IDENTIFICATION OF MULTIPLE SEVERE CONTINGENCIES IN A POWER SYSTEM

Abstract

To avoid the system to be insecure, critical corridors i.e. subsets of those transmission lines that are most severe to extreme events are identified. It is computationally difficult to take into account all possible multiple outages of lines hence a screening method is used to find most critical contingencies. This thesis presents various efficient contingency selection algorithms; which required reducing the computational effort and having higher level of accuracy. For contingency analysis the computation time is very important aspect as it is to be implemented online. The work presented in this dissertation is divided into two main parts. The first part presents the optimization techniques to solve for contingency screening problems. These optimization techniques minimize the severity indices and maximize the effect of events (like, load shedding, loading factor and expandability of lines) for screening of critical contingencies. In the second part methods based on . artificial intelligence techniques are presented to predict or calculate the severity indices for first order contingency. In this work the Hybrid Neural Network, Support Vector Machine and Relevance Vector Machine are employed to predict the performance indices to select the severe contingencies. Different test systems and IEEE 30 bus system are considered to demonstrate the effectiveness of presented methodologies.