Abstract
Abstract
The rapid decay of electrochromic performance of V2O5 limits its widespread application, which has been proven to be attributed to the presence of ion traps. detrapping operation is an effective strategy to overcome ion traps and restore the electrochromic performance of V2O5. This process frees the ions embedded in V2O5 from shallow defects, but the effect and mechanism of action are still unclear. Therefore, this study attempts to fit this issue using data-driven machine learning (ML) methods, predicting the varying performance regeneration levels of V2O5 electrochromic materials due to different electrical parameters. Six different machine learning methods were employed in the experiment, and the models were trained using five-fold cross-validation. The results showed that the Random Forest (RF) method had the highest coefficient of determination score (R2 = 0.9) and lower root mean square error (MSE = 0.0054) for predicting material performance recovery, indicating its effectiveness in predicting the degree of material performance recovery. Furthermore, data-driven sensitivity analysis indicates that the extracted charge amount during constant detrapping procedure is a crucial factor determining the restoration effect. These results can serve as a reference for research in the field of electrochromism.