Load Shedding Technique for Power System Using Neural Network Improved by Cuckoo Search Algorithm

Abstract

The present paper introduces a load shedding methodology that leverages an upgraded neural network that relies on the Cuckoo search (CS) optimization algorithm to compare the efficiency and applicability with other methods in terms of speed and feasibility. The proposed method will be tested on the IEEE-37 bus system. The results of the method are compared with other optimization methods. Thereby, this method gives good results and feasibility in application. The criteria of voltage are considered, specifically the sensitivity index dV/dQ is proposed to find weak buses in the system that need to be relieved of the active power burden. Then, the shedding priority bus ranking is created to ensure the most favorable load shedding plan for the system to maintain voltage stability. Besides, the frequency parameter is also considered to calculate the optimal amount of shed load. The model system was tested by using POWERWORLD software. After comparing the results with other methods outlined in the paper, it has been determined that the proposed approach is highly effective for optimizing grid shedding in the system.