A carregar...
Publicação
Plataforma de suporte à gestão de cargas no setor residencial suportada por RNA
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.77 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Nos últimos 70 anos o consumo de energia foi superior aos 12 mil anos anteriores
[1]. Seguindo a mesma tendência o consumo de energia elétrica também tem
aumentado. Este facto é justificável pela utilização generalizada de energia elétrica
nos mais variados setores, os setores industrial e doméstico apresentam a maior
quantidade de energia consumida nos últimos anos.
Atualmente, essa dependência prende-se com o facto de existirem inúmeros
equipamentos que necessitam de eletricidade ininterruptamente para a sua
correta operação, como exemplo temos: sistemas de segurança, equipamentos
frigoríficos, entre outros. O aumento de equipamentos presentes nas habitações
também é um fator relevante, acarretando maiores consumos ao nível de energia
elétrica e consequentemente um aumento da fatura elétrica.
Todas as mudanças nos hábitos de consumo obrigaram a repensar os
sistemas de fornecimento de energia aos locais de consumo. Ao longo dos últimos
anos, a liberalização do mercado português, associada ao mercado de energia
elétrica, tem culminado no surgimento de novos tipos de tarifas, nomeadamente o
mercado de tarifas dinâmicas.
Com o intuito de obter melhorias durante a utilização de energia elétrica,
foram utilizadas Redes Neuronais que permitiram uma aprendizagem constante
com a utilização de preços precedentes para a realização dessa mesma
aprendizagem. Desta forma, recorreu-se a softwares de Inteligência Artificial, cada
vez mais procurados e que têm permitido a realização de inúmeras operações tais
como a previsão de acontecimentos.
Tendo em conta os pontos apresentados anteriormente, surgiu a ideia de
criar uma ferramenta que permitisse prever preços de energia elétrica e,
concludentemente, realizasse uma gestão de cargas que seriam deslocadas e
alocadas em espaços temporais onde o preço de eletricidade fosse o mais reduzido.
Esta ferramenta tem como principal objetivo proporcionar uma redução da
fatura de eletricidade, tendo por base uma boa gestão de cargas baseada na
previsão proveniente da rede neuronal implementada.
In the last 70 years energy consumption has been higher than in the previous 12,000 years [1]. Following the same trend, the consumption of electrical energy has also increased. This fact is justified by the widespread use of electricity in the most varied sectors, among them the industrial and domestic sectors adding up to the largest amount of energy consumed in recent years. Currently, this dependence happens because there are several equipments that need electricity uninterruptedly for correct operation, such as security systems, refrigerating equipment and others. The increase in equipments present in homes is also a relevant factor, leading to greater consumption of electricity and consequently an increase in the electric bill. All the changes in consumption habits have forced a change in the way energy reaches our homes. Over the past few years, the liberalization of the Portuguese market, associated with the electricity market, has culminated in the emergence of new types of tariffs, namely the dynamic tariff market. In order to obtain improvements during the use of electricity, Neural Networks were used, which allowed constant learning and the use of previous prices. In this way, Artificial Intelligence software was used, which is increasingly sought after, and which has allowed for numerous operations such as event forecasting. Due to the points presented above, the idea arose of creating a tool that would make it possible to predict electricity prices and, conclusively, to manage loads that would be displaced and allocated in temporal spaces where the price of electricity was lower. The implementation of this tool has as its main objective to provide a reduction in the electricity bill whenever good load management occurs based on the forecast coming from the implemented neuronal network.
In the last 70 years energy consumption has been higher than in the previous 12,000 years [1]. Following the same trend, the consumption of electrical energy has also increased. This fact is justified by the widespread use of electricity in the most varied sectors, among them the industrial and domestic sectors adding up to the largest amount of energy consumed in recent years. Currently, this dependence happens because there are several equipments that need electricity uninterruptedly for correct operation, such as security systems, refrigerating equipment and others. The increase in equipments present in homes is also a relevant factor, leading to greater consumption of electricity and consequently an increase in the electric bill. All the changes in consumption habits have forced a change in the way energy reaches our homes. Over the past few years, the liberalization of the Portuguese market, associated with the electricity market, has culminated in the emergence of new types of tariffs, namely the dynamic tariff market. In order to obtain improvements during the use of electricity, Neural Networks were used, which allowed constant learning and the use of previous prices. In this way, Artificial Intelligence software was used, which is increasingly sought after, and which has allowed for numerous operations such as event forecasting. Due to the points presented above, the idea arose of creating a tool that would make it possible to predict electricity prices and, conclusively, to manage loads that would be displaced and allocated in temporal spaces where the price of electricity was lower. The implementation of this tool has as its main objective to provide a reduction in the electricity bill whenever good load management occurs based on the forecast coming from the implemented neuronal network.
Descrição
Palavras-chave
Algoritmos de Previsão Inteligência Artificial Consumo de Energia Doméstica Redes Neuronais Previsão de Preços