Power consumption abnormity identification model training method and device and computer equipment

The invention relates to an electricity consumption anomaly recognition model training method and device and computer equipment. The method comprises the following steps: clustering each piece of historical electricity consumption data to obtain a clustering abnormal data cluster and a clustering normal data cluster; performing abnormal data identification based on the data distribution of each historical electricity consumption data to obtain an identification abnormal data cluster and an identification normal data cluster; determining a target abnormal data cluster based on the clustering abnormal data cluster and the identification abnormal data cluster, and determining a target normal data cluster based on the clustering normal data cluster and the identification normal data cluster; obtaining a training data set and a training label based on the target abnormal data cluster and the target normal data cluster; and inputting the training data set into the initial electricity consumption anomaly recognition