Improvements in classification of power system security

Author

Andersson, Christian ; Solem, Jan Erik

Author_Institution

Sch. of Technol. & Soc., Malmo Univ.

fYear

0

fDate

0-0 0

Abstract

In this paper power system security is estimated with a linear support vector machine (SVM). The SVM is used to calculate a hyperplane which separate states of the power system that sustain an (n-1) fault from states that do not. The hyperplane can be used on-line in order to determine if the system is moving into insecure conditions. The proposed method is developed to support operators at control centers and make it easier to make decisions in stressed operating situations. This paper shows that the scaling of variables is important in order to obtain good results. A new method to scale and reduce the number of measured variables is presented. The problem with a data set consisting of both continuous and binary variables is also discussed

Keywords

power engineering computing; power system security; support vector machines; SVM; binary variables; control centers; linear support vector machine; power system security classification; Binary codes; Power system faults; Power system measurements; Power system modeling; Power system security; Power system stability; Real time systems; Statistical learning; Support vector machine classification; Support vector machines; Power system stability; decision support system; support vector machines;

fLanguage

English

Publisher

ieee

Conference_Titel

Power Engineering Society General Meeting, 2006. IEEE

Conference_Location

Montreal, Que.

Print_ISBN

1-4244-0493-2

Type

conf

DOI

10.1109/PES.2006.1709330

Filename

1709330