Planning and Operation Assessment of a Microgrid

This book introduces planning and evaluation of grid-connected photovoltaic (PV) systems-based microgrid to supply Assiut University main campus with electricity. The microgrid system requires a planning policy that can anticipate how much electricity will be consumed to meet future power demands. So, the work of this book starts with the electrical energy consumption forecasting. The forecasting process, in this book, adopts two machine learning tools that are Gaussian process (GP) tool and neural networks technique. The forecasting methodology is divided into two sub-models. The first one is a neural network model in the context of nonlinear autoregressive (NAR) model that can predict future values of a set of exogenous variables affecting electrical energy consumption. The second one is a GP model, which can be trained for relating the predicted exogenous variables to the electrical energy consumption in the process of future electrical energy consumption forecasting. In this book, the GP approach has demonstrated reasonable forecasting for one year ahead with a mean absolute percentage error (MAPE) of 4.9 %.