Study On The Transient Stability Assessment Method Based On Association Rules

Supercapacitor Transient stability assessment(TSA)is a key issue to ensure the secure and operation of power systems.The traditional time-domain simulation method and the direct method are difficult to overcome themselves bottleneck problem.With the gradual expansion of the scale of the power system,the amount of computation is increased rapidly,which leads to the difficulty of online simulation and control.At the same time,it is difficult for direct method to apply to the complex models.And this limitation restricts its application in power systemIn recent years,with the rapid development of computer science and information technology,such as cloud computing and big data,TSA based on machine learning techniques has got much more attention and showed much promise.This kind of method has the advantages of strong generalization ability,fast evaluation speed and the ability to find the key operation information,which has broad prospects in the field of power system on-line transient stability assessment.In this paper,the association rules algorithm is introduced into the study of TSA.On the basis of analyzing and summarizing the previous work,the reliable and easy to understand stability criteria are excavated from the grid operation data,which can provide support for the realization of intelligent decision-making in the operation of large power grid.Firstly,the operation information database is established.On the one hand,the fast automatic simulation platform based on the PSD-BPA software was developed,which generated massive simulation samples the new England 39 bus test system.On the other hand,a month historical data of a grid system across the region was organised and cleared as another study data sets.Secondly,this paper puts forward a feature selection method based on weighted random forest and recursive feature elimination strategy to find out key features that can influence of the level of stability,and remove redundant input features.Therefore,the efficiency of association rules and the interpretation of rules are improved.Thirdly,the discretization of continuous characteristics is one of the essential steps of data preprocessing because most of variables of power system is continuous.In this chapter,the deficiencies of the Chi Merge discretization algorithm were summarized,then the improved algorithm was used to discrete the continuous data into discrete intervals.Finally,on the basis of the previous chapters,the FP-Growth algorithm was used to generate a preliminary rule database of TSA,and some association rules were summarized and analyzed,which can provide operators with reference information and adjustment measures that can improve the security level of power system.