjectivity in the modeling process. To address this, this paper integrates deep learning techniques by introducing the classic Multi-Layer Perceptron (MLP) into a sparse neural network framework, forming the Sparse MLP model, and incorporates the Gaussian Error Linear Unit (GeLU) activation function for equivalent circuit selection. In this study, the model's performance is comprehensively evaluated using multiple metrics, including precision, recall, F1 score, and AUC. Experimental results demonstrate that, compared to traditional MLP and other benchmark models, the Sparse MLP model exhibits improved classification performance. The sparsification design reduces model complexity and training time, not only enhancing classification accuracy but also providing an efficient and reliable analytical approach for electrochemical impedance spectroscopy research.