An intelligent alarm analysis method for power grid dispatching

Abstract

Disclosed are an intelligent alert analysis method for power grid scheduling, the method comprising the steps of: extracting alert information of a power grid, and pre-processing the alert information; de-noising the pre-processed alert information, and obtaining a de-noising classification rule; generalizing and de-duplicating the de-noised alert information to generate an alert combination, determining the alert type of each alert combination; mining correlated rules on the alert combination and the alert type corresponding thereto to obtain a power grid alert inference rule. The present invention improves the filtration rate of noise data in the alert information and the comprehensiveness of alert types.

Description

technical field [0001] The invention relates to an alarm analysis method, in particular to an intelligent alarm analysis method oriented to power grid dispatching, and belongs to the technical field of electric power system dispatching. Background technique [0002] With the continuous expansion of the grid scale, the generated alarm information is also increasing. Thousands or even tens of thousands of alarm information flood into the control center every day. Due to abnormalities in the operation of measuring equipment and necessary equipment maintenance will generate noise data, resulting in insufficient accuracy of alarm information. At the same time, the current alarm information is directly presented to the dispatcher at an extremely fast speed, and no power grid inference rules with correlations have been formed, which makes the dispatcher unable to quickly determine the type of alarm. [0003] In order to ensure the accuracy of alarm information and quickly determin...

AI Technical Summary

Problems solved by technology

[0012] 1) When denoising, only the history event table is used without combining other relevant information, resulting in long denoising time

[0013] 2) The decision tree algorithm was not improved during denoising, resulting in a decrease in classification accuracy

[0014] 3) When using the rough set algorithm for rule reasoning, the alarm type is only set to the common alarm type, resulting in insufficient comprehensive power grid alarm reasoning rules

[0015] When mining a large amount of alarm information, the noise filtering rate of the alarm information cannot be effectively improved, and comprehensive power grid alarm reasoning rules can not be excavated, which cannot effectively meet the needs of the power system.

Images