Progressive entropy based contingency grouping for deriving decision trees for multiple contingencies

This paper deals with decision tree based power system security assessment for multiple contingencies. In this paper, we propose a contingency grouping technique to produce good, but reduced number of decision trees for multiple contingencies. The grouping is based on the degree of overlap in class boundary regions or post-contingency performance measures of every contingency, as captured by the graphical index progressive entropy. The proposed method has been demonstrated on French EHV system for a voltage collapse problem in west France, Brittany region.

► Proposes decision tree-based security assessment for multiple contingencies.
► Devised a novel graphical metric, progressive entropy, for deriving common trees.
► Illustrated on the Brittany region of French power system.
► Obtained lesser operating rules that performed well for many contingencies.