Adaptive network reliability analysis: Methodology and applications to power grid

Flow network models can capture the underlying physics and operational constraints of many networked systems including power grid transportation water networks. However, analyzing reliability using computationally expensive flow-based faces substantial challenges, especially for rare events. Existing actively trained meta-models, which present a new promising direction in analysis, are not applicable to networks due inability these methods handle high-dimensional problems as well discrete or mixed variable inputs. This study presents first adaptive surrogate-based Network Reliability Analysis Bayesian Additive Regression Trees (ANR-BART). approach integrates BART Monte Carlo simulation (MCS) via an active learning method that identifies most valuable training samples based on credible intervals derived by over space predictor variables proximity points estimated limit state. Benchmark grids IEEE 30, 57, 118, 300-bus their flow cascading failure analysis considered investigate ANR-BART, MCS, subset simulation, passively-trained optimal deep neural BART. Results indicate ANR-BART is robust yields accurate estimates probability, while significantly reducing computational cost analysis.