Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa

Financiador

Organização

Publicações

Photovoltaic Power Converter Management in Unbalanced Low Voltage Networks with Ancillary Services Support

Publication . Barrero-González, Fermín; Pires, Victor Fernão; Sousa, José L.; Martins, João F.; Milanés-Montero, María Isabel; González-Romera, Eva; Romero-Cadaval, Enrique; CTS - Centro de Tecnologia e Sistemas; UNINOVA-Instituto de Desenvolvimento de Novas Tecnologias; DEE2010-B1 Energia; DEE - Departamento de Engenharia Electrotécnica e de Computadores; MDPI - Multidisciplinary Digital Publishing Institute

The proliferation of residential photovoltaic (PV) prosumers leads to detrimental impacts on the low-voltage (LV) distribution network operation such as reverse power flow, voltage fluctuations and voltage imbalances. This is due to the fact that the strategies for the PV inverters are usually designed to obtain the maximum energy from the panels. The most recent approach to these issues involves new inverter-based solutions. This paper proposes a novel comprehensive control strategy for the power electronic converters associated with PV installations to improve the operational performance of a four-wire LV distribution network. The objectives are to try to balance the currents demanded by consumers and to compensate the reactive power demanded by them at the expense of the remaining converters’ capacity. The strategy is implemented in each consumer installation, constituting a decentralized or distributed control and allowing its practical implementation based on local measurements. The algorithms were tested, in a yearly simulation horizon, on a typical Portuguese LV network to verify the impact of the high integration of the renewable energy sources in the network and the effectiveness and applicability of the proposed approach.

2019Artigo científico

Acesso aberto

Ver mais

Soil Classification Resorting to Machine Learning Techniques

Publication . Dias, Didier Narciso; Pires, João; Martins, Bruno

Soil classification is the act of resuming the most relevant information about a soil profile into a single class, from which we can infer a large amount of properties without extensive knowledge of the subject. These classes then make the communication of soils, and how they can best be used in areas such as agriculture and forestry, simpler and easier to understand. Unfortunately soil classification is expensive and requires that specialists perform varied experiments, to be able to precisely attribute a class to a soil profile. This master’s thesis focuses on machine learning algorithms for soil classification mainly based on its intrinsic attributes, in the Mexico region. The data set used contains 6 760 soil profiles, the 19 464 horizons that constitute them, as well as physical and chemical properties, such as pH or organic content, belonging to those horizons. Four data modelling methods were tested (i.e., standard depths, n first layers, thickness, and area weighted thickness), as well as different values for a k-Nearest Neighbours imputation. A comparison between state of the art machine learning algorithms was also made, namely Random Forests, Gradient Tree Boosting, Deep Neural Networks and Recurrent Neural Networks. All of our modelling methods provided very similar results, when properly parametrised, reaching Kappa values of 0.504 and an accuracy of 0.554, with the standard depths method providing the most consistent results. The k parameter for the imputation showed very little impact on the variation on the results. Gradient Tree Boosting was the algorithm with the best overall results, closely followed by the Random Forests model. The neuron based methods never achieved a Kappa score over 0.4, therefore providing substantially worse results.

2019-11Dissertação de mestrado

Acesso aberto

Ver mais

Entidade financiadora

Fundação para a Ciência e a Tecnologia

Programa de financiamento

6817 - DCRRNI ID

Número da atribuição

UID/CEC/50021/2019