Blackout Prediction in Smart Grids using Machine Learning

Even though there has been significant improvement in the development and implementation of smart grid, the use of Machine Learning (ML) in modeling cascading failures in the grid is not yet prevalent. Here we use various ML algorithms to categorize cascading failure in a grid. Once we classify regression models are used to predict the cascading effect. Since access to real time data on the grid parameters was not available, data generated by a synthetic failure simulator framework created using MATLAB has been used. Features like the number of failed lines, power loading level, capability of load shedding are included. An IEEE 118-bus network consisting of 186 transmission lines, 118 nodes and 54 generators was used to generate the dataset. The major objective of this paper is to portray the effect of cascading failure using regression and learning models using which we’ll be able to predict blackouts as early as possible.