Neural network applied to Monte Carlo Method based Probabilistic Power Flow

The world is increasing the share of renewable energy. This change of tendency influences the way how planning and operation of the electrical grid work. Natural resources such as wind and solar bring uncertainties to the network, so the traditional load flow which is a deterministic problem may not work anymore. This change of tendency comes with a substantial growth of the use of smart meters, which allows real-time monitoring of the electrical grid and provide a huge amount of data that can be used for system planning. One of the techniques that takes advantage of these big data is the neural network. This one consists of creating an algorithm or model that learns from the data by figuring out patterns to give a prediction when a new data is entered. In essence, this project implements a neural network model that will learn from a probabilistic power flow problem, to create a model capable of giving accurate predictions taking into account the uncertainties of the system. In this way, there no need to do the calculations to solve a probabilistic power flow, as the model just find a pattern from the historical data and will give a result according to it. This will reduce significantly the simulation time