Load Indices Based Voltage Profile Assessment of Real Time Distribution System Using Generalized Regression Neural Network

Abstract

The secure operation of distribution system primarily relies on the bus voltages. In the conventional power system with no distributed generation, it is necessary to estimate the voltage magnitudes using the forecasted power demands. The distribution load flow needs to be repeated with different load values so that the preventive and the corrective measures can be planned in order to ensure the system operates reliably. Hence, there is the need for an accurate method in order to estimate voltage magnitudes with lesser computation time. This paper proposes an Artificial Intelligence (AI) based method to estimate the voltage magnitudes. One of the major limitations of the AI based methods is their number of inputs. Especially if the voltages of practical power system need to be estimated, the number of inputs will be equal to the number of buses. This paper proposes a method for bus voltage estimation of distribution system using few indices. The advantage of the proposed method is that the number of inputs will not increase proportionally with the total number of buses. The proposed method is tested on 52 bus Tirunelveli region Distribution Substation. The voltage obtained using the proposed Generalized Regression Neural Network (GRNN) method is compared with the results of conventional load flow. From the test results, it is observed that the proposed method estimates the voltage magnitudes in less computation time with good accuracy.
Copyright © 2019 Praise Worthy Prize - All rights reserved.