Reliability evaluation of composite generation and transmission systems via binary logistic regression and parallel processing

Abstract

In power system planning studies that involve the search for optimal investments with low risks for the power grid, probabilistic reliability assessments provide very useful tools and indices. A challenge faced in the application of these tools is related to the computational effort demanded by the evaluation process, especially when the combined effects of failure of generation and transmission equipment are considered. In this context, the present work proposes a new method for efficient estimation of the main composite reliability indices by combining Binary Logistic Regression (BLR) technique, a machine learning tool used for binary data classification, with the non-Sequential Monte Carlo simulation (NS-MCS) method. In addition, a computational parallelization strategy is incorporated to the proposed method to improve even more the efficiency of the composite reliability assessment. The performance of the proposed approach is analyzed by evaluating composite reliability indices for the IEEE-RTS considering two different generation and load scenarios, in addition to a real large-scale power system. The results obtained are compared with those using the NS-MCS method in its conventional version.