SEMI-SUPERVISED LEARNING-BASED ABNORMAL ELECTRICITY UTILIZATION USER DETECTION METHOD

摘要

The present invention relates to the technical field of detection. Disclosed is a semi-supervised learning-based abnormal electricity utilization user detection method. The method comprises the following steps: data preprocessing; generation of a first grade grey list based on clustering analysis; generation of a second grade grey list based on outlier degree calculation; and generation of a third grade grey list based on similarity calculation. An abnormal electricity utilization user detection model based on semi-supervised learning provided in the present invention aims at forming a user dubiety degree ordered list, so that a key detection list is provided for manual detection, and accuracy and efficiency of on-site detection are improved.