Fault detection and classification in six-phase transmission system using k-nearest neighbour algorithm

Abstract

The present scenario have witnessed the day-by-day increasing demand of electrical energy. To fulfil considerably rising demand, six-phase transmission system has proved to be serious contender in comparison to the conventional three-phase transmission system. Because it is advantageous to transmit 73 % more power with the existing network than that of the conventional three phase double circuit without any major modification. Protection of six-phase transmission system has been a very complex and tough task since the number of possible faults in six-phase is much more as compared to three phase. A k-nearest neighbour (KNN) based protection scheme is proposed in this paper for detection and classification of all possible faults in six-phase transmission system. KNN algorithm is simple and quite preferment. It follows lazy strategy i.e. zero effort at training time and full effort at the prediction time. The system under study is composed of 138 KV, 60Hz six-phase transmission system of 68 km line length and has been modeled with reference to the Springdale-McCalmont line of Allegheny Power System by using Simpower system toolbox of MATLAB. The effectiveness of proposed protection scheme have been examined through testing. The testing data set consist of all possible shunt faults with variation in location of fault, fault resistance and angle of inception. It is quite clear with the results obtained through simulation of KNN based fault detector/classifier that the considered algorithm appropriately detects/classifies all types of faults within one cycle time. The test result indicates that the proposed protection technique is impassive against the change in fault parameters.