Accurate voltage sag-source location technique for power systems using GACp and multivariable regression methods

Voltage sags and outages affect power quality (PQ) in terms of service continuity and disturbance propagation. Many methods have been adopted for locating the source of voltage sags in power systems; however, most of the methods can only identify the relative location of the sag source. This paper presents a new method to identify the exact voltage sag-source location in a power system based on the multivariable regression (MVR) model. In the proposed method, the number and placement of the PQ monitors are first determined by genetic algorithm and the Mallow's Cp index. By considering the monitoring buses as independent variable in the MVR model, suitable regression coefficients are obtained from the training data to estimate the unmonitored bus voltages. The fully trained MVR models are then used to determine the maximum voltage deviation and minimum standard deviation, which in turn identify the exact voltage sag-source location. To validate the proposed method, the IEEE 9bus and 30bus test systems are used. The results show that the MVR model provides good accuracy in locating the voltage sag source.