Random Forest Regression-Based Fault Location Scheme for Transmission Lines

The location of the fault is a very important issue with electric power systems, as it allows the system to be isolated as soon as possible, and to be recovered as quickly as probable. This allows electric equipment to function without an overload and ensures that buyers are satisfied. This paper proposes a random forest (RF) and Teager-Kaiser energy operator (TKEO)-based fault location scheme tested using an IEEE 14-bus system. The Teager energy of both voltage and current signals of pre- and post-fault signals has been evaluated and given as an input to the RF regressor modules. For different types of faults, four regressor models were developed. Testing of the proposed scheme under a variety of fault conditions has demonstrated the ability of the scheme to accurately determine the fault site. This fault location scheme retains its accuracy even if there are changes in fault characteristics such as fault initiation angle and fault resistance. Another advantage of this scheme is its ability to determine the location of high impedance faults (HIF) accurately. The absolute error in fault location estimation is well within 1 km for most of the test cases.

Automated summary

The location of faults is crucial in electric power systems, and this paper presents a fault location scheme using random forest and Teager-Kaiser energy operator. The proposed scheme has been tested and proven to accurately determine fault locations, including high impedance faults.