Abordagem auto-adaptativa baseada no conceito de expectativa de vida aplicada aos métodos Particle Swarm Optimization e máquinas kernel para previsão da velocidade do vento e geração eólica

Detalhes bibliográficos
Ano de defesa: 2022
Autor(a) principal: Bezerra, Erick Costa
Orientador(a): Não Informado pela instituição
Banca de defesa: Não Informado pela instituição
Tipo de documento: Tese
Tipo de acesso: Acesso aberto
Idioma: por
Instituição de defesa: Não Informado pela instituição
Programa de Pós-Graduação: Não Informado pela instituição
Departamento: Não Informado pela instituição
País: Não Informado pela instituição
Palavras-chave em Português:
Link de acesso: http://www.repositorio.ufc.br/handle/riufc/69605
Resumo: The increase of wind power generation has presented challenges to the electrical networks, requiring a decision-making tool capable of dealing with its intermittence. For this purpose, a self-adaptive approach based on the concept of life expectancy is introduced. The concept of life expectancy is applied to the PSO which is used for training a neural network and compared with other training tools. Then, it is adapted to kernel machines through an auto-adaptative multiple kernel learning algorithm, which is successfully used to produce very short-term wind power forecasts at eight wind farms in Australia. The proposed method is based on a competitive tracking method, and the algorithm deals with some common difficulties of PSO and kernel methods, e.g., local optima clustering and the increasing kernel matrix size associated with time and memory complexities and the overfitting problem. The proposed method always considers the new information received by the model, thus identifying changes in the time series, avoiding abrupt loss of information and maintaining a controlled number of examples, since there is an adaptive selection of particles and active kernels. As a result, reducing the probability of overfitting in both applications, and working with the smallest dictionary possible when compared to others kernel machines. The new method, compared to others, such as backpropagation, DE, an online version of the extreme learning machine, persistence and different kernel machines, obtains lowest errors and a reduction on execution time in general.