Aplicação da mineração de dados em análise de resistência à compressão de concreto com adição de cinzas residuais de proceddo térmico de geração de energia

• Ano de defesa: 2022
• Autor(a) principal: Schafer, Adalberto Gularte
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Pampa; UNIPAMPA; Mestrado em Ciência e Engenharia de Materiais; Brasil; Campus Bagé
• Palavras-chave em Português: Concreto; Cinza volante; Resistência à compressão; Mineração de dados; Aprendizagem de máquina; Concrete; Gray steering wheel; Compressive strength; Data mining; Machine learning; CNPQ::ENGENHARIAS
• Link de acesso: https://repositorio.unipampa.edu.br/jspui/handle/riu/7724

The amount of data available on the internet has made it necessary for its proper processing, one of the most used techniques is data mining. In this technique the data can be processed through machine learning algorithms, optimizing the analysis and reducing or even discarding the need for experiments in some cases. In this context, concrete, as the second most consumed material in the world, appears in several publications, with the compressive strength of concrete being the main object of study, as it is a critical factor in the technical specifications and use of concrete in civil construction. Concrete production generates significant environmental impact due to the extraction of materials, so waste materials have been widely used as a concrete component. Many publications are the subject worldwide, which motivated the creation of databases with a significant number of observations. This work uses a database available on the Kaggle platform, augmented with data from articles and publications in the area of concrete with added waste, to which it already has 1100 observations. The R programming language is being used to develop data mining and machine learning algorithms, branches of artificial intelligence based on the idea that systems can "learn" with data, identify patterns, and make decisions with minimum of human intervention. As a result, this work obtained a behavior pattern of the compressive strength of concrete with addition through the data outputs of the machine learning algorithm and data mining, which were used decision trees, the graph of the most significant variables, 5D graph, graphs of correlations of concrete components and statistical analysis of results. By analyzing the results, we can obtain the essential parameters in compressive strength, attest to the reliability of the data through suitable metrics, and obtain the pattern trace. The work seeks to carry out laboratory tests to certify the results, with a comparative analysis between data from data mining and the experiment.