Gaussian Process Regression-Based Predictive Fault Location Algorithm for Three-Terminal Lines Using Two-Terminal PMUs

A novel fault location algorithm for three-terminal line (TTTL) via two-terminal phasor measurement units (PMUs) is presented in this paper. A proposed angular compensation factor (ACF) is imposed into the fault loop equations to compensate for the angular displacement between the tie-point and the fault path currents. To predict the ACF, a trained model that uses the Gaussian process regression (GPR) technique is proposed. The regression space is defined by the response variable ACF and four featured variables. Moreover, the training data points mapped ACFs to the four featured variables for different fault scenarios that are defined by the fault type, location, and resistance. Furthermore, training of the data points is performed via GPR, and the results of the root mean squared errors (RMSEs), which are accuracy scores, indicate that the rational quadratic (RQ) and squared exponential (SE) kernel functions have the best estimations. The suggested approach accurately identifies the faulty section and fault class and reduces the computational burden by targeting relevant equations. Extensive case studies are executed with PSCAD/EMTDC and analyzed with MATLAB. The changes in the fault location, inception angles, synchronization errors, signal noise, interphase fault resistance, measurement errors, network configuration and CT saturation were incorporated.