Power System Fault Diagnosis Using Fuzzy Decision Tree

Accurate and rapid fault identification has indispensable role in the power system operation. After the fault event the power outage zone will outspread to the neighboring regions. For a power system to regain a healthy state, the exact and timely fault identification is necessary. Traditionally, the fault zone identification is based on operators' expertise. Later on, expert systems and artificial intelligence have been proposed to either diminish the operator's intervention in fault zone identification or boost the speed of determining the fault zone. This paper presents an unconventional methodology for diagnosing the power system fault, based on fuzzy decision tree (FDT). FDT introduces fuzzy rule base to conventional decision tree by incorporating multi-class decision at the terminal nodes with degree of probability for each participating classes. The FDT are trained in offline mode for different operating conditions using power system analysis. Performance of developed scheme validated on IEEE 9-bus system, IEEE 14 bus systems.

I. Introduction

Traditionally, the operators identify the fault zone based on their expertise and thinking. Thus, expert systems [1], [2] and artificial intelligence [3], [4] have been proposed to replace the operator's job. Other model and optimization-based methods [5]–[7] have also been reported for fault identification