基于混合健康因子和改进BiGRU神经网络的锂电池SOH估计OA

Accurate estimation of the state of health(SOH)of lithium-ion batteries is crucial for enhancing their safety and extending their service life.To address the limitations of existing data-driven methods in capturing health indicators and achieving sufficient model generalization,an SOH estimation method based on hybrid degradation features and an improved bidirectional gated recurrent unit(BiGRU)neural network was proposed.First,by integrating charge-discharge curve features with parameters of a second-order equivalent circuit model,a multidimensional hybrid health indicator was constructed to comprehensively characterize battery degradation behavior.Then,an improved Harris Hawks optimization(IHHO)algorithm combining chaotic initialization,nonlinear decreasing factor,and Cauchy-Gaussian hybrid mutation was employed to optimize the BiGRU hyperparameters.Meanwhile,a self-attention(SA)mechanism was incorporated to build the IHHO-BiGRU-SA model.Finally,the proposed model was validated using CALCE and NASA battery datasets.Experimental results show that compared with traditional methods,the proposed IHHO-BiGRU-SA model averagely reduces the mean absolute error(MAE)and root mean square error(RMSE)by 46.0%and 43.9%on the CALCE dataset,and by 39.3%and 38.7%on the NASA dataset,respectively.These results demonstrate that the proposed method achieves high estimation accuracy and excellent generalization performance in SOH estimation.