Abstract

Crude oil can have great uses and applications, to achieve this it must undergo a process of conversion of primary to secondary energy called refining. Refining is the set of processes that are applied to crude oil in order to separate its useful components and adapt its characteristics to the needs of society. Among these products obtained from the refining process is gasoline, which is obtained using various types of catalysts. In this paper, we propose to use the Parallel System of Classification based on the Ensemble of Mixture of Experts (PCEM) developed in C using MPI (Message Passing Interface) that guarantees the obtaining of results that reflect the performance of a set of evaluated catalysts and thus proceed to the election of one that meets the industrial requirements of this process or propose improvements to this based on their behavior in the process. To carry out this system, it is proposed to use the Data Mining process on a repository of data obtained from a Catalytic Cracking Microactivity Reactor. Within the process of Data Mining is the task of classification of data, which was selected to be the engine of operation of the system proposed in this paper. We implemented a series of classifiers to compare the operation of the PCEM, that can predict new data between three different types of gasoline grades, obtaining in all the tests that the PCEM high rates in the performance measures with respect to the traditional classifiers.