The rapid growth of renewable energy sources

such as wind and solar power

together with global carbon neutrality targets

is driving the transformation of the energy system. To mitigate the intermittency inherent in renewable energy

the integration of energy storage systems has become imperative. In China

the expansion of electric vehicles (EVs) has positioned them as mobile energy storage units

with the stock of new energy vehicles (NEVs) reaching 31.4 million by 2024. While retired EV batteries retain 70% to 80% of their original capacity

they are suitable for second-life applications

such as grid peak shaving and distributed storage

offering both environmental and economic benefits. However

safety concerns persist

requiring accurate predictions of state of health (SOH) for safe operation and optimal utilization of these batteries. To address this challenge

this paper proposes an improved Transformer model

where discrete wavelet transform (DWT) is first employed to deal with the inherent noise during charge/discharge cycles. The serial Convolutional Neural Network (CNN) structure is utilized to mine local health factors and position information based on residual connections encoded into the Transformer network. The trend fusion module is added to improve the network integration capability. Evaluations using both public center for advanced life cycle engineering (CALCE) and experimental lifetime battery datasets B_X demonstrate the superiority and effectiveness of the DWT-CNN-Transformer model. It showcases faster convergence speed and higher optimization accuracy compared with other baseline approaches

significantly bolstering the precision and robustness of SOH predictions.