Deep Hierarchical Learning for Estimating High Penetration of Renewable Energy Sources in Smart Grid Environments

A prototype model for deep hierarchical learning and information for highly penetrating renewable energy resources (HPRES) is presented in this study. Because the stochastic nature of renewable energy increases the potential for hazardous situations, the growing deployment of renewable generators has posed a significant challenge to power system stability. As a result, a learning algorithm is used based on labelled databases and other complex data to learn how to reliably predict HPRES in real time without knowing the incoming renewable power data. The suggested method uses various historical data, such as wind direction, sunshine duration, and power distribution operations data, to produce modest, real-time results on the predictability of the data. MA does a rudimentary spectral analysis before starting to build the training dataset. The strategy's effectiveness is evaluated using a typical 16-machine, 64-bus testing setup.