Alarm root cause identification method based on causal network mining and graph attention network

Abstract

The invention discloses an alarm root cause identification method based on causal network mining and a graph attention network, and solves the problem of rapid and accurate fault positioning of a large-scale complex communication network. Starting from the reality of network equipment alarms, a maximum and minimum hill climbing method (MMHC) is used for mining causal trigger relationships among the alarms, and on the basis, a graph attention network is used for accurately positioning the alarms. The model has certain fault tolerance for the mined alarm relationship, and the weight influence ofdifferent neighbor nodes is adjusted through an Attention mechanism, so that the identification of the root cause alarms is more accurate, and 93% of identification accuracy is achieved.

Description

technical field [0001] The invention belongs to the field of intelligent operation and maintenance (AIOPS), and in particular relates to an alarm root cause identification method based on causal network mining and graph attention network (GAT). Background technique [0002] In a large-scale network operation and maintenance environment, when a network device fails, a large amount of alarm information will be generated, and due to the correlation between devices, it is very likely to trigger an alarm for the device associated with it in a short time. In the current Huawei wireless field scenario, the occurrence of a fault often triggers multiple alarm events, so that the equipment and business processes related to the fault will generate alarm information. At the same time, these alarm information (alarm streams) are likely to be superimposed together, submerging the real fault alarms, making fault identification very difficult. Therefore, it is of great practical significanc...

AI Technical Summary

Problems solved by technology

Due to the complexity, unreliability and non-determinism of the communication system, the problem of fault location is inherently complicated

The current fault location and delimitation diagnosis process relies heavily on the knowledge and experience of engineers. As the service scale and complexity grow, the process will become more tedious and error-prone

Initially relying on a single expert system technology could not achieve good results, and the algorithm gradually moved closer to the rule-based expert system + other automatic / semi-automatic algorithms. In order to solve faults more efficiently and quickly to improve user experience, it has been used in recent years Direct root cause location based on machine learning and data-driven methods, but the accuracy of fault identification and rapid recovery of faults still need to be improved

Images