In order to accurately predict the discharge trajectory of the battery, this paper proposes a lithium-ion batteries (LiBs) discharge trajectory prediction method based on digital twin aggregation. First, partial life cycle discharge cycle data of a set of batteries is utilized and the data is sequence-aligned. A Transformer-L based long sequence discharge trajectory prediction model is constructed based on this data. This model has the ability to be applied in real-time on the same type of battery, and it can accurately predict the full voltage discharge curve of the real battery in the current cycle based on limited voltage samples. Then, the effectiveness of the current model is checked in real time and the model is updated when it drifts. Experimental results on MIT's battery dataset show that the proposed method has significant performance in real-time prediction of discharge trajectory.