The variational mode decomposition(VMD)method is one of the highly effective approaches for extracting multiscale health features of lithium batteries.However

it still has the problem of inadequate decomposition adaptability.To address this problem

a new state of health(SOH)estimation model is proposed

which fuses improved VMD method based on weighted mean of vectors(INFO)optimization algorithm and bi-directional long short-term memory(BiLSTM)network.Firstly

with the minimum envelope entropy as the fitness function

the INFO method is introduced to adaptively optimize the core parameters of VMD

thereby achieving optimal decomposition of capacity data and quantitative discrimination of the regularity of decomposed components.Then

the BiLSTM network is used to respectively predict the decomposed components

which contain original capacity decay and regeneration information.Finally

the SOH estimates are obtained by reconstructing the prediction results.Simulation experiments on the SOH of four groups of lithium batteries under different working conditions in the NASA dataset are conducted.The results show that compared with baseline models such as ABMS-CEEMDAN-LSTM

CEEMDAN-LSTM

root mean square error(RMSE)of the new model is reduced by an average of 40.0% and 26.4%

respectively.