| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 4.7 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
O conhecimento dos hábitos de consumo de energia elétrica tem-se mostrado uma ferramenta importante nos diversos setores elétricos. A liberalização do setor elétrico, em Portugal e no resto do mundo, culminou no surgimento de novos agentes, que aumentaram a competitividade, sobressaindo aqueles que conseguem fornecer serviços de qualidade a preços baixos.
Nesta dissertação é proposto e avaliado um modelo de caracterização de curvas típicas de carga para consumidores de baixa tensão. A identificação dos padrões de consumo é baseada na aplicação de algoritmos de agrupamento. A base de dados consiste em dados de consumo de energia elétrica de 194 clientes de baixa tensão, localizados nas cidades do Porto, Matosinhos e Vila Real. Com o conhecimento obtido na etapa de agrupamento é elaborado um modelo de classificação, capaz de classificar novos consumidores de acordo com seus dados de consumo. A metodologia de agrupamento é baseada em sete algoritmos particionais e hierárquicos, juntamente com seis índices de validação de agrupamento, capazes de identificar a melhor partição dos dados. Para finalizar o ciclo do reconhecimento de padrões é utilizado um modelo de classificação baseado em árvores de decisão para classificar novos consumidores. Para tornar o modelo simples cada curva de carga é representada por cinco índices capazes de representar o formato das curvas de carga. A metodologia proposta nesse trabalho demonstra ser uma ferramenta eficaz que pode ser utilizada nos mais diversos setores, destacando-se a utilização do conhecimento na otimização da contratação de energia para clientes de baixa tensão. Os dados dos consumidores podem ser constantemente atualizados na tentativa de melhorar o modelo obtido nesse trabalho, obtendo estimativas que consigam representar melhor os consumidores e seus hábitos de consumo.
The knowledge of electricity consumption’s habits has been an important tool in electrical sectors. The constant liberalization of the electricity sectors, in Portugal and the rest of the world, culminated in the emergence of new agents, which increased the competitiveness, standing those that can provide quality services at low prices. In this dissertation, a characterization model of typical load curves for low voltage consumers is proposed and evaluated. The identification of consumption patterns is based on clustering analysis. The database consists of electricity consumption data of 194 low voltage consumers, located in the cities of Porto, Matosinhos and Vila Real. With the knowledge obtained in clustering analysis, a classification model is used to classify new consumers according to their consumption data. The clustering methodology is based on seven algorithms, partitional and hierarchical, six clustering validity indices are used to identify the best data partition. To complete the cycle of pattern recognition, the classification model based on decision trees is used in the classification of new consumers. To make the model simple, each load curve is represented by five indices, each index is capable to represent the shape of the load curves. The methodology proposed in this work demonstrates to be an effective tool, and can be used in most diverse sectors, highlighting the use of knowledge in the optimization of the energy contracting for low voltage consumers. Consumer data can be constantly updated in attempt to improve the model obtained in this work, finding estimates that can better represent consumers and their consumption habits.
The knowledge of electricity consumption’s habits has been an important tool in electrical sectors. The constant liberalization of the electricity sectors, in Portugal and the rest of the world, culminated in the emergence of new agents, which increased the competitiveness, standing those that can provide quality services at low prices. In this dissertation, a characterization model of typical load curves for low voltage consumers is proposed and evaluated. The identification of consumption patterns is based on clustering analysis. The database consists of electricity consumption data of 194 low voltage consumers, located in the cities of Porto, Matosinhos and Vila Real. With the knowledge obtained in clustering analysis, a classification model is used to classify new consumers according to their consumption data. The clustering methodology is based on seven algorithms, partitional and hierarchical, six clustering validity indices are used to identify the best data partition. To complete the cycle of pattern recognition, the classification model based on decision trees is used in the classification of new consumers. To make the model simple, each load curve is represented by five indices, each index is capable to represent the shape of the load curves. The methodology proposed in this work demonstrates to be an effective tool, and can be used in most diverse sectors, highlighting the use of knowledge in the optimization of the energy contracting for low voltage consumers. Consumer data can be constantly updated in attempt to improve the model obtained in this work, finding estimates that can better represent consumers and their consumption habits.
Description
Keywords
Descoberta do Conhecimento em Banco de Dados Mineração de Dados Agrupamento de Dados Classificação Perfis Típicos de Carga Knowledge Discovery in Databases Data Mining Clustering Classification Typical Load Profiles