|
Descripción
|
|
|---|---|
| The present electrical systems have not evolved since its inception. This fact has triggered the emergence of different problems which are necessary to tackle in order to enhance the grid performance. The demand growth is one of these problems, while others, such as the Information and Communications Technology (ICT) or Distributed Generation (DG), have been recently developed inside the grid without their proper integration. This Thesis addresses the problems arising from the management and operation of existing electrical grids and their evolution to what is considered the grid of the future or Smart Grid (SG). The SG is born from the convergence of ?ve aspects: i) the grid, ii) ICTs, iii) renewable energies, iv) Electrical Energy Storages (EESs) and v) Demand Side Management (DSM). This Thesis consists of a first step towards the SG by linking and integrating the key aspects for its development and deployment in the near future. To this end, the enhancement of the grid status is achieved by the smoothness of the aggregated consumption. In order to fulfill this objective, an algorithm has been proposed that processes the data gathered from the ICTs to benefit all the parts of the grid. Some of these benefits are: better use of the infrastructure, reduction in its size, greater operational efficiency, cost reductions and integration of the DG, among others. The proposed algorithm is based on a decentralized approximation in which the users are made participants in their decisions, being able to manage their power ows into this objective. It is implemented by following DSM techniques combined with an automatic control of demand that helps to integrate Distributed Energy Resources (DER) (renewable energies and EESs), which leads to an innovative concept called Active Demand Side Management (ADSM). In this Thesis, an Artificial Intelligence (AI) approach is used to implement the proposed algorithm. This algorithm is built based on Artificial Neural Networks (ANNs), specifically Recurrent Neural Networks (RNNs). The use of ANNs is motivated by the advantages of working with distributed, adaptive and nonlinear systems. The election of RNNs is based on their dynamic behavior, which perfectly fits with the nonlinear dynamic behavior of the grid. In addition, a neural controller is used to operate in each element of the grid to increase the global efficiency by smoothing the aggregated consumption and to maximize the local self-consumption of the available DER. Furthermore, there is no communication among the users and the sole available information is the aggregated consumption of the grid. Finally, the enhancement of the grid is achieved collectively by using the proposed algorithm to coordinate the responses of the neural controller ensemble. | |
|
Internacional
|
Si |
|
ISBN
|
|
|
Tipo de Tesis
|
Doctoral |
|
Calificación
|
Apto cum laude |
|
Fecha
|
13/07/2016 |