Abstract

As soon as there occurs any fault in a buried power cable, it becomes impossible to determine the fault's position, and a significant amount of time gets wasted while trying to find out where it is located. This paper suggests an economical and compact method of solving this problem. The voltage and current values of the faulty cable at the sending end will be generated from the sequence networks by using equations that will be obtained for six states of cables: normal operation, line-to-ground fault, line-to-line fault, double line-to-ground fault, three-phase fault, and open circuit state. The classifier will decide about the faulty cable state, whereas the regressor will predict the fault distance. All these methods are implemented within a Flask service where all events are stored in the SQLite database and where the explanation of conclusions is also provided together with alert notifications by means of an n8n workflow, as well as by an Arduino UNO along with the LCD and relay modules as a local output. Random forest algorithm correctly classified all test events and detected faults with an average error of 0.65 km on a 5 km long line.

Keywords

How to Cite This Article

Rights