Inovasi Proteksi DOCR: Pemanfaatan Kecerdasan Buatan dalam Koordinasi Sistem Pembangkit Tersebar

Abstract

Distributed Generation is an approach involving the integration of decentralized power generators into the distribution network. This approach can reduce transmission losses, enhance energy supply reliability, decrease carbon emissions, and promote active consumer participation in energy production. However, the addition of distributed generators presents new challenges to the electrical system's reliability and security of operation. Disruptions such as short circuits or overcurrents may occur in the system, requiring an appropriate protective response to prevent more significant damage. This research utilizes modeling and simulation of distributed generation systems under various operational conditions. The simulation results are employed in training a neural network to understand the relationship patterns between Directional Over Current Relay (DOCR) parameters and system operational conditions. The backpropagation algorithm is applied in training the Artificial Neural Network (ANN), using inputs such as maximum short-circuit current (ISC), fault location, and fault type. Time Dial Setting (TDS) and Ipickup values are used as training targets for the ANN. After testing, the results align with the target data. The effectiveness of this method is further demonstrated through ETAP simulations, ensuring that the ANN is a suitable approach for modeling adaptive and optimal relay coordination systems.