Peak Power Prediction of Lithium-Ion Battery Considering Long Discharge Time for Trains Under Emergency Traction

When an electric multiple unit (EMU) loses catenary supply, the on-board energy-storage system must simultaneously sustain auxiliary loads and provide sufficient traction power for the train to reach a safe stopping location. In this situation, the available battery power is not constant over a long discharge interval because the state of charge (SOC), terminal voltage, and equivalent-circuit parameters evolve continuously. This study develops a long-horizon state-of-power (SOP) prediction framework for emergency traction. An emergency power-flow model first links battery power to traction-motor demand and auxiliary consumption. A second-order RC model is then combined with fast recursive least squares (FRLS) for online SOC and parameter estimation, while offline SOC-dependent parameter functions are used to update the battery model inside the SOP prediction horizon. On this basis, the peak discharge current is determined under battery-design, SOC, and terminal-voltage constraints by an iterative binary-search procedure. The resulting SOP is further mapped to the available emergency traction force. The validation includes the 25 °C LTO-cell tests together with low-SOC, multi-temperature, method-comparison, ablation, route-stress, and pack-level analyses. These results show that updating the equivalent-circuit parameters inside the prediction horizon materially changes the long-horizon power limit relative to a fixed-parameter formulation, while the binary-search implementation retains a computational structure suitable for train-borne use.