Classification Approaches to Foster the Use of Distributed Generation with Improved Remuneration

There are currently efforts to implement the concept of smart grids throughout the electric sector. This will bring radical changes to the entire management of the sector. The energy market does not run away from the rule. In this way, virtual power players will be required to update their business models to introduce all the concepts that the context of smart grids imposes. Thus, in this article is proposed a method that aggregates distributed generation and consumers who belong to demand response programs. Optimized scheduling, resource aggregation and classification of possible new resources, rescheduling, and remuneration are the phases of the methodology proposed and presented in this article. The focus will be on classification phase and the main objective is to create rules, through a previously trained model, to be able to classify the new resources and help with the challenges that virtual power players may face. Thus, five classification methods were tested and compared: neural networks, Bayesian naïve classification, decision trees, k-nearest neighbor method, and lastly support vector machine method.The present work was done and funded in the scope of the following projects: CONTEST Project (P2020-23575), and UID/EEA/00760/2013 funded by FEDER Funds through COMPETE program and by National Funds through FCT