System downtime intelligent prediction method, device and equipment and readable storage medium

The invention discloses a system downtime intelligent prediction method, device and equipment and a readable storage medium, and the method comprises the steps: inputting a first feature vector corresponding to real-time log information into an anomaly detection model, obtaining a detection result, and determining real-time abnormal log information based on the detection result; the anomaly detection model comprises an isolated forest model and a density clustering model, and output information of the isolated forest model is input information of the density clustering model; and predicting the time and probability of system downtime based on the real-time abnormal log information. According to the method, the real-time abnormal log information is generated through the abnormal detection model, the time and probability of system downtime are predicted through the real-time abnormal log information, operation and maintenance personnel maintain the system according to the prediction result, and the stability of