Ricardo Almeida Barbosa de Sá

Neural Network Models For The DevelopmentAnd Evaluation Of New Fuels

In this paper data mining techniques are used to model the non-trivial relationships between dependent variables (mass exhaust emissions) and independent variables (engines’ technologies and fuel properties). Models based on experiments to predict pollutant emissions from gasoline properties and engine technologies can improve the design process. An applied adaptive neural network model structure provides for efficient combining of available expert knowledge with existing experimental data. Efficient learning methods for neural network were applied, and genetic optimization of the models was also explored.

A Data Mining Approach to Support the Development of New Fuels and Technology

In the present work, data mining techniques are used to model the non trivial relationships between properties that characterize the fuels, engine technologies and car emissions. Using models to predict car emissions from fuel properties and technologies engines can improve their development process. To support models of relational data, using the Object Linking and Embedding Database for Data Mining technology, a Simple Naive Bayes Incremental classifier was implemented in Microsoft® SQL Server™, supporting numeric input attributes, multiple prediction attributes and incremental update of data. Computational experiments using real word data sets were made to evaluate the results obtained by this classifier.