Adaptive Underfrequency Load Shedding design based on generation losing estimation using Artificial Neural Network method

Abstract

Under Frequency Relay (UFR) is used to prevent the frequency instability of a generation outage by balancing the load with the supply. The approach of this paper is to design an Adaptive Under Frequency Load Shedding (AUFLS) model in power system simulation software by estimating the magnitude of losing generation using artificial neural network (ANN) based on the early response of frequency after the disturbance occurred then the shed loads will be obtained based on this value estimation. The New England IEEE 39 bus system is used for this dynamic simulation. There are three operation scenarios planned with different dispatches of the generators, each generator is simulated by removing it from the system then the frequency system response is used as training data for the ANN model. The analysis is performed by comparing the frequency response of the system using Conventional 4 Steps Under Frequency Relay Load Shedding (UFLS) and the Adaptive Under Frequency Load Shedding (AUFLS). The results show that the frequency response after disturbance using the Adaptive Under Frequency Load Shedding (AUFLS) revert to the initial frequency closer and faster than using Conventional 1 Steps Under Frequency Relay Load Shedding (UFLS).