The present invention relates to the technical field of train traction energy consumption anomaly monitoring, and discloses a subway train traction energy consumption anomaly monitoring method based on ensemble learning, said method comprising the following steps: pre-processing historical operation data of subway train operation;Use the obtained data to establish an integrated learning model for abnormal monitoring of metro traction energy consumption; pre-process the real-time data collected by the acquisition equipment; use the pre-processed real-time data as the input to the abnormal monitoring model to monitor the traction energy consumption of metro trains in real time; effectively identify the abnormal values of metro train traction energy consumption by setting up the ensemble learning model, which provides a reliable basis for the safe operation and maintenance of the metro train, and is of great practical engineering significance.