Anomaly online clustering method, device and electronic device

The present invention discloses an anomaly online clustering method and belongs to the computer technological field. According to the prior art, when a clustering model is trained, clustering rules are required to be updated irregularly, so that new anomaly categories can be discovered, and as a result, the problem of the lag of the discovery of the new anomaly categories is brought about, however, with the method of the invention adopted, the problem in the prior art can be solved. The method includes the following steps that: if a system is abnormal, the feature of an online anomaly is acquired; the online anomaly is classified through a current online clustering model on the basis of the feature, so that an anomaly category can be determined; and if a preset update condition is met, thecurrent online clustering model is updated according to the feature of the online anomaly and the anomaly category corresponding to the online anomaly. According to the anomaly online clustering method of the invention, the