Unsupervised KPI Anomaly Detection Method Based on Serialized Autoencoder

Abstract

The invention discloses an unsupervised KPI anomaly detection method based on a serialized autoencoder, aiming at improving the accuracy and robustness of KPI anomaly detection. The technical solution is to build an unsupervised KPI anomaly detection system based on a serialized autoencoder; the input module extracts the KPI value vector; the data preprocessing module preprocesses the KPI value vector; the offline training module splits the standardized KPI value vector , build an anomaly detection model and train the model; the automatic threshold selection module calculates the threshold; the effective detection window setting module obtains the effective window value; the input module, the data preprocessing module, the automatic threshold selection module, the effective detection window setting module, and the online detection module Cooperate with each other to detect the online KPI sequence and obtain the detection result; the display module displays the detection result. The invention can solve the problems of high false positive rate and false negative rate, and effectively improve the accuracy rate of KPI abnormal detection.

Description

technical field [0001] The invention belongs to the technical field of intelligent operation and maintenance of information technology systems, and specifically relates to a serialized autoencoder (an autoencoder is a deep neural network with a special structure, including two symmetrical structures of an encoder and a decoder). Supervise KPI (Key Performance Indicator, key performance indicators) anomaly detection method. Background technique [0002] In recent years, with the widespread deployment of cloud computing services, the continuous expansion of data center scale, and the continuous development of network and communication systems, the number and scale of internal IT architectures of Internet companies, operators, and financial institutions have continued to expand. The number of servers, storage devices, network devices, etc. is increasing, and the system structure is becoming more complex and diverse. The operation and maintenance work to ensure the stability, av...

AI Technical Summary

Problems solved by technology

First of all, this method provides a serialized anomaly detection method for KPI. By improving the existing time series anomaly detection model and using KPI as sequence data for training and detection, higher accuracy and robustness can be achieved. Secondly, on the premise of ensuring the accuracy, this method supports the operation and maintenance personnel to set the effective detection time period, so that the operation and maintenance personnel can more flexibly set the effective detection period according to the needs, and reduce the number of system alarms to a certain extent, thereby slowing down the alarm storm Finally, the method uses the linear interpolation (Linear Interpolation) method to automatically fill in the missing points in the data preprocessing stage, and uses the extreme value theory (Extreme Value Theory, EVT) to realize the automatic threshold value selection function before the detection stage, avoiding artificial The problem of false positive rate / missing negative rate trade-off caused by setting the threshold

Images