A novel fractional system grey prediction model with dynamic delay effect for evaluating the state of health of lithium battery

Abstract

Accurately predicting the state of health (SOH) of lithium batteries is critical to improving the energy storage technology of batteries. However, most research focuses solely on the performance degradation trends of lithium batteries during cycling currently, while ignoring the dynamic time lag effects of the influencing factors. Therefore, a novel fractional system grey prediction model with dynamic delay effect is proposed in this paper (abbreviated as FHMGM(1, N)). The fractional-order Huasdorff operator is introduced into the grey prediction model to reflect the dynamic delay between variables in the process of charging and discharging. Meanwhile, the selected grey prediction model with system structure can better characterize the relationships among multiple variables in the system and analyze several main variables that have parallel relationships. In addition, using genetic algorithm to determine the optimal nonlinear parameters can effectively improve the accuracy and stability of the model in the prediction process. Finally, the new model is applied to two types of lithium battery datasets for fitting and prediction, which are compared with the existing grey models and artificial intelligence models. The results show that the new model can better adapt to lithium battery data with different structures, showing good robustness and prediction performance.