Composite Power System Reliability Evaluation Using Artificial Neural Networks

This paper uses Deep learning and Monte Carlo Simulation (MCS) to speed up composite power system reliability evaluation. Due to recurring optimum power flow (OPF) solutions, reliability evaluation approaches for large integrated power grids are computationally demanding. Machine learning can avoid OPF in reliability assessment by identifying system states as successful or failed. They can only assess energy and power indices by solving OPF for all failures. This research calculates minimal load curtailments without OPF, except during training. Power, energy, probability, and frequency indices are evaluated. This paper presents a neural network based classification technique and linear regression based regression for evaluating composite power systems reliability. The proposed framework is illustrated through IEEE RTS 79 and 96 test systems. Findings show that the proposed method calculates composite system reliability indices accurately and efficiently.