A Data-driven Method for DRT Function Reconstruction Based on Ridge Regression and Genetic Algorithm

Electrochemical impedance spectroscopy (EIS) technology is widely applied in the field of lithium-ion battery research. The distribution of relaxation times (DRT) method, as a novel approach for EIS analysis, has received considerable attention due to its capability to non-destructively interpret EIS data. However, the reconstruction of the DRT function is inherently a data-driven inverse problem, presenting issues of ill-posedness. This study proposes a new method for the dynamic reconstruction of the DRT function that combines ridge regression with genetic algorithm (GA) for collaborative optimization, aiming to enhance the stability of DRT function reconstruction. To address the ill-posed problem, ridge regression is used to reconstruct the DRT function, with cross-validation ensuring the quality of the reconstruction. Moreover, to overcome the limitations of traditional regularization methods that rely on empirical parameter selection and lack noise resistance, a GA is introduced to achieve adaptive optimization of the regularization parameters, effectively balancing the complexity of the model and the accuracy of the fit. Experimental results show that this method can accurately separate the relaxation time characteristic peaks corresponding to the equivalent circuit, demonstrating good robustness under Gaussian noise interference, with superior peak separation compared to traditional parameter selection methods. This research breaks through the limitations of manual parameter tuning and provides a high-precision EIS analysis tool for interpreting the aging mechanism and assessing the health status of lithium-ion batteries, particularly suitable for the extraction of electrochemical process characteristics in complex noise environments.