Abnormality detection method and device, equipment and storage medium

Abstract

The invention discloses an anomaly detection method, device and equipment and a storage medium, and the method comprises the steps: carrying out the feature integration of a sample time sequence feature corresponding to each anomaly type event set and a sample text feature corresponding to each sample event in each anomaly type event set, and obtaining a comprehensive feature vector corresponding to each anomaly type event set, training the lightweight gradient elevator model in combination with a sample processing strategy to obtain an anomaly detection model; and inputting the comprehensive feature vector corresponding to the target time period into the anomaly detection model to obtain an anomaly detection result in the prediction time period. By means of the mode, the accuracy and generalization ability of the model during anomaly detection are remarkably improved, dependence on a large amount of marked data is reduced, meanwhile, more comprehensive exceptional event understanding is provided for the model, and therefore efficient and accurate prediction on various exceptional events in a complex scene is achieved; and the prediction accuracy and the coverage rate of the prediction scene are greatly improved.

Images