46results about How to "Reduce the number of alerts" patented technology

The invention provides a warning association analysis method based on a knowledge base. The method includes the steps that 1, warning information existing in a warning bus is acquired; 2, warning absorption association analysis is conducted on the warning information, and a root warning is obtained; 3, warning deduction association analysis is conducted on the warning information, and a new warning is deducted to serve as a root warning; 4, the result obtained after association analysis is identified in the warning bus in a association identifier setting mode; 5, after the root warning is determined, automatic confirmation is conducted on sub-warnings, and all the relevant sub-warnings are closed synchronously as along as the root warning is closed. According to the warning association analysis method, real-time rule match is conducted on multiple warnings in the warning bus according to the association rules of warning absorption and warning deduction stored in the knowledge base, the absorption relation and the deduction relation are built in sequence to form a tree-shaped structure, so that the root warning is analyzed in real time, the derivative warnings are absorbed, or a new warning is obtained according to the existing warning deduction.

The invention discloses a method of a network element management system reporting the warning number message to a network management system, which aims at reporting the warning occurrence number message and reducing the warning number at the same time. The method is characterized in recoding the warning occurrence total number which is received by the network element management system, and the warning number which is received in a time interval; wherein, the warning occurrence total number is the unrestored warning total number which is received by the network element management system; the network element management system takes the warning occurrence total number and the warning number as the content of warning number message to report to the network management system. By adopting the method provided by the invention, the EMS can report the warning number message to the NMS while reduce the warning number at the same time, thus facilitate the NMS to obtain more warning related messages, and also facilitate the management and maintenance work of the managers.

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.

A method for analyzing alarm data and a system thereof are disclosed. The method, which is applied in a system having a peer-to-peer model indicating connection relation of network elements, includes: dividing a complete dataset of historical alarms into a plurality of alarm subsets according to network element types and alarm reasons (S101); combining any two divided alarm subsets, and calculating support degree of two alarm subsets in each kind of combination within the complete dataset of historical alarms (S102); if the support degree larger than a first threshold, according to each alarm subset of these two alarm subsets, calculating conditional probabilities of present of alarms in another alarm subsets in the simultaneous present of alarms in each alarm subset, respectively (S103); if the conditional probability larger than a second threshold, determining that alarms in the alarm subset, which function as conditions, are root alarms with respect to alarms in another alarm subsets (S104).